Research And Implementation Of Model Predictive Control For Multiple Gas Collectors Of Coke Ovens

Since coke is an important raw material of the steel industry, cokingprocess becomes an important part of steel industry. The main by-product ofthe coking process is raw gas, so the gas collecting process is related to thestability of the whole coking process, the quality of coke and the life of theequipments. The pressure of gas collector pipe is an important industrialparameter, which marks the running status of the gas collecting process, eventhe whole coking process. Therefore, investigation of pressure controlstrategy for the gas collecting process is of significance to the steel industry.Normally, several coke ovens are connected together asymmetric, whichlead the multiple gas collectors pressure system to a complex industrialprocess that is constrained, strongly coupled, asymmetric and nonlinear.Model predictive control (MPC) is an advanced process control. MPCfocus on the complicated industril process, like multiple gas collectorspressure system, and has achieved successful and mature applications. ThenMPC is introduced to the multiple gas collectors pressure system. Based onsystem analysis, this dissertation gets the simplified model, investigate,simulate and realize the MPC strategy for the system.The main work of thisdissertation can be summarized as the following four parts:1) Development of the multiple gas collectors pressure system modelingand simulation software. Based on the mechanism model of the system, themodular simulation environment is built by using Matlab/Simulink. A model simplifying method is raised for MPC.Then the software is developed.2) Research of model predictive control for multiple gas collectors ofcoke ovens. An extended state space model and a feedback-feedforwardcompounded structure are chosen for MPC. The feasibility and effectivenessof the MPC strategy is verified by comparing with PID via simulation.3) Design and implementation of a generic simulation platform forMPC.The generic simulation platform is for both the MPC for multiple gascollectors pressure system and the general MPC simulations.The processingabilities of the platform are verified by several types of MPC simulations.4) Development of the software running on an optimization slave stationfor multiple gas collectors pressure optimization control system.Theoptimization station is a slave station of the control system. The softwarerunning on the slave station is developed.