Intelligent Fault Diagnosis And Performance Optimization Of Hydrogen-fueled Engnie

With the current global energy crisis and increasingly serious environmental pollution problem,it is urgent to look for clean energy to replace the traditional fuel for internal combustion engines.Hydro gen energy is a kind of renewable,non-polluting,high calorific value and abundant clean energy.As the alternative fuel of internal combustion engines,hydrogen has its own obvious advantages.Compared with traditional fuels,the combustion of hydrogen fuel in the internal combustion engine is often accompanied by the abnormal combustion phenomenon such as pre-ignition and backfire,which seriously affects the normal operation of the system.Optimizing and controlling its operating parameters and suppressing abnormal combustion are the key technologies to be solved urgently.In this paper,based on the existing experimental conditions,various artificial intelligence optimization algorithms and intelligent fault diagnosis algorithms are deeply studied and applied to the optimal control and fault diagnosis of hydrogen internal combustion engine.The optimal control model of hydrogen internal combustion engine is built.It's a nonlinear optimization model based on artificial neural network.Firstly,the BP ne ural network is used to solve the nonlinear mapping relationship between the ignition advance angle and the working conditions(speed and load).Then the MATLAB software is used to simulate and optimize the ignition advance angle to optimize the engine operation.In view of the deficiency of BP algorithm in calculating speed and precision,three improved algorithms are proposed: particle swarm optimization fuzzy neural network algorithm,L-M algorithm and Powell algorithm,and then simulation research is done respectively.The particle swarm optimization fuzzy neural network algorithm has the global search ability of intelligence,and it can be used for efficient parallel search with high computational efficiency.L-M algorithm is an improved algorithm based on the Newton method and the gradient algorithm and it has the advantages of the two algorithms.L-M algorithm's network converges fast and also has high computational efficiency.Powell algorithm is a kind of unconstrained optimization algorithm,which avoids the computation of derivative and accelerates the computation speed.Through the experiment,for the ignition advance angle optimization and control modeling of hydrogen internal combustion engine,the L-M neural network has the highest convergence speed and accuracy,and is much higher than the traditional artificial neural network.Finally,the MAP diagram of the ignition advance angle of the hydrogen internal combustion engine of full working conditions is built by using the L-M neural network algorithm,and the purpose of optimal control of hydrogen internal combustion engine is achieved.The pre-ignition fault diagnosis of hydrogen internal combustion engine is achieved based on intelligent algorithm.According to the cylinder pressure rise rate signal of the hydrogen internal combustion engine under pre-ignition and normal combustion,the EMD algorithm is used to decompose the original signal and extract the fault feature and the conjugate neural network is selected to realize the identification of fault feature.In this paper,the method of computer signal processing is used to realize the real-time identification of the pre-ignition fault of the hydrogen internal combustion engine,and the diagnosis conclusion is directly output.The accuracy of fault diagnosis is high with high calculation speed,so it is convenient to realize the correction control of the hydrogen internal combustion engine.