Fault Diagnosis Of Hydraulic Pump Based On Multi-source Sensor Information Fusion

The hydraulic pump is an important part of industrial transmission and control.Whether it can work normally and stably is very important for the entire system,so it is very necessary to monitor the operating status and trouble.It is difficult to directly and effectively observe its operating status,and the internal failure mechanism is complicated.It is difficult to distinguish the type of failure by detecting a single physical signal.The built-in multi-source sensor system detects different types of physical parameters and uses intelligent algorithms to analyze these data.Dig deep to achieve the purpose of accurately identifying the failure mode of the hydraulic pump.This paper mainly researches the precise identification of the complex fault mode of the hydraulic pump.According to the JDL system model,a fault diagnosis system with threelevel multi-source sensor information fusion for hydraulic pump fault diagnosis is designed.The data level performs feature extraction and normalization on the data collected by the sensor;the feature level uses the PSO-BP network optimized by the adaptive adjustment method proposed in this paper to classify the faults of each sensor subnet;the decision level uses improved DS evidence The local fault diagnosis results are merged to make a decision,and finally a highly accurate joint diagnosis and recognition result is output to achieve accurate fault diagnosis of the hydraulic pump.Based on the actual failure of the plunger pump,the correlation between the data type of the sensor and the failure is studied,and the sensors of temperature,pressure and vibration signals are selected to detect the failure of the plunger pump.Further study the extraction of characteristic parameters,discuss the sensitivity of various dimensionless amplitude data to various faults of the plunger pump,select parameters that can fully reflect the operating conditions of the plunger pump,and finally serve as the input of the fault diagnosis network.For the feature layer,a PSO-BP local fault diagnosis network for the plunger pump is constructed.Aiming at the shortcomings of the basic PSO particle swarm optimization algorithm,which easily converges to local optimum,precocity,and low search accuracy at the later stage,based on the iterative update formula of the PSO algorithm,two local fault diagnosis optimization methods are proposed,which are dynamic acceleration constant and adaptive speed.The feasibility of the optimized algorithm is verified through the diagnosis of the plunger pump.The result shows that the optimized PSO-BP local diagnosis algorithm can quickly and effectively complete the diagnosis and identification of each subnet at the feature level.Aiming at the defect that single sensor local diagnosis cannot achieve accurate diagnosis,a multi-sensor combined diagnosis method is introduced.Aiming at the shortcomings of DS evidence theory,an index function-based model is proposed to modify the data source,and a new algorithm to measure the degree of mutual exclusion between evidence is proposed to avoid local diagnosis data fusion failure.Finally,it is verified through a case that the improved D-S fault data fusion algorithm has a good effect on the classification of plunger pump fault diagnosis.