| The coordinated system of thermal power unit is a typical nonlinear system,which has large inertia and time delay,and is affected by unknown disturbances.With the increase of capacity,the dynamic characteristics of the unit under variable conditions change more significantly.For ultra supercritical units,under the current requirements of deep peak shaving,the traditional control methods can not achieve good control effect due to the large range of load variation.Therefore,it is necessary to establish an accurate model applicable to the full operating range of ultra supercritical units,and to design an advanced control system with good control performance on this basis.The main contents of this paper are as follows:(1)Introducing the standardized Euclidean distance to calculate the population similarity,an improved differential evolution(IDE)algorithm based on diversity evaluation index is proposed.The population similarity is used as a diversity evaluation index to guide the adaptive adjustment of parameters,selection of mutation strategy and judgment of population reconstruction,while IDE algorithm also uses population size reduction strategy to reduce the calculation amount,thereby improving the efficiency of the algorithm.The performance of the proposed algorithm is tested by standard functions,and IDE algorithm is used to identify the transfer function of superheated steam temperature system and to tune the parameters of the predictive controller,which verifies the effectiveness of its application in the thermal system.(2)Analyzing the operation characteristics of ultra supercritical units,a third-order state space nonlinear model of the coordinated system is established by using gray box method.Firstly,the physical equations of each subsystem are listed to help establish the nonlinear model structure of the coordinated system.Then,according to the steady-state data of the unit,the corresponding static parameters are obtained,and the undetermined functions are fitted.Then,the proposed IDE optimization algorithm is used to identify the dynamic parameters of the model with the step response test data at different load points,and the final model is obtained.A step response simulation experiment of the established model verifies the accuracy of the model.(3)Considering the nonlinear characteristics of the coordinated system of ultra supercritical units,the improved fuzzy c-means algorithm(IFCM)is used to construct the model set,and a multi-model predictive control strategy based on weighting objective functions of the sub model set is proposed.IFCM algorithm introduces dimension difference weight to modify the objective function,and the optimal number of clusters is determined according to the effectiveness evaluation index of the results.IFCM algorithm is used to divide the operation data into different working conditions,and the model set is constructed by linearization.Some of the sub models are selected to construct the sub model set,and the corresponding sub model set is called to participate in the calculation according to the matching degree of the working conditions.The weights are calculated based on the weight distances between the current working condition and the sub models,thus the global objective function is obtained to directly form the global control.The control method can effectively reduce the amount of on-line calculation,while reducing the fluctuation of control variables and improving the stability of the system.The improved multi-model predictive control method is used to carry out a simulation of load changes in the full range of deep peak shaving,which verifies the adaptability of the method to the wide range of load changes. |
