Modeling, Optimization And Filter Estimation Of Complex Industrial Processes

System simulation, optimization and control of complex industrial processes must build on modeling, accurate models make optimization successful, control effective. Filter estimation and optimization are effective methods of increase economic efficiency. The final purpose is to get stable production and maximum economic benefit. The paper took FCC reactor and regenerator system and selective hydrogenation of acetylene reactor as examples, optimized, modeled, and filter estimated, got excellent optimization results, precise mathematical model and accurate estimated results.The paper researched three optimization methods, including sequential unconstrained minimization method, variable polyhedron method, improved particle swarm optimization method, analysis the advantages and disadvantages. Besides, the paper took kalm filter and particle filter to modify the soft-sensing model, it's excellent in solving non-Gaussian problems.The paper modified FCC correlation model on the basis of Blanding equation from one regeneration to two regeneration, made the model in accordance with actual industrial process, simulated actual production status exactly.On the basis of model, the paper took three methods optimize operation conditions of FCC reactor and regenerator, which could increase economic efficiency significantly, provided basis for developing operation and controlling program, proved the feasibility and effectiveness of steady-state operation optimization.According to over hydrogenation phenomenon of Qilu Petrochemical acetylene hydrogenation reactor, it needs to implement advanced process control scheme. The paper estimated acetylene output molar percentage and selectivity using soft-sensing method. Collected data through OPC, built first principle model, took least square method to calculate model, tested and verified model offline, the accuracy can meet the need of advanced process control, which proved the effectiveness and usability. Otherwise, the paper took particle filter to modify the model, compared to kalm filter estimation, particle filter was better. Modified the model with PF was a useful method and provided a reference scheme to soft-sensing correction.