| Today, the industry requires the manufacturing equipments to be more automatic, optimized, flexible, integrated, intelligent, accurate and high-speed. So they must have the characters of monitoring and fault-diagnosis. So far, to enhance the system of monitoring and fault-diagnosis with real tune property and veracity, many studies have been done and many principles have been mentioned.Upon a lot of domestic and overseas literatures, we looked back the history, characters and requirements of monitoring and fault diagnosis for equipments. We summarized the traditional identification theories, and discussed some intelligent algorithms existed.Monitoring and fault diagnosis is a pattern recognition task. The natural immune system can distinguish a lot of foreign substances using a bit of antibodies, and it has the feature of the pattern recognition.This paper retrospected the studies on monitoring and fault diagnosis based on the artificial immune system. Then we improved this algorithm and executed it for numerical control machine. The real-time monitoring and fault diagnosis, as a pattern recognition process, needs multi-module in line, hi this paper, we modeled the system based on the theory of multi-agent.This paper was organized as the following:In the first chapter, as the exordium, we firstly reviewed the evolution and the feature of modern manufacturing systems. Then the importance to monitor and diagnose faults for those systems was introduced with their demands. We investigated the developments and features of the monitoring and fault-diagnosis. Moreover the traditional identification theories and intelligent algorithms existed were reviewed. In the end, the aim and primary work of this research is brought forward.In Chapter 2, we introduced the natural immune mechanism firstly. Then the artificial immune systems which are widely used today, as self/nonself and immune networks, weredescribed. And we laid out the actuality about exertion of artificial immune system in monitoring and fault diagnosis, and pointed out the direction for the following study.Chapter 3 was the core and the innovation of this paper. It illuminated the exertion of the real-time monitoring and fault diagnosis in the numerical control machine in details.In Chapter 4, the system model based on the theory of multi-agent was built for simulation experiment. 3 groups of agents, data-gathering agent, data-disposing agent and decision-making agent were brought forth with their frames, functions and key technologies to realize them.Chapter 5 expatiated on several key points in realizing experimental systems, and validated the feasibility of the theory and its model through 2 groups of experiments. Chapter 6 summarized and presented the orientation for the future.Numerical control machine is a complicated, nice system. Thus this makes the monitoring and fault diagnosis a system project, which relates to not only the veracity and the real tune of diagnosis, but also the forecast to the future. Artificial immune system was introduced to monitoring and fault diagnosis for numerical control machine in this thesis as the beginning, and a lot of work is to be done in the future. |
