ANN-based modelling and estimation of daily global solar radiation data:A case study
M.Benghanem a,1,A.Mellit b,*,1,S.N.Alamri a
a Department of Physics,Faculty of Sciences,Taibah University,P.O.Box 344,Medina,Saudi Arabia
b
Department of LMD/Electronics,Faculty of Sciences Engineering,LAMEL,Jijel University,Ouled-aissa,P.O.Box 98,Jijel 18000,Algeria
a r t i c l e i n f o Article history:
Received 18August 2008Accepted 21March 2009
Available online 24April 2009Keywords:
Global solar radiation Correlation Modelling Estimation
Neural networks
a b s t r a c t
In this paper,an arti?cial neural network (ANN)models for estimating and modelling of daily global solar radiation have been developed.The data used in this work are the global irradiation H G ,diffuse irradiation H D ,air temperature T and relative humidity H u .These data are available from 1998to 2002at the National Renewable Energy Laboratory (NREL)website.We have developed six ANN-models by using dif-ferent combination as inputs:the air temperature,relative humidity,sunshine duration and the day of year.For each model,the output is the daily global solar radiation.Firstly,a set of 4?365points (4years)has been used for training each networks,while a set of 365points (1year)has been used for testing and validating the ANN-models.It was found that the model using sunshine duration and air temperature as inputs,gives good accurate results since the correlation coef?cient is 97.65%.A comparative study between developed ANN-models and conventional regression models is presented in this study.
ó2009Elsevier Ltd.All rights reserved.
1.Introduction
Long-term average values of the instantaneous (or hourly,daily,monthly)global and diffuse irradiation on a horizontal surface are needed in many applications of solar energy designs.The measured values of these parameters are available at a few places.When the measurement data are not available,the usual practice is to esti-mate them from theoretical or empirical models that have been developed based on measured values.Knowledge of the amount of solar radiation falling on a surface of the earth is of prime impor-tance to engineers and scientists involved in the design of solar-en-ergy systems.In particular,many design methods for thermal and photovoltaic systems require monthly average daily radiation on a horizontal surface as an input,in order to predict the energy pro-duction of the system on a monthly basis [1].In practice,it is very important to appreciate the order of measurements prior to any modelling study for both solar radiation and sunshine duration or daylight.There is a relative abundance of sunshine duration data and therefore it is a common practice to correlate the solar radia-tion to sunshine duration measurements.In many countries,diur-nal bright sunshine duration is measured at a wide number of places [1].Solar radiation data and its compound play very impor-tant role in designing,sizing and performance of energy and renewable energy systems [2].
An arti?cial neural network (ANN)provides a computationally ef?cient way of determining an empirical,possibly nonlinear rela-tionship between a number of inputs and one or more outputs.ANN has been applied for modelling,identi?cation,optimization,prediction,forecasting and control of complex systems.In solar radiation modelling and prediction,many studies have been per-formed using an ANN.Most of them use the geographical coordi-nate and meteorological data such as relative humidity,air temperature,pressure,sunshine duration,etc ...as inputs of the network for estimation of global solar radiation,and only few works were interested by using only the meteorological data for estimation of solar radiation.
Mohandes et al.[3]used data from 41collection stations in Saudi Arabia.From these,the data for 31stations were used to train a neu-ral network and the data for the others 10stations for testing the network.The input values to the network are latitude,longitude,altitude and sunshine duration.The results for the testing stations obtained are within 16.4%and indicate the viability of this approach for spatial modelling of solar radiation.Alawi and Hinai [4]used ANNs to predict solar radiation in areas not covered by direct mea-surement instrumentation.The input data to the network are the location,and the monthly values of data as pressure,temperature,relative humidity,wind speed and sunshine duration.The monthly-predicted values of the ANN-model compared to the ac-tual global radiation values for this independent dataset produced an accuracy of 93%and a mean absolute percentage error of 5.43.Mohandes et al.[5]used radial basis function (RBF)networks for modelling monthly mean daily values of global solar radiation on horizontal surface and compared its performance with that of
0196-8904/$-see front matter ó2009Elsevier Ltd.All rights reserved.doi:10.1016/j.enconman.2009.03.035
*Corresponding author.Tel.:+213551998982.
E-mail addresses:a.mellit@https://www.doczj.com/doc/6918202378.html, ,amellit@ictp.it (A.Mellit).1
Present address:The International Centre for Theoretical Physics (ICTP),Strada-Costiera,1134014Trieste,Italy.
Energy Conversion and Management 50(2009)
1644–1655
Contents lists available at ScienceDirect
Energy Conversion and Management
j o u r n a l h o m e p a g e :w w w.e l s e v i e r.c o m /l o c a t e /e n c o n m a
n
a multilayer perception(MLP)model and a classical regression model.The proposed network employs as inputs the latitude,lon-gitude,altitude and sunshine duration.According to the authors, the results on locations that are not included in the modelling indi-cate viability of the neural network methods to solve such prob-lems when compared with a classical regression model.Although the data sample is relatively small,representing only1year from each of32locations,it demonstrates the concept.The average mean absolute error(MAPE)for the MLP network is12.6and the average MAPE for radial basis function(RBF)networks is10.1.An ANN based model for estimation of monthly daily and hourly val-ues of solar global radiation were proposed by Reddy and Manish [6].Solar radiation data from13stations spread over India have been used for training and testing the ANN.The maximum mean absolute error between predicted and measured hourly global radi-ation is 4.07%.The results indicate that the ANN-model show promise for predicting solar global radiation at places where mon-itoring stations are not established.
Sozen et al.[7]used a neural network for the estimation of solar potential based on geographical coordinates(latitude,longitude and altitude),meteorological data(sunshine duration and mean temperature)and the corresponding month as inputs of the net-work.The measured data from seventeen stations in Turkey col-lected between the years2000and2002were used.One set with data for11stations was used for training a neural network and the other dataset from six stations was used for testing.According to the authors,the maximum mean absolute percentage error was found to be less than 6.7%and a correlation coef?cient about 99.89%for the testing stations.The predictions from the ANN-mod-els could enable scientists to locate and design solar-energy sys-tems in Turkey and determine the appropriate solar technology. Mellit et al.[8]proposed a simpli?ed hybrid model for generating sequences of total daily solar radiation,which combine a neural network and Markov chain.This model is called ANN–MTM(Mar-kov transition matrix).The inputs of the proposed model are the geographical coordinates while the outputs are the daily global so-lar radiation.It can be used for generating sequences of solar radi-ation at long term and it was applied for Algeria.The unknown validation data set produced very accurate prediction with a root mean square error(RMSE)not exceeding8%between the mea-sured and predicted data.A correlation coef?cient ranging from 90%to92%has been obtained.
Hontoria et al.[9]used a MLP for developing a solar radiation map for Spain.The inputs are the previous irradiation,clearness in-dex(K t)and the hour order number of the K t.The classical methods are unable to generate solar radiation series in places where no solar information is available.Nevertheless,the methodology proposed is able to do the generation;it is more versatile than the classical methods and so is able to draw maps of the zone.This methodology is easily extendable to other places.The only require-ment is the knowledge of the hourly solar radiation from only one site of the zone where the map is going to be drawn.Tymvios et al.
[10]presented a comparative study of Angstroms and arti?cial neural network methodologies in estimating global solar radiation, where several models have been proposed.The parameters used as inputs were the daily values of measured sunshine duration,theo-retical sunshine duration,maximum temperature and the month number.According to the authors,the best ANN-model was the one with all inputs except the month number and the results showed a MBE and RMSE of0.12%and0.67%,respectively.The ANN methodology is a promising alternative to the traditional ap-proach for estimating global solar radiation,especially in cases where radiation measurements are not readily available.
Mellit et al.[11]proposed a new model for the prediction of dai-ly solar radiation.This model combines neural network and fuzzy logic(ANFIS).The inputs of this model are the mean daily air tem-perature and sunshine duration.The correlation coef?cient ob-tained for the validation data set is98%and the mean relative error(MRE)was found less than1%.A methodology for estimating of daily global irradiation on station located in complex terrain is proposed by Bosch et al.[12],the proposed technique is based on the using of neural network.The ANN-model developed provides a satisfactory performance with an RMSE of6.0%when comparing the estimated values with the measured ones over the whole val-idation data set.On the other hand,model performance for each station has presented no dependence with the distance to the ref-erence station or with the altitude,with RMSE below7.5%and mean relative error(MBE)lower than1%for most of the stations. In addition,this methodology can be applied to other areas with a complex topography.
Mellit et al.[13]also proposed a new hybrid model based on neuro-fuzzy and Markov chain for predicting the sequences of dai-ly clearness index,the generating solar radiation data have been used for the sizing of a PV-system.The RMSE between measured and estimated values varies between0.0215and0.0235and the mean absolute percentage error(MAE)is less than2.2%.In addi-tion,a comparison between the results obtained by the ANFIS model and arti?cial neural network(ANN)models was presented in order to show the advantage of the proposed hybrid model.
Senkal and Kuleli[14]used an ANN for the estimation of solar radiation in Turkey.Meteorological and geographical data(lati-tude,longitude,altitude,month,mean diffuse radiation and mean beam radiation)are used in the inputs layer of the network,and so-lar radiation is the output.Additionally,the authors used a physical method for estimating the solar radiation from Meteosat-6satellite C3D data.According to the author,the monthly mean daily sum values were found as54W/m2and64W/m2(training cities), 91W/m2and125W/m2(testing cities),respectively.
Recently,Rehman and Mohandes[15]used an ANN for estima-tion of daily solar radiation from air temperature and relative
Nomenclature
ANFIS adaptive new fussy inference system ANN arti?cial neural network
FFNN feed-forward neural network
MAE mean absolute error.
MBE men bias error
MLP multilayer perception
MPE mean percentage error
MRE mean relative error
RBF radial basis function
RMSE root mean square error
a,b,c coef?cient of regression models H G global solar irradiation(Wh/m2/day)
H0extraterrestrial global solar radiation(Wh/m2/day) H u relative humidity(%)
K t clearness index
Lat latitude(°)
Lon longitude(°)
r correlation coef?cient
S sunshine duration(h)
S0extraterrestrial sunshine duration(h)
SS fraction sunshine
T air temperature(°C)
M.Benghanem et al./Energy Conversion and Management50(2009)1644–16551645
humidity at Abha City(Saudi Arabia).A set of4years has been used for training the network,while a set of1year is used for testing and validating the model.Results show that using the relative humidity along with daily mean temperature outperforms the other cases with absolute mean percentage error of4.49%.The absolute mean percentage error for the case when only day of the year and mean temperature were used as inputs was11.8%. This error is about10.3%when maximum temperature is used in-stead of mean temperature.
Our aim is to develop a best model with a few parameters as in-puts to estimate the solar radiation data,by using an arti?cial neu-ral network.We have tried different combination of meteorological data as inputs of the ANN-models.These data are chosen due to their correlation with the global solar radiation.
The?rst model,called ANN-S model,which has the sunshine duration(S)as input.The second model(ANN-ST)has the sunshine duration and air temperature(T)as inputs,while in the third mod-el(ANN-STH u),we use the sunshine duration,air temperature and relative humidity(H u)as inputs.For all models,the output is the daily solar radiation data,in particularly the global irradiation (H G).Since the sunshine duration S could not be available in some stations,we have developed three others model,by using as inputs the air temperature T(ANN-T model),the relative humidity H u (ANN-H u model)and the air temperature T and relative humidity (ANN-TH u model).
This paper is organized as follows the next section provides a database description.A correlation between different solar radia-tion components is presented in Section3.Section4gives a brief introduction on the neural networks.Section5deals with the implementation of the ANN-models for estimating daily global solar radiation.Results and discussion are given in the?nal section.
1646M.Benghanem et al./Energy Conversion and Management50(2009)1644–1655
2.Database
The data used in this work are the global irradiation H G,diffuse irradiation H D,air temperature T and relative humidity H u.These data are available from1998to2002at the National Renewable Energy Laboratory(NREL)website.The global irradiation H G on horizontal surface has been collected each5min since1998until 2002.Therefore,we have deduced the sunshine duration(S)which is the duration time when the energy received on horizontal sur-face is above120W/m2.These data are normalized by dividing each of them by corresponding extraterrestrial value.Fig.1a shows a typical example of a daily radiation sequence H G and daily sun-shine duration received on a horizontal surface at Al-Madinah site [16].Fig.1b shows the air temperature and relative humidity mea-sured on horizontal surface during5years.
Fig.2a illustrates the evolution of daily irradiation for the year 2002at AL-Madinah site.It also shows the values of H0,which rep-resents the extraterrestrial radiation.Fig.2b shows the evolution of daily sunshine duration for the year2002.This?gure shows clearly that there is seasonal trend with super imposed?uctuation day to day of the daily values of solar radiation data,corresponding curves of clearness indexes(K t=H G/H0)values and sunshine duration frac-tion(SS=S/S0)are presented in Fig.2c.The distribution of clearness index K t is around the yearly average clearness index0.7281.This shows that the global irradiation at Medina site is higher and many applications of solar energy will be done with good results.
3.Correlation between different solar radiation components
In linear regression model,the dependent variable comprises the ratio of the global solar irradiation to the available radiation at the top of the atmosphere(H0);and the independent variable comprises the ratio of the measured sunshine duration to the the-oretical available sunshine duration(S0).
We have investigated empirical relationships and calculated the values of diffuse solar irradiation under the weather conditions of Al-Madinah location[16].The measured data of global irradiation, the corresponding sunshine duration,the air temperature and rel-ative humidity are used in linear,multi-linear and polynomial regression analysis.
3.1.Correlation between global irradiation and sunshine duration
The relations between the fraction(H G/H0)and sunshine dura-tion(S/S0)are given by the following relations:
H G
0?atb
S
e1T
H G
0?atb
S
tc
S
2
e2T
where a,b and c are the coef?cients of regression.Fig.3a shows the correlation between daily global irradiation and sunshine duration for Al-Madinah site.The correlation coef?cient is94%.
3.2.Correlation between global irradiation and air temperature
Fig.3b shows the correlation between the global irradiation and the air temperature at Al-Madinah.In order to evaluate the model of correlation between the air temperature and the global irradiation received on horizontal surface,we have considered the data mea-sured from sunrise until midday and the data from midday until sunset[16].The linear regression for experimental data is given by:
H G H0?a1tb1
T
T Max
e3T
a1and b1are the coef?cients of linear regression.
The correlation coef?cient is68%which is less than the above
(in the case of global radiation and sunshine duration).
3.3.Correlation between global irradiation and relative humidity
Fig.3c shows the correlation between the global irradiation H G
and the relative humidity H u at Al-Madinah site.The correlation
model is given by:
H G
H0
?a1tb1á
H u
H uMax
e4T
a2and b2are the coef?cients of linear regression.The correlation
coef?cient is72%.
4.Arti?cial neural networks
Arti?cial neural networks have been used widely in many appli-
cation areas.Most applications use a feed-forward neural network
(FFNN)with the back-propagation(BP)training algorithm.There
are numerous variants of the classical BP algorithm and other
training algorithms[17].All these training algorithms assume a
?xed ANN architecture and during training,they change the
weights to obtain a satisfactory mapping of the data.
The main advantage of the feed-forward neural networks is that
they do not require a user-speci?ed problem solving algorithm(as
is the case with classic programming)but instead they‘‘learn’’
from examples,much like human beings.
Another advantage is that they possess inherent generalization
ability.This means that they can identify and respond to patterns
which are similar but not identical to the ones with which they
have been trained.On the other hand,the development of a
feed-forward ANN-model also poses certain problems,the most
important being that there is no prior guarantee that the model
will perform well for the problem at hand.
A typical feed-forward neural network is shown in Fig.4.The
training data set consists of N training patterns{(x p,t p)},where p
is the pattern number.The input vector x p and desired output vec-
tor t p have dimensions N and M,respectively;y p is the network
output vector for the p th pattern.The thresholds are handled by
augmenting the input vector with an element x p(N+1)and setting
it equal to one.The simpli?ed diagram of the back-propagation can
be seen in Fig.5.
5.ANN-based implementation of solar radiation models
The main objective of this study is to model and estimate the
global solar radiation from other parameters such as sunshine
duration,air temperature and relative humidity,using arti?cial
neural network.So far,we try to develop three global solar radia-
tion models.In the?rst model,we try to estimate the global solar
radiation H G,from only the sunshine duration S and the day t of
year as inputs,so:
H G?e fet;STe5T
However,in the second model we try to estimate the global so-
lar radiation from the day t of year,the sunshine duration S and the
air temperature T as inputs,so:
H G?e fet;S;TTe6T
While in the third model,we add to the last parameters other
parameter,which is the relative humidity H u,so:
H G?e fet;S;T;H uTe7T
M.Benghanem et al./Energy Conversion and Management50(2009)1644–16551647
where e f is depend on the weight and the bias of the neural network (for the optimal architecture).
Otherwise,some parameters could not be available like sun-shine duration.For this,we have reduced the number of inputs
1648M.Benghanem et al./Energy Conversion and Management 50(2009)1644–1655
M.Benghanem et al./Energy Conversion and Management50(2009)1644–16551649
parameters by using only the air temperature,relative humidity and the day t of year as inputs of the ANN-model,so:
H G?e fet;TTe8TH G?e fet;H uTe9TH G?e fet;T;H uTe10T
The feed-forward neural network shown in Fig.4is used in this work.Where x i,(i=2,3,4)correspond,the time t,the sunshine duration S,the air temperature T and the relative humidity H u, respectively,while y i represents the output which correspond in our study to the global horizontal solar radiation.
The described data set will be divided into two sets,the?rst set of4years(365?4)will be used for training the feed-forward neu-ral network,while the second set which consists of365data,will be used for testing and validating each models.The inputs and the output for each model are?xed previously,while the number of hidden layers and neurons within each layer will be adjusted during the training process.Also,the error for stopping the process is?xed at10à4.The pseudo-code of feed-forward network trained by Levenberg–Marquardt(LM)algorithm,which is an effective BP training algorithm,is shown in Appendix A.
6.Results and discussion
The computer codes for each ANN-model were developed in the MATLAB software(version7.5).Therefore,three feed-forward neu-ral networks are trained until the best performance is obtained (the cost error should less or equal to the?xated error).Once,this criterion is achieved the optimal weights and bias are saved,that
Table1
Statistical test for different simulation between measured and estimated ANN-models
(H G?e fet;ST;H G?e fet;S;TTand H G?e fet;S;T;H uT).
ANN architecture(MLP)MPE(%)RMSE MBE r(%)
First model H G?e fet;ST
h2?5?1i 2.43660.046767 1.368597.34
h2?3?1i 2.43390.045524 1.354197.33
h2?2?1i 2.20770.049515 1.251497.44
h2?7?1i 2.34740.046759 1.524797.20
h2?7?1i 2.25010.046891 1.458297.16
h2?5?3?1i 2.46780.047431 1.541497.11
Second model H G?e fet;S;TT
h3?2?1i 2.25880.046997 1.301497.40
h3?3?1i 2.29030.044262 1.021597.65
h3?5?1i 2.43320.046039 1.195497.48
h3?7?1i 2.50300.046555 1.364597.35
h3?9?1i 2.37010.045740 1.198597.54
h3?11?1i 2.84390.047357 1.651897.03
h3?5?3?1i 2.52270.045738 1.185497.55
Third model H G?e fet;S;T;H uT
h4?2?1i 2.52270.045738 1.184797.55
h4?3?1i 2.80720.047032 1.354197.35
h4?5?1i 2.52000.044121 1.178997.54
h4?7?1i 2.91350.047847 1.545197.16
h4?9?1i 2.75070.048074 1.658497.00
h4?5?3?1i 2.65210.045862 1.596597.14 1650M.Benghanem et al./Energy Conversion and Management50(2009)1644–1655
M.Benghanem et al./Energy Conversion and Management50(2009)1644–16551651
will be used for testing and validating the models.Fig.6a–c shows a comparison between measured and estimated daily global solar radiation for the?rst ANN-model,the second and the third ANN-model,respectively.
As it can be seen,good agreement is obtained for all models. However,in the point of view of the correlation coef?cient r,the second model Eq.(6)presents better accurate results than others ANN-models in this study.
In order to test and validate each model,we have done a statis-tical test(root mean square error‘RMSE’,the correlation coef?cient ‘r’,mean bias error,‘MBE’and mean percentage error,‘MPE’)be-tween measured and estimated global solar radiation by the
1652M.Benghanem et al./Energy Conversion and Management50(2009)1644–1655
ANN.The used statistical test is presented in Appendix B .Obtained results are summarized in Table 1.For several simulations,it was found that the best performance is obtained for the second ANN-model according to the correlation coef?cient between both series.The obtained r is 97.65%,which is higher than other models.The MPE is 2.2903and the RMSE is 0.044251.
According to this table,we can note that in all ANN-models,one hidden layer is suf?cient for modelling and estimation the daily global solar radiation.Therefore,we do not need to add more than one layer.In addition,the number of neurons in hidden layer can be arranged in the interval from two to ?ve neurons for the best
performance,in this case all developed ANN-model can be con-verge very fast.If we increase or decrease the number of neurons from this interval,the correlation coef?cient decreases and the process needs more computing time for converging.
From Table 1,we can observe that the correlation coef?cient for each simulation is arranged between 97.00%and 97.65%,however,we opted for the second ANN-model,and this model need in his in-put two parameters (T ,S )and day of year that are always available,and they can be measure easily.Therefore,the developed model can be given by the following approximate formula:
e y ?
X M k ?121texp àP M j ?1P N
i ?1w 1ei ;j Tx ei TeTtb 1ej T
à124350@1A w 2ek Tb 2e11T
where w 1,w 2,b 1and b 2are,respectively,the weights and the bias of
the networks,x ,represents the inputs data which can take the sun-shine duration,air temperature,relative humidity and the day of year (according to the model).M and N are the number of neurons in the hidden layer and in the input layer,respectively.The exper-imental values of w 1,w 2,b 1and b 2are shown in Appendix C .
In addition,we have developed others three ANN-models,the ?rst ANN-model has as input the day of year and air temperature,the second has as input the day of year and relative humidity while
Table 2
Statistical test between measured and estimated global solar radiation by the
developed ANN-models (H G ?e f et ;T T;H G ?e f et ;H u Tand H G ?e f et ;T ;H u T).ANN architecture (MLP)MPE (%)RMSE MBE (%)r (%)First model H G ?e
f et ;T Th 2?5?1i
2.09580.063534 2.541989.20Second model H G ?e f et ;H u Th 2?5?1i
4.00120.067655 2.842187.00Third model H G ?e f et ;T ;H u Th 3?2?1i
3.958
0.065374
2.7012
88.99
M.Benghanem et al./Energy Conversion and Management 50(2009)1644–1655
1653
the third model combine both relative humidity,temperature and the day of year.Therefore,we develop these ANN-models in order to show the in?uence of each parameter for estimating daily global solar radiation.
Fig.7a–c shows a comparison between measured and estimated daily global solar radiation by using neural networks for each mod-el.From these curves an acceptable agreement between measured and estimated daily global radiation is obtained by ANN.
The correlation coef?cient for each model is arranged between 87%and89%.Table2resumes the statistical test between mea-sured and estimated H G.It is clearly shown that,the best perfor-mance is obtained by the?rst model r=89.20%,this is proves the correlation between the global solar radiation and air temper-ature.Therefore,in the case,when we use as input only the relative humidity,the correlation coef?cient is decreases to87%,Also, when we mixed both air temperature and relative humidity the correlation coef?cient is weakly decreases to88.99%.
In any case the best model of these three ANN-models cannot provides accurately such as the above second ANN-model devel-oped(ANN-ST model),which need in his inputs air temperature and the sunshine duration.Therefore,the last three ANN-models can be used in the case when we have not the sunshine duration, because this parameter play very important role for obtaining good accurate results.
In order to show the potential of the proposed ANN-models,we have made a comparative study between designed ANN-models and conventional correlation models Eqs.(1)–(4).
Therefore,the calculated parameters a,b and c for Al-Madinah are given as follows:
H G H0?à0:3824t:2786
S
S0
e12T
H G
0?0:1166à0:2202
S
t1:0723
S
2
e13T
H G H0?0:6369t0:037
T
T Max
e14T
H G
0?0:7556à0:1353
H u
uMax
e15TA comparison between measured and calculated H G by the
above formulas is illustrated in Fig.8a–d as can be seen,the second correlation formulate Eq.(13)gave better results than the others models,since the good value obtained for correlation coef?cient (r=97.48%).
Table3illustrates a comparison between different ANN-models and conventional regression models.According to this table,we can notice that the second model,with S and T as inputs presents better accurate results than others ANN-models done in this study. All models exhibited low MBE values(Table3).For most of the models,the MBE values are comparable to the experimental error for the ANN-models proposed by this research and it cannot be considered as decisive for the prevalence of any one of the models.
7.Conclusion
In this paper,ANN-models for estimating of solar radiation in Al-Madinah(Saudi Arabia)have been developed.Measured daily global solar radiation was compared with those obtained by the different designed ANN-models.Obtained results indicate that the second ANN-model(ANN-ST model)has better accurate results than the others ANN-models.
However,for each developed ANN-models the correlation coef-?cient r is grater than97%.In addition,in this study case,only one hidden layer is suf?cient for estimating the daily global solar radi-ation from other parameters,and the number of neurons in the hidden layer is arranged between three and?ve neurons.
It should be noted that the sunshine duration play very impor-tant role for obtaining high accurate results,therefore,this has been proven in the case where we have eliminated the sunshine duration from the inputs of the three ANN-models.Also it has been demonstrated that,the ANN-models which use only the air tem-perature and day of year as inputs can give a good results to the others models in the point view correlation coef?cient.
Comparing the RMSE values for all model presented in this work,it can be seen that the model ANN-ST exhibits the best re-sults.The results obtained render the ANN methodology as a prom-ising alternative to the traditional approach for estimating the global solar radiation.However,in the case where we have not the sunshine duration we can use the three developed model with-out sunshine duration.
Acknowledgement
The authors would like to thank the International Centre for Theoretical Physics,Trieste(Italy)for providing material for achieving the present work.
Appendix A
Pseudo-code of the Levenberg–Marquardt algorithm
While not stop-criterion do
{
Calculates E p(w)for each pattern p
E1<à
P N
p?1
E pewTT E pewT;
Calculates J p(w)for each pattern p
Repeat
Calculates D w;
E2<à
P N
p?1
E pewtD wTT E pewtD wT;
If E1E2then
l<àl?b;
EndIf
Until E2 l<àl/b; w<àw+D w; EndWhile } Appendix B The mean bias error:MBE?1 P N i?1 H GeiTàe H GeiT The root mean square error:RMSE?1 ??????????????????????????????????????????????? P N i?1 H GeiTàe H GeiT 2 r Table3 Comparative study between developed ANN-models and conventional regression models. Models r(%)RMSE ANN-models H G?e fet;ST97.440.049515 H G?e fet;S;TT97.650.044262 H G?e fet;S;T;H uT97.540.044121 H G?e fet;TT89.200.143534 H G?e fet;H uT87.000.167655 H G?e fet;T;H uT88.990.165374 Conventional regression models H G H0?à0:3824t1:2786S S0 97.280.05120 H G H0?0:1166à0:2202S S0 t1:0723S S0 2 97.480.04410 H G H0?0:6369t0:037T T Max 89.500.12154 H G H0?0:7556à0:1353H u H uMax 86.590.25181 1654M.Benghanem et al./Energy Conversion and Management50(2009)1644–1655 The mean percentage error:MPE?1 N P N i?1 H GeiTàe H GeiT àá H GeiT eT 100 The correlation coef?cient:r? ??????????????????????????????????????????????????????????????????? P N i?1 H GeiTàH GeiT eTe H GeiTàe H GeiT 2 P N i?1 H GeiTàH GeiT eT2 P N i?1 e H GeiTàH GeiT 2 v u u u t Appendix C The weights and bias for the?rst model H G=f(t,S)(one hidden layer within two neurons) w1?à11:6496à8:614 1:90730:1642 ! ;w2? 0:1133 189:2765 !T ; b1?15:8106 à4:7057 ! ;and b2??189:1878 The weights and bias for the second model H G=f(t,S,T)(one hidden layer within three neurons) w1? à2:1318à0:33550:1904 à24:8997à15:8987à1:9004 à4:1366à5:7393à4:0259 2 64 3 75;w 2 ? à44:0741 0:0939 0:1156 2 64 3 75 T ; b1? 4:1163 33:2602 5:0252 2 64 3 75and b 2 ??43:9390 The weights and bias for the third model H G?fet;S;T;H uT(one hidden layer within three neurons) w1? 8:5171à6:528à9:29202:5949 6:22220:2803à0:9850à0:3057 à65:039975:129618:8369à21:39990 3:42700:86530:8688à0:0143 1:98618:9220à0:65820:1375 2 66 66 66 4 3 77 77 77 5 ; w2? 0:0325 7:2711 0:0351 0:3359 à0:1491 2 66 66 66 4 3 77 77 77 5 T ;b1? à2:3850 à6:7339 à1:8850 à2:8539 à8:1378 2 66 66 66 4 3 77 77 77 5 and b2??7:5913 References [1]Zekai S.Solar energy fundamentals and modeling techniques: atmosphere,environment,climate change and renewable energy.Springer; 2008. [2]Mellit A,Kalogirou SA.Arti?cial intelligence techniques for photovoltaic applications:a review.Progr Energy Combust Sci2008;34:574–632. [3]Mohandes M,Rehman S,Halawani TO.Estimation of global solar radiation using arti?cial neural networks.Renew Energy1998;14(1–4):179–84. [4]Alawi SM,Hinai HA.An ANN-based approach for predicting global radiation in locations with no direct measurement instrumentation.Renew Energy 1998;14(1–4):199–204. [5]Mohandes M,Balghonaim A,Kassas M,Rehman S,Halawani https://www.doczj.com/doc/6918202378.html,e of radial basis functions for estimating monthly mean daily solar radiation.Sol Energy 2000;68(2):161–8. [6]Reddy KS,Manish R.Solar resource estimation using arti?cial neural networks and comparison with other correlation models.Energy Convers Manage 2003;44:2519–30. [7]Sozen A,Arcakly′og￠lu E,Ozalp M.Estimation of solar potential in Turkey by arti?cial neural networks using meteorological and geographical data.Energy Convers Manage2004;45(18–19):3033–52. [8]Mellit A,Benghanem M,Hadj Arab A,Guessoum A.A simpli?ed model for generating sequences of global radiation data for isolated sites:using arti?cial neural network and a library of Markov transition matrices.Sol Energy 2005;79(5):468–82. [9]Hontoria L,Aguilera J,Zu?ria P.An application of the multilayer perceptron: solar radiation maps in Spain.Sol Energy2005;79:523–30. [10]Tymvios FS,Jacovides CP,Michaelides SC,Scouteli https://www.doczj.com/doc/6918202378.html,parative study of Angstroms and arti?cial neural networks methodologies in estimating global solar radiation.Sol Energy2005;78:752–62. [11]Mellit A,Hadj Arab A,Khorissi N,Salhi H.An ANFIS-based forecasting for solar radiation data from sunshine duration and ambient temperature.In:IEEE power engineering society general meeting;24–28June2007,Florida(USA).p. 1–6. [12]Bosch JL,Lopez G,Batllesa FJ.Daily solar irradiation estimation over a mountainous area using arti?cial neural networks.Renew Energy 2008;33(7):1622–8. [13]Mellit A,Kalogirou SA,Shaari S,Salhi H,Hadj Arab A.Methodology for predicting sequences of mean monthly clearness index and daily solar radiation data in remote areas:application for sizing a stand-alone PV system.Renew Energy2008;33(7):1570–90. [14]Senkal S,Kuleli T.Estimation of solar radiation over Turkey using arti?cial neural network and satellite data.Appl Energy2009;86(7–8):1222–8. [15]Rehman S,Mohandes M.Arti?cial neural network estimation of global solar radiation using air temperature and relative humidity.Energy Policy 2008;36:571–6. [16]Benghanem M,Joraid AA.A multiple correlation between different solar parameters in Medina,Saudi Arabia.Renew Energy2007;32(14):2424–35. [17]Haykin S.Neural networks,a comprehensive foundation.New York:MacMillan;1999. M.Benghanem et al./Energy Conversion and Management50(2009)1644–16551655 centrifugemodelling离心模拟centrifuge离心机centripetalacceleration向心加速度centripetalforce向心力centripetal向心的centrode瞬心轨迹centroidaxis重心轴线centroidofarea面心centroid质心centroidalprincipalaxesofinertia重心诌性轴centrosymmetric中心对称的ceramicbearing陶瓷轴承ceramicbond陶瓷结合剂ceramiccoating陶瓷涂层ceramicengine陶瓷发动机ceramicfilter陶瓷过滤器ceramicindustry陶瓷工业ceramici ulater陶瓷绝缘子ceramictip陶瓷刀片ceramictool陶瓷车刀ceramic 陶瓷的ceramics陶瓷cermettool金属陶瓷刀具cermet金属陶瓷cesium铯cetanenumber十六烷值CF 吸顶型风机CeilingMountedTypeFancfrp碳纤维增强塑料cg ystemcgs单位制cgsunit厘米克秒单位chai eltconveyor链带输送机chai elt链带chai lock链动滑轮chai rake链闸chai ridge链桥nondime ional无量纲的nondirective非定向的nondi ersivewave非弥散波nonelasticbuckling非弹性屈曲nonelasticscattering非弹性散射nonelastic非弹性的nonequilibriumplasma非平衡等离子体nonequilibriumproce非平衡过程nonequilibriumstate非平衡态nonequilibriumstate非平衡状态nonequilibriumsurfacete ion非平衡表面张力nonequilibriumthermodynamics非平衡态热力学nonequilibrium非平衡nonevanescentwave无阻尼波nonferrousalloy非铁合金nonferrousmetal有色金属nonferrousmetallurgy非铁金属冶炼术nonflatne不平面度nonfreepoint非自由质点nongeostrophicflow非地转怜nonholonomicco traint非完整约束nonholonomicsystem非完整系统nonholonomicvelocitycoordinate非完整速度坐标nonhomogeneou oundary非齐次边界nonidealflow非理想怜noninertia无惯性noninertialsystemofcoordinates非惯性坐标系noninertialsystem非惯性系nonisentropicflow非等熵怜nonisotropicmaterial非蛤同性材料nonisotropy非蛤同性nonius游标nonlinearaerodynamics非线性空气动力学nonlineardistortion非线性失真nonlinearelectrodynamics非线性电动力学collarbearing环状止推轴承collarheadscrew带缘螺钉collarjournal有环轴颈collarnut凸缘螺collarpin凸缘销collarthrustbearing环状止推轴承collarvortex涡环collar凸边collateralmotion次级运动collectivemodeofmotion集体运动模式collectivemotion集体运动collectorefficiency集热僻率collectorring滑环collector集电器colletchuck弹簧夹头collier运煤船collimation视准collimation准直collimator视准仪Collimator准直仪collisionchain碰撞链collisioncro ection碰撞截面collisiondiameter 碰撞直径collisiondiffusion碰撞扩散collisionexcitation碰撞激发collisionfrequency碰撞频率 外阴白色病变考试试题 一、A1型题（本大题11小题．每题1.0分，共11.0分。每一道考试题下面有A、 B、C、D、E五个备选答案。请从中选择一个最佳答案，并在答题卡上将相应题号的相应字母所属的方框涂黑。） 第1题 外阴硬化性苔癣的早期病理改变是 A 表层细胞过度角化 B 表层细胞增生 C 真皮乳头层水肿 D 毛囊角质栓塞 E 底层细胞增生 【正确答案】：C 【本题分数】：1.0分 第2题 女，50岁。因外阴瘙痒而就医，组织病理为增生型，营养不良，下列治疗中哪项是正确的 A 因有恶变趋向，应及早手术治疗 B 全身治疗 C 补充多量维生素 D 活检有非典型增生时手术治疗 E 全身+局部治疗 【正确答案】：D 【本题分数】：1.0分 第3题 关于外阴瘙痒下列哪项是正确的 A 外阴瘙痒是外阴癌的早期表现 B 外阴瘙痒最常见的原因是蛲虫病 C 外阴瘙痒严重时用肥皂液清洗会有所好转 D 外阴瘙痒经治疗无效应作单纯性外阴切除术 E 外阴瘙痒不是一种疾病而是多种疾病可引起的一种症状 【正确答案】：E 【本题分数】：1.0分 第4题 女，37岁。外阴奇痒，分泌物不多。妇检：两侧小阴唇增厚，外阴粘膜不红，阴道畅，皱襞正常，无异常分泌物，宫颈柱状，光滑，I°肥大，子宫前位，常大，双附件(-)，为确诊应选用 A 外阴活检 B 阴道分泌物涂片 C 宫颈涂片(CG D 阴道镜 E 盆腔B超 【正确答案】：A 【本题分数】：1.0分 第5题 属于癌前病变的外阴白色病变是 A 增生型营养不良 B 硬化苔癣型营养不良 C 混合型营养不良 D 营养不良伴有上皮不典型增生 E 白癜风 【正确答案】：D 【本题分数】：1.0分 第6题 精神分裂症的病因及发病机理 精神分裂症病因：尚未明，近百年来的研究结果也仅发现一些可能的致病因素。（一）生物学因素1．遗传遗传因素是精神分裂症最可能的一种素质因素。国内家系调查资料表明：精神分裂症患者亲属中的患病率比一般居民高6.2倍，血缘关系愈近，患病率也愈高。双生子研究表明：遗传信息几乎相同的单卵双生子的同病率远较遗传信息不完全相同 的双卵双生子为高，综合近年来11项研究资料：单卵双生子同病率（56.7%），是双卵双生子同病率（12.7%）的4.5倍，是一般人口患难与共病率的35-60倍。说明遗传因素在本病发生中具有重要作用，寄养子研究也证明遗传因素是本症发病的主要因素，而环境因素的重要性较小。以往的研究证明疾病并不按类型进行遗传，目前认为多基因遗传方式的可能性最大，也有人认为是常染色体单基因遗传或多源性遗传。Shields发现病情愈轻，病因愈复杂，愈属多源性遗传。高发家系的前瞻性研究与分子遗传的研究相结合，可能阐明一些问题。国内有报道用人类原癌基因Ha-ras-1为探针，对精神病患者基因组进行限止性片段长度多态性的分析，结果提示11号染色体上可能存在着精神分裂症与双相情感性精神病有关的DNA序列。2．性格特征：约40%患者的病前性格具有孤僻、冷淡、敏感、多疑、富于幻想等特征，即内向 型性格。3．其它：精神分裂症发病与年龄有一定关系，多发生于青壮年，约1/2患者于20～30岁发病。发病年龄与临床类型有关，偏执型发病较晚，有资料提示偏执型平均发病年龄为35岁，其它型为23岁。80年代国内12地区调查资料：女性总患病率（7.07%。）与时点患病率（5.91%。）明显高于男性（4.33%。与3.68%。）。Kretschmer在描述性格与精神分裂症关系时指出：61%患者为瘦长型和运动家型，12.8%为肥胖型，11.3%发育不良型。在躯体疾病或分娩之后发生精神分裂症是很常见的现象，可能是心理性生理性应激的非特异性影响。部分患者在脑外伤后或感染性疾病后发病；有报告在精神分裂症患者的脑脊液中发现病毒性物质；月经期内病情加重等躯体因素都可能是诱发因素，但在精神分裂症发病机理中的价值有待进一步证实。（二）心理社会因素1．环境因素①家庭中父母的性格，言行、举止和教育方式（如放纵、溺爱、过严）等都会影响子女的心身健康或导致个性偏离常态。②家庭成员间的关系及其精神交流的紊乱。③生活不安定、居住拥挤、职业不固定、人际关系不良、噪音干扰、环境污染等均对发病有一定作用。农村精神分裂症发病率明显低于城市。2．心理因素一般认为生活事件可发诱发精神分裂症。诸如失学、失恋、学习紧张、家庭纠纷、夫妻不和、意处事故等均对发病有一定影响，但这些事件的性质均无特殊性。因此，心理因素也仅属诱发因 外阴疾病 外阴：阴道口外边的外露部分肛门、阴道口及尿道邻近，经常受阴道分泌物及尿液的浸渍，容易发炎。 常见病症：外因瘙痒、外阴炎、外阴白色病变、外因溃疡、外阴癌 外阴疾病 外阴：阴道口外边的外露部分肛门、阴道口及尿道邻近，经常受阴道分泌物及尿液的浸渍，容易发炎。 常见病症：外因瘙痒、外阴炎、外阴白色病变、外因溃疡、外阴癌 外因瘙痒 外阴瘙痒是妇科疾病中很常见的一种症状，外阴是特别敏感的部位，妇科多种病变及外来刺激均可引起瘙痒，使人寝食难安、坐卧不宁。外阴瘙痒多发生于阴蒂、小阴唇，也可波及大阴唇、会阴和肛周 病因：1.慢性局部刺激，外阴、阴道、宫颈炎症的异常分泌物的刺激； 2.外阴不清洁及紧身化纤内裤、卫生巾等致通透不良； 3.外阴寄生虫病，如阴虱、蛲虫、疥疮等； 4.各种外阴皮肤病和外阴肿瘤等； 5.全身性疾病的外阴局部症状，如糖尿病、尿毒症、维生素缺乏等。 症状：外因皮肤瘙痒、烧灼感和疼痛瘙痒部位多生与阴帝、小阴唇、也可波及大阴、会阴、甚至肛周围 危害：(1)性外阴部瘙痒严重时，不但使人坐卧不宁，影响工作、学习、生活和睡眠。 (2)由于女性外阴瘙痒，会影响夫妻生活，所以极有导致夫妻感情不和，严重的 甚至造成感情破裂，婚姻走向终点。 (3)诱发生殖器感染，盆腔炎、肾周炎、性交痛等，日久不愈还可导致多种疾病 同时发生，疾病的危害严重的会危害女性健康，甚至还会造成女性不孕等严重后果。 (4)女性外阴瘙痒严重时，不易根治，易反复，引发早产、胎儿感染畸形等，造 成终身遗憾。 治疗1.外阴涂药 使用有止痒作用的洗剂、膏霜等，如炉甘石洗剂、苯海拉明软膏、皮质醇类软膏等。 2.局部封闭或穴位注药 如皮质醇激素、维生素B12、非那根等。 3.针对病因治疗。 4.预防1. 注意经期卫生，勤清洗。 2.不要冲洗阴道，因为阴道有自清的功能，如果刻意冲洗反而不利 3.忌乱用、烂用药物，忌抓搔及局部摩擦。 4.忌酒及辛辣食物，不吃海鲜等及易引起过敏的药物 6 .久治不愈者应作血糖检查。少吃糖类可避免常常感染霉菌，如少吃淀粉类、糖类以及刺激性的食物(例如酒、辛辣物、油炸类)，多吃蔬菜水果类，水份要充足。 5、不穿紧身兜裆裤，内裤更须宽松、透气，并以棉制品为宜。 6.就医检查是否有霉菌或滴虫，如有应及时治疗，而不要自己应用“止痒水”治疗。 8.保持外阴清洁干燥，尤其在经期、孕期、产褥期，每天用女性护理液清洗外阴更换内裤。 9.不穿化纤内裤、紧身裤，着棉织内衣裤。局部坐浴时注意溶液浓度、温度及时间、注意事项。 10.外阴瘙痒者应勤剪指甲、勤洗手，不要搔抓皮肤，以防破溃感染从而继发细菌性感染。 11.上完厕所请记得由前往后擦，因为肛门可能会带来不少细菌，所以如厕后请不要由肛门擦到阴部，才能减少感染的机会。 12.内裤要和其他的衣物分开洗，最好暴晒，可以减少细菌的滋生。如果患有霉菌性阴道炎的话，最好内裤都有热水煮 外阴溃疡外阴溃疡是发生于外阴部的皮肤黏膜发炎、溃烂、缺损。病灶多发生于小阴唇和大阴唇内侧，其次为前庭黏膜及阴道口周围。病程有急性及慢性。 大小阴唇、阴道口周围、阴蒂等处（外阴疾病发展中出现的一个过程，不是一个独立的疾病，有急性和慢性）急性外阴溃疡：非特异性外阴炎病情较轻，多在搔抓之后出现一般比较表浅，但疼痛比较厉害 慢性外阴溃疡：持续时间较长，或者反复发作 癌症引起的溃疡，与结核性溃疡很难鉴别，需做确诊 Fifth Generation Communication Automotive Research and innovation Deliverable D3.2 Report on Channel Modelling and Positioning for 5G V2X Version: v1.0 2018-11-30 This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 761510. Any 5GCAR results reflects only the authors’ view and the Commission is thereby not responsible for any use that may be made of the information it contains. http://www.5g-ppp.eu Deliverable D3.2 Report on Channel Modelling and Positioning for 5G V2X Abstract 5GCAR has identified the most important use cases for future V2X communications together with their key performance indicators and respective requirements. One outcome of this study is that accurate positioning is important for all these use cases, however with different level of accuracy. In this deliverable we summarize existing solutions for positioning of road users and justify that they are not sufficient to achieve the required performance always and everywhere. Therefore, we propose a set of solutions for different scenarios (urban and highway) and different frequency bands (below and above 6 GHz). Furthermore, we link these new technical concepts with the ongoing standardization of 3GPP New Radio Rel-16. An important prerequisite for this work is the availability of appropriate channel models. For that reason, we place in front a discussion of existing channel models for V2X, including the sidelink between two road users, their gaps, as well as our 5GCAR contributions beyond the state of the art. This is complemented with results from related channel measurement campaigns. Comparison of Artificial Neural Network and Regression Models in Software Effort Estimation I.F. Barcelos Tronto, J.D. Sim?es da Silva, N. Sant'Anna 1Laboratory for Computing and Applied Mathematics - LAC Brazilian National Institute for Space Research - INPE C. Postal 515 – 12245-970 – S?o José dos Campos - SP BRAZIL E-mail: {iris_barcelos, demisio, nilson}@lac.inpe.br Keywords: software effort, artificial neural network, regression analysis, software development estimate Abstract: Estimating development effort remains a complex problem attracting considerable research attention. Improving the estimation techniques available to project managers would facilitate more effective control of time and budgets in software development. In this paper, predictive Artificial Neural Network and regression based models are investigated, comparing the performance of both methods. The results show that ANNs are effective in effort estimation. 1. Introduction The continuous hardware and software development, jointly with the world economical interaction phenomenon has contributed to the competitiveness increase between producing and delivering companies of software product and services. In addition, there has been a growing need to produce low cost high quality software in a short time. A quality level and international productivity can be achieved through the use of effective software management process, focalizing people, product, process, and project. The project requires planning and accompaniment supported by a group of activities, among which the estimates (effort, resources, time, etc.) are fundamental, because they supply a guide for the other activities. The predictive process involves the set of procedures presented in the Figure 1 [1]. Software size estimates are important to determine the software project effort [2], [3,] [4], [5]. However, according to the last research reported by the Brazilian Ministry of Science and Technology - MCT, in 2001, only 29% of the companies accomplished size estimates and 45,7% accomplished software effort estimate [6]. There is not a specific study that identifies the causes of the effort low estimates index, but the reliability level of the models can be a possible cause. These data presented by MCT evidences the importance to use an effort estimate alternative approach, through which one can have reliable estimates with simple execution model. Figure 1. The predictive process [1] Biometrics59,614–623 September2003 Partial AUC Estimation and Regression Lori E.Dodd1,?and Margaret S.Pepe2,?? 1Biometric Research Branch,National Cancer Institute,6130Executive Blvd,MSC7434, Rockville,Maryland20892,U.S.A. 2Department of Biostatistics,University of Washington,Box357232,Seattle,Washington98195,U.S.A. ?email:doddl@https://www.doczj.com/doc/6918202378.html, ??email:mspepe@https://www.doczj.com/doc/6918202378.html, Summary.Accurate diagnosis of disease is a critical part of health care.New diagnostic and screening tests must be evaluated based on their abilities to discriminate diseased from nondiseased states.The partial area under the receiver operating characteristic(ROC)curve is a measure of diagnostic test accuracy.We present an interpretation of the partial area under the curve(AUC),which gives rise to a nonparametric estimator. This estimator is more robust than existing estimators,which make parametric assumptions.We show that the robustness is gained with only a moderate loss in e?ciency.We describe a regression modeling framework for making inference about covariate e?ects on the partial AUC.Such models can re?ne knowledge about test accuracy.Model parameters can be estimated using binary regression methods.We use the regression framework to compare two prostate-speci?c antigen biomarkers and to evaluate the dependence of biomarker accuracy on the time prior to clinical diagnosis of prostate cancer. Key words:Diagnostic testing;Mann-Whitney U-statistic;Regression;Receiver operating characteristic curve. 1.Introduction Screening and diagnostic tests are familiar and ever-evolving tools of modern medicine.Populations of healthy individu- als characterized as at-risk are commonly screened for dis- eases such as cancer and heart disease.Early detection by screening is considered essential to alleviate disease burden, and considerable resources have been devoted to developing new screening tests.New diagnostic tests that are less inva- sive,less expensive,and more accurate than existing proce- dures are sought for diagnosis of many conditions.Technolo- gies that measure gene and protein expression,as well as new imaging procedures,all hold promise in this regard.Prior to widespread application,however,rigorous evaluation of test accuracy and of factors that e?ect it is compulsory. Inherent in the analysis of screening and diagnostic tests are costs and bene?ts,both monetary and nonmonetary,associ- ated with true-positive and false-positive diagnoses.Consider a continuous test result Y for which Y>c indicates a positive test result,and let D andˉD denote diseased and nondiseased states,respectively.The true-positive rate at a threshold c, TPR(c),is de?ned as P(Y>c|D)≡S D(c).The corresponding false-positive rate,FPR(c),is P(Y>c|ˉD)≡SˉD(c).Costs and bene?ts are associated with any given{FPR(c),TPR(c)} pair.The receiver operating characteristic(ROC)curve plots {FPR(c),TPR(c)}for all possible thresholds c,and provides a visual description of the trade-o?s between TPRs and FPRs as one changes the threshold stringency(Figure1).We can write the ROC curve as a function of t=SˉD(c)as follows: ROC(t)=S D{S?1ˉ D (t)}.An uninformative test is represented by a straight line from the(0,0)vertex to(1,1),while a curve pulled closer towards the(0,1)vertex indicates a better- performing test. Frequently,the best threshold is not known when a test is under evaluation,and it may vary depending on the setting in which the test is implemented.A summary measure that aggregates performance information across a range of pos- sible thresholds is desirable.The area under the ROC curve (AUC),de?ned as 1 ROC(t)dt,summarizes across all thresh- olds,and is the most commonly used measure of diagnostic accuracy for quantitative tests.However,the AUC summa- rizes test performance over regions of the ROC space in which one would never operate,i.e.,for{FPR(c),TPR(c)}values of no practical interest.In population screening,large monetary costs result from high false-positive rates;hence the region of the curve corresponding to low false-positive rates is of pri- mary interest.In diagnostic testing,it is critical to maintain a high TPR in order not to miss detecting subjects with dis- ease.In this case,interest is in the region of the ROC curve corresponding only to acceptably high TPRs.In this article, we consider a summary index for the ROC curve restricted to a clinically relevant range of false-positive rates.The partial AUC is AUC(t0,t1)= t1 t0 ROC(t)dt,(1) where the interval(t0,t1)denotes the false-positive rates of in- terest.The analogue that restricts to a range of true-positive rates will also be discussed.Selecting the interval(t0,t1)is an 614 外阴白色病变的症状表现有哪些 外阴白色病变是慢性外阴的营养不良。属于营养不良的一种。而这也有分为好几个类型，混合型、增生型和硬化苔藓型等等都是外阴白色病变的类型。 外阴奇痒为主要症状，搔痒时间从发病到治疗有2～3月之内，也有达20 年之久，搔痒剧烈程度不分季节与昼夜，如伴有滴虫性或霉菌性阴道炎，分泌物会更多，局部烧灼感，刺痛与搔痒所致的皮肤粘膜破损或感染有关，局部有不同程度的皮肤粘膜色素减退，常有水肿，皲裂及散在的表浅溃疡。 一、增生型营养不良 一般发生在30～60岁的妇女，主要症状为外阴奇痒难忍，抓伤后疼痛加剧，病变范围不一，主要波及大阴唇，阴唇间沟，阴蒂包皮和后联合处，多呈对称性，病变皮肤增厚似皮革，隆起有皱襞，或有鳞屑，湿疹样改变，表面颜色多暗红或粉红，夹杂有界限清晰的白色斑块，一般无萎缩或粘连。 二、硬化苔藓型营养不良 可见于任何年龄，多见于40岁左右妇女，主要症状为病变区发痒，但一般远较增生型病变为轻，晚期出现性交困难，病变累及外阴皮肤，粘膜和肛周围皮肤，除皮肤或粘膜变白，变薄，干燥易皲裂外，并失去弹性，阴蒂多萎缩，且与包皮粘连，小阴唇平坦消失，晚期皮肤菲薄皱缩似卷烟纸，阴道口挛缩狭窄，仅容指尖。 幼女患此病多在小便或大便后感外阴及肛周不适，外阴及肛周区出现锁孔状珠黄色花斑样或白色病损，一般至青春期时，病变多自行消失。 三、混合型营养不良 主要表现为菲薄的外阴发白区的邻近部位，或在其范围内伴有局灶性皮肤增厚或隆起。 四、增生型或混合型伴上皮非典型增生 一般认为在增生型及混合型病变中，仅5、10例可出现非典型增生，且此非典型增生仅限于增生的上皮细胞部分。非典型增生多无特殊临床表现，局部组织活体组织检查为唯一的诊断方法。但如外阴局部出现溃疡。或界限清楚的白色隆起时，在该处活检发现非典型增生，其癌变的可能性较大。 精神分裂症的发病原因是什么 精神分裂症是一种精神病，对于我们的影响是很大的，如果不幸患上就要及时做好治疗，不然后果会很严重，无法进行正常的工作和生活，是一件很尴尬的事情。因此为了避免患上这样的疾病，我们就要做好预防，今天我们就请广州协佳的专家张可斌来介绍一下精神分裂症的发病原因。 精神分裂症是严重影响人们身体健康的一种疾病，这种疾病会让我们整体看起来不正常，会出现胡言乱语的情况，甚至还会出现幻想幻听，可见精神分裂症这种病的危害程度。 (1)精神刺激：人的心理与社会因素密切相关，个人与社会环境不相适应，就产生了精神刺激，精神刺激导致大脑功能紊乱，出现精神障碍。不管是令人愉快的良性刺激，还是使人痛苦的恶性刺激，超过一定的限度都会对人的心理造成影响。 (2)遗传因素：精神病中如精神分裂症、情感性精神障碍，家族中精神病的患病率明显高于一般普通人群，而且血缘关系愈近，发病机会愈高。此外，精神发育迟滞、癫痫性精神障碍的遗传性在发病因素中也占相当的比重。这也是精神病的病因之一。 (3)自身：在同样的环境中，承受同样的精神刺激，那些心理素质差、对精神刺激耐受力低的人易发病。通常情况下，性格内向、心胸狭窄、过分自尊的人，不与人交往、孤僻懒散的人受挫折后容易出现精神异常。 (4)躯体因素：感染、中毒、颅脑外伤、肿瘤、内分泌、代谢及营养障碍等均可导致精神障碍，。但应注意，精神障碍伴有的躯体因素，并不完全与精神症状直接相关，有些是由躯体因素直接引起的，有些则是以躯体因素只作为一种诱因而存在。 孕期感染。如果在怀孕期间，孕妇感染了某种病毒，病毒也传染给了胎儿的话，那么，胎儿出生长大后患上精神分裂症的可能性是极其的大。所以怀孕中的女性朋友要注意卫生，尽量不要接触病毒源。 上述就是关于精神分裂症的发病原因，想必大家都已经知道了吧。患上精神分裂症之后，大家也不必过于伤心，现在我国的医疗水平是足以让大家快速恢复过来的，所以说一定要保持良好的情绪。 外阴白色病变的检查诊断方法是什么 外阴奇痒是外阴白色病变的主要症状，搔痒时间从发病到治疗有2～3月之内，也有达20年之久，搔痒剧烈程度不分季节与昼夜。专家提示，一旦发现自己有类似于外阴白色病变的这种，应立即到医院进行确诊。早期的诊断及治疗对我们早日恢复健康并且尤为重要。 外阴白色病变的检查： 多点活检送病理检查，确定病变性质，排除早期癌变，活检应在有皲裂，溃疡，隆起，硬结或粗糙处进行，为做到取材适当，可先用1%甲苯胺蓝(toluidine blue)涂病变区，待白干后，再用1%醋酸液擦洗脱色，凡不脱色区表示该处有裸核存在，提示在该处活检，发现非典型增生或甚至癌变的可能性较大，如局部破损区太广，应先治疗数日，待皮损大部愈合后，再选择活检部位以提高诊断准确率。 外阴白色病变的诊断： 1、症状判断：外阴白斑一般根据症状就可以判断，比如，外阴局部粘 膜发白，瘙痒、粗糙、脱屑等现象的出现，都会诊断为外阴白斑，当然，外阴白斑有很多的类型，如果外阴白斑属于增生型，也就是说局部的皮肤粘膜增厚了，弹性变差了，而且也出现了相应的溃疡等不适症状。这是主要的外阴白斑的诊断方法。 2、细胞活检：有时外阴白斑的诊断需要进一步的做细胞活检，观察有 没有癌细胞，以便于确诊。活检病理检查确定病变性质，排除早期癌变。活检应在有皲裂溃疡、隆起、硬结或粗糙处进行为做到取材适当，外阴白斑的诊断方法可先用1%甲苯胺蓝涂病变区，待白干后再用1%醋酸液擦洗脱色。凡不脱色区表示该处有裸核存在，提示在该处活检发现非典型增生或甚至癌变的可能性较大。如局部破损区太广，应先治疗数日待皮损大部愈合后，再选择活检部位以提高诊断准确率。 3、病理诊断依据：除了解疾病的主要临床症状外，还应对疾病的发病 机理有一定的认识，因为导致外阴白斑皮肤瘙痒及色素的减退或脱色的疾病有很多种，不只是外阴白斑一种，它们的表现虽有些不同，但用肉眼不易区别开来，所以当遇到外阴有病损不典型或慢性皲裂、局限性增厚、溃破等症状的患者时，必须依靠活组织病理检查确诊。 精神分裂症的病因是什么 精神分裂症是一种精神方面的疾病，青壮年发生的概率高，一般 在16～40岁间，没有正常器官的疾病出现，为一种功能性精神病。 精神分裂症大部分的患者是由于在日常的生活和工作当中受到的压力 过大，而患者没有一个良好的疏导的方式所导致。患者在出现该情况 不仅影响本人的正常社会生活，且对家庭和社会也造成很严重的影响。 精神分裂症常见的致病因素： 1、环境因素：工作环境比如经济水平低低收入人群、无职业的人群中，精神分裂症的患病率明显高于经济水平高的职业人群的患病率。还有实际的生活环境生活中的不如意不开心也会诱发该病。 2、心理因素：生活工作中的不开心不满意，导致情绪上的失控，心里长期受到压抑没有办法和没有正确的途径去发泄，如恋爱失败， 婚姻破裂，学习、工作中不愉快都会成为本病的原因。 3、遗传因素：家族中长辈或者亲属中曾经有过这样的病人，后代会出现精神分裂症的机会比正常人要高。 4、精神影响：人的心里与社会要各个方面都有着不可缺少的联系，对社会环境不适应，自己无法融入到社会中去，自己与社会环境不相 适应，精神和心情就会受到一定的影响，大脑控制着人的精神世界， 有可能促发精神分裂症。 5、身体方面：细菌感染、出现中毒情况、大脑外伤、肿瘤、身体的代谢及营养不良等均可能导致使精神分裂症，身体受到外界环境的 影响受到一定程度的伤害，心里受到打击，无法承受伤害造成的痛苦，可能会出现精神的问题。 对于精神分裂症一定要配合治疗，接受全面正确的治疗，最好的 疗法就是中医疗法加心理疗法。早发现并及时治疗并且科学合理的治疗，不要相信迷信，要去正规的医院接受合理的治疗，接受正确的治 疗按照医生的要求对症下药，配合医生和家人，给病人创造一个良好 的治疗环境，对于该病的康复和痊愈会起到意想不到的效果。 财务模型培训 今天的议题 财务模型可以用来做什么？ 财务模型可以用来做什么？ 战略规划项目中的财务预测模型是什么？用来干什么？ 战略规划项目中的财务预测模型是什么？用来干什么？centrifugemodelling离心模拟centrifuge离心机
外阴白色病变考试试题
精神分裂症的病因及发病机理
外阴疾病
欧洲车联网项目 5GCAR_D3.2-Channel Modelling and Positioning for 5G V2X
Effort Estimation
AUC Estimation
外阴白色病变的症状表现有哪些
精神分裂症的发病原因是什么
外阴白色病变的检查诊断方法是什么
精神分裂症的病因是什么
financial modelling training
财务预测模型介绍 工具与数据 构建简单的财务预测模型 构建复杂的财务预测模型
预算Budgeting 成本估算Cost Estimating 销售预测Sales Forecasting 销售预测 市场表现预测Market Share Forecasting 市场表现预测 投资项目的选择Project Selection & Management 投资项目的选择 业务/产品组合管理－ 业务 产品组合管理－组合方案与优化 产品组合管理 实时的市场分析Time-to-Market Analysis 战略决策的评估Real Options Valuation 并购M & A
战略规划项目中的财务预测模型是 模拟各业务在不同运营策略/ 各业务在不同运营策略 模拟各业务在不同运营策略/业务模 式下及不同市场环境下各项业务的 式下及不同市场环境下各项业务的 大体财务表现及发展趋势。因此， 大体财务表现及发展趋势。因此，
在模拟业务表现时，先不考虑股权比 例、投资收益、少数股东权益等问 题；在最后并总表时，可以综合考虑 最后计数单位宜采用“千元”、“百 万元”等较大计量单位，切忌采用 元、角、分等极其精准的单位计量
财务预测模型绝对不能直接作为： 财务预测模型绝对不能直接作为： 直接作为
–
精准的预算 绩效考核的标准
–
但是，可以作为制定预算时的初步参 照，且其中部分有关运营方面的假设 项，例如产能利用率、损耗率等也可在 绩效考核中作为参照
在战略项目中， 财务预测模型将用于 在战略项目中，财务预测模型将用于 在战略项目中， 在战略项目中， 比较不同战略备选方案的经济价值 比较不同战略备选方案的经济价值 促进战略规划、行动方案的细化，并进行验证，反过来指导战略的制定 促进战略规划、行动方案的细化，并进行验证，反过来指导战略的制定 根据财务计划对未来进行资源安排提供依据 根据财务计划对未来进行资源安排提供依据
相关主题
文本预览