The study of South African distribution (Dx.) network’s load forecasting using recent and state of the art AI (machine learning, deep learning and ensemble deep learning) techniques, is limited. The impact of weather parameters on load forecasting performance of AI techniques in forecasting South African large power users is not well understood. This paper proposes a novel distribution network load forecasting system. The paper further introduces deep learning and ensemble deep learning techniques in forecasting the power consumption of large South African power users. The paper introduces these techniques through an investigation of their performance against that off state of the art machine learning techniques, ANFIS and OP-ELM. The impact of temperature on the performance of these techniques is also investigated. This investigation was conducted on three case studies, with three different industrial large power consumer loads. LSTM-RNN proved to be a more efficient load forecasting technique for the proposed load forecasting system, achieving the lowest load forecasting error in all three case studies. Ensembles of LSTM were found to overall achieve lower errors than the individual techniques’ models. This improvement was less than 1%. The inclusion of temperature was found to generally improve the load forecasting performance of ML and DL techniques’ models.

Unplanned power plant failures have been seen to be a major cause of power shortages, and thus customer power cuts, in the South African power grid. These failures are measured as the unplanned capability loss factor (UCLF). The study of South Africa’s UCLF is almost non-existent. Parameters that affect the future UCLF are, thus, still not well understood, making it challenging to forecast when power shortages may be experienced. This paper presents a novel study of South African UCLF forecasting using state-of-the-art deep learning techniques. The study further introduces a novel deep learning ensemble South African UCLF forecasting system. The performance of three of the best recent forecasting techniques, namely, long short-term memory recurrent neural network (LSTM-RNN), deep belief network (DBN), and optimally pruned extreme learning machines (OP-ELM), as well as their aggregated ensembles, are investigated for South African UCLF forecasting. The impact of three key parameters (installed capacity, demand, and planned capability loss factor) on the future UCLF is investigated. The results showed that the exclusion of installed capacity in the LSTM-RNN, DBN, OP-ELM, and ensemble models doubled the UCLF forecasting error. It was also found that an ensemble model of two LSTM-RNN models achieved the lowest errors with a symmetric mean absolute percentage error (sMAPE) of 6.43%, mean absolute error (MAE) of 7.36%, and root-mean-square error (RMSE) of 9.21%. LSTM-RNN also achieved the lowest errors amongst the individual models.