Research And Application Of The Fast MPC Algorithm Based On Quasi-equivalent Amplitude Decaying Aggregation Strategy

Model predictive control,which has been massively applied in industrial fields,can reflect the control object's dynamic characteristic accurately by using its predictive model and have excellent constrain solving and steady control ability at the same time.However,when facing large scale or requiring high real-time systems,the problem of large computation burden can not be simply ignored.We proposed a kind of quasi-equivalent aggregation strategy based on the decaying amplification aggregation to reduce the number of the optimization variables of the on-line optimization problem and make the design more practical.Based on this quasi-equivalent strategy,this paper studied its application in temperature and humidity control of air enthalpy lab project and parallel inverters droop control.The main works of this paper include:1)The design of quasi-equivalent model predictive control.By researching the theorem of aggregation strategy and its performance optimization,we proposed a quasi-equivalent aggregation strategy based on simulated annel method to reduce on-line computation and have decent performance at the same time.2)Ident and modeling of the air enthalpy lab.Studied the structure and working environment of air enthalpy lab,measured the data of model,ident the ARX model of the temperature and humid decouplingly.Derived the state space model based on ARX model.3)Simulated the temperature and humide control program,deployed our model predictive controller in scene.Studied the expression of model predictive control problem in air enthalpy lab,simulated the fast model predictive control performance in Matlab.Finished coding air enthalpy lab scene deploying C++ program,debugged and optimized control performance in scene and achieved good performance in steady time and overshoot.4)Applied the fast model predictive control in parallel inverters.Analysed the working principle of three-phase 2 voltage level inverter.Modeled and concluded the model predictive control problem,combined the active power – frequency and reactive power – amplitude droop control strategy to divide load and reduce circulating current.Simulated in Matlab/Simulink,achieved good divide performance in our model predictive driven parallel inverters system.