N
N
naumovvr2019-02-25 06:00:08
Neural networks
naumovvr, 2019-02-25 06:00:08

What type of neural network should be taken to predict the failure of elements?

Good day to all. There was a problem if anyone can help I will be very grateful. The problem is that I can’t understand what type of neural network to take in order to predict the failure of some objects depending on any factors ... More precisely, there is a rail (x) that fails ... There are several characteristics : 1) Tonnage passed on the rails for each kilometer of the track. 2) The length of the curves in the section. 3) Descent or ascent.
There is data for a section of the track of a certain length ... There is a rail exit for the past 5 years at every km, and all the data about which I spoke above! And I need to predict for the next year the output of the rails at each km, depending on these characteristics. This is not a time series!
Question!!!: what grid to write? what kind of architecture should it have (in neural networks, a teapot) in the code, too, but I'm learning quickly. Thank you)

Answer the question

In order to leave comments, you need to log in

2 answer(s)
C
CHolfield, 2019-02-25
@CHolfield

to predict the failure of elements, there is such a science "Reliability and product quality management", it considers such reliability indicators as "Mean time between failures", "Mean time between failures", "Mean service life", "Mean time between repairs ". What the hell is the grid, where do you climb with your neural networks? Do you think that there are some buttons in the world that allow you to get results in such areas of knowledge, without studying this very knowledge? Fie on you, ignoramuses.

D
dmshar, 2019-02-25
@dmshar

As already mentioned, neural networks have nothing to do with it. Science, which has been dealing with such issues for 50 years, is named above.
If you really want to get acquainted with the methods of machine learning using the example of this task, then I give a tip, the corresponding section is called "Building Regression Models". As an additional reading - "Analysis of survival time" (yes, that's such an "optimistic" title). And no neural networks!
But without understanding what, why, where and why (in particular, why the neural network is not working here), a person is doomed to " drive formulas into Excel ". So I suggest you start with the basics.

Didn't find what you were looking for?

Ask your question

Ask a Question

731 491 924 answers to any question