Development of a method for diagnosing and forecasting power supply systems for mining enterprises

The object of the study is the power supply system of the mining enterprise Nova-Zinc LLP, located in Central region of Republic of Kazakhstan, which specializes in the extraction and enrichment of lead-zinc ores. Reliable power supply in mining is challenged by heavy mechanical loads, vibration, and severe weather; this case study is used only as an example of conditions common in many mining regions worldwide. This study proposes a diagnostic-and-forecasting method for distribution power systems based on routinely available operational records and climatic indicators. The method was tested using outage data from 2020–2024. Using least squares, a multivariate regression model was obtained for feeder emergency outage duration as a function of cable damage (F5), transformer failure (F6), and the climatic factor (Climate). The model is significant overall (F-test p < 0.01) and explains 68.7% of downtime variation (R2 = 0.687); residual diagnostics indicate normality and no autocorrelation. The average marginal effects are 7.561 h for cable failures, 3.314 h for transformer failures, and 2.418 h for climatic impacts, highlighting cables as the dominant driver of prolonged outages. To assess energy performance, a separate model was built for the loss share in the power system as a function of outage duration, phase clashing (F3), and Climate. This loss model has low explanatory power (R2 = 0.2013) and non-significant factor coefficients (p > 0.05). Finally, bivariate regressions show that Climate strongly affects phase clashing (F3) (R2 = 0.793) and moderately affects ground faults (F1) and insulator chipping (F2) (R2 = 0.533 each). The proposed method supports maintenance prioritization, climate-mitigation measures, and continuous updating as new outage records are added, strengthening decision-making and system robustness