Optimization Solution Of Process Control Configuration Design

Chemical processes are usually high-dimensional multivariable systems.When designing the bottom regulatory PID control system,control engineers must consider the interaction between control loops and implement the control loop configuration and variable pairing.With the dimension increasing of multivariable systems,the possible variable pairings are rapidly increasing and the computation load will be insufferable.The traditional subjective analysis methods relying on control engineers are difficult to be applied and fail to the optimal control configuration,so a simple and fast computer optimization algorithm is urgently needed.In order to solve the above problems,this paper proposes some new control configuration design methods based on the idea of genetic algorithm.Firstly,the genetic algorithm is applied for the control configuration design of fully decentralized control system.The concepts of relative energy gain array and controllability are introduced.The structure matrix column vector is encoded in string form,and the variable pairing is converted into an array form for crossover,mutation,selection and other genetic operations.The genetic operator can be adjusted appropriately according to the constraints,and the optimal control configuration can be obtained for fully decentralized control system.The algorithm is also of fast calculation speed.When the system dimension is too large,the accuracy of optimization is significantly improved compared with the observation method.The algorithm lays the foundation for the optimization algorithm of the block control configuration design.Secondly,the genetic algorithm is applied for the control configuration design of block decentralized control system with given block pattern.Several sets of fully decentralized control structure with better controllability are used as the starting points of genetic algorithm.Combined with the overall performance evaluation index,the column vector and the row vector of the given block structure matrix are simultaneously re-encoded in the form of strings.Under the given block constraints of structure matrix,the optimal control configuration of the block decentralized control system is obtained through a series of genetic operations and comparative analysis.Finally,in order to find the optimal block decentralized control configuration with optimal block pattern,a searching method that can greatly reduce the calculation is proposed.Based on the optimal fully decentralized control configuration,the block corresponding to single control loops is gradually added to structure matrix.Under each block pattern,the control configuration design with given block pattern is used.This method finds out the correlation between the overall performance evaluation index and different block pattern.Then the optimal block decentralized control configuration will be found with this searching method step by step.