Research on the seagull optimization algorithm-based convolutional neural network rolling bearing fault diagnosis method

Abstract

It is crucial to understand the rolling bearing fault diagnosis procedure since rolling bearings are frequently used in rotating machinery and if a failure occurs, it will interfere with the proper operation of the entire piece of machinery or piece of equipment. Deep learning is increasingly being used in mechanical fault diagnosis, with convolutional neural networks(CNN) being the most common type. In recent years, the rapid growth of artificial intelligence has caused fault diagnosis methods to evolve as well. A CNN fault diagnostic approach based on seagull optimization algorithm (SOA) is suggested and applied to the fault diagnosis of rolling bearings in order to address the issues with convolutional neural networks, such as high data requirements, unstable gradients, and optimization challenges. The seagull optimization approach is used to design a deep learning model by selecting the structural hyperparameters in the CNN model as efficiently as possible. The accelerated life experimental dataset for rolling bearings is trained using the optimized CNN model. The training outcomes of the CNN model on the dataset before and after optimization are compared, and the findings demonstrate that the suggested method has a superior optimization effect. As can be seen, the seagull optimization algorithm and CNN can be organically combined and applied to the fault diagnosis method for rolling bearings.