Diesel Misfire Fault Diagnosis Algorithm And Real-time Diagnosis System

China is a large country using internal combustion engines.By the end of 2021,the number of motor vehicles in China ranks first in the world,and the exhaust emissions produced by motor vehicles will cause certain damage to the ecological balance of nature and people’s health.In order to monitor and control automobile emissions,China has put forward increasingly strict emission regulations.In the National IV Emission Regulations of 2008,OBD(On Board Diagnostics)system was required for class I vehicles in China to monitor the operating status and corresponding emissions in real time.And the monitoring project of OBD system in the subsequent emission standards is further increased.Misfire fault is an important detection item in OBD system.Diesel engine misfire is a phenomenon that the engine cylinder cannot ignite and burn normally.It mainly due to insufficient air intake or failure of oil supply system.The misfire of diesel engine will lead to output torque decline,combustion efficiency reduction and the corresponding deterioration of emissions.If it is in the state of misfire for a long time,it will cause certain damage to the exhaust aftertreatment device,and even cause the deformation or fatigue damage of crankshaft.Therefore,in the process of engine operation,if the misfire fault can be diagnosed and eliminated in time,theexhaust emissions can be reduced and unnecessary economic losses can be avoided.By analyzing the misfire mechanism of a 6-cylinder in-line diesel engine,this paper puts forward a fault diagnosis method of complete misfire and partial misfire of diesel engine by using the speed signal or vibration signal of diesel engine combined with deep learning algorithm.The algorithm overcomes the shortcomings of traditional manual fault feature extraction and uncertain feature selection criteria.It can automatically extract fault features and accurately map the operating state parameters of the engine to misfire fault.For the complete misfire fault of diesel engine,the combination of speed signal and convolutional neural network is adopted,which can achieve higher diagnosis accuracy under wide speed and load conditions.The trained CS-CNN was transplanted into STM32 microcontroller,and combined with the corresponding acquisition program,the misfire fault diagnosis system was built by combining the collection and diagnosis.The real-time diagnosis test is carried out on a 6102 diesel engine to verify the real-time performance and accuracy of the diagnosis system.For partial misfire diagnosis of diesel engines,a convolution neural network with similar structure is constructed based on the speed signals and vibration signals of diesel engines respectively.It is found that the vibration signal with higher sampling rate has more advantages than the speed signal in the partial misfire fault diagnosis.Finally,the fault diagnosis of partial misfire of diesel engine is realized by combining the vibration signal of diesel engine with residual network.In this paper,the misfire fault of diesel engine is studied systematically,and the misfire diagnosis method combined with deep learning is proposed,and the diagnosis program is used in the real-time diagnosis system for real-time diagnosis test,which provides a theoretical reference for the research of misfire fault diagnosis of diesel engine.