Study On The Integration Of Artificial Immune System And Evidence Theory Used In Rotating Machinery Concurrent Fault Diagnosis

Rotating machinery is widely used in industry, it is composed by the rotor, the bearing system, the stator or machine casing, couplings and other components. It complete the work by rotation. Rotating machinery, as some of the key equipment in the industrial sector, and its direct impact on the health of the production. In the event of downtime, it will cause huge economic losses and even catastrophic consequences. Fault diagnosis technology is to understand the mechanical state in the process of running; to predicte its reliability; to identificate the reason for mechanical failure, location and the degree of risk; to forecast the development trend of failures, and to make maintain decisions in the special situation. Its purpose is to reduce the losses caused by failure. With the development of technology, machinery and equipment is becoming more complex, concurrent fault diagnosis has become a common problem. The complexity, level, relevance, uncertainties and other characteristics of concurrent fault lead to great difficulties in the correct diagnosis of concurrent failure. To solve this problem, this paper construct concurrent fault diagnosis model by integrating artificial immune system with evidential theory. The model is proved that it can effectively achieve diagnosis of the concurrent fault through experiments.Artificial immune system, which is inspired by biological immune system, has brought new ideas for fault diagnosis. The negative selection algorithm which is derived from the "self" and "non-self" recognition mechanism in the biological immune system, making the artificial immune system can effectively apply to the fault diagnosis of mechanical equipment and identify the working status. The load, working conditions, speed and other working conditions will not affect the dimensionless index. Non-dimensional-parameter immune detectors are build by using these advantages. These Non-dimensional-parameter immune detectors can process the vibration signal which is obtained from the sensor in the machinery and equipment for the range of dimensionless index, as the characteristics of a variety of fault, will be analyzed. Evidence theory that has advantages in uncertainty, measurement and combination make it the introduction of many scholars to concurrent fault diagnosis. The evidence theory fuse the fault information which is obtained by Non-dimensional-parameter immune detectors in the artificial immune system to achieve the final determination of fault. This is the the idea of rotating machinery fault diagnosis apply the Integration of artificial immune system and evidence theory which is proposed in this paper.This paper proposed the optimized dimensionless index by using genetic algorithm. In contrast with the original dimensionless index is only sensitive to some fault defects, the new optimized dimensionless index is the better solution. This paper extended the Evidence theory and fully considered the characteristics of the different indexs for different fault diagnosis capability and sensitivity. Evidence theory fusion fault information by using weighted means to improve the reliability and sensitivity of diagnosis. The way of the Integration of artificial immune system and evidence theory is used to rotating machinery fault diagnosis is feasible, which is proved through Experiments.