Leakage Prediction Of Plunger Pump Based On HP Filter And ARIMA Model

The hydraulic system is one of the key systems in the equipment manufacturing industry.As the "heart" of the hydraulic system,the plunger pump has the advantages of high rated pressure,high speed,long life,large driving power,light weight per unit power,compact structure and flow rate adjustment and convenient maintenance,etc.At the same time,as the core power component of the hydraulic system,plunger pumps are widely used in fields such as broaching machines,die-casting,forging,heavy industry,mobile machinery,military equipment,mining and metallurgical machinery.During operation,temperature rise,abrasion and mixing of impurities will aggravate the contamination of the hydraulic oil of the plunger pump,causing damage to the surface of the friction pair and performance degradation.As the running time increases,the accumulated wear formed by the high-speed operation of the friction pair will gradually increase the internal leakage of the pump,which will cause decrease of the system oil pressure,and the flow supply and volumetric efficiency cannot meet the requirements.Therefore,the working performance of the plunger pump has an important influence on the hydraulic system,mastering the working characteristics of the plunger pump and the characteristics that affect the performance index of the plunger pump is an important basis for the implementation of health management of the hydraulic system.In this paper,the leakage rate is used as an indicator of the performance degradation of the axial piston pump to construct its short-term leakage prediction model.First of all,the failure mechanism of the axial piston pump and the leakage of the key friction pairs that have an important impact on performance degradation are analyzed and the reasons for the leakage as a performance degradation index are explained.Secondly,HP filter decomposition is performed on the leakage data to lay a data foundation for the application of filtered data for modeling.Finally,analyze the prediction process of the ARIMA model,analyze the leakage data as a time series based on the data processing method and model determination,perform predictive analysis on its fitting ARIMA model,further seek measures to optimize the model,use the comprehensive method of fitting the ARIMA model to the leakage data processed by HP filtering to carry out an example analysis,and at the same time compare the errors of the two sides.The results show that,by comparing the mean square error and relative error of the prediction results comprehensively,the synthesis method of prediction results of the filtered leakage data fitting model are better than those of the unfiltered leakage data fitting model.At the same time,due to the characteristics of the model,adjusting the amount of training data and the amount of prediction to the ARIMA model will also have an impact on the prediction effect.The more accurate result will resulted when the smaller the amount of prediction while keeping the amount of training data unchanged.The better result will resulted when the larger amount of training data while keeping the amount of prediction unchanged.Due to the characteristics of the HP filter itself,changing its smoothing parameters will also have an impact on the prediction results.The selection of the best smoothing parameters should not be too large or too small.Therefore,the method proposed in this paper provides a certain reference for the short-term prediction method selection of the axial piston pump leakage and the later health management.