Diagnosis method of aluminum electrolytic cell condition based on principal component similarity measure

Abstract

The invention discloses an aluminum electrolysis cell condition diagnosis method based on principal element similarity measure. The method is characterized by comprising the following steps of: 1. briefing original features, namely calculating m principal elements Bj of which the contribution rates are on the top of the list by a core principal analysis method, sequentially checking the contribution rate of each original feature to an aluminum electrolysis cell condition, deleting the original features of which the contribution rates are lower than a contribution degree threshold value, and realizing feature briefing; and 2. constructing a classification model of the aluminum electrolysis cell condition by taking the briefed features as input variables of a probability nerve network, and taking the type of the aluminum electrolysis cell condition corresponding to the maximum output value of the model as a diagnosis result. By the aluminum electrolysis cell condition diagnosis method, shortcoming that the core principal elements have no definite physical meaning is overcome; the number of sensors and the calculation amount are reduced; furthermore, a fault diagnosis model is constructed by utilizing the probability nerve network with short one-time training time and high diagnosis precision; and the method is suitable for on-line diagnosis for the aluminum electrolysis cell condition.

Description

technical field [0001] The invention belongs to the technical field of fault diagnosis, and in particular relates to a method for diagnosing the condition of an aluminum electrolytic cell based on a principal component similarity measure. Background technique [0002] Fault diagnosis is to make judgments on the operating status and abnormal conditions of the system, and make judgments based on the diagnosis results to provide a basis for system fault recovery, the most important of which are fault detection and type judgment. Fault detection means that after establishing a connection with the system, it periodically sends a detection signal to the lower computer, and judges whether the system has a fault through the received response data frame; fault type judgment means that after the system detects a fault, it analyzes the cause and determines The type of system failure. Due to the time-varying characteristics of the system, it is difficult to establish accurate online fa...

AI Technical Summary

Problems solved by technology

However, the obtained pivot is still the linear combination of the original variables in the feature space, which only achieves the purpose of dimensionality reduction, and does not eliminate the number of original features, and the number of original features reflects the number of sensors. The existing kernel principal component analysis Cannot reduce production cost by reducing the number and types of sensors

[0005] In the fault diagnosis of aluminum electrolytic cells, BP neural network is the most widely used modeling method. However, due to the large number of characteristics that characterize the condition of aluminum electrolytic cells and their strong correlation with each other, the amount of data involved in modeling is large, and the calculation is relatively complicated. It is difficult to establish an accurate diagnosis model with BP neural network

[0006] The defect of the existing technology is: when diagnosing the condition of the aluminum electrolytic cell, it is necessary to install a large number of different types of sensors, and to calculate the detection data of all the sensors to realize the correct fault diagnosis. The amount of data involved in the modeling is large, and the calculation is relatively complicated. Complicated, resulting in low production efficiency and high production costs

Images