Research On Fault Diagnosis Method Of Electro-hydraulic Servo Valve Using Machine Learning

With the continuous development of science and technology,Electro-hydraulic servo valve as an important part of hydraulic servo system is widely used in iron and steel smelting,construction machinery and other industries,its performance is a direct impact on the safety of the entire hydraulic system.Once the Electro-hydraulic servo valve fails,the whole production line will stop running,bringing huge losses to the enterprise.Because the price of Electro-hydraulic servo valve itself is high and the maintenance cost is very expensive,therefore,the fault diagnosis of Electro-hydraulic servo valve has high engineering application value.From the perspective of machine learning and based on the historical fault data,this paper carries out research on the fault diagnosis of Electro-hydraulic servo valve.The main research contents are as follows:(1)In view of the complex nonlinear relationship between the failure phenomenon and the cause of the Electro-hydraulic servo valve,a fault diagnosis model based on Dragonfly optimization limit learning machine is constructed.The input weight and hidden layer threshold of limit learning machine are optimized by using Dragonfly algorithm in speed and global optimization ability.The model is applied to the fault diagnosis of Electrohydraulic servo valve.The simulation results show that the model is effective and the test speed and accuracy of the fault diagnosis of the Electro-hydraulic servo valve are improved.(2)Aiming at the disadvantage that some types of Electro-hydraulic servo valve have fewer fault samples,a transfer learning fault diagnosis model based on Markov metric is proposed.The distribution of source and target domains is bridged by instance weight,and the distance between samples is measured by using Markov distance,and the instance weight and Markov distance are updated at the same time.In view of the disadvantage that single source migration learning is easy to produce negative transfer,a new method based on Multi-source dynamic TrAdaBoost is proposed.The weight of source domain is adjusted by using multiple source data and adding a dynamic factor at the same time.The simulation results show that the algorithm can effectively improve the fault identification rate of Electro-hydraulic servo valve.(3)The design of portable servo valve detection platform is completed.Based on the in-depth analysis of the functional requirements of the detection platform,the overall design framework of the detection platform is given,and the detailed scheme of component selection and structural layout of the detection platform is completed;The software flow chart of upper computer and lower computer and the software design framework of portable servo valve detection platform are given.Finally,the effectiveness of the detection platform is proved by experiments.The above research results provide technical support for the fault diagnosis of Electrohydraulic servo valve,and it is helpful to be widely used in various military and civil Electro-hydraulic servo control systems.