Font Size: a A A

Research Of Predictive Control Based On Mismatch Information Under Model-Plant Mismatch

Posted on:2012-04-17Degree:MasterType:Thesis
Country:ChinaCandidate:J Q XinFull Text:PDF
GTID:2268330425490475Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
Predictive control is an important control strategy, but as a control method based on the characteristic information of the controlled object, predictive control needs an accuracte model. As a control method based on the feature information of the plant, a high degree of the accuracy of the model is required in predictive control method. In the event that bias of prediction model is on the small side, the predictive control method will obtain satisfactory control effect. But, in contrast, if the prediction model is mismatch with the plant, control effects will be influenced directly. So studying of predictive control method for model-plant mismatch is important on theoretical and realistic signification.The basis of the paper is that there are already some methods to accurately evaluate, test and diagnose the performance of the model predictive controller. And the result shows the reason of a bad performance of control system is because of model mismatch. Based on this, the contect of this paper is as follows:Firstly, analyzed predictive control principle and DMC in details. By the simulation of predictive control when the mismatch of model-plant is larger, it is necessary to propose to adjust the model parameters. Scondly, briefly introduce the classification of model-plant mismatch. For the model with only one mismatched parameter, proposes one method of constantly online adjusting mismatched parameters based on the combination of error iteration and one dimension search method, to make the frequency of output less as far as possible to meet the control requirements. And then do some improvements based on it. Put forward a method to online correct the search step length ak by using a PI regulator. Through this, the frequency of output is less. Thirdly, Introduces the fuzzy control in detail. When there are more than one mismatched parameters, fuzzy control is been used to had a trying. By using δ%, tr and ts as inputs, mismatched parameters are gained, and record the constant of the mismatch parameters. Then using△K and△T are outputs, by establishing a reasonable rule library, searches the two mismatched parameters to let the outputs meet the requirements with less batchs online. The last, verifying the method is effective by experiments.
Keywords/Search Tags:predictive control, DMC, model-plant mismatch, robust, fuzy rules
PDF Full Text Request
Related items