Research On Gearbox Fault Diagnosis Based On Machine Learning

With the rapid development of high-speed railways,the number of track equipment in operation has increased dramatically.As one of the key components in the track equipment drive system,the gearbox failure or failure will not only bring significant economic losses,but also cause huge hidden dangers to people's life and the smooth and normal operation of the track equipment.The rail transportation industry has adversely affected.The structure of the gear box is more complicated,and it is often a component of equipment with frequent failures.Especially in harsh environments,typical faults such as broken teeth,pitting,tooth surface wear,and shaft bending are often prone to occur.And the gear box's running state directly affects the performance of the entire machine,so the fault diagnosis technology of the gear box has far-reaching significance.Now that artificial intelligence technology is booming,gearbox fault diagnosis methods have gradually transitioned from the original manual diagnosis to the current intelligent diagnosis.Manual fault diagnosis requires a high level of professionalism of the technical staff.The technical staff only evaluates the running state of the gearbox based on their own experience and knowledge,making the diagnostic results inaccurate.Intelligent diagnosis combines fault diagnosis with artificial intelligence,machine learning,and other fields.It obtains vibration signals through sensors,and digs out information that can reflect the fault characteristics of gears,and then passes these fault characteristic information to the machine.The purpose is to pass these already The obtained relevant data or relevant parameters allow the machine to learn and discover the correlation between the data and the result,and obtain relevant experience and knowledge,and then continuously enrich the machine's own knowledge reserve and improve its performance,so as to make the machine as similar to human as possible.Ability.Then the trained machine is applied to the gearbox fault diagnosis,instead of the traditional process of artificial analysis and judgment,so that the gearbox fault diagnosis has accuracy and objectivity,thereby overcoming the shortage of manual diagnosis.Therefore,the paper will focus on signal noise reduction,feature parameter extraction analysis and classification based on machine learning methods for fault vibration signal processing methods.The proposed method will be applied to the fault diagnosis of gearboxes.The signal is the research object,and the gear failure signal is comprehensively analyzed by methods such as time domain analysis,frequency domain analysis,and wavelet transform,and it is converted into the form of feature vectors.Finally,using the matlab platform,a set of gearbox fault diagnosis programs based on vibration signals is constructed.The program can effectively extract the fault characteristic parameters of the gearbox and diagnose and identify the fault type of the gearbox.