Early Detection of Coil Failure in Solenoid Valves
Jesper Liniger, Søren Stubkier, Mohsen Soltani, Henrik C. Pedersen
Abstract
This article describes the development and application of a model-based scheme for detecting the early signs of coil failure in solenoid valves. Contrary to other works, the proposed method simply isolates the fault-induced change in the coil resistance from thermal effects and provides a cost-effective solution with no comprehensive equipment demand. The model-based detection method employs a simple thermal model of the solenoid and an extended Kalman filter (EKF) for generating coil current residuals. The EKF utilizes measurements of coil voltage, coil current, ambient, and fluid temperature. The CUmulative SUM of the current residual is found to be sensitive to the early signs of coil failure. An experimental setup is utilized for evaluating the robustness of the detection method in the event of changing ambient temperature, convection, and fluid flow conditions. Results show that the method is able to detect the early signs of coil failure, and robust behavior and high detection probabilities are achieved for realistic fault sizes. The realistic fault sizes and the number of fault instances before failure are determined through extensive accelerated tests, where solenoids are operated at high temperature until failure.