Adaptive control algorithms for electric drives of irrigation pumping units in the context of agricultural digitalization: bench studies and on-farm verification

Abstract

The article presents the development and experimental study of adaptive control algorithms for electric drives of irrigation pumping units based on the analysis of the wattmeter diagram, i.e. the time dependence of the active power consumed by the electric drive. The algorithm provides automatic formation of a load reference at start-up (a preliminary reference after 30 minutes and a final one after 60 minutes of steady operation), tolerance control of the “less — normal — more” type relative to the reference, and a “two out of three” majority voting logic that excludes false trips. The study was carried out on a specialized test bench of Kazan State Power Engineering University with a loading system based on an induction motor with direct torque control (DTC) and a digital PID torque controller (overshoot below 10%, settling time 0.1 s); the loading profiles were formed from wattmeter diagrams recorded at an operating drip irrigation pumping station. In 140 bench experiments, the normal mode, clogging of filters and drip lines, dry running and cavitation were reproduced; the probability of correct mode identification amounted to 96.4%, the mean detection time was 14 ± 3 s for dry running and 85 ± 12 s for cavitation. Onfarm verification at the pumping station of an agricultural enterprise of the Republic of Tatarstan (2025 irrigation season, 720 h of operation) confirmed the correctness of the bench tuning without readjustment of the algorithms: six episodes of progressive clogging and two episodes of dry running were identified in a timely manner and three false disturbances were filtered out. The reduction in the specific energy consumption of the station by 11.3% recorded over the season refers to the transition from throttling to frequency control of water supply as a whole and represents a calculated estimate valid provided that the hydraulic conditions of water supply are preserved; the contribution of the adaptive algorithms to this result was not evaluated separately. The possibility of integrating the algorithms into the digital platform of the enterprise via Modbus TCP and MQTT protocols as an element of agricultural digitalization is shown.