search

chickenbestlover / RNN-Time-series-Anomaly-Detection

RNN based Time-series Anomaly detector model implemented in Pytorch.
Apache License 2.0
1.17k stars 316 forks source link

How to use this code for my robot anomaly detection? #8

Open yunzhonghe2017 opened 5 years ago

yunzhonghe2017 commented 5 years ago

After some change, the code has successfully been run on my computer. Then,......... If I want use this code for my robot anomaly detection, how should I prepare my data.

I have a basic idea. I want to do the anomaly detection based on multi-sensor data of the joint motors ( ie,. rotate speed, torque, and temperature). How should I prepare my data according to the code you provide?

Then, I have tried to understand the data of the nyc_taxi.csv data，can you tell me what each column of the table means? default

chickenbestlover commented 5 years ago

A: DateTime B: NewYork taxi passenger demand C: time of the day (in minutes) D: day of the week (1: Monday ~7: Sunday maybe?)

How about pre-processing the robot's multisensor data like any other data set in this code and then learning it on the network?

yunzhonghe2017 commented 5 years ago

Thanks for your constructive reply. @chickenbestlover If the robot's data is arranged as follow, can we learning it on the network? how can we preprocess the data? 1