炼化变工况下多变量过程的智能PID协调整定OA

In petroleum refining and chemical production processes,situations frequently arise where the automatic control mode of the control loop fails after significant changes in operating conditions,necessitating manual intervention.Additionally,the decentralized control structure adopted in the field control loops fails to adequately consider the inter-loop interference caused by coupling effects.Based on the closed-loop architecture of"perception-decision-execution",an intelligent coordinated proportional-integral-derivative(PID)tuning method for multivariable processes under variable operating conditions is proposed.Firstly,collect the data of manipulated variables and controlled variables inside the control loop.Adopt the idea of segmented modeling and use the gated recurrent unit(GRU)neural network that based on the model order to construct the process dynamic model of the closed-loop control loop.Secondly,the magnitude of network updates is taken as the trigger mechanism for intelligent tuning after a significant change in working conditions to achieve the trigger tuning of the slow rate mechanism.Then,both the time-domain performance and weak-coupling performance indicators are connected through appropriate weights.The improved particle swarm optimization algorithm is adopted to coordinate and tune the parameters of multiple PID controllers.Finally,the comparative experiments were completed using the heavy distillate oil tower model of Shell Company.It has been verified that this method can reduce the coupling influence between control loops and can timely achieve PID coordinated retuning after operational condition changes.