Research On Rolling Process Fault Diagnosis Approach Based On Information Fusion

Along with the rapid development of automation technology in steel and iron field in our country, the request of reliability and security of rolling control system is getting higher and the request of product quality is becoming stricter. The fault diagnosis of strip mill has become one of the most significant research directions in the process automation field. In the actual rolling production line, the working environment of equipments is harsh, the failure mechanism is complex and diverse, and the fault characteristic parameters obtained by a single sensor are often vague, uncertain and even wrong sometimes. In order to achieve the comprehensive and accurate rolling process diagnosis, it needs to make full use of the fault information from multi-source synthetically.This thesis focuses on researching the fault diagnosis of looper system, monitoring AGC and rolling dynamic temperature control system which are important components of finishing mill. Through constructing an effective sensor network and taking advantage of the characteristic information of looper angle, rolling force and some other signals, combining with the failure characteristics of the rolling process, this thesis designs a three-stage information fusion diagnosis system. The data level is mainly to extract the failure biting process dates which could represent the fault characteristic information of rolling process and then normalized. The feature level conducts the local diagnosis by building four parallel ANFIS fusion diagnosis centers. The decision-making level fuses the local diagnosis from each ANFIS fusion diagnosis center based on the improved D-S evidence theory. Finally, it realizes the fault diagnosis.In the local diagnosis of feature level, according to the problems of the gradient down learning methods of error back propagation algorithm which is used in ANFIS is easy to fall into local minimum value in the network learning and training process, this thesis uses the method of F-R conjugate gradient to prove the property of ANFIS. At last, it verifies that the method of improved ANFIS is more effective by a numerical example.In the fusion diagnosis of decision-making level, in order to solve the problem that D-S evidence theory can’t combine highly conflicting evidences, this thesis takes the evidence sources themselves into consideration and according to the decision idea that the minority submit to the majority, a new improved D-S algorithm based on the weight of evidence is proposed by analysis of existing improved methods. The numerical example shows that the method can deal with the highly conflicting evidences efficiently, the convergence speed is faster and the diagnosis results are more desirable than other improved algorithms.