Application Of Multi-model Predictive Algorithm In Large-scale Thermal Power Unit Control

For the complex industrial control system,especially the large-inertia,large-delay system and multi-model predictive control of the power plant have the advantages of low accuracy,easy transplantation and good control effect.Based on EDPF-NT + system,the control strategy of the main steam temperature system is studied.Mainly include the following aspects:(1)Based on the T-S fuzzy model,a multi-model set is established.Based on the analysis of the transfer function model of the main steam temperature system,the sub-model sliding switching and the controller weighting are adopted to achieve bumpless switching.Main steam temperature control scheme of stepwise generalized predictive control.On the MATLAB / SIMULINK simulation platform,this control scheme is applied to a large range of load while setting value tracking and perturbation experiment,and the control effect and PID are analyzed and compared.(2)EDPF-NT + system OPC package through the system-specific NETASK interface from the MMI's virtual DPU read real-time data,and read data from the OPC client package for the MODBUS packet to MODBUS slave transmission to the virtual DPU.Real-time two-way communication between MATLAB and EDPF-NT + is realized by using the system 's OPC software package and MATLAB' s OPC toolbox.The simulation results show that the advanced control algorithm can be used in EDPF-NT + system for real-time simulation or real-time control.The advanced control algorithm written in EDPF-NT + can also simulate the MATLAB model in real time..This provides an efficient way to develop and validate the EDPF-NT + advanced control algorithm.(3)In order to measure the control quality of the designed main steam temperature control scheme in EDPF-NT + system,the real-time communication method of OPC is used to simulate the lift load on EDPF-NT + simulation system platform.The main steam temperature system Control program to control the effect of analysis and comparison.The results show that the effectiveness of the sliding multi-model stepwise generalized predictive control is better than that of the PID control with parameter tuning.