Study On Fault Diagnose Of Electronic Control Common-Rail System Of High-Power Marine Diesel Engine Based On Built-in-test Technology

The marine diesel engine electronic common rail system is quite different from the vehicle system in terms of the composition type and operation mode,which caused that the failure diagnosis technology for vehicle electronic common rail system is difficult to be directly applied to the marine electronic common rail system.At the same time,the existing research work lacks failure reasoning and isolation technology of the electronic common rail system and it is difficult to effectively guide the maintenance.In this paper,the research on the failure diagnosis technology for high power marine diesel engine electronic common rail system based on the online test technology is applied,which has great significance for improving the reliability and maintainability of marine high power diesel engines.Based on the FMEA(Failure Model Effectiveness Analysis)method,the potential failure model of the high-power marine diesel engine electronically controlled common rail system is analyzed,and the online failure testing method is provided.The paper established a testability model based on the multi-signal flow graph technology,analyzed the “failure-testing” consistency matrix of the electronically controlled common rail system and finally gave the online diagnosis algorithm based on the consistency matrix model.The failure detection and coverage rates both can reach 100% and the isolation rate can reach 94.44%,which are able to meet the requirements of the engineering application.According to the features of the electronically controlled injector that its metrological characteristics slowly decaying with time as well as the requirements for consistent engineering practices,this paper put forward the online self-learning network prediction algorithm targeting the “group” characteristics.The research on the consistency overproof failure online testing algorithm is completed on the condition that the “group” metrological characteristics are decaying slowly based on the curve for pressure changes in the accumulator of injector during injection with the SPC principle taken into consideration.This algorithm can quickly follow the characteristic changes of the electronically controlled injector in 5s and the measured online learning accuracy in ±1%,which are able to effectively diagnose the electronically controlled injector with the metrological characteristics different among the electronically controlled injectors.Based on the PNN neural network technology,the classification model is established for the failure and normal states of the fuel limiting valve,which can identify 100% fuel limiting valve failures by the curve of pressure changes in the accumulator of the electronically controlled injector after the injection has ended.The oil pump failures can be detected by online measuring the injector injection flow rate,controlling oil flow rate and system leakage flow rate.Online testing the common rail pressure unloading process after shut down can identify the parameters of the effective flow area of the common rail system leakage,with the error within-6.24%~1.34%.In view of the common rail pressure signal processing module,this paper the mutual check diagnosis algorithm of the physical redundant sensor based on the consistency relation matrix is studied.For the common rail system researched in this paper,when the set credibility function has its threshold up to 0.7,this algorithm can reliably detect any faulty sensor with the positive or negative deviation with the rail pressure mean value over 6MPa or 4MPa.In view of the speed and phase signals,this paper emphatically analyzes the self-check testing algorithm for the speed and phase signals.For the drive modules of the electronically controlled injector and electronically controlled fuel pump inlet metering solenoid valves,this paper determined the reasonable range of the characteristic parameters according to the Monte Carlo method.The engineering practices are then tested whether the drive module has failed by online testing current waveform eigenvalue.Online diagnosis system prototype on the basis of the distributed architecture is developed.Based on the Dspace simulator hardware board and the online diagnosis system prototype,this paper sets up a hardware-in-loop simulation verification platform,based on the semi-physical simulation technology.The simulation verification results show that the developed online diagnosis system can provide correct and reasonable diagnosis conclusions,which are able to be applied to real machines.