Application Of Neural Networks To Coal-bed Methane Engine Control System

Application of advanced control theory to coal-bed methane (CBM) engine control system is very important and significant, because it can well improve CBM engine's power performance and emission properties. Study on Corresponding characteristics for related Sensors and Air-Fuel-Ratio Control Strategies of CBM Engine is is an important task in analysis and designing process of CBM engine control system, it lays the groundwork for the design and simulation of the engine electronic control system.The nonlinear dynamic modeling methods for a hot wire type mass air flow (MAF) sensor were studied based on the experiment data in order to design efficiently the engine air fuel ratio feed-forward controller in Electronically Controlled Unit. The parametric and linear neural network non-parametric model were established and validated, the results show that The neural network non-parametric model is more accurate than parametric model, and can correctly describe the dynamic characteristics of the sensor in both positive and negative working mode with one model, so it is better than parametric model when they are used for simulation sub-model in engine control system.Based on the modular design concept, initially established the average coal-bed methane engine model in Matlab/Simulink environment using experimental data obtained on CBM engine experiment bench. The combination of mechanism and testing modeling method was adopted in the process. To improve identification accuracy neural network which has strong nonlinear approximation ability and a high degree of learning ability was used to identify model parameter. The simulation results of the model validation show that the system model after setting parameters and neural network correction can basiclly predictive the engine output trend and range in both steady-state and dynamic state.The air-fuel ratio control precision is the key factor of improving engine performance, the neural network based feedforward-feedback control strategy for the coal-bed methane engine air-fuel ratio was proposed in view of the gasoline engine air-fuel ratio control strategy. Feed-forward control was performed by the adjustable weights Neural Networks, and feedback control was completed by the conventional PID controller. The simulation results show that proposed method has better control effect than that only using PID controller.Finally, kinds of methods were studied including BP-network based tuning PID control algorithm, CMAC - PID control algorithm and neural network predictive control algorithm for coal-bed methane engine speed control, the study results show that predictive control is more suitable for speed control algorithm, which is helpful to improve testing and measurement system on test bench and to develop electronic control throttle.