2007 IEEE/WIC/ACM International Conference on Intelligent Agent Technalogy Introducing Commodity Flow to an Agent-Based Model
E-commerce System
Tomasz Serzysko
Department of Mathematics and Information Science,
Warsaw University of Technology,Poland
Maria Ganzha,Maciej Gawinecki,Pawel Kobzdej,Marcin Paprzycki
System Research Institute,Polish Academy of Sciences,Poland
Costin Badica
Department of Computer Science,University of Craiova,Romania Abstract
In our model agent-based e-commerce system [2] we have assumed that a certain number of items of a given product is available for sale. In this note we introduce a model logistics subsystem and discuss how it will be integrated with the system. Keywords: e-commerce;logistics system;agent
1.Introduction
Currently, we are developing and implementing a model agent-based e-commerce system (see [2] and references collected there). In this system multiple buyer agents attempt at making purchase by participating in price negotiations in e-stores and selecting the best o?er, while e-stores attempt at maximizing profit resulting from product sales.Thus far our attention was focused on buyer-seller interactions. By assuming that products are in the warehouse we have omitted the question where do they come from. The aim of this work is to describe how our system can be extended to include product restocking processes.
Before proceeding let us make a few observations. First, the proposed logistics subsystem is not ―stand alone‖(e.g. similar to that considered in [3, 1, 6]). Instead, it has been created within the context of our e-commerce system, which has directly
influenced its design. Second, while somewhat similar, processes involved in e-store restocking a warehouse differ from client making a purchase in an e-store (e.g. in product demand prediction, interactions with wholesalers, methods of price negotiations that involve more ―cond itions,‖ offer selection criteria, etc.) Therefore we have created a separate logistics subsystem (instead of reusing already modeled functions; e.g. price negotiations). Third, this note is devoted to the agent structure and agent interactions and, due to the space limitations, we omit important topics like:forecasts derivation, o?er evaluation etc.However, these functions can be encapsulated into modules that can become a part of an appropriate agent. Therefore readers should envision that, for instance, when we write that ―rece ived o?ers are evaluated,‖then their favorite method of o?er evaluation has been utilized.
To proceed, first, we briefly describe our e-commerce system.We follow with assumptions that underline the logistics subsystem and description of new agents that were introduced. Finally, we present the sequence diagram of restocking and use it to discuss in detail how this process will take place in our system.
2. System Description
Our system is a distributed marketplace in which software agents perform e-commerce functions (see [2] for details, the Use Case diagram in particular). User-Client is represented by the Client Agent (CA). The CA is autonomous and when a purchase order is communicated by the User-Client, it works until either it is completed, or purchase is abandoned. The CA communicates with the Client Information Center (CIC), which facilitates information which e-stores sell which products (yellow-page matchmaking).For each store that sells the requested product, the CA delegates a Buyer Agent (BA) to participate in price negotiations and if successful, possibly attempt at making a pur- chase (successful price negotiations result in a product reservation for a specific time; after which products that have not been purchased are available for sale again).Since multiple BAs representing the same CA can win price negotiations the CA makes the decision if either of available o?ers is good enough to make a purchase. Buyer Agents can participate in negotiations only if the Gatekeeper Agent (GA) admits them (if they are trusted; e.g. BAs that win price negotiations but do not make a purchase may be barred from subsequent negotiations). The GA represents a given e-store and is created by the Shop Agent (SA). The SA is the central manager and facilitating the selling process it utilizes the GA, and a set of Seller Agents (SeA) that
negotiate price with incoming BAs, as well as a Warehouse Agent (WA) that is responsible for inventory and reservation management. Thus far, the WA was responsible for managing product reservations and the inventory. Specifically, (1) before a new price negotiation the WA ―re served‖ a given product—so that if negotiation ended successfully there was an item that could be sold; (2) when a reservation ended in purchase, it adjusted product counters; and (3) when a reservation expired it also adjusted products counters. However, the WA was always envisioned as the ―gat eway‖between the store and suppliers, which is one of the foci of this note.
3. Assumptions behind the logistics system
Let us now specify assumptions that underline design of the logistics subsystem:
1.Nowadays, except of largest store-chains (e.g. TESCO, WalMart), companies outsource transportation activities (considered non-core business activities) to specialists (e.g. UPS, Mayflower). However, we assume here that suppliers are still re sponsible for facilitating transportation. Therefore, we omit transportation related processes and focus only in interactions between e-stores and suppliers
2. As a result of (1), transportation costs are assumed to be paid by the supplier and included directly in product price (e.g. discount on delivery costs, will manifest itself in the total price).
3. While in the original system auction-based price negotiations were used, here we opted for the simplicity of the FIPA Contract Net Protocol [4]. Therefore, in the logistics subsystem, a single round of negotiations consisting of a call for proposals and evaluation of responses, is used.
4. New functions and agents
In order to perform logistics-related tasks, several new roles were introduced; some of them have been delegated to agents existing in the system, while others warranted adding new agents. Specifically:
?demand estimation—to draw information from sales data and/or external premises to predict future sales of products,
?warehouse monitoring—to observe supply levels and react in case of a risk of
dropping below values considered su?cient to satisfy estimated demand,
?order management—to coordinate issuing orders for goods, to assist in
evaluating received o?ers,
?ordering goods—to contact suppliers for their of- fers and to select the best o?er,
?selling goods—in the system suppliers were also modeled; while goods are
acquired without extensive price negotiations, ―someone‖ has to deliver proposals to ordering components,
?logistics yellow pages—the role of the ―l og istics CIC‖is very similar to the original CIC ([2]); it has to provide lists of potential suppliers of products; obviously, it is possible for a shop to become a sup- plier for another shop and to suggest that the two CIC s could be joined, but we decided to clearly separate the client-side from the shop and from the supply-side. Another reason for this separation was that while some shops may not be interested in becoming wholesaler, we would have to make changes to the original CIC data structure (e.g. wholesaler—yes/no). Finally, since the logistics subsystem does not involve auctions, the separation is even more warranted.
Let us now see how these tasks/roles could be placed in our system. The demand estimation role was attributed to the existing Shop Decision Agent (SDA), responsible for the ―kn owledge managemen t‖functions (e.g. trust management, sales trend data mining, etc.) in the shop.
The warehouse monitoring role is already a part of the existing WA. The di?erence is that now WA becomes a proactive manager of supplies; acting on predictions supplied by the SDA.
Fulfillment of the order management role required introduction of the Logistics Agent (LA), which became the ―centra l manager‖ of the logistics subsystem. It is responsible for contacting the logistics CIC for the list of potential suppliers and managing a pool of agents responsible for ordering goods from ―wholesalers.‖ Finally, it collects and manages data related to supplier reliability. This data, in turn, will be one of factors in selecting the supplier.
The ordering goods process is facilitated by the Ordering Agent (OA), which is also a new agent. Its task is to issue a call for proposals, collect responses and select the best o?er taking into account factors such as: price, delivery time, reliability etc. Let us recall that due to the modularity of agent design ([2]), our system is flexible enough so that any method of selecting an o?er can be applied (it can be encapsulated in a module and plugged into the OA).
The selling goods role is realized by a very simple Wholesale Agent (WhA). Its role is to respond to CFP’s incoming from OA s.Currently we assume that WhA s receive instructions in what way to generate a stream of responses to the CFPs.
Finally, logistics yellow pages are facilitated by the logistics CIC Agent. Its role is to
store a complete list of suppliers and products that they sell. Obviously, the logistics CIC uses the original product ontology ([2]), extended by the logistics ontology. When the system is initialized, each WhA registers with the logistics CIC and provides it with a list of products for sale.
What was described thus far is summarized in an UML use case diagram presented in Figure 1.
Figure 1. Use Case of the logistics subsystem
5. Typical Product Restocking Process
Let us now describe the processes involved in restocking the warehouse. Here, we skip the description of system initialization, and start with the Shop Decision Agent sending a forecast to the Warehouse Agent. The sequence of actions resulting form such a forecast is depicted, as a UML sequence diagram in Figure 2.
The SDA communicates the forecast to the WA by sending a FIPA Inform message containing the PredictionDescription (which contains all necessary data such as: product ID, amount of predicted sales, standard deviation of sales, expected purchase price, period for which this forecast is valid, etc.). We assume that the SDA forecasts are of the type: until a new fore- cast, weekly sales are expected to be 45 items of a given product. Forecasts can be issued at specific times (e.g. once a week or once a month) and their frequency de- pends on the information found in data analyzed by the SDA to derive forecast(s).
The WA starts by examining current stock of a given product, and if current supplies are su?cient, it sets up to check their levels at the end of the time unit specified in the forecast (i.e. forecasts specified on weekly basis are checked once a
week). If stocks are insu?cient, the WA utilizes the FIPA Request Protocol (FIPA specification SC00026) and FIPA SL language [4] (used in all agent interactions), to com- municate with the Logistics Agent. The initial message from the WA is the FIPA Request message sent to the LA and it contains OrderRequest action with the Or- derRequestDescription. The OrderRequestDescription contains the necessary information specifying the order to be made: product ID, preferred delivery time, amount and maximum price. Delivery time and amount are computed based on the current product level, predicted delivery time and an overall inventory strategy.
Upon receiving the request, the LA dispatches a query to the logistics CIC to obtain a list of suppliers of a given product. Ensuing conversation conforms to the FIPA Query Protocol, starting with the FIPA Query-Ref message containing the CICQuery action with the Product ID. The logistics CIC responds with the FIPA Inform-Ref message containing the CICResponse with a list (possibly empty) of suppliers. Empty list results in a FIPA Failure message (with OrderRequestResult set to failure) send by the LA to the WA. Similar response is sent when the logistics CIC cannot be contacted.
When the non-empty list was received, the LA removes these suppliers that have their reliability value below a certain threshold. Then the LAAgentDescriptions list is formed by supplementing each CICAgentDescription received from the logistics CIC with the reliability information. If a given WhA’s is not known a default trust value is used.
After preparing the list, the LA utilizes the FIPA Request Protocol to find a free OA. Busy agents will re- spond with FIPA Refuse messages. If all agents respond in such a way, this process may need to be repeated un- til a free OA is found and responds with the FIPA Agree message. The LA then sends the ordering request to the selected OA and awaits for the result of the ordering pro- cess. LA’s message contains the IssueOrder action with the OrderDescription and the LAAgentDescriptions.
After obtaining the request from the LA, the OA engages in the FIPA ContractNet Protocol interactions with WhAs from the list. It sends the FIPA CallForProposal message, containing CFPRequest with OrderDescription to the WhAs. WhAs evaluate the CFP and submit their o?ers by sending FIPA Propose messages containing the CFP Response action
with O?erDescription or, if terms contained in the CFP are unacceptable/not interesting, respond using the FIPA Refuse message.
Responses must arrive within a timeframe speci- fied by the OA, after which the OA proceeds to evalu- ate them. First, it filters unacceptable o?ers. Note that it is possible that
some WhAs may respond knowingly with proposals that violate some of the conditions
and in special circumstances when no better o?ers were found the OA may need to accept such o?ers. In the next step, o?ers are ranked and the winner is determined. Winner is
sent a FIPA AcceptProposal message containing the ConfirmationRequest action with its offer quoted. The winning WhA must in turn reply with the FIPA Inform message containing Confirmation Response action with the OrderConfirmation which has unique orderID generated by the supplier. This successfully completes the ordering process. The winner can also withdraw the o?er by sending a FIPA Failure message. In this case,
runner-ups are contacted in an iterative manner. In case when there are no more o?ers left or there were no o?ers to begin with, the OA sends a FIPA Failure message to the LA,
which, in turn, forwards it to the WA. When the winner confirms the order, the OA sends to the LA a FIPA Inform message containing the InformResult action with the WhA-received Or der Confirmation, thus completing the protocol. At this time the LA sends information to the WA, inside a message of the FIPA Agree type. This performative is used in compliance with the protocol to indicate that the LA is performing the desired task (ordering), but its e?orts do not guarantee success (ordering success order success), and thus sending the final response (FIPA Inform) is inappropriate at this stage. Meanwhile the OA returns to the pool of available Ordering Agents.
Now the purchase enters the delivery monitoring stage. Here, the LA waits for the delivery from the WhA to be registered with the WA. When a delivery arrives the WA sends (to the LA) a plain FIPA Inform message containing the WADelivery action with the DeliveryDescription, which has supplier’s AgentID and the already mentioned orderId. The LA does not need to respond to this message, but it checks the messages to see if it is currently awaiting a delivery with the given orderId coming from a supplier AgentID. If it finds a match, the ordering process is completed. As a result, the reliability value of supplied AgentID is increased. If a delivery notification does not come within time agreed in the OrderConfirm, actions must be undertaken (recall, that receiving supplies is vital to the e-Shop as its warehouse is likely to run out of stock Those actions are: (1) retry the ordering (sending reminder to the WhA / choosing new WhA), if there is still time, and (2) marking that a retry has been made.
If there is still time before the deadline (established by the WA), then order can be retried. If it is the first time an attempt to retry the order is made, a reminder is sent to the
WhA. To this end, LA contacts a free OA with a FIPA Request Protocol message with a Reminder action containing AgentID of the WhA and the orderId. The OA accepts the job (the FIPA Agree) and contacts the WhA (also using FIPA Request Protocol), sending it the exact same action. The WhA is expected to reply within a timeframe using either a FIPA Failure (o?er is withdrawn) or a FIPA Inform providing new Order Confirmation with a new delivery time, which is forwarded to the LA unchanged. In the case of an agreement, the LA returns to awaiting delivery, in the case of failure, the LA removes this WhA from the LA Agent Description list and locates an OA to perform entirely new search for a supplier. New search is also ordered if a reminder to the supplier whose delivery we were waiting resulted in a failure. The monitoring stage ends when: (1) delivery is received, or (2) reminder to the supplier was made, but it was refused, while deadline has already passed. Note that we assume that the actual order failure occurs only when the delivery deadline has passed and the reminder failed.This is because it is possible that there is an order delay and goods may arrive late. This information can be obtained from the WhA, and thus the need for the reminder.
Figure 2. Restocking process: sequence diagram
When the monitoring stage ends, the WA is notified about the result by the FIPA Inform or the FIPA Failure message to complete the FIPA Request protocol. The message will contain the OrderRequest action with the OrderRequestResult set appropriately. Furthermore, at this stage the reliability bonuses and/or penalties are calculated and applied. Finally, in the case of a successful order, the WA sends to the SDA a FIPA In- form message containing status information about the re-stocking of the warehouse.
6. Concluding remarks
In this note we have discussed the way in which the logistics subsystem is being introduced into our model agent-based e-commerce system. We have presented used UML’s use case and sequence diagrams to formally depict and discuss the most important features of our approach. Due to the lack of space, we have focused our attention on agents and their interactions. The proposed system has been implemented and is in the final testing phase.
References
[1]AgentisSoftware,2007.https://www.doczj.com/doc/2983436.html,/
[2] C. Bádicá, A. Báditá, M. Ganzha, M. Paprzycki, Developing a Model Agent-based E-commerce System. In: Jie Lu et. al. (eds.) E-Service Intelligence—Methodologies, Technologies and Applications, Springer, Berlin, 2007,555–578
[3] C. A. Butler and J. T. Eanes. Software agent technology for large scale, real-time logistics decision support. US Army Research Report ADA392670, 2001. 23 pages.
[4] FIPA: Foundation for Physical Agents.https://www.doczj.com/doc/2983436.html,
[5] JADE: Java Agent Development Framework.http：//jade.cselt.it
[6]W.Ying and S.Dayong. Multi-agent framework for third party logistics in e-commercestar. Expert Systems with Applications, 29(2):431–436, August 2005.
International Conference on Intelligent Agent Technalogy,2007, 78(7):
294-298引入第三方物流企业的以代理人为基础的电子商务系统
模型
Tomasz Serzysko
数学信息科学部门，波兰华沙工业大学
Maria Ganzha,Maciej Gawinecki,Pawel Kobzdej,Marcin Paprzycki
波兰研究院
Costin Badica
罗马尼亚大学计算机科学部门
摘要
在我们的模型以代理人为基础的电子商务系统[2]的基础上,我们假定一定数量的物品的某一特定的产品是适合出售的。摘要本文我们介绍了模型的物流系统并且讨论了如何将它与系统结合。
关键词:电子商务；物流系统；代理
1.简要
目前,我们正在开发和实施模式以代理人为基础的电子商务系统(参见[2]的基础上,参考文献收藏)。在该系统中当网上商店试图获得产品销售利润最大化的时候，多买方代理试图通过在电子商店参与价格谈判做出购买价格并且选择最好的供货者。到目前为止我们的注意力都集中在计划之间的相互作用。假定, 我们忽略产品来自哪里，产品在仓库里。这个工作的目的是描述我们的系统可以扩展到包括产品进货过程。
之前,让我们做一些评论。首先,本文提出的物流系统不是―独立‖的（例如与【3,1,6】中提到的相似）相反，它已经建立在我们的电子商务系统中，对它的设计有直接影响。第二，虽然有些类似，电子商店补库存的过程与顾客在电子商店购买物品的过程是不一样的。（例如在产品的需求预测,与批发商的交流,涉及更多条件的议价方法,提供更多的选择标准等等）因此我们已经创造了一个单独的物流系统而不是模仿功能的重用（例如议价）第三，这个笔记是专注于代理结构和代理交互的，由于
空间的限制，我们省略了重要的信息控制系统，比如推导预测,提供评估等等。可是，这些功能可以封装成模块,可以成为一个合适的代理的一部分。因此,读者应展望,例如,当我们写的是―报价评估‖,那么提供评价最好的方法会被利用。
继续,首先简要介绍一下我们的电子商务系统。假设我们同意假定物流系统以及代理人的介绍。最后,我们给出了进货的序列图,并且用它来详细讨论这个过程在我们的系统如何作用的。
2.系统描述
本系统是在分布式市场中进行商务功能的软件代理(有关详情参见[2],用例图)。使用者客户端是为客户代理(CA)表现的。客户代理是有自主性的、当采购订单是由使用者客户端沟通,它一直工作直到购买完成或者被放弃。客户代理与顾客信息中心交流(CIC),顾客信息中心能介绍哪个电子商店销售哪种商品(黄页配对)。每个商店出售被要求的产品,客户代理代表买房代理进行价格谈判,如果成功的话,就可能购买(成功的价格谈判导致产品预订一个特定的时间,之后,并没有被购买的产品都会再出售）。因为代表同一个客户代理许多买方代理能够赢得价格谈判，客户代理作出决定是否有足够好的供应商来购买产品。只有在GA允许的情况下，买方代理才可以参与价格谈判（如果买房代理是可信的，例如赢得了价格谈判，但是不买将会禁止后续的谈判）。客户代理代表一个给定的电子商店,是由商店代理人创造的(SA)。SA是中央管理和销售过程,它运用GA,建立一系列的卖方代理(SeA),价格谈判的买房代理与负责库存和管理的仓库代理。迄今为止,仓库代理是负责管理产品预订及库存。确切地说,(1)在一个新的价格谈判之前WA―保留‖一个给定的商品-所以如果谈判最后成功，还有一件产品可以卖掉;(2)当购买时保留结束,调整产品数量;(3)当保留过期，也调整产品数量,可是仓库代理总是被设想为商店和供应商之间的门户,这也是本说明的重点。
3. 物流系统背后的假定
现在让我们指定强调一下物流系统设计的假设。
1 如今，除了最大的连锁店（例如乐购，沃尔玛），公司将业务外包运输活动（被认为是非核心业务活动）交给专家（例如UPS，五月花）。然而，我们假定，供应商负责推动交通。因此，我们忽略交通相关流程，并且把焦点集中在电子商店和供应商的互动上。
2 作为1的结果、运输费用被认为是由供应商支付的,包括直接在产品的价格(比如运费折扣,表现它自己在总价)。
3 而在原系统中，采用基于拍卖价格谈判,我们选择了这个简单的FIPA合同网协议[4]。因此,在整个物流系统中,包含提议和评价反应的一轮谈判被使用。
4.新功能和代理
为了完成物流相关工作,介绍了几种新的角色,他们中的一些已经被委托给系统存在的代理人,而另一些角色则授权添加新的代理人。特别地:
需求估计--从销售数据绘制资讯及/或外部场所来预测未来的产品销售,
仓库监控--观察供应水平和遇到降低到不能满足预计需求的风险时的反映
订单管理--协调订货,来协助评估产品报价,
订货--联系供应商订货并且选择最佳的提议,
销售货物--在系统中，供应商也被效仿了;当商品没有广泛的价格谈判而被收购时，―有人‖已经提供建议订购部件,
物流黄页簿--―物流CIC‖ 的这个角色是非常相似于原CIC([2]);它具有潜在的供应商提供的产品清单;显然,一个店成为另外一个店的供应者是可能的,表明这两个CIC 是可以联合在一起的,但我们决定要明确区分顾客的一方和供应者一方。要做这种区分的另外一种原因是,有些商店,不得有兴趣成为批发商,我们就得改变原CIC数据结构(如批发商-是/否)。最后,由于物流系统并不需要拍卖，分离是更被批准的。
现在让我们来看看这些任务/角色是如何放置在我们的系统的。需求估计角色归于购物决策代理人(SDA), 在商店里负责―知识管理‖功能(如信托管理、销售趋势数据挖掘等)。
仓库监控角色已是仓库代理的一部分。不同的是,现在仓库代理成为一个积极的供应管理者;通过购物决策代理人作用于预测供应。
完成订单管理角色要求引进作为物流系统中心的物流代理(LA)。负责联系物流CIC，获得潜在的供应商和管理的代理商的产品清单，以便从―批发商‖订购货物。‖最后,它收集和管理与供应商的可靠性相关的数据。这个数据,反过来,将成为选择供应商的其中一个因素
这个商品订购过程是由一种新型的代理-订购代理(OA)促成的。它的任务是要提出建议,收集反应和考虑一些因素如:价格、交货时间、可靠等来选择最佳的提议,。让我们回想一下,由于模块化的代理设计([2]),我们的系统有足够的灵活性,以便任何提供选择的方法可以应用(它可以在一个模块也可以插入到OA)。
销售商品的角色是用非常简单的批发代理(WhA)来实现。它的作用是为了回应来自订购代理的CFP。目前我们认为批发代理通过某种方式收到指示产生CFP的一些反
应。
最后,物流黄页是由物流CIC代理促成的。它的主要作用是储存的一张完全的供应商表,以及他们出售的产品表。很明显,物流CIC利用原有的产品本体([2])、延长物流本体。当系统初始化后,每个批发代理登记物流CIC，并且提供这一系列的产品销售。
迄今为止所描述的是总结了在图1中的UML用例图。
5. 典型的产品进货过程
现在让我们来描述涉及到进货的过程,我们跳过描述系统的初始化,并从商店决策代理人到仓库的代理发送一个预测开始。行为的结果是产生预测的序列, 在图2中作为一种统一建模语言(UML)的顺序图。
图1.物理系统的用例图
商店的决策代理人通过发送包含预测描述的FIPA通知消息传达预测到仓库的代理人(包含所有必要的资料,如:产品ID,预测销售数量,标准偏差的销售,期望购买价格,这一预测的有效期等)。我们认为这个商店的决策代理人的预测是这种类型，在这种预测下，每周销售预计将45件产品。预测在特定的时间可以发布(例如每周或每月一次)，他们的频率取决于通过商店决策代理人的数据分析中得到的信息来获得预测。
通过检查仓库的经纪人开始流通股票的产品,如果现在的供应是充足的,它建立指定的单位来检查他们的在某一时间最后的水平(例如预测每周指定检查一周一次)。如果股票的不足、仓库代理要求采用FIPA请求协议（FIPA SC00026规格）和FIPA SL 语言[4](适用于所有的代理人交流)来和物流代理交流。最初的来自批发商代理的消息是发送到物流代理的FIPA请求消息，它包含带有订购要求的描述订购要求行动。订购要求描述包含来做订购的必要的信息：产品ID,首选的交货时间、数量和最大的价格。基于当前的产品水平和预测的交货时间和全面的库存策略，交货时间、数量被确
定下来。
在收到请求后,物流代理询问物流顾客信息中心以获得包含某一特定的产品供应商的一个列表。随后谈话符合FIPA查询协议,则开始通过包含用产品ID从顾客中心信息查询的FIPA查询信息。物流CIC响应包含有供应商列表（可能为空）的CIC回应的FIPA通知消息。空的名单导致一个从物流代理发送到批发商代理的FIPA失败的消息(存在失败的订购要求结果集)。当物流CIC无法联络的时候，类似的反应被发送。
当收到空清单,物流代理移除可靠性的价值低于某个阈值的这些供应商,然后LA 代理描述列表是由从有可靠性信息的物流顾客信息中心得到的顾客信息中心代理描述来补充组成的。如果给定的批发商是不被知的，那么一个预设的信赖被使用。
在准备这份清单后,物流代理利用FIPA请求协议找到一个免费的订购代理。繁忙的代理人会报以FIPA拒绝的讯息。如果所有的代理反应都是这样一种方式,这个过程可能需要重复,直到找到一个免费的订购代理和响应FIPA同意的信息。物流代理服务器发送订购要求到选定的订购代理并且会等待着结果.物流代理信息包括有订单描述的问题订单处理以及物流系统代理描述。
从物流系统得到请求后,订购代理参加FIPA合同网协议同在列表中的批发商代理们交流。订购代理发送FIPA需要建议讯息,包括请求批发商代理的订购描述。批发商代理们评价CFP并且通过发送包含提供描述的CFP回应的FIPA协议信息来发送他们能提供的。或者CFP中的信息是不可接受的或者是不让人感兴趣的，那么会通过FIPA 的拒绝信息来回应。
回复必须通过订购代理在一个时间段内到达,然后订购代理收益来评价他们。首先,它接受报价。注意,在没有更好的供应商时，订购代理可能需要接受这钟供应商，这些批发代理提供建议时故意违反一些条件和特殊情况。在下一步,供应商被排名并且成功者被确定。成功者会发送一个包含提供报价的确认要求的FIPA接受协议信息。。获胜的批发商必须依次答复含有订购确认通知的的FIPA通知消息。这个订购确认通知具有从供应商获得的特殊的订购ID。这个成功完成订购过程。胜利者也可以通过发送FIPA失败消息撤回要约。在这种情况下,执行者进行迭代的联系。万一没有更多的供应商存在，或者没有供应商开始,订购代理会发送一个FIPA失败的消息给物流代理。并且按顺序发送给仓库代理。当获胜者确认订单,订购代理发送一个含有批发商接受订单确认的通知结果的FIPA通知消息。从而完成了协议。这时,物流代理发送信息给仓库代理,在信息里面有FIPA允许类型的消息。这种行为是用于符合协议表明物流代理执行预期的任务(订购),但是它并不能保证成功(订购成功订购成功),因此发最后的回应（FIPA通知）在这个阶段是不合适。同时订购代理返回可用订购代理池。
现在购买进入发货监测的阶段。在这里,物流代理等待来自批发代理的发货来登记仓库代理。当一个送货到达时，仓库代理发送(到物流代理)一个包含带有发货描述
的仓库代理发送行动的FIPA通知协议。这个发货描述具有供应商代理人ID和先前提到的订单的ID。物流代理不需要对这个消息进行回复,但是它会检查这个讯息,看看它是否正在等待一个来自供应商代理ID与交货的订单Id。若发现相符,订购过程完成。作为结果,供应商代理ID的可靠性增加。如果一个交货通知不能在订购确认允许的时间内到达，,必须进行召回行动，接收供应对于电子商店是至关重要的，作为仓库可能缺货。这些行为是:(1)再订货(发送提示批发代理或者选择新批发代理),如果还有时间,可以(2)标志一个重试。
如果截止日期（由仓库代理确定）前还有时间，然后订单能再审。如果这是第一次尝试再订购,警告被发送到批发商代理。为了达到这一目的,物流代理通过包含一个有批发商代理的代理ID和订购ID的提醒的FIPA请求协议消息来与一个免费的订购代理接触。订购代理接受这份工作(FIPA许可)并且（同样使用FIPA请求协议）和批发商代理接触，发送精确同样的行动。批发商代理被期望在预计的时间框架内答复,或者使用一个FIPA失败（报价撤销）或者通过一个FIPA通知提供有新的发货时间的新的订购确认，这个新的发货时间不变的发给物流系统。在达成协议,物流代理重新返回等待发货，如果失败,物流代理从物流代理描述列表上消除这个批发商代理，并且为一个订购代理定位来进行全新的寻找供应商。新的搜索还命令如果我们收到正在等候的供应商交货导致失败的一个提醒,，监控期结束，当(1):交货,或(2)提示付款给供应商,但它是被拒绝,而期限已经过了。注意,我们假定, 只有当交货期限已过或提醒失败时，实际的订购失败,这是因为它有可能出现订购延迟和货物到达时间推迟的情况。这些信息可从批发商处得来,因而需要提醒。
图2进货过程：顺序图表
当监控期结束后,仓库代理人通过FIPA通知或FIPA失败信息完成FIPA请求协议来告知结果。这个消息将包含含有合适地命令-要求-结果集合的命令-要求。另外,在这一阶段，可靠性和/或处罚的奖金被计算及应用。最后,在一个成功的命令,仓库代理人把FIPA通知消息发送给商店决策代理人,其中包含了仓库的进货状态信息。
6.结论
本文探讨了,物流系统以何种方式被引入以代理人为基础的电子商务系统模型。我们将采用统一建模语言(UML)的使用情况和时间图来正式地描述和讨论了我们方法的最重要的特征。由于缺乏空间,我们把注意力集中在代理人及其交互作用上。该系统已经实施,并且在最后的测试阶段。
参考文献
[1]代理软件,2007.https://www.doczj.com/doc/2983436.html,
[2] C. Bádicá, A. Báditá, M. Ganzha, M. Paprzycki. 以代理人为基础电子商务系统的开发模式. 卢杰(音) 等编著的电子服务的智能方法,技术及应用,柏林,施普林格, 2007,555–578
[3] C. A. Butler and J. T. Eanes. 大规模的软件代理技术, 实时物流决策支持. 美国军队研究报告ADA392670, 2001:23 页.
[4] FIPA: 依据物理代理.https://www.doczj.com/doc/2983436.html,
[5] JADE: Java.代理开发框架.http：//jade.cselt.it
[6]W.Ying and S.Dayong. 第三方物流的电子商务的多智能体框架.专家系统与应
用.29(2):431–436, 2005.8.
Statistical hypothesis testing Adriana Albu，Loredana Ungureanu Politehnica University Timisoara,adrianaa@aut.utt.ro Politehnica University Timisoara,loredanau@aut.utt.ro Abstract In this article,we present a Bayesian statistical hypothesis testing inspection, testing theory and the process Mentioned hypothesis testing in the real world and the importance of, and successful test of the Notes. Key words Bayesian hypothesis testing; Bayesian inference;Test of significance Introduction A statistical hypothesis test is a method of making decisions using data, whether from a controlled experiment or an observational study (not controlled). In statistics, a result is called statistically significant if it is unlikely to have occurred by chance alone, according to a pre-determined threshold probability, the significance level. The phrase "test of significance" was coined by Ronald Fisher: "Critical tests of this kind may be called tests of significance, and when such tests are available we may discover whether a second sample is or is not significantly different from the first."[1] Hypothesis testing is sometimes called confirmatory data analysis, in contrast to exploratory data analysis. In frequency probability,these decisions are almost always made using null-hypothesis tests. These are tests that answer the question Assuming that the null hypothesis is true, what is the probability of observing a value for the test statistic that is at [] least as extreme as the value that was actually observed?) 2 More formally, they represent answers to the question, posed before undertaking an experiment,of what outcomes of the experiment would lead to rejection of the null hypothesis for a pre-specified probability of an incorrect rejection. One use of hypothesis testing is deciding whether experimental results contain enough information to cast doubt on conventional wisdom. Statistical hypothesis testing is a key technique of frequentist statistical inference. The Bayesian approach to hypothesis testing is to base rejection of the hypothesis on the posterior probability.[3][4]Other approaches to reaching a decision based on data are available via decision theory and optimal decisions. The critical region of a hypothesis test is the set of all outcomes which cause the null hypothesis to be rejected in favor of the alternative hypothesis. The critical region is usually denoted by the letter C. One-sample tests are appropriate when a sample is being compared to the population from a hypothesis. The population characteristics are known from theory or are calculated from the population.
Inventory management Inventory Control On the so-called "inventory control", many people will interpret it as a "storage management", which is actually a big distortion. The traditional narrow view, mainly for warehouse inventory control of materials for inventory, data processing, storage, distribution, etc., through the implementation of anti-corrosion, temperature and humidity control means, to make the custody of the physical inventory to maintain optimum purposes. This is just a form of inventory control, or can be defined as the physical inventory control. How, then, from a broad perspective to understand inventory control? Inventory control should be related to the company's financial and operational objectives, in particular operating cash flow by optimizing the entire demand and supply chain management processes (DSCM), a reasonable set of ERP control strategy, and supported by appropriate information processing tools, tools to achieved in ensuring the timely delivery of the premise, as far as possible to reduce inventory levels, reducing inventory and obsolescence, the risk of devaluation. In this sense, the physical inventory control to achieve financial goals is just a means to control the entire inventory or just a necessary part; from the perspective of organizational functions, physical inventory control, warehouse management is mainly the responsibility of The broad inventory control is the demand and supply chain management, and the whole company's responsibility. Why until now many people's understanding of inventory control, limited physical inventory control? The following two reasons can not be ignored: First, our enterprises do not attach importance to inventory control. Especially those who benefit relatively good business, as long as there is money on the few people to consider the problem of inventory turnover. Inventory control is simply interpreted as warehouse management, unless the time to spend money, it may have been to see the inventory problem, and see the results are often very simple procurement to buy more, or did not do warehouse departments . Second, ERP misleading. Invoicing software is simple audacity to call it ERP, companies on their so-called ERP can reduce the number of inventory, inventory control, seems to rely on their small software can get. Even as SAP, BAAN ERP world, the field of
吉林化工学院理学院 毕业论文外文翻译English Title（Times New Roman ，三号） 学生学号：08810219 学生姓名：袁庚文 专业班级：信息与计算科学0802 指导教师：赵瑛 职称副教授 起止日期：2012.2.27～2012.3.14 吉林化工学院 Jilin Institute of Chemical Technology
1 外文翻译的基本内容 应选择与本课题密切相关的外文文献（学术期刊网上的），译成中文，与原文装订在一起并独立成册。在毕业答辩前，同论文一起上交。译文字数不应少于3000个汉字。 2 书写规范 2.1 外文翻译的正文格式 正文版心设置为：上边距：3.5厘米，下边距：2.5厘米，左边距：3.5厘米，右边距：2厘米，页眉：2.5厘米，页脚：2厘米。 中文部分正文选用模板中的样式所定义的“正文”，每段落首行缩进2字；或者手动设置成每段落首行缩进2字，字体：宋体，字号：小四，行距：多倍行距1.3，间距：前段、后段均为0行。 这部分工作模板中已经自动设置为缺省值。 2.2标题格式 特别注意：各级标题的具体形式可参照外文原文确定。 1．第一级标题（如：第1章绪论）选用模板中的样式所定义的“标题1”，居左；或者手动设置成字体：黑体，居左，字号：三号，1.5倍行距，段后11磅，段前为11磅。 2．第二级标题（如：1.2 摘要与关键词）选用模板中的样式所定义的“标题2”，居左；或者手动设置成字体：黑体，居左，字号：四号，1.5倍行距，段后为0，段前0.5行。 3．第三级标题（如：1.2.1 摘要）选用模板中的样式所定义的“标题3”，居左；或者手动设置成字体：黑体，居左，字号：小四，1.5倍行距，段后为0，段前0.5行。 标题和后面文字之间空一格（半角）。 3 图表及公式等的格式说明 图表、公式、参考文献等的格式详见《吉林化工学院本科学生毕业设计说明书（论文）撰写规范及标准模版》中相关的说明。
Advantages of Managed Code Microsoft intermediate language shares with Java byte code the idea that it is a low-level language witha simple syntax , which can be very quickly translated intonative machine code. Having this well-defined universal syntax for code has significant advantages. Platform independence First, it means that the same file containing byte code instructions can be placed on any platform; atruntime the final stage of compilation can then be easily accomplished so that the code will run on thatparticular platform. In other words, by compiling to IL we obtain platform independence for .NET, inmuch the same way as compiling to Java byte code gives Java platform independence. Performance improvement IL is actually a bit more ambitious than Java bytecode. IL is always Just-In-Time compiled (known as JIT), whereas Java byte code was ofteninterpreted. One of the disadvantages of Java was that, on execution, the process of translating from Javabyte code to native executable resulted in a loss of performance. Instead of compiling the entire application in one go (which could lead to a slow start-up time), the JITcompiler simply compiles each portion of code as it is called (just-in-time). When code has been compiled.once, the resultant native executable is stored until the application exits, so that it does not need to berecompiled the next time that portion of code is run. Microsoft argues that this process is more efficientthan compiling the entire application code at the start, because of the likelihood that large portions of anyapplication code will not actually be executed in any given run. Using the JIT compiler, such code willnever be compiled.
软件专业毕业论文外文文献中英文翻译 Object landscapes and lifetimes Tech nically, OOP is just about abstract data typing, in herita nee, and polymorphism, but other issues can be at least as importa nt. The rema in der of this sect ion will cover these issues. One of the most importa nt factors is the way objects are created and destroyed. Where is the data for an object and how is the lifetime of the object con trolled? There are differe nt philosophies at work here. C++ takes the approach that con trol of efficie ncy is the most importa nt issue, so it gives the programmer a choice. For maximum run-time speed, the storage and lifetime can be determined while the program is being written, by placing the objects on the stack (these are sometimes called automatic or scoped variables) or in the static storage area. This places a priority on the speed of storage allocatio n and release, and con trol of these can be very valuable in some situati ons. However, you sacrifice flexibility because you must know the exact qua ntity, lifetime, and type of objects while you're writing the program. If you are trying to solve a more general problem such as computer-aided desig n, warehouse man ageme nt, or air-traffic con trol, this is too restrictive. The sec ond approach is to create objects dyn amically in a pool of memory called the heap. In this approach, you don't know un til run-time how many objects you n eed, what their lifetime is, or what their exact type is. Those are determined at the spur of the moment while the program is runnin g. If you n eed a new object, you simply make it on the heap at the point that you n eed it. Because the storage is man aged dyn amically, at run-time, the amount of time required to allocate storage on the heap is sig ni fica ntly Ion ger tha n the time to create storage on the stack. (Creat ing storage on the stack is ofte n a si ngle assembly in structio n to move the stack poin ter dow n, and ano ther to move it back up.) The dyn amic approach makes the gen erally logical assumpti on that objects tend to be complicated, so the extra overhead of finding storage and releas ing that storage will not have an importa nt impact on the creati on of an object .In additi on, the greater flexibility is esse ntial to solve the gen eral program ming problem. Java uses the sec ond approach, exclusive". Every time you want to create an object, you use the new keyword to build a dyn amic in sta nee of that object. There's ano ther issue, however, and that's the lifetime of an object. With Ian guages that allow objects to be created on the stack, the compiler determines how long the object lasts and can automatically destroy it. However, if you create it on the heap the compiler has no kno wledge of its lifetime. In a Ianguage like C++, you must determine programmatically when to destroy the
英文翻译 英语原文： . Introducing Classes The only remaining feature we need to understand before solving our bookstore problem is how to write a data structure to represent our transaction data. In C++ we define our own data structure by defining a class. The class mechanism is one of the most important features in C++. In fact, a primary focus of the design of C++ is to make it possible to define class types that behave as naturally as the built-in types themselves. The library types that we've seen already, such as istream and ostream, are all defined as classesthat is,they are not strictly speaking part of the language. Complete understanding of the class mechanism requires mastering a lot of information. Fortunately, it is possible to use a class that someone else has written without knowing how to define a class ourselves. In this section, we'll describe a simple class that we canuse in solving our bookstore problem. We'll implement this class in the subsequent chapters as we learn more about types,expressions, statements, and functionsall of which are used in defining classes. To use a class we need to know three things: What is its name? Where is it defined? What operations does it support? For our bookstore problem, we'll assume that the class is named Sales_item and that it is defined in a header named Sales_item.h. The Sales_item Class The purpose of the Sales_item class is to store an ISBN and keep track of the number of copies sold, the revenue, and average sales price for that book. How these data are stored or computed is not our concern. To use a class, we need not know anything about how it is implemented. Instead, what we need to know is what operations the class provides. As we've seen, when we use library facilities such as IO, we must include the associated headers. Similarly, for our own classes, we must make the definitions associated with the class available to the compiler. We do so in much the same way. Typically, we put the class definition into a file. Any program that wants to use our class must include that file. Conventionally, class types are stored in a file with a name that, like the name of a program source file, has two parts: a file name and a file suffix. Usually the file name is the same as the class defined in the header. The suffix usually is .h, but some programmers use .H, .hpp, or .hxx. Compilers usually aren't picky about header file names, but IDEs sometimes are. We'll assume that our class is defined in a file named Sales_item.h. Operations on Sales_item Objects
本科生毕业设计（论文）外文科技文献译文 译文题目(外文题目)学院(系)Socket网络编程的设计与实现A Design and Implementation of Active Network Socket Programming 机械与能源工程学院 专学业 号 机械设计制造及其自动化 071895 学生姓名李杰林 日期2012年5月27日指导教师签名日期
摘要：编程节点和活跃网络的概念将可编程性引入到通信网络中，并且代码和数据可以在发送过程中进行修改。最近，多个研究小组已经设计和实现了自己的设计平台。每个设计都有其自己的优点和缺点，但是在不同平台之间都存在着互操作性问题。因此，我们引入一个类似网络socket编程的概念。我们建立一组针对应用程序进行编程的简单接口，这组被称为活跃网络Socket编程（ANSP）的接口，将在所有执行环境下工作。因此，ANSP 提供一个类似于“一次性编写，无限制运行”的开放编程模型，它可以工作在所有的可执行环境下。它解决了活跃网络中的异构性，当应用程序需要访问异构网络内的所有地区，在临界点部署特殊服务或监视整个网络的性能时显得相当重要。我们的方案是在现有的环境中，所有应用程序可以很容易地安装上一个薄薄的透明层而不是引入一个新的平台。 关键词：活跃网络；应用程序编程接口；活跃网络socket编程
1 导言 1990年，为了在互联网上引入新的网络协议，克拉克和藤农豪斯[1]提出了一种新的设 计框架。自公布这一标志性文件，活跃网络设计框架[2，3，10]已经慢慢在20世纪90 年代末成形。活跃网络允许程序代码和数据可以同时在互联网上提供积极的网络范式，此外，他们可以在传送到目的地的过程中得到执行和修改。ABone作为一个全球性的骨干网络，开 始进行活跃网络实验。除执行平台的不成熟，商业上活跃网络在互联网上的部署也成为主要障碍。例如，一个供应商可能不乐意让网络路由器运行一些可能影响其预期路由性能的未知程序，。因此，作为替代提出了允许活跃网络在互联网上运作的概念，如欧洲研究课题组提出的应用层活跃网络（ALAN）项目[4]。 在ALAN项目中，活跃服务器系统位于网络的不同地址，并且这些应用程序都可以运行在活跃系统的网络应用层上。另一个潜在的方法是网络服务提供商提供更优质的活跃网络服务类。这个服务类应该提供最优质的服务质量（QOS），并允许路由器对计算机的访问。通过这种方法，网络服务提供商可以创建一个新的收入来源。 对活跃网络的研究已取得稳步进展。由于活跃网络在互联网上推出了可编程性，相应 地应建立供应用程序工作的可执行平台。这些操作系统平台执行环境（EES），其中一些已 被创建，例如，活跃信号协议（ASP）[12]和活跃网络传输系统（ANTS）[11]。因此，不 同的应用程序可以实现对活跃网络概念的测试。 在这些EES 环境下，已经开展了一系列验证活跃网络概念的实验，例如，移动网络[5]，网页代理[6]，多播路由器[7]。活跃网络引进了很多在网络上兼有灵活性和可扩展性的方案。几个研究小组已经提出了各种可通过路由器进行网络计算的可执行环境。他们的成果和现有基础设施的潜在好处正在被评估[8，9]。不幸的是，他们很少关心互操作性问题，活跃网络由多个执行环境组成，例如，在ABone 中存在三个EES，专为一个EES编写的应用程序不能在其他平台上运行。这就出现了一种资源划分为不同运行环境的问题。此外，总是有一些关键的网络应用需要跨环境运行，如信息收集和关键点部署监测网络的服务。 在本文中，被称为活跃网络Socket编程（ANSP）的框架模型，可以在所有EES下运行。它提供了以下主要目标： ??通过单一编程接口编写应用程序。 由于ANSP提供的编程接口，使得EES的设计与ANSP 独立。这使得未来执行环境的发展和提高更加透明。
毕业论文（设计）外文翻译 题目：中国上市公司偏好股权融资：非制度性因素 系部名称：经济管理系专业班级:会计082班 学生姓名：任民学号： 200880444228 指导教师：冯银波教师职称:讲师 年月日
译文： 中国上市公司偏好股权融资：非制度性因素 国际商业管理杂志 2009.10 摘要：本文把重点集中于中国上市公司的融资活动，运用西方融资理论，从非制度性因素方面，如融资成本、企业资产类型和质量、盈利能力、行业因素、股权结构因素、财务管理水平和社会文化，分析了中国上市公司倾向于股权融资的原因，并得出结论，股权融资偏好是上市公司根据中国融资环境的一种合理的选择。最后，针对公司的股权融资偏好提出了一些简明的建议。 关键词：股权融资，非制度性因素，融资成本 一、前言 中国上市公司偏好于股权融资，根据中国证券报的数据显示，1997年上市公司在资本市场的融资金额为95.87亿美元，其中股票融资的比例是72.5％，，在1998年和1999年比例分别为72.6％和72.3％，另一方面，债券融资的比例分别是17.8％，24.9％和25.1％。在这三年，股票融资的比例，在比中国发达的资本市场中却在下跌。以美国为例，当美国企业需要的资金在资本市场上，于股权融资相比他们宁愿选择债券融资。统计数据显示，从1970年到1985年，美日企业债券融资占了境外融资的91.7％，比股权融资高很多。阎达五等发现，大约中国3/4的上市公司偏好于股权融资。许多研究的学者认为，上市公司按以下顺序进行外部融资：第一个是股票基金，第二个是可转换债券，三是短期债务，最后一个是长期负债。许多研究人员通常分析我国上市公司偏好股权是由于我们国家的经济改革所带来的制度性因素。他们认为，上市公司的融资活动违背了西方古典融资理论只是因为那些制度性原因。例如，优序融资理论认为，当企业需要资金时，他们首先应该转向内部资金（折旧和留存收益），然后再进行债权融资，最后的选择是股票融资。在这篇文章中，笔者认为，这是因为具体的金融环境激活了企业的这种偏好，并结合了非制度性因素和西方金融理论，尝试解释股权融资偏好的原因。
本科毕业设计（论文）中英文对照翻译 院（系部）电气工程与自动化 专业名称电子信息工程 年级班级 04级7班 学生姓名 指导老师
Infrared Remote Control System Abstract Red outside data correspondence the technique be currently within the scope of world drive extensive usage of a kind of wireless conjunction technique,drive numerous hardware and software platform support. Red outside the transceiver product have cost low, small scaled turn, the baud rate be quick, point to point SSL, be free from electromagnetism thousand Raos etc.characteristics, can realization information at dissimilarity of the product fast, convenience, safely exchange and transmission, at short distance wireless deliver aspect to own very obvious of advantage.Along with red outside the data deliver a technique more and more mature, the cost descend, red outside the transceiver necessarily will get at the short distance communication realm more extensive of application. The purpose that design this system is transmit cu stomer’s operation information with infrared rays for transmit media, then demodulate original signal with receive circuit. It use coding chip to modulate signal and use decoding chip to demodulate signal. The coding chip is PT2262 and decoding chip is PT2272. Both chips are made in Taiwan. Main work principle is that we provide to input the information for the PT2262 with coding keyboard. The input information was coded by PT2262 and loading to high frequent load wave whose frequent is 38 kHz, then modulate infrared transmit dioxide and radiate space outside when it attian enough power. The receive circuit receive the signal and demodulate original information. The original signal was decoded by PT2272, so as to drive some circuit to accomplish
长江大学工程技术学院 毕业设计（论文）外文翻译 外 文 题 目 Matlab Based Interactive Simulation Program for 2D Multisegment Mechanical Systems 译 文 题 目 二维多段机械系统基于Matlab 的 交互式仿真程序 系 部 化学工程系 专 业 班 级 化工60801 学 生 姓 名 李泽辉 指 导 教 师 张 铭 辅 导 教 师 张 铭 完 成 日 期 2012.4.15 顶层配置在管路等，要求设备，所有设要求，对调整使案，编是指机确保机组中资料试
外文翻译 二维多段机械系统基于Matlab 的交互式仿真程序 Henryk Josiński, Adam ?witoński, Karol J?drasiak 著；李泽辉 译 摘要：本文介绍了多段机械系统设计原则，代表的是一个模型的一部分的设计系统，然后扩展 形成的几个部分和模型算法的分类与整合的过程，以及简化步骤的过程叫多段系统。本文还介绍了设计过程的二维多段机械系统的数字模型，和使用Matlab 的软件包来实现仿真。本文还讨论测试运行了一个实验，以及几种算法的计算，实现了每个单一步骤的整合。 1 简介 科学家创造了物理模型和数学模型来表示人类在运动中的各种形式。数学模型 使创建数字模型和进行计算机仿真成为可能。模型试验，可以使人们不必真正的实 验就可以虚拟的进行力和力矩的分解。 本文研究的目的是建立一个简单的多段运动模型，以增加模型的连续性和如何 避免不连续为原则。这是创建一个人类运动模型系统的冰山一角。其使用matlab 程 序包创建的数字模型，可以仿真人类运动。 文献中关于这一主题的内容很广泛。运动的模式和力矩的分解在这些文献中都 有涉猎。动态的平面人体运动模型，提出了解决了迭代矩阵的方法。还值得一提的 是这类项目的参考书目，布鲁贝克等人提出了一个模型——人腿模型，这个以人的 物理运动为基础的平面模型仿真了人腿——一个单一的扭簧和冲击碰撞模型。人腿 模型虽然简单，但是它展示人类的步态在水平地面上的运动特征。布鲁贝克等人还 介绍，在人腿模型的双足行走的基础上，从生物力学的角度而言，符合人体步行的 特征。这个模型具有一个躯干，双腿膝盖和脚踝。它能够合理的表现出人多样的步 态风格。一个仿真人类运动的数学模型反应出了人的部分运动状态。 图1. 力的分解 2 力的分解