Short-term load forecasting (STLF) is necessary for system operators; however, its difficulty has been increasing since distributed resources, particularly behind-the-meter (BTM) PV resources, have been introduced to power systems. This study proposes a framework for STLF for holidays considering the four major factors that affect the net load profiles —calendar, trend, weather, and BTM PV. The target holiday is first paired with historical holidays following its calendar factor, which are defined as “similar days.” Subsequently, in terms of the remaining three factors, the differences between the historical holidays and target holidays are calculated, and their effects on load differences (factor-induced load differences) are quantified and reflected. Finally, for each pair, the modified load profiles are generated and combined to obtain a daily load profile of the target holiday. The proposed framework was implemented on a case study of Korean national holidays, and its forecasting accuracy was compared with conventional forecasting methods. The accuracy metrics show that the proposed framework outperforms conventional methods. The results suggest that the proposed framework can be applied to STLF for holidays to improve forecasting accuracy.

Advanced metering infrastructure (AMI) is spreading to households in some countries, and could be a source for forecasting the residential electric demand. However, load forecasting of a single household is still a fairly challenging topic because of the high volatility and uncertainty of the electric demand of households. Moreover, there is a limitation in the use of historical load data because of a change in house ownership, change in lifestyle, integration of new electric devices, and so on. The paper proposes a novel method to forecast the electricity loads of single residential households. The proposed forecasting method is based on convolution neural networks (CNNs) combined with a data-augmentation technique, which can artificially enlarge the training data. This method can address issues caused by a lack of historical data and improve the accuracy of residential load forecasting. Simulation results illustrate the validation and efficacy of the proposed method.

As photovoltaic (PV) power plants are an essential component of modern smart grids, the PV generation forecasting of such plants has recently been gaining interest. The forecasting results of PV power often suffer from large errors because of unusual weather conditions. In a learning-based forecasting model, the forecasting accuracy can be enhanced by using carefully selected data for training rather than all the data without any screening. That is, using a training set that only contains information obtained from similar days can help enhance the accuracy of learning-based PV forecasting. This paper proposes a forecasting method for small-scale PV generation. This method is based on long short-term memory; further, it detects similar days considering the different impacts of weather variables on PV power according to the day. This method can address issues caused by unnecessary learning from non-similar historical days. The simulation results demonstrate that the proposed method exhibits better performance than do existing similar day detection methods.

Jeju Island announced the “Carbon Free Island (CFI) Plan by 2030” in 2012. This plan aims to replace conventional generators with distributed energy resources (DERs) up to a level of 70% by 2030. Akin to Jeju Island, as DERs have been expanded in islanded power systems, variable renewable energy (VRE) has become a significant component of DERs. However, VRE curtailment can occur to meet power balance, and VRE curtailment generally causes energy waste and low efficiency, so it should be minimized. This paper first presents a systematic procedure for estimating the annual VRE curtailment for the stable operation of the islanded power systems. In this procedure, the VRE curtailment is estimated based on the power demand, the grid interconnection, the capacity factor of VRE, and conventional generators in the base year. Next, through the analysis of the hourly net load profile for the year in which the VRE curtailment is expected to occur, a procedure was proposed to find the season and hour when VRE curtailment occurs the most. It could be applied to revised Time-of-Use (ToU) tariff rates as the most cost-effective mitigation method of VRE curtailment on the retail market-side. Finally, price elasticity of electricity demand was presented for applying the revised ToU tariff rate scenarios in a specific season and hour, which found that VRE curtailment occurred the most. Considering self- and cross-price elasticity of electricity, revised ToU tariff rate scenarios were used in a case study on Jeju Island. Eventually, it was confirmed that VRE curtailment could be mitigated when the revised ToU tariff rates were applied, considering the price elasticity of demand.

Occasionally, wind curtailments may be required to avoid an oversupply when wind power, together with the minimum conventional generation, exceed load. By curtailing wind power, the forecast uncertainty and short-term variations in wind power can be mitigated so that a lower spinning reserve is sufficient to maintain the operational security of a power system. Additionally, the electric vehicle (EV) charging load can relieve the oversupply of wind power generation and avoid uneconomical wind power curtailments. This paper presents a stochastic generation scheduling method to ensure the operation security against wind power variation as well as against forecast uncertainty considering the stochastic EV charging load. In the paper, the short-term variations of wind power that are mitigated by the wind curtailment are investigated, and incorporated into a generation scheduling problem as the mixed-integer program (MIP) forms. Numerical results are also presented in order to demonstrate the effectiveness of the proposed method.

Power systems face new challenges owing to the rapid expansion of photovoltaic (PV) power. Recent trends indicate a decline in demand coupled with extensive solar power utilization during the spring season, resulting in increased behind-the-meter (BTM) PV generation, subsequently reducing the net demand. This reduction introduces stability issues in the power grid. Accurate minimum demand forecasting in the spring season is pivotal for grid stability and directly influences essential power system security services. However, minimum demand forecasting in the spring season is challenging owing to the fluctuating net demand affecting BTM PV generation, decreased temperature sensitivity in heating and cooling, diverse consumer behavior during holidays, and shifting industrial activities post-COVID-19. These complexities significantly hinder the forecasting accuracy. In this paper, a medium-term minimum demand forecasting framework is proposed, which considers the various features that affect the load profiles: temperature, economic indicators, seasonal changes, and BTM PV capacity. The optimal input variables are determined annually using feature selection. In particular, the proposed framework improves the forecasting accuracy by using forecasted energy consumption information. Furthermore, the proposed parallel long short-term memory (LSTM)-multi layer perception (MLP) model extracts annual temporal characteristics, learns nonlinear relationships, and captures patterns. A validation using historical load demand data demonstrates its superiority over conventional methods through a comparative analysis.