Research On The Identification Method Of Diesel Engine High-pressure Common Rail System Operation State Based On Rail Pressure Signal

The high-pressure common rail system is one of the key subsystems of high-power-density diesel engine,with complex hydraulic transmission process and strong transient flow characteristics.Due to the existence of hydraulic shock,cavitation erosion and other phenomena within the system,the probability of high pressure leakage,component wear,abnormal injection and other operating conditions in the high pressure common rail system is relatively high.Abnormal operating conditions will directly affect the operational safety of the entire diesel engine.Since the high-pressure common rail tube realizes system pressure accumulation,liquid flow buffering and pressure limitation,its rail pressure signal is directly related to its operating status.Compared with other signals,the rail pressure signal has obvious advantages in identifying the operating status of the common rail system.In the signal processing process,feature vector construction and signal classification methods are becoming increasingly mature and widely used in operating status identification.However,there are few studies on operating status identification of high-pressure common rail systems based on rail pressure signal analysis.Currently,they are mainly confined to the problems of insufficiently complete methodology system,poor pertinence of the operation state identification method,and low accuracy of the operation state identification model,etc.At the same time,problems such as the sampling frequency,sample diversity,and sample interference of the rail pressure signal are also the main factors affecting the accuracy of operating status identification of the high-pressure common rail system.Therefore,the theoretical study of common rail system operating state identification based on the strong transient characteristics of the rail pressure signal,improving the method system,and constructing a model with strong targeting and high identification rate are of great significance for the identification of the operating state of the high-pressure common rail system and the improvement of the stability of high power density diesel engines.Based on the rail pressure signal,this paper adopts the method of experimental research and theoretical analysis to analyze the rail pressure signal and injection consistency of 72 operating conditions,including 4 abnormal states(oil leakage at the inlet of common rail pipe,oil leakage at the inlet of injector,delayed injection of injector,and wear of high-pressure oil pump plunger)and their corresponding 6 degrees of occurrence under 3 operating conditions(100% load condition,75% load condition,and idling condition),to clarify the influence of different degrees of occurrence on the rail pressure and injection consistency of common rail system.Analyze the rail pressure signals and injection consistency of 72 operating conditions,and clarify the influence of different abnormalities of the common rail system on the rail pressure and injection consistency;the operating state identification method of the highpressure common rail system and the sampling frequency,sample diversity,sample interference and other issues that affect the operating state identification accuracy are studied: The most suitable research strategy for operating status identification of high-pressure common rail systems was proposed;the influence of sampling frequency on the accuracy of operating status identification was clarified,and the optimal sampling frequency was obtained;A small sample diversity prediction model for operational state recognition and an anti-interference and training set update model are constructed.The accuracy of the common rail system operation state recognition model is improved.The main research work and results are as follows:(1)Starting from the composition of high-power density diesel engine high-pressure common rail system,analyze the functions of its main components,build a test bench on this basis,determine the common rail system operating state identification test test method,collect the rail pressure signal and fuel injection under different operating conditions.The rail pressure signal and fuel injection volume under the condition were analyzed to analyze the influence of different abnormal state occurrence levels on fuel injection performance,which showed that the rail pressure signal can accurately represent the occurrence of abnormal states in the highpressure common rail system.(2)Regarding the eigenvector construction and signal classification methods in the process of rail pressure signal processing,this paper studies two time scale decomposition methods(variational modal decomposition and ensemble empirical modal decomposition),two signal separation methods(energy method and zero-crossing point slope method),four eigenvalue selection methods(energy,sample entropy,fuzzy entropy and time domain characteristics)and two state classification methods(random forests and neural network),a total of 86 types of high-pressure common rail system operating status identification method,by analyzing the advantages and disadvantages of different operating status identification strategies,the most suitable method for carrying out high-pressure common rail system operating status identification is obtained,which provides a theoretical basis for the subsequent selection of high-pressure common rail system operating status identification methods.(3)Aiming at the adaptability of rail pressure signal sampling frequency when identifying the operating status of the high-pressure common rail system.This paper builds a high-pressure common rail system operating state identification model based on Ensemble Empirical Mode Decomposition(EEMD),energy method and neural network,and analyzes the eigenvalue vector distribution and eigenvalue vector dispersion obtained at different sampling frequencies.An analysis study was conducted to explore the relationship between sampling frequency and operating status identification accuracy.The analysis results show that the operating state recognition accuracy rate increases with the decrease of sampling frequency,from 88.4% at10000 Hz to 92.8% at 5000 Hz,and reaches a maximum of 100% at 3000 Hz and 2500 Hz,and then decreases to 98% at 1000 Hz.When the sampling frequency is reduced from 10,000 Hz to3000 Hz or 2500 Hz,the operating state recognition accuracy increases by 13.1%.(4)When identifying the operating status of the high-pressure common rail system,the number of rail pressure signal samples is insufficient,resulting in a decrease in the accuracy of the calculated diversity index,resulting in a decrease in the identification accuracy.This paper proposes that by carrying out research on the chaotic characteristics and diversity of rail pressure signals,analysis and research on the chaos characteristics of Lyapunov exponents,analysis and research on the chaos characteristics of rail pressure signals in different operating states,and research on the relationship between the diversity index of rail pressure signals and the total Lyapunov exponent.A diversity prediction model based on Lyapunov index was developed.Combined with bench tests,the prediction accuracy of the above prediction model under different operating conditions was verified,and compared with the traditional diversity calculation method.The results of the two methods were The average accuracy rates are 92.336%and 81.111% respectively.After using the prediction model,the accuracy rate increased by13.830%.(5)In view of the problems encountered in the operation status identification of highpressure common rail systems,the untrained status cannot be classified and the training set update is difficult,resulting in low operating status identification accuracy.This paper analyzes the anti-interference ability and the training set update ability of the common rail system operating status identification model and the training set update ability,and proposes an antiinterference ability of the high-pressure common rail system operating status identification model based on the alpha shapes algorithm and K-means(k-means clustering algorithm)algorithm.Combined with bench tests,verified the recognition accuracy of the above classification model in different operating states.The operating state recognition model adding the anti-interference classification model achieved a classification accuracy of 90% and 83.3%for the untrained state and the subordinate state respectively.Compared with the operating state identification model without the anti-interference classification model,the above model increases the operating state identification accuracy of the high-pressure common rail system from 97.5% to 98.75%,improving the operating state identification accuracy.