The use of adaptive algorithms with dynamic delay to improve the reliability of control systems for construction and road robotic complexes

Abstract

The article discusses the problem of ensuring the reliability of control systems for construction and road robotic complexes (SDCS) in conditions of degradation of the quality of sensory data – partial loss, noise, time shifts and systematic shifts. An approach based on the use of adaptive algorithms with dynamic delay, developed as part of a dissertation study on robust fault tolerance analytics systems, is proposed. Such algorithms allow you to dynamically adjust the depth of the time window of the analysis depending on the current level of uncertainty of the input data, which increases the stability of the closed control loop. The mathematical formalization of the algorithm based on a modified recurrent equation with variable memory is given, and the metric of the adaptive robustness coefficient is introduced. The experimental verification was performed on a stand simulating an autonomous paver with inertial navigation and local GNSS failures. The results showed that the application of the proposed algorithm makes it possible to reduce the average trajectory deviation by 31% and reduce the variance of the control effect by 27% while simulating 20% omissions in RTK-GNSS data. Under conditions of prolonged GNSS failure (more than 8 s), the system remained operational without an emergency shutdown, unlike the classic PID controller with a fixed filtering window. The results obtained confirm the effectiveness of the proposed approach for improving the reliability and fault tolerance of robotic road construction systems.