Modeling, simulation & fault detection in excavators with time-varying loading

Excavators experience large variable forces due to the interaction between the ground and the machine. The time-varying nature of these interaction forces causes stresses in machine components and may result in damage. Such faults can lead to a complete system failure, which delays the excavating process and incurs maintenance costs and opportunity costs of lost production; however, this situation can be avoided if faulty components are identified and isolated in their early stages. The focus of this study is on the machine-ground interaction period within a digging cycle, when the bucket is engaged with the material, cutting and pushing it forward. A laboratory-scale test rig has been developed and commissioned to emulate the time-varying behavior of the excavator while cutting and pushing soil. The rig is equipped with sensors that acquire data from different machine parts to monitor the process as well as machine condition. The whole system is modeled as a rigid-body, lumped-parameter, dynamical system. This model is used to represent the nominal plant so that the response of the physical system can be compared to the response from the model. Sensor signals are processed and features are extracted that relate to fault signatures for scenarios in which faults are artificially induced in the machine. Results are presented for fault signatures that were successfully identified for certain types of fault in the rig components. The eventual goal of the work is to develop a set of fault signatures that can be used to identify and isolate the fault in machine parts during time-varying operation.