An equipment fault diagnosis method base on improved D-S evidence theory

Abstract

The invention relates to an equipment fault diagnosis method based on an improved D-S evidence theory. The method comprises the following steps: step S1, acquiring vibration signals collected in the operation process of the equipment, and extracting time-domain analysis characteristic parameters and time-frequency analysis characteristic parameters; step S2, obtaining the time-domain analysis characteristic parameters and time-frequency analysis characteristic parameters; step S2, based on the pre-constructed RBF, BP and Elman neural network models, taking the extracted time-domain analysis feature parameters and the time-frequency analysis feature parameters as the inputs of the three neural network models for preliminary diagnosis; step S3, fusing the output results of the three neural network models; step S4: diagnosing the fault according to the fusion result. Compared with the prior art, the method calculates the weight correction evidence source, improves the algorithm of the combination rule on the basis of reserving the advantages of the original combination rule, and fully utilizes the high-speed operation of the computer and the complementarity of the multi-source information to improve the quality of the information through the related data and information.

Description

technical field [0001] The invention relates to a fault diagnosis method, in particular to a device fault diagnosis method based on improved D-S evidence theory. Background technique [0002] During the operation of the equipment, the vibration signal contains rich operating information of the system, so the vibration signal is often used as the main basis for equipment fault monitoring and diagnosis. Due to the complex structure of some equipment, the acquisition of its vibration signals requires multiple sensors to collect, and the amount of information is large. In addition, each sensor is different and the measurement points are different, which will lead to instability of diagnostic results. To achieve accurate judgment of equipment failures, it is necessary to use information fusion to process data transmitted from different information sources, combine these data according to certain rules, and then make comprehensive, efficient, accurate and reasonable judgments to e...

AI Technical Summary

Problems solved by technology

In terms of modifying the combination rules, most of the improved methods ("Li Bicheng, Wang Bo, Wei Jun, Qian Zengbo, Huang Yuqi. An effective evidence theory synthesis formula [J]. Data Acquisition and Processing, 2002, (01): 33-36.", "Sun Quan, Ye Xiuqing, Gu Weikang. A New Synthetic Formula Based on Evidence Theory [J]. Journal of Electronics, 2000, (08): 117-119.", "Deng Yong, Shi Wenkang .An improved combination rule of evidential reasoning[J].Journal of Shanghai Jiao Tong University, 2003,(08):1275-1278. ") are improved on the basis of Yager's formula, and a new synthetic formula is proposed, but because the original formula is removed The normalization factor in , assigns most of the support rate to the uncertain item, which increases the difficulty of decision-making, and the situation of "one-vote veto" will still appear

In terms of correcting evidence sources, some improved methods ("Du Feng, Shi Wenkang, Deng Yong. Evidence feature extraction and its application in the improvement of evidence theory [J]. Journal of Shanghai Jiao Tong University, 2004, (S1): 164-168 .", "Gao Jinqiu, Liu Jinglin. Improvement of DS Evidence Theory and Its Application in Fault Diagnosis of Aviation Alternator [J / OL]. Micro & Special Electric, 2016,44(06):37-40.", "Fei Xiang, Zhou Jian. A D-S evidence weight calculation method for dealing with conflicting evidence[J]. Computer Engineering, 2016, 42(02):142-145.") Use the similarity, conflict degree or contradiction coefficient between evidence to determine the revision The weighting factors of different evidence sources, and the closeness of evidence are processed. These methods have certain improvement effects, but they do not make good use of the correlation between evidence, and there is a certain degree of subjectivity in the determination of weight.

Images