Research On Fault Diagnosis Approaches Of Excavator's Hydraulic System

With the automation progress of the excavator, fault diagnosis of hydraulic system has become one of the key technologies for the modern excavator. The development of such techniques has great significance in improving reliability and working efficiency of the excavator. Based on theory research, simulation modeling and experiment, fault detection and diagnosis approaches for the excavator's hydraulic system are systematically studied in this paper. The main contents are as follows:1. On the basis of study on the fault mode and the fault mechanism of the hydraulic system, model parameters of the modular hydraulic components of the excavator are properly corresponded to the specific fault mode and the fault mechanism. As a result, a fault diagnosis research strategy of the excavator's hydraulic system is proposed, which is taken as the guidance of fault diagnosis research, simulation testing and experiment testing.2. With the study of the dynamic PCA, a dynamic principal component model of the hydraulic subsystem loop of the excavator is established. Combined with multivariate statistical testing, a fault detection approach of the excavator's hydraulic system based on dynamic PCA is proposed. In addition, taking account into the practical application, an online modeling method and an online fault detection method are proposed.3. A fuzzified ARX model, called FARX model, is proposed for the fault diagnosis of the excavator's hydraulic system.(1) With the study of the FARX model structure and corresponding fault feature extraction, a fault feature extraction approach of the excavator's hydraulic system based on FARX model is proposed.(2) A diagnosis approach based on FARX model and FCM is proposed. On the basis of target fault features, FCM is served as fault classifier and the output of the FCM is the result of diagnosis.(3) A fault diagnosis approach of the excavator's hydraulic system based on FARX model and RBF network is proposed. Before diagnosis, the RBF network is firstly trained with target fault features and a fault classifier is obtained. The output of this fault classifier is the result of diagnosis.4. The dynamic GRNN model with multi-model diagnosis is applied to the excavator's hydraulic system.(1) A dynamic GRNN model is proposed by introducing the global feedback to the GRNN. The structure of dynamic GRNN model and a corresponding multi-step prediction method are studied.(2) A fault detection approach of the excavator's hydraulic based on dynamic GRNN model is proposed, in which the sum of residuals' square is developed to test model's residual. Incorporating with multi-model diagnosis, a diagnosis approach based on dynamic GRNN model is proposed for the excavator's hydraulic system.5. For testing the diagnosis approaches, a simulation model of the SWE50 manipulator and corresponding hydraulic system are established in AMESim simulation environment. Various fault cases are simulated. In addition, with the SWE50 experimental excavator, sample data is generated from five single fault cases including piston wear, spool strake, spool wear, loose slipper and port plate wear, and three multiple fault cases including spool wear plus spool stroke, spool wear plus piston wear and loose slipper plus port plate wear. The detection and diagnosis approaches are tested with sample data from simulation and experiment. The results show that the fault detection and diagnosis approaches could effectively applied to the excavator's hydraulic system.