Research On Hierarchical Predictive Control Strategy And Its Applications In Industrial Process

It is difficult to hold a fully balanced state between commodity supply and demand during an increasingly competitive market time, in this case commodity prices will frequently fluctuate, and optimal operating points will shift, which result in a decline in the economic efficiency of enterprises and control quality. In order to obtain the maximum economic benefit of the enterprise production, the production process operation needs timely access to the optimum operating point under the economic environment, then make the MPC controllers to track the operating point.As a kind of optimization control strategy, model predictive control(MPC) can achieve better control effect and improve the utilization rate of energy compared to the traditional PID control strategy, thus MPC brings more significant economic benefits to the enterprise. Generally, the optimal trajectory of MPC is based on the experience of the equipment operator, which deviates from the actual optimal operating point, and is not sensitive to the change of the economic environment. In order to self-optimize the optimum operating point and track control operating point, this research focused on Hierarchical Model Predictive Control(HMPC) which combine of optimization and control followings:(1) This thesis summarizes the previous research results of the optimization control strategy and great achievements of advanced control technology in the economic benefits, and summarizes the research status of HMPC strategy at home and abroad.(2) For research on HMPC strategy, firstly need to research the process of conduct market, build objective economic function, and turn the economic optimization problem into Linear programming(LP) and Non-linear programming problem(NLP) problem under the constraints of process model. Use LP and Sequential Quadratic Programming(SQP) algorithm to compute the optimal value of variable and as the optimal operating point, when economic factor price fluctuate frequently, then passed to the control layer, where control layer select MPC algorithm to track the operating point. Then applied HMPC strategy to CSTR process, the simulation results show that HMPC can steady CSTR run at the optimum operating point always, at the same time improve economic efficiency and ensure the stability of the system.(3) As for nonlinear objects, its operation modes corresponding to a plurality of fixed linear model which is valid only in the vicinity of the equilibrium point, when the balance changes, the linear model fails, and the system generates shock overshoot, or even destabilize the system, so it can’t meet the control requirements. In order to solve this problem, this paper proposes a multi-layered model predictive control(HMMPC) strategy, where the entire operating range was divided into several sub-intervals according to the change of operating points, then establish the linear model using small gain method in each sub-section, next switch sub-model by output error when operating point is jump, and design MPC controller with the sub-model switched. The comparing of simulation shows that, the HMMPC strategy can control system effective under the premise of ensuring the stability of the system, and can improve the control quality of transition process when operating point is jump.(4) Applied HMPC to the control of heavy oil fractionators, and verify its validity by simulation, indicating it’s advantages in industrial applications.(5) Finally, summarize the work of this paper, and the problem of future research and development trends of HMPC were discussed.