Study On Fault Diagnosis Of Diesel Engine Based On PSO Optimized BP Neural Network

Diesel engines have a series of advantages such as good fuel economy,stable operation,durability,large output power,and low exhaust pollution,and have been widely used in various types of large-scale machinery.However,these machines are highly task-intensive and have a high frequency of use.The failure rate of diesel engines also suddenly increases.This leads to a gradual increase in the forecasting and diagnostic processing requirements for diesel engine faults,heavy engine diagnostic tasks,and high technical barriers.Since large machines need to maintain good working conditions for a long period of time and cannot be dismantled frequently for diagnosis,new fault diagnosis and detection techniques are needed.Based on this,this paper proposes an artificial intelligence method for engine fault prediction.It uses the PSO algorithm to optimize the parameters of the BP neural network and collects engine exhaust data,and selects the number of input neurons as 4,and the designed network contains 6 output neurons.Fault diagnosis through the trained model.The main content of this article is as follows.Firstly,the research status of diesel engine fault detection is reviewed and the significance of diesel engine fault diagnosis is clarified.The status quo and development trend of diesel engine fault diagnosis system are analyzed,and the research status of artificial neural network is analyzed.Then,the diesel engine fault model of BP neural network is analyzed.The prediction model without PSO algorithm optimization is discussed.The fault prediction mechanism of diesel engine is analyzed.The basic structure of BP neural network and the function of each module are introduced.The model of fault diagnosis of diesel engine based on BP neural network is discussed.Secondly,the PSO-optimized BP neural network is studied in detail and the new algorithm is used for fault detection.The basic information of the PSO algorithm is introduced and the algorithm principle and algorithm flow are analyzed.Then the process of combining the PSO algorithm and the BP neural network is studied.The new algorithm is simulated and the effectiveness of the PSO-BP algorithm is verified.Finally,the exhaust emissions（CO,CO₂,NOx,HC,and O₂）of the SC4H140Q4 diesel engine at different speeds are used as the training sample set.The samples are normalized and preprocessed before the training sample.The neural network is created and trained,and the trained neural network uses the test sample to diagnose the working state of the engine cylinder,thereby obtaining a specific fault condition in which the cylinder is in a normal working state,a light fire state,or a severe fire state.The accuracy of the fault detection by the training data network is verified by the verification data set,and the reliability of the proposed method is judged.