Research On Multi-Model Control Of Superheated Steam Based On Artificial Immunity

Dynamic characteristic of power plant thermal process is close related with operation conditions of power unit. For peak load unit,its process model and structure parameters have obviously fast time-varying characteristic while load changing. It cannot get satisfactory control result using robust or adaptive control scheme for controlled objects of this sort. Multi-Model adaptive control scheme is exclusively designed for this sort of objects.Biological immune system has been proven to be of diversity,massive distribution, robustness,adaptation,immune memory etc. In this paper,artificial immune principles are introduced to Multiple Model adaptive control from two different angles for improving Multi-Model control. The main contents of this paper are as follows:1 Reviewing the history of generation,development,application of Multi-Model adaptive control and Artificial immune; summarizing some research methods of Multi-Model adaptive and prospecting chief direction of future research of Multi-Model adaptive control; introduing approaches and achievements of artificial immune applications in process control and prospecting further artificial immune applications in process control.2 Based on charateristic of immune response and immune cells' structures, a new pure immune algorithm is presented and be validated by a multim odal function. The algorithm this paper present is applied to model identification and controller's parameters tuning to optimizing each model control of model class, which optimize Multi-Model adaptive control from algorithm angle. The algorithm can be applied to other areas because of its general applicability.3 In Multi-Model adaptive control, model identification and controller's parameters tuning are done by the algorithm this paper present in chapter 2. Multi-Model control methods is researched combined with superheated steam. A smooth switching scheme has been designed and its stability has been proved. Considering frequent switching may be happening in practical running conditions, a scheme named switching delay is designed.4 Because of immune network's juvenility, using neural network that simulating immune trait construct Multi-Model controller, which develop a new way for Multi-Model control being intelligentized.