Research On Gas Pressure Regulation And Optimal Control Strategy Of Power Generation System

Improving energy efficiency has become a hot research topic.A large amount of pressure energy is not used in the process of gas transportation in China.Combining pressure energy with an generator to form a gas power generation system can effectively recover pressure energy for power generation,and whether the gas power generation system can operate stably depends on the gas Control quality of the pressure regulating system.At present,the gas pressure regulating system usually uses the PID control method to control the outlet pressure.Due to the non-linear and hysteretic characteristics of the gas system,affected by the fluctuation of the gas load of downstream users,the traditional PID control cannot guarantee the dynamic performance of the system,which is likely to cause gas outlet Large pressure fluctuations will not only affect the system's stable gas supply to downstream users,but also affect the stable operation of the power generation system.For the power generation system,the usual practice is to reduce the pressure fluctuations by adjusting the flow regulating valve at the inlet of the expander,thereby stabilizing the generator speed,but the energy will be greatly lost after passing through the throttle valve,and this part cannot be fully recovered.Pressure energy.In addition,the current common control method of the generator/ grid-side converter of the power generation system is direct torque/power control,which can only be selected according to the error between the initial torque,flux linkage or power detection value of the control cycle and its set value Voltage vectors,because it takes a certain amount of time to calculate and query the voltage vector table,and there are multiple voltage vectors to be output when the system is running,the voltage vector determined according to the table lookup method is not necessarily the optimal control amount at the current time.It may cause large ripples in torque,flux linkage or power.Aiming at the current problem of large fluctuations in the outlet pressure of gas pressure regulating systems,this paper proposes a nonlinear predictive control strategy for gas pressure regulating systems based on load prediction.Taking a Beijing gas pressure regulating station system as the research object,two neural networks are used respectively Constructed a system load prediction model and a gas outlet pressure prediction model;in order to solve the problem of large calculation amount of existing nonlinear optimization algorithms,this paper uses neural network as an optimized feedback controller to use the nonlinear processing capability of the neural network to achieve nonlinearity Predictive control.The paper uses the system objective function as the optimization performance index of the neural network controller.Based on the Lagrangian variation method,the gradient descent method is used to train the neural network controller online,and the impact of load disturbance is overcome in time using load prediction and rolling optimization.Maintain a stable outlet pressure.The neural network gradient descent training algorithm based on Lagrangian variation method takes up moderate storage space and small amount of calculation,which ensures the convergence of the controller weights and the stability of the system,which is easy to implement in engineering.On the other hand,this paper uses dual PWM converter power electronic components instead of the original throttle speed control device to control the generator speed and achieve grid connection,combined with the pressure control of the pressure regulating system,so that the system can realize the voltage regulating function at the same time.Use pressure energy to generate electricity.In order to solve the problem of large fluctuations in motor-side torque and grid-side power,this paper studies the model prediction direct torque/power control strategy,establishes a prediction model for torque,flux,and power,and performs delay compensation to eliminate the algorithm The control delay caused by the calculation will reduce the deviation between the above variables and its set value as the optimization goal,and predict the change of torque,flux linkage or power under the action of different voltage vectors in the predicted time domain.The optimized objective function evaluates the prediction results of each voltage vector,selects the voltage vector sequence that optimizes the overall objective function in the prediction time domain,and outputs the first term of the sequence to improve the dynamic performance of the system,reduce torque,Flux or power pulsation.The simulation experiment results show that the neural network nonlinear predictive control strategy of the gas pressure regulation system proposed in this paper reduces the control overshoot by 8% and the adjustment time by 60% compared with the PID algorithm after parameter tuning.The proposed model predicts a direct torque/power control strategy.Compared with the traditional direct torque/power control method,the engine-side speed and torque ripple are reduced by 3% and 15%,and the grid-side power ripple is reduced by 8%.The harmonic rate of the current is reduced by 6.92%,and the overall dynamic response speed is increased by 50%.