Exporting remaining useful life models with MATLAB Coder

12 次查看（过去 30 天）

Simon Mählkvist 2019-5-29

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/464577-exporting-remaining-useful-life-models-with-matlab-coder

评论： Simon Mählkvist 2019-6-3

I want to generate a MATLAB function which i can export into C code via MATLAB coder.

I've been working on an remaining useful life (RUL) model which I would like to export onto an Raspberry Pi (RPi) using MATLAB Coder to generate C code. I've encountered some difficulties doing this and failed to find any documentation in the respective products which would help me.

The RUL model is created with the predictive maintenance toolbox and the residualSimilarityModel() function. I then use fit() to trim the model.

Using whos on the mdl file output the following:

whos mdl

Name Size Bytes Class Attributes

mdl 1x1 8 residualSimilarityModel

What i want to know is how to embed the RUL model into function script so that i can export one single file.

2 个评论
显示无隐藏无

Alec Stothert 2019-5-29

Unfortunately residualSimilarityModel does not support code generation. We are working on adding code generation support for the RUL models but it is not yet available and I would be interested in discussing your application with you in more detail. One challenge with deploying similarity models is that the training data (data used for fit) is also needed for prediction and that can place restrictions on deployment.

Simon Mählkvist 2019-6-3

The quantity of the training data is not large and embedding it along with the script for training the model should not be an issue.

Follow up question: How should the training data be setup so that it can be exported with MATLAB Coder

If it is possible to provide an explanation or link to documentation describing this process I consider your response to be a valid answer to the questions and would recommend submitting it as an answer.

I am grateful for the assistance and if there are any details missing from my part don’t hesitate to ask.

请先登录，再进行评论。

请先登录，再回答此问题。