Optimization Of Spinning Process Based On Orthogonal Collocation

Polyester fibers are in increasing demand in various fields for their superior properties.The technologies related to fiber production are constantly being developed and updated,with predictable differentiation and optimal control of high-performance fiber properties gradually developing into one of the important research directions for future fiber production.The polyester fiber mechanism model is complex and nonlinear,and the solution of the differential-algebraic equations is the basis for the dynamic optimization of spinning production.The existing algorithm will decrease the solution rate and increase the error due to the problem size increase for large-scale industrial continuous output.Moreover,the selection of state initial values and control inputs in actual production is random,which can cause the unpredictable and uncontrollable steady-state performance of fibers.The current research on dynamics fiber production planning involves initial value optimization.Still,it is inadequate in terms of accuracy and speed performance,and no further research has been done on control input optimization strategies for fiber production.Through solving the nonlinear programming problem with the discretization of the continuous mechanism model for polyester fibers,this thesis provides an in-depth study of the dynamic problem of fiber production and proposes a suitable optimal control strategy.The main research content can be summarized as follows:（1）For the polyester fiber model solving problem,a fully associative Radau orthogonal collocation algorithm based on finite element discretization and nonlinear programming is proposed in this paper.The algorithm discretizes the dynamic model in the continuous domain and solves the discrete model using nonlinear programming methods.Extensive experimental results show that the model-solving algorithm proposed in this paper can solve the fiber mechanism model accurately and efficiently and outperforms the current algorithm in terms of all-around solving performance.（2）For optimizing fiber state primaries and decision variables,a two-layer optimization strategy based on multiple targeting and receding horizon control is proposed in this paper.The initial state values are first sought using gradient descent-based multiple targeting strategies to obtain the best initial process input values.Then the future control inputs are optimized online using receding horizon control based on the model’s current output predictions.The strategy can realize the single target accurate product performance tracking and achieve the best production efficiency control for the multi-objective optimization problem in actual production.