The Research And Application Of Fault Diagnosis Of Ship Rotation Machinery Based On Data Driven Methods

Rotating machinery is one of the most widely used mechanical equipment in the industrial field, which is also an important part of the ship system. Due to the process defects, improper installation and the long-term adverse effects of the working environment, the faults of the rotating machinery in the ship system can occur frequently, which can degrade the performance of shipping, and even lead to the accidents or casualties. At present, the most used methods for fault diagnosis of the ship system dependent on the experts’ experience, which can bring to the high misdiagnosis. Therefore, it is of both theoretical importance and practical significance to study the fault diagnosis method for rotating machinery.In this thesis, the data driven based fault diagnosis methods are investigated for the ship rotating machinery. At the beginning, an overview of fau lt diagnosis methods is introduced, especially the development of the data driven based fault diagnosis methods for rotating machinery. Also, the advantages and shortcomings of the existing results are analyzed.To use the date-driven based fault diagnosis, the amount of history data is obtained through the real-time monitoring and experiments. The data is collected during the shipping task, which includes data under normal condition and faulty condition. Experimental data is derived from the faul t simulation platform of rotating machinery, which contains the data of common faults and data under normal working condition.The application of empirical mode decomposition method in rotating machinery fault diagnosis is studied. The traditional denoise method is improved, and sample entropy theory is combined with empirical mode decomposition method to realize fault diagnosis of sh ip rotating machinery. According to the principle of the correlation of component, the genetic algorithm is put forward to combined with the denoising method of empirical mode decomposition threshold value. The result is the correlation between denoising signal and the original signal are enhanced. By combining empirical mode decomposition with sample entropy theory, the fault diagnosis of shi p rotating machinery is realized.Based on texture analysis, a new fault diagnosis method for ship rotating machinery is proposed. The rotating machinery vibration signal is converted to t he gray level image. The texture image analysis methods is presented to achieve the fault identification. The feasibility and practicability of the method are verified by theory and experiments.Finally, based on Matlab graphical, the user interface of the fault diagnosis system for ship rotating machinery is de veloped. The development environment and t he overall structure of the system are introduced with the implementation methods of each module being described in detail, and the validity and practicability of the system being tested.