Research On Disturbance Rejection For Multivariable Processes With Time Delays Based On Coupling Analysis

Process industry plays an important role in industrial production.Most process control systems can be described as multivariable systems with time delays.The inherent complex couplings and the existence of time delays bring some difficulties to the design of such control systems.In process control,the set-point of the system is relatively fixed,and disturbance rejection is the primary goal of the control system.The existence of disturbance usually deteriorates the control performance of the industrial system and even endangers the stability of the system.Therefore,it is of great practical significance to carry out research on disturbance rejection methods for multivariable systems with time delays.At present,a large number of disturbance rejection designs for multivariable systems with time delays are based on the state-space method,which is complex and difficult to implement.Besides,most of the existing methods treat couplings as internal disturbances,and rarely use them for disturbance rejection design.Given this,establishing the relationship between couplings and system disturbance rejection performance,and developing a control system design method with stronger disturbance rejection capacity,simple design process and easy engineering implementation,is the key problem to be solved.In this paper,from the perspective of disturbance rejection,the influence of disturbance on the system under complex couplings is analyzed by using the multivariable frequency domain control methods,and is utilized to guide the design of disturbance rejection controller and inverse model.In addition,analytical forms of the disturbance rejection controller and disturbance observer filter are given based on the internal model control and optimal control theory.The specific work can be summarized as follows:1.The design method of decentralized control system for disturbance rejection based on partial effective open-loop transfer function.Simple structure and easy implementation are the outstanding features of the decentralized control system.Given this,a main channel selection method based on coupling analysis and partial effective open-loop transfer function is proposed to improve the disturbance rejection performance of the system without changing the existing decentralized controller structure.Considering the complex form of the effective open-loop transfer function,the necessary model reduction and simplifications are performed first,and the single variable internal model control strategy is then employed to design the controller.At last,the controller parameters are tuned to satisfy the robustness requirements and the controller is transformed into the PID form that is easy to implement.2.The design method of disturbance rejection control system based on partially decentralized control.From the perspective of coupling analysis instead of dealing with couplings directly,this paper analyzes the relationship between the multivariable controller structure and disturbance rejection,and transforms the treatment of coupling implicitly to the selection of the controller structure.Firstly,by using the disturbance analysis index and the equivalent internal model framework,the partially decentralized controller structure with stronger disturbance rejection capability is selected.And then,the partially decentralized control system is transformed into the corresponding non-square decentralize control system to facilitate the controller design.When designing the controller,a pseudo-inverse factor is introduced to avoid calculating the complex pseudo-inverse matrix,and the value range of the pseudo-inverse factor is given,from the perspective of ensuring system stability.Finally,the partial decentralized controller is obtained in its analytical form.The advantage of the designed partially decentralized control system is that the control structure is relatively simple and can be implemented on the basis of the original decentralized control system,which is of practical importance.3.The design method of disturbance observer-based control system with modified inverse model.In view of the difficulty of obtaining a sufficiently accurate inverse model in the design of multivariable disturbance observer,an improved inverse model calculation method is proposed in this paper.Firstly,the problem that the inverse model elements cannot be realized is solved by introducing a diagonal compensator,and the ranges of the compensator parameters are given based on rigorous mathematical derivation.Secondly,the Nyquist set is used to describe the dynamic characteristics of the compensated inverse model,and the curve fitting technique is utilized to obtain the simplified inverse model elements that are easy to implement,avoiding complex dynamic matrix inversion.In addition,to further reduce the disturbance estimation error caused by the inverse model approximation and improve the disturbance rejection performance of the system,the disturbance observer filter is designed to minimize the H? norm of the disturbance estimation error.After obtaining the optimal form of the filter,parameters are determined by satisfying both the performance and robustness requirements.4.The design method of disturbance observer-based control system with partially coupled inverse model.In a disturbance observer-based control system,the existence of uncertainty and model approximation error makes it impossible for us to obtain a completely accurate inverse model,and the resulting disturbance estimation error has a limited impact within a certain range.However,when the disturbance estimation error caused by the inaccuracy of the inverse model cannot be ignored,the process dynamics will affect the system outputs through the disturbance estimation error.In this case,for better disturbance suppression performance,the design of the inverse model must consider the dynamic characteristics of the process.To this end,firstly,the relative input disturbance gain index is proposed to analyze the disturbance rejection capability of the system under different inverse models.With this index,the structure and concrete form of the partially coupled inverse model can be determined.Then,a single iteration strategy is presented to deal with the cross terms of the filter elements in the outputs.Meanwhile,optimal forms of the filter elements are obtained by minimizing the H2 norm of the outputs.Finally,parameters of the filter are tuned to meet the robustness and performance requirements of the system.The proposed MDOB design with partially coupled inverse model is an innovation and development of the existing MDOB design methods,which provides a new idea for further improvement of the inverse model and disturbance rejection capability of the system.