Research On Equipment Fault Diagnosis Based On Full Vector EEMD And Case-based Reasoning

As rotating machinery is widely used in industrial fields such as electric power,petrochemical metallurgy,and aerospace,the safety of its equipment has become the focus of scholars at home and abroad because it is related to the safety of people’s lives and property.Other failures are the most common,accounting for about 70% of total failures.Nowadays,with the rapid development of Industry 4.0 intelligent industry,fault diagnosis technology based on a large amount of industrial data is emerging,and the data-driven intelligent fault diagnosis method has gradually become a rising star with its efficient and rapid characteristics.In this paper,the rotor imbalance fault is taken as the research object,and the extreme learning machine algorithm combining time-frequency domain features with additional features and the sparse autoencoder algorithm with adaptive feature extraction are compared.Aiming at the situation that the single-channel vibration signal information is incomplete and the diagnosis result will be misjudged,the Full Vector spectrum information fusion technology is used to fuse the two-channel signals.The structural parameters adopt experience and reference methods to compare the two models of single and double channels with a total of four models,and use genetic algorithms to optimize the models.The main research work of the article is as follows: An imbalanced fault diagnosis method based on Full Vector spectrum technology and extreme learning algorithm(Full Vector ELM)was established and verified through experiments.The process is to first use the Full Vector spectrum technique to fuse the X and Y channel signals to generate a more complete information fusion signal,extract the time-frequency domain features,Full Vector energy features,and additional features of the fusion signal as inputs to construct a complete Full Vector limit Learning the model and constructing a single-channel extreme learning model for comparison.Experiments show that the model based on the full-vector ELM can recognize and distinguish imbalanced faults better than the singlechannel model.An imbalanced fault diagnosis method based on Full Vector spectrum technology and extreme learning algorithm and an imbalanced fault diagnosis algorithm based on Full Vector spectrum and support vector machine are established and verified through experiments.The process is to first use the Full Vector spectrum technology to fuse the two-channel signal to generate a more complete information fusion signal,extract the fusion signal time-frequency domain features,Full Vector energy features,and additional features as inputs to construct a complete Full Vector Machine Learning model At the same time,a single-channel extreme learning model is constructed for comparison.The experiment shows that the recognition ability of imbalanced faults based on the Full Vector Machine Learning model is higher than that of the singlechannel model.Using experimental data and noise data to test the two models,the results show that the Full Vector Machine Learning fault diagnosis model has strong adaptability to the experimental data,and the anti-noise ability is better than the single-channel data,but the performance is when the signal-to-noise ratio is less than 1.Not good.For the Full Vector Machine Learning algorithm,the problem of identifying inaccuracy and the redundancy of input features when the noise signal is strong,the EEMD algorithm is used to filter the noise signal,and the genetic algorithm GA is used to optimize the feature for the redundant feature to remove unnecessary redundancy.For extra features,a genetic algorithm combined with correlation feature selection for rotor imbalance fault feature optimization method is proposed to filter the features,reduce the redundancy between features,and significantly improve the operation efficiency.At the same time,in order to make up for the problem of decline in diagnostic accuracy,a genetic algorithm weighted ELM imbalanced fault feature optimization method is proposed.By adding weight coefficients,the error between feature quantities due to excessive size difference is reduced.Experimental results show that this method can significantly improve the accuracy and calculation efficiency of imbalanced faults.