基于MAS的故障诊断系统自适应模型与协作机制研究

发布时间：2018-01-25 17:09

本文关键词： 故障诊断多Agent系统(MAS) 协作机制合同网协议自适应模型　出处：《太原理工大学》2011年硕士论文　论文类型：学位论文

【摘要】：随着现代工业的迅速发展,机械设备的规模越来越庞大,自动化程度越来越高,集成化管理方式也越来越普遍,一旦设备出现问题将会带来重大的事故和巨大的财产损失。传统的采用简单仪器和人工经验的方法己不能满足现代复杂设备的故障诊断。近年来,机械故障诊断技术在国内外受到高度的重视,在机械学、通信学、计算机学和人工智能等科学的基础上迅速发展成为一门新兴学科。利用智能系统对设备进行故障诊断,及时发现故障,以保障关键设备的正常运行具有重大的意义。 MAS(多Agent系统)作为一种良好的解决分布式问题的技术,其方法和理论在很多领域都有广泛应用。针对现代机械故障诊断系统的复杂性、分布性和动态性,MAS能利用模块化的设计思想和并行的分布式处理技术,将复杂的系统分解成相对独立的子系统,而后通过各Agent的交互和协作共同完成相对复杂的任务。当然,复杂机电设备诊断的过程中会出现大量新的信息和知识,面对动态的诊断环境,MAS故障诊断系统必须能适应其周围环境,根据不同的诊断对象构建相应的模型。因此,研究自适应的诊断模型有着很重要的理论和现实意义。一个好的多Agent系统,必须有良好的协作机制的支持。多Agent间的协作问题研究是伴随着多Agent系统的出现就产生的。协作不仅能使整个系统解决问题的能力得到提高,而且增加了多Agent系统的灵活性。针对系统的分布性特征,本文选择了基于合同网的协作机制,并且在传统合同网的基础上,提出了一种改进的合同网模型PC-CNP (Postpone Commitement-Contract Net Protocol,推迟确认的合同网协议),改善了传统合同网在资源分配和通信问题上缺陷。实验证明,在基于PC-CNP的协作机制下,任务完成效率明显高于基于传统合同网的机制。本文以某大型选煤厂的复杂机电设备为背景,建立了基于MAS的故障诊断自适应模型AMFD (Adaptive Model in Fault Diagnosis)。该模型的层次结构与复杂机电设备的层次化组织特点相匹配,同时具有分布性和自适应性,能够根据诊断对象的变化,通过在拓扑结构、Agent组织、诊断流程和知识领域方面的重构,从而得到优化的诊断系统。为了规范系统的描述,本文采用BNF(巴科斯范式)形式化语言对该AMFD的结构组成等方面进行了形式化定义。最后,在JADE平台上实现了AMFD的原型系统,其中,其动态的知识管理和层次化的体系结构使增强了系统可扩展性和可伸缩性。
[Abstract]:With the rapid development of modern industry, machinery and equipment more and more large scale, more and more high degree of automation, integrated management is becoming more and more popular, once the equipment problems will lead to major accidents and huge property losses. Methods using simple instrument and artificial experience of the traditional can not meet the modern fault diagnosis of complex equipment in recent years, machinery fault diagnosis technology has been highly valued at home and abroad in mechanics, communication science, computer science and artificial intelligence based on the rapid development of science has become a new subject. Fault diagnosis of equipment based on intelligent system, timely detection of failure, to guarantee the normal operation of the key equipment is of great significance.
MAS (multi Agent system) as a good solution to the problem of distributed technology, its theories and methods are widely used in many fields. Because of the complexity of modern mechanical fault diagnosis system, distributed and dynamic, MAS can use the modular design method and parallel distributed processing technology, the complex the system is decomposed into relatively independent subsystems, and then through the interaction and collaboration of each Agent to complete complicated tasks. Of course, the emergence of a large number of new information and knowledge will be the process of complex electromechanical equipment diagnosis, diagnosis in the dynamic environment, the MAS fault diagnosis system must be able to adapt to the surrounding environment, according to the corresponding model the different diagnosis target construction. Therefore, there is an important theoretical and practical significance to study the adaptive diagnosis model.
A good multi Agent system, there must be a good mechanism for collaboration support. The research collaboration between multi Agent is accompanied by the emergence of multi Agent system is generated. The cooperation ability can not only make the whole system to solve the problem is improved, but also increase the flexibility of the multi Agent system. According to the characteristics of distribution system in this paper, the coordination mechanism based on contract net, and based on the traditional contract net, proposes an improved contract net model (PC-CNP Postpone Commitement-Contract Net Protocol contract net protocol, delayed confirmation) to improve the traditional contract net, in resource allocation and communication problems on defects. The experimental results show that based on the the coordination mechanism of PC-CNP, task completion rate was significantly higher than that of the traditional mechanism based on contract net.
Based on the background of complex electromechanical equipment of a large coal preparation plant, established the fault diagnosis model based on adaptive AMFD MAS (Adaptive Model in Fault Diagnosis). The characteristics of hierarchical organization hierarchical structure of the model and the complex electromechanical equipment matching, which has distributed and adaptive, can according to the change of the diagnosis object, through in the topology, Agent, reconstruction of diagnosis process and knowledge, in order to get the diagnosis system optimization. In order to standardize the description of the system, this paper uses BNF (Backus paradigm) formal language for the formal definition of the AMFD structure. Finally, a prototype system of AMFD based on JADE based on the knowledge management and hierarchical structure of the dynamic enhances the expansibility and scalability.

【学位授予单位】：太原理工大学
【学位级别】：硕士
【学位授予年份】：2011
【分类号】：TH165.3

【参考文献】