Publication Details
Abstract
This paper examines the problem of improving the reliability of adaptive control in water supply and heating systems under conditions of noise, measurement asynchronousness, and residual estimation errors. A protection mechanism is proposed between the estimation unit and the fuzzy adaptation unit for the PID controller coefficients. After normalizing the current state and parameter estimates, a vector of invalidity indicators is generated. If the information is correct, the controller coefficients are determined using the Mamdani fuzzy inference mechanism. If the admissibility limit is violated, the adaptation is disabled, and control continues with fixed coefficients. This approach eliminates erroneous controller reconfiguration and maintains the continuity of the computational cycle.