APPLICATION OF DIAGNOSTIC SYSTEMS TO DETECT FAULTS IN TECHNOLOGICAL PROCESS CONTROL SYSTEMS

Abstract

Conducting online diagnostics of industrial technological processes is an effective procedure for enhancing the functional safety of the process itself and reducing economic losses caused by failures or equipment malfunctions, which increase downtime for repairs. Early recognition of emerging faults requires process operators to take appropriate actions that could allow for relatively quick and successful diagnostics to prevent failures. Thus, the blocking and safety measures foreseen and implemented in the control systems, with the right approach to diagnostics, will not be activated, and therefore, the process is not interrupted.
Thus, diagnostic procedures guarantee, among other things, significant savings in the execution of technological processes. Accurate and high-quality diagnostics allow for isolating faults and applying appropriate solutions and repair actions. The accuracy of diagnostics can be determined by adding up the number of faults detected in each elementary block. The fewer faults collected in the elementary block, the more accurate the diagnostic procedure. Diagnostic tests reveal a set of characteristic symptoms associated with a fault or a set of faults. These can be defined as a pattern of specific values (symptoms) of diagnostic signals, as well as certain sequences (orders) of these signals.
The study analyzed the application of diagnostic systems for detecting faults in process control systems. The modeling of the technological process of turbine operation for electricity production was conducted. The modeling was carried out using the MATLAB/Simulink application package. Linear ascending signal functions were used as input signals. The system components included modules of the compressor, turbine, controller, and burner. Obtained transient process graphs and residual functions were used to detect faults. Fault detection has a time delay, which can be reduced by implementing additional algorithms for analyzing residual functions.
The obtained results of fault detection in various sensors are sufficient for industrial diagnostic applications, considering also that the minimum number of detected faults can be reduced if additional algorithms for residual analysis are applied.