| Planetary gearbox has many advantages such as large transmission ratio,strong carrying capacity,and high transmission efficiency.Therefore,it is widely used in mechanical transmission systems in wind power generation,aviation,shipbuilding,metallurgy,and other industries.Since planetary gearboxes mostly work under complex working conditions and harsh environments,they often lead to serious wear and fatigue cracks on key components such as the sun gear and planetary gear.If the planetary gearbox fails,and no measures are taken in time,it may cause the transmission system to fail,and even more serious consequences.Therefore,it is of great practical significance to carry out fault diagnosis for this part.In this subject,a new set of fault diagnosis model has been developed for the problems of planetary gearbox fault feature extraction,fault feature dimensionality reduction and fault identification classification.The model mainly includes the following three steps: First,the time-shift weighted multiscale fuzzy entropy is used to fully mine the fault information of the planetary gearbox signal and construct a fault feature set with higher dimensions.Then,the improved generalized regularized coplanar discriminant analysis is used to reduce the dimensionality of the high-dimensional fault feature set to obtain a low-dimensional,easyto-identify feature set.Finally,the coyote optimization algorithm is used to optimize the support vector machine to classify the faults of the feature set after dimensionality reduction.And the effectiveness of the method was verified through experiments.The main tasks of the subject are as follows:(1)In order to fully dig out the fault information of the vibration signal of the planetary gearbox,the multiscale fuzzy entropy is combined with the ideas of time shifting and weighting,and the time-shift weighted multiscale fuzzy entropy is developed to extract the fault features of planetary gears.The effectiveness of this method is verified by simulation and experimental data of planetary gearbox.In addition,it is compared with the existing multiscale fuzzy entropy and time-shift multiscale fuzzy entropy.(2)In order to avoid the redundant interference of the features proposed by TSWMFE,the IGRCDA algorithm is used to screen high-dimensional features to obtain a sensitive lowdimensional feature set.The effectiveness of the method is verified by using the experimental data of planetary gearbox,and the dimensionality reduction effect of IGRCDA algorithm and common dimensionality reduction methods are compared in details.(3)In order to better classify and identify the fault category of planetary gearbox,the proposed COA-SVM method is used to identify and classify the low dimensional fault feature set of planetary gearbox,and the advantages and effectiveness of this method are verified by simulation experiments and experimental data of planetary gearbox.(4)Based on the above methods,a planetary gearbox fault diagnosis model based on TSWMFE,IGRCDA and COA-SVM is developed.The effectiveness and generalization performance of this method are verified by the fault diagnosis experiments of planetary gearbox and rolling bearing testing machine. |
