The SP Group has more than 11,000 substations delivering electricity to Singapore's industrial, commercial, and residential consumers. Part of the work includes checking for insulation degradation in substations, which is highly labour-intensive as operators have to use a handheld device to extract patterns and waveforms measured by the equipment at the substations. This is followed by manually scanning through the pattern and waveform data for potential abnormalities. Our SIT researchers have devised a machine learning platform that analyses the waveform data to flag out suspected partial discharge to make the process more efficient. Using a unique algorithm called adaptive clustering, this removes noise of varying levels to accurately isolate partial discharge data points for analysis.