Multi-class fault detection in electro-hydraulic servo systems using support vector machines

Electro-hydraulic servo systems are used as actuators for positioning and force exertion in heavy-duty and relatively precise industrial applications. In addition, to achieve the best performance of the system, all of the components must work properly. These systems are often subjected to some problematic issues such as leakage, fluid property variation, and actuator seal damage. These problems are called “faults” which degrade the control performance. This paper presents a novel approach for fault detection in electro-hydraulic servo systems using pressure signals analysis. The leakage fault influences many aspects of the fluid pressure signals. The proposed method is based on pressure signals analysis in one of the cylinder chambers in response to periodic square inputs to a proportional valve. For this purpose, propounded features are extracted from supply pressure signals. These features are considered as height, width, and location of peaks. In the proposed scheme, a support vector machine classifier has been exploited to classify systems into multiple classes including faultless systems and systems with different levels of fault. Moreover, this method does not need the model of actuator or leakage. The performance of proposed fault detection scheme was verified through experimental results.