HYBRID FFT-MCSA APPROACH FOR EARLY FAULT DETECTION IN PUMP ELECTRIC DRIVE SYSTEMS: A CASE STUDY FROM UZBEKISTAN IRRIGATION NETWORKS
Abstract
ABSTRACT. Uzbekistan operates 1,698 pumping stations that deliver 32.5 billion m³ of water yearly. These stations consume 7.2 billion kWh, representing 16-21% of the country’s total electricity. The problem? Most pumps are 50+ years old. Their efficiency dropped from 85-92% down to 50-70%. Water losses reach 30-40% in some systems. We looked at whether combining two diagnostic methods - Fast Fourier Transform for vibration and Motor Current Signature Analysis for electrical signals - could catch faults earlier than traditional methods. Using data from the Amu-Bukhara cascade where pumps handle sandy water (1.25 mg/m³) and extreme heat (+60°C), we built simulations that mirror real conditions. Our hybrid method detected bearing problems with 94.7% accuracy, rotor bar breaks at 91.2%, and eccentricity issues at 89.5%. That’s much better than using FFT alone (78.3%) or MCSA alone (72.6%). False alarms dropped 63% compared to standard threshold monitoring. If adopted across Uzbekistan’s network, this approach could cut energy use by 30-40% and reduce unexpected shutdowns by 70-80%. The system works without constant internet - important for remote pump stations. These factors matter because Presidential Decree PQ-265 mandates modernizing 518 aggregates by 2028, and 200 stations are transitioning to public-private partnerships requiring reliable monitoring. KEYWORDS: Condition monitoring, Electric drive, FFT analysis, MCSA, Pump diagnostics, Spectral methods.