How to make a dynamic state-space model for MPC controller?

2 次查看（过去 30 天）

Mustafa Rawash 2018-8-17

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/415334-how-to-make-a-dynamic-state-space-model-for-mpc-controller

评论： Haritha 2018-8-19

pred_mod_simple_1.m

Hello everyone, I'm building a stability controller for a vehicle using Model Predictive control (MPC). I'm using Simulink to simulate the vehicle and the controller, and using a the "MPC controller" block from the Simulink library. This block requires an MPC controller object that includes the state-space prediction model. I create this object before the simulation by running the attached m.file. My problem is as follows: my prediction model must have a state-space matrices with variable elements, for example the "A" matrix includes elements that are provided to the controller during the simulation because they are dynamic THEY CAN'T BE CONSTANT. I tried using "realp" as shown in the m.file to create generalized matrices and generalized state-space models, and placed "to workspace" blocks in the simulink model that captures the value of these elements, but for some reason the values of these elements do not update during the simulation. How can I build MPC object with dynamic state-space model for the "MPC controller" block?

3 个评论
显示 1更早的评论隐藏 1更早的评论

Mustafa Rawash 2018-8-18

do you mean I should customize the QP solver for the already used "MPC controller" block, or should I create my own MPC controller algorithm with a custom QP solver?

Haritha 2018-8-19