Research On The Algorithm Of Adaptive Speed Control Of Diesel Engine Based On Rbf Neural Network

The electronic governor is a core component of the diesel engine.It is the main technical means to improve the performance of the diesel engine speed control unit,the fuel injection quantity control rack position of the electromagnetic actuator and control of fuel injection pump,so as to control the speed of diesel engine.At present,most of the double closed loop control scheme of position loop and speed loop is adopted in the speed control strategy,and the common method of engineering application is the conventional PID control.This method is simple and effective,but also has the flaw,because the diesel engine is nonlinear and time-varying and the optimal PID parameters will be changed,therefore it is very important to study a control strategy that can automatically adapt to the operating conditions of the diesel engine.Radial basis function neural network algorithm is a new type of algorithm,which has a strong self-learning ability,it can achieve the function of parameter self tuning,and greatly improve the learning efficiency of PID control.In order to compare the superiority of this algorithm,the common PID,BP neural network and radial basis function neural network are used to verify the three algorithms respectively.In this paper,the research object is the D6114 type diesel generator,mathematical modeling in Simulink,take the average value modeling scheme,controller strategy to S-function programming to write module,respectively build controller model three different control algorithms,first off-line simulation,validated the superiority of intelligent algorithms,and radial basis intelligent algorithm robustness and control performance of the BP neural network has better.Then,the gradual completion of semi physical simulation,respectively,three kinds of control algorithms under the model into the d SPACE,the use of D6114 diesel engine mean value model,the real actuator,through the semi physical simulation debugging to further verify the reliability of the control algorithm of radial basis.The electronic governor general using MCU as the processor,but the cycle is long,spend too much time and energy in hardware,and using d SPACE as the controller can save time and effort,will focus on algorithm development.In this paper,as the controller to complete the experiment verification using d SPACE's DS1103 single board,which in starting,stable operation and dynamic loading operation stage respectively complete the experiment of three different control algorithms.The results show that compared with the common PID intelligentalgorithm has obvious superiority,and RBF intelligent algorithm than the BP neural network is more robust and reliability.