Fault prediction method for penicillin fermentation process based on multivariate time series attention-lstm

Abstract

The invention discloses a penicillin fermentation process fault prediction method based on Attention-LSTM of multivariate time series. Firstly, a plurality of relevant process variables in the penicillin fermentation process are selected through Pearson correlation analysis, and then input to LSTM through the Attention mechanism The hidden vectors at different times of the sequence are given different weights, which makes the neural network prediction model more effective in processing long-term sequence inputs, and realizes fault prediction modeling for the penicillin fermentation process. The present invention adopts the failure prediction model combining the attention mechanism and LSTM to predict the failure of the penicillin fermentation process, overcomes the problem that the existing LSTM tends to ignore important timing information when processing long sequence input, and makes the failure prediction based on LSTM more accurate.

Description

technical field [0001] The invention belongs to the field of industrial fermentation production process failure prediction modeling and application, in particular to a penicillin fermentation process failure prediction method based on multivariate time series Attention-LSTM. Background technique [0002] In the actual production process, the fault diagnosis of the system is usually only based on the current situation. However, if the system fault can be predicted early, the fault can be found and eliminated in time before the system fails, which can reduce the number of faults. impact on the system. Therefore, the fault prediction of the system is of great significance to ensure the safety of the system. [0003] A time series (or dynamic sequence) refers to a sequence in which the values ​​of the same statistical indicator are arranged in the order of their occurrence time. Time series analysis is a time domain method that uses parametric models to process ordered random ...

AI Technical Summary

Problems solved by technology

However, in the face of long-term sequences, the cyclic neural network is prone to problems such as gradient disappearance, gradient explosion, and insufficient long-term memory capacity.

Images