Model Predictive Control For Accelerated Cooling And Controlled Process

The accelerated cooling process takes use of the control of plate temperature evolution to massively produce plates with expected performance. It is a complicated system with many characteristics: a large number of system inputs and outputs, various plate guages, universal and complicated influencing factors, hardly measured transient temperature of plate inside the cooling, etc., and as the consequence the traditional control strategies can not apply for the high quality steel requirements. Taking acceletated cooling process as the object, this dissertation uses predictive control to study the system model identification and controller design. It includes:1) A pre-calculation and post-calculation method for accelerated cooling process is proposed. The pre-calculation uses local study strategy to determine model parameters of the energy balance partial differencial equation in plate cooling process based on the plate physical parameters, process requirements and history cooling database. Due to the difference between the pre-calculate and the real values, we take use of post-calculation to modify the model parameters with the measurment of the final temperature after the cooling process, which can continously update the database and increase the pre-calculation accuracy. This method can precisely indentify process models for every guage of plate with various influencing factors and large ranges of parameters changing, and automaticly adapt to the environment. It also provides the basis for controller design and presents the whole process temperature information. 2) A novel real-time optimization method is proposed to employ the plate velocity to control thefinal temperature. It simplifies the control system by considering the process target as the plate temperature in specific spot of cooling aparatus, which avoids the complex resolve expression between the plate speed and spot temperature. On this basis, we propose a horizon time-varying model predictive control method based on the plate speed pre-estimation, in which the optimization horizon varies with the plate position and speed (control inputs) to ensure the whole plate temperature optimization, and design a iteration algorithm based on the estimation of future plate velocity to solve this complex optimization. The numeric results show that the accuracy of final temperature can be greatly improved. 3) Propose a control strategy by taking the cooling water fluxs of the cooling water headers as the manipulated variables, which increases the control dimensions and improves the flexibility and accuracy of the controller compared to the method that takes the plate speed as the control inputs. We also propose a predictive control algorithm based on the set value recomputation. It dynamically determines the reference curves of every local controller with the information of starting temperature, and employs the appoximate non-linear state space model of the original partial differential equation as the predictive model, as well as linearizes the predictive model at the current operating points to decrease the optimization complexity. This method can guarantee the control accuracy and the real-time property, as well as control the complicated cooling curves.4)Propose a distributed predictive control for plate cooling process. It divides the cooling process into many subsystems interacting with inputs and states, and each subsystem employs model predictive control, among which the neighborhood optimization strategy is used for coordination. This approach utilizes the input and state information during optimization and sufficiently considers the neighborhood performance index to impove the system global performance. We give the solved results and stability condition of the proposed method and analyze the reason why this method can improve the system performance.