Linearization potential well wall path expansion stochastic resonance weak feature extraction method

Abstract

The present invention provides a linearization potential well wall path expansion stochastic resonance weak feature extraction method. The method comprises: performing preprocessing of collected vibration signals with different fault degrees of a machinery device to allow the vibration signals to satisfy the small parameter signal input conditions in the heat insulation approximation hypothesis; taking the preprocessed signals as the input of a linearization potential well wall path expansion stochastic resonance system, and selecting the output signal to noise ratio of the stochastic resonance system to be the fitness function of a genetic algorithm, performing adaptive searching optimization of the system parameters, and realizing the optimal dynamic matching between the system and the input signals; and finally, inputting signals to systems with different faults, and taking the outputs of the path expansion stochastic resonance system with the optimal matching parameters as the optimal enhancement result to perform fault feature extraction and qualitative analysis according to the optimal dynamic system matching parameters. The linearization potential well wall path expansion stochastic resonance weak feature extraction method is able to improve the potential well walls of the two sides of a traditional bi-state stochastic resonance system so as to enhance the weak fault feature enhancement extraction capability.

Description

technical field [0001] The invention belongs to the technical field of mechanical fault diagnosis, and in particular relates to a method for extracting weak features of path expansion stochastic resonance of a linearized potential well wall. Background technique [0002] Due to the complex multi-time-varying transmission path, the mutual coupling of excitation and response of multiple vibration sources, and the interference of the external environment and many other factors, the characteristics of mechanical faults are often overwhelmed by serious background noise, which will inevitably lead to the signal noise of the collected fault vibration signals. The ratio is extremely low and the fault features are extremely weak, so it is difficult to directly extract and diagnose. The signal processing method based on the traditional noise elimination theory will inevitably cause more or less damage to the weak fault features while eliminating the noise, resulting in mutual damage, ...

AI Technical Summary

Problems solved by technology

[0003] The traditional bistable stochastic resonance system has been widely used in mechanical fault diagnosis and has shown certain advantages, but it often results in steep potential well walls on both sides when the optimal barrier height is obtained by adjusting parameters, resulting in restricted particle movement paths In the narrow potential well width range, it cannot continue to change with the change of the system input signal, causing the output signal of the stochastic resonance system to reach saturation, thus limiting the ability to enhance the extraction of weak fault features of the traditional bistable stochastic resonance system. One shortcoming may cause the traditional bistable stochastic resonance system to greatly reduce the ability to enhance the extraction of weak fault features of mechanical equipment, and cannot perform qualitative analysis on the degree of fault development

Images