Modular Signal-Based Condition Monitoring of a Hydraulic Servo-System

Abstract

This paper investigates the use of the standard available measured signals in a hydraulic system (the pressure signals of the two hydraulic cylinder chambers) to develop a modular signal-based monitoring system. The developed system has to detect and classify the typical faults that can occur in the hydraulic systems, e.g. internal leakage, dynamic friction, or undissolved air in the hydraulic cylinder chambers. The proposed monitoring system consists of different detection modules as well as a decision fusion module. Short Time Fourier Transform (STFT) and other statistical feature extraction tools, such as arithmetic mean value, are realized to extract the relevant features of the pressure signals. The extracted features undergo a further filtration process to classify the possible occurred faults. A decisionfusion module is used to provide a final decision about the system state as well as to identify the type of the possible occurred fault. The developed approach is able to detect the state of the hydraulic system under all considered operational conditions. Furthermore it is able to identify the type of most of the possible occurred faults.