Simulink Parameter Estimation Error Nonlinear Least Squares

Question

Guilherme Pereira 2017-8-29

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/354434-simulink-parameter-estimation-error-nonlinear-least-squares

评论： Changjiang Wang 2022-3-6

Hi,

I'm trying to run the parameter estimation example showed on this video (https://www.mathworks.com/videos/estimating-parameters-of-a-battery-68957.html).

I've done all the setups exactly as in the video and when I start the estimation of the parameters, I get the following message: "the levenberg-marquardt algorithm does not handle bound constraints and the trust-region-reflective algorithm requires at least as many equations as variables: aborting". I'm completly lost at what's the reason for that, even more because I use the exact same model as in the video and I can't find the source of the problem.

Does anyone know how to solve it?

The model and codes I use are the ssc_lithium_cell_1RC_estim.slx and ssc_lithium_cell_1RC_estim_ini.mat and the data used for the estimation is the one from LiBatt_PulseData.mat that comes together with the files when you download it.

PS.: I've had to change the solver type in the configurations manually to ode15s. The simscape model also comes with the ssc_lithium_cell_1RC_estim_ini.mat in the PreLoadFnc Callback. I'm working with Matlab 2014a.

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Guilherme Pereira 2017-8-30

I've somehow managed to make it work. I still don't know the exact cause of the problem, but I think there's some kind of setup on the Simulink model that must be done before someone starts to work with the parameter estimation tool.

The reason for that hypothesis is because different from the file used for the parameter estimation of the lithium ion battery on the video I mentioned before, the one I got from the download section seems to be missing some kind of model configuration that wasn't shown on the video, since for example, I had to manually change the solver type, the simulation time isn't the same and there are no inputs configured to the inport of the model, so I can't run the simulation like on the video.

The second reason for the hypothesis is that I've managed to make it work, by putting the battery's circuit model in a copy of a different Simulink model I've downloaded from Matlab, in which I've successfully performed the parameter estimation during previous work.

Does anyone know what am I missing when configuring my model for parameter estimation, so I can avoid the same mistake in the future?

请先登录，再进行评论。

请先登录，再回答此问题。