I want to develop an ML model to predict machine failure prior failure. I have a time series dataset from various sensors(10 sensors) of machine. In this dataset we have 10 days dataset before each failure. How do i start by using Machine learning model and python.

I wants this model will predict the failure 1 day before the failure. Please help. Thanks You.

CodePudding user response：

There are a lot of ways you can try, I will suggest you to create a lag feature by one day, where you shift your time by one step.

for example if your data:

you can make it like this

you can check here for quick information about time series I learned this from their first lesson https://www.kaggle.com/learn/time-series

CodePudding user response：

I would suggest you some resources about predictive maintenance (which is exactly predicting "machine failure prior failure"). This source should be adapted to get started.

Then, you may want to learn about survival analysis and especially Weibull distributions that can be adapted to your modeling. This is a usual way to represent the time to failure of a component. You can find many resources online about Weibull models, more or less technical, along with implementations in Python or R. For a deep learning approach, you can consult this source.