A demand side prediction method for persistent scheduled power-cuts in developing countries

Irregularity of scheduled power-cut induces consumer´s inefficient activity and therefore the consumer expects to know power-cut occurrence in advance. This paper proposes 3-heuristics which enable consumers to predict starttime of power-cuts for next day: SBP (Start-time of power-cut Based Prediction) using historical power-cut start-time data, FBP (Frequency Based Prediction) using historical frequency fluctuation data and ADSP (Adaptive Data Selection Prediction) which is a hybrid exploiting advantages of SBP/FBP with appropriate data period for overcoming changes of power-cut pattern. The evaluation results with power data of Chennai India showed that SBP totally achieved higher prediction success ratio than FBP and SBP has the advantage on regular power-cut pattern instead FBP has the same on the irregulars. Data period to maximize prediction success ratio depends on power-cut pattern as for SBP/FBP. The highest prediction success ratio was marked by ADSP which adaptively combined start-time/frequency data and determined data period on the basis of power-cuts pattern.