No abstract

With the increase of electrical equipment, the hidden danger of its failure also rises. Aiming at the problem that the classification accuracy of the basic traditional classification algorithm is not high when judging whether the electrical failure occurs, this paper proposes a Gray Wolf Optimization Support Vector Machine model to improve the recognition rate of electrical diagnosis, referred to as GWO-SVM. We first collected the most common linear (incandescent lamps) and non-linear (microwave ovens) household electrical appliances in normal working and arc fault waveform signals in real life, and then carried out frequency domain feature extraction. Finally, compared with no-optimized SVM and BP neural network, we employed the gray wolf optimization algorithm to optimize support vector machine, and the accuracy of GWO-SVM reaches 90%.

Rolling bearing components are broadly utilized in major mechanical fabricating businesses and guaranteeing the secure and steady operation of rolling heading could be a basic necessity of the fabricating prepare. Engineering for intelligent manufacturing has grown significantly in importance during the past several years in the manufacturing sector. The method for identifying mechanical faults based on “frequency domain analysis plus intelligent model” has developed rapidly. In this study, methods such as envelope spectrum analysis and spectral kurtosis are applied to process and analyze fault data to improve the service life of rolling bearing production equipment. In addition, we perform grid search tuning of the hyperparameters in spectral kurtosis, enabling faster frequency band selection for envelope spectral bandpass filtering.

No abstract

No abstract