Research And Design Of Hydraulic Fault Intelligent Diagnosis System For Injection Molding Machine Based On Data Driven

With the gradual improvement of industrial Internet layout and the steady implementation of intelligent manufacturing strategy in China,in the field of industrial equipment operation condition monitoring and health management,intelligent fault diagnosis technology has gradually become the research focus of intelligent application of various industrial equipment.Injection moulding machine,as the main equipment of polymer material processing,is widely used,but it is still widely used in the way of post-maintenance.Especially when the internal hydraulic system of the equipment fails,because of the strong concealment of the fault,it brings tremendous difficulties to the investigation and maintenance work,but also seriously affects the production efficiency.It is urgent to improve the intelligent level of the existing injection moulding machine.The fault diagnosis expert system can improve the efficiency of diagnosis and maintenance to a certain extent,but the realization of this diagnosis technology depends on the subjective experience of experts,and does not have self-renewal function,so the accuracy of fault diagnosis is low.Aiming at the above problems,this paper designs a data-driven intelligent diagnosis system for hydraulic faults of injection moulding machine,simulates the leakage state of hydraulic cylinder in actual injection process through non-destructive human operation,obtains fault data samples,completes fault feature extraction and fault diagnosis model training using data mining algorithm,and realizes remote intelligent fault diagnosis combined with Internet technology.Break.The research work of this paper includes the following aspects:(1)Analyzing and investigating the requirement of fault diagnosis and historical fault maintenance record of injection moulding machine.Based on the organizational structure of data-driven diagnosis system,the software model and structure design of the system are determined.(2)The object of fault diagnosis is set as leakage fault in injection cylinder with high frequency,and a fault simulation experiment platform is built.The data samples of injection moulding machine under normal,slight leakage,general leakage and serious leakage are successfully collected.(3)A method combining the process characteristics of injection moulding machine was proposed to extract different fault features between injection stage and packing stage.Four feature variables were extracted by exploratory analysis method in injection stage and five feature variables were extracted by time domain statistics and wavelet packet transform energy method in packing stage.(4)XGBoost algorithm is adopted to train the fault feature data with small sample data,and PSO algorithm is used to optimize the super-parameters.The final accuracy rate of the diagnosis model on the test set is 88.9%.The intelligent diagnosis system for hydraulic fault of injection moulding machine developed by Java Web technology and Python script can realize many functions,such as data upload,data visualization,feature extraction,remote fault intelligent diagnosis,self-renewal of diagnosis model,etc.On 12 sets of field simulation data,the system fault recognition rate is 100% and all kinds of response time are stable,which shows that the system runs stably and reliably.