Industrial equipment vibration characteristic value trend prediction method based on data model

Abstract

An industrial equipment vibration characteristic value trend prediction method based on a data model comprises the following steps: step 1, collecting characteristic values of a plurality of data typesignals of industrial equipment through a vibration sensor; 2, observing the stability of the training data; 3, forming an autoregressive moving average model; 4, determining the number of parametersaccording to a Bayesian information criterion function method; 5, checking the randomness of the prediction error sequence of the optimal model; 6, verifying the model; step 7, model prediction. Characteristic values of a plurality of data type signals of the industrial equipment are collected through the vibration sensor to serve as parameters, the acquisition way of the parameters is simple, the efficiency is high, and an over-fitting or under-fitting state does not exist.

Description

technical field [0001] The invention belongs to the field of state monitoring of industrial equipment, in particular to a data model-based method for predicting the vibration characteristic value trend of industrial equipment. Background technique [0002] Industrial equipment maintenance is the core content of the daily management of the factory. The healthy and safe operation of equipment can not only increase product output and optimize production efficiency, but also reduce consumption, ensure the safe production of the enterprise, and help the factory maximize economic benefits. [0003] The current methods of industrial equipment maintenance in the factory mainly include: after-the-fact maintenance, preventive maintenance (that is, regular maintenance) and predictive maintenance. These three different maintenance methods are also the three stages of equipment management development. For a long time, my country's industrial equipment has generally implemented a mainten...

AI Technical Summary

Problems solved by technology

If the network structure is too large, the efficiency in training is not high, and overfitting may occur, resulting in low network performance and reduced fault tolerance; if the network structure is too small, the network may not converge, and the network structure directly affects the performance of the network. Approximation ability and generalization properties, therefore, the difficult choice of network structure restricts the application of BP neural network algorithm in trend prediction

Images