Cascade Large Time Delay Dispersion Predictive Control

Cascade process is widely existed in nonferrous metallurgy manufacturing processes. This process consists of a class of serially connected processes, the related action exists only in the direction of process flow and the whole system is unidirectional and irreversible. There is long time delay in the process due to the transmission of material and energy exchange, response time of the measurement instrument or device, execution time of actuator, etc, thus enhance the difficulties of stability of the process control. Besides, the actual manufacturing processes largely consist of complex dynamic process, and the precise prediction of predictive control is largely influenced by the model error due to the mismatch between mathematical model of the object and the actual object characteristics.In order to overcome the adverse effects caused by the varied process parameters, a new Adaptive Predictive Control method based on modified Particle Swarm Optimization (PSO) is presented to overcome the model error. All parameters of system model are identified on-line using PSO, thus the predictive model mismatch can be effectively overcome. Also a PSO based predictive control is proposed in which the PSO is used for iterative optimization, which can solve the complicated optimization with various kinds of constraints. Simulation result show that this method proposed has satisfied dynamic performance and robustness.For the long dead-time characteristics and structure feature of such industrial process, a decentralized predictive control algorithm is proposed in this paper. For each sub-process, the inputs of its related sub-process are treated as measurable disturbances, thus the object model can be decomposed into process model and interference model. Based on the process model and interference model, predictive control with feed-forward compensation and rolling optimization strategy is adopted for each sub-process. The industrial application result show that this method proposed can improve the effect of system coupling and long dead-time obviously, and on-line calculation was reduced greatly.