This paper presents the analysis of an on-line fault detection method for proportional directional hydraulic valves applied on speed governors of hydroelectric power plants. This application area is very sensitive for unexpected maintenance or long duration stops since most of power plants are interconnected on an electrical power grid. A plant stop must be programmed previously and approved by a regulatory agency. Consequently, the implementation of a fault detection and monitoring system can reduce maintenance and operational costs as well as safety risks of equipment and operators. The developed method is based on monitoring both the valve supply current and spool position related to an input control signal. Therefore, it is applicable for so called servoproportional valves, it means, those including spool position measurement and with embedded electronics. Valve static and dynamic behaviour depends on spool friction, flow forces, solenoid force and valve closed loop controller. Such characteristics are influenced by the spool size, overlapping and manufacturing tolerances. In this paper, the effectiveness of the method to monitor and detected faults in valves with different sizes and constructive parameters is demonstrated experimentally using five proportional valves.
A Partial Robust M-Regression-Based Prediction and Fault Detection Method