A practical system for online diagnosis of control valve faults

In this paper, we present a new control valve monitoring system using the nonparametric statistical hypothesis tests for the diagnosis of backlash, deadband, leakage, and blocking, four common faults found in many control systems. To make our system practical and inexpensive, we utilize the sensor measurements available in most control systems. For the classification of individual faults, we extract the geometric features from the measurement signals and detect the faults from their temporal trends. Based on the features, the statistical hypothesis tests are applied to discriminate the faults.