Short term load forecasting and the effect of temperature at the low voltage level

Haben, Stephen; Giasemidis, Georgios; Ziel, Florian; Arora, Siddharth

doi:10.1016/j.ijforecast.2018.10.007

Cited by 72 publications

(63 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A comparative study between several benchmarks e.g. an autoregressive model and Holt-Winters-Taylor (HWT) are compared in terms of point and quantile forecasting in [30]. As deep learning-based quantile forecasting, LSTM and CNN are extended via a pinball loss function to extract the quantile in [31] and [32], respectively.…”

Section: B Brief Literature Reviewmentioning

confidence: 99%

Deep-Based Conditional Probability Density Function Forecasting of Residential Loads

Afrasiabi

Mohammadi

Rastegar

et al. 2020

IEEE Trans. Smart Grid

111

View full text Add to dashboard Cite

This paper proposes a direct model for conditional probability density forecasting of residential loads, based on a deep mixture network. Probabilistic residential load forecasting can provide comprehensive information about future uncertainties in demand. An end-to-end composite model comprising convolution neural networks (CNNs) and gated recurrent unit (GRU) is designed for probabilistic residential load forecasting. Then, the designed deep model is merged into a mixture density network (MDN) to directly predict probability density functions (PDFs). In addition, several techniques, including adversarial training, are presented to formulate a new loss function in the direct probabilistic residential load forecasting (PRLF) model. Several state-of-the-art deep and shallow forecasting models are also presented in order to compare the results. Furthermore, the effectiveness of the proposed deep mixture model in characterizing predicted PDFs is demonstrated through comparison with kernel density estimation, Monte Carlo dropout, a combined probabilistic load forecasting method and the proposed MDN without adversarial training.Index Terms-Residential load forecasting, conditional probabilistic load forecasting, deep mixture network, convolutional neural network, gated recurrent unit

show abstract

Section: B Brief Literature Reviewmentioning

confidence: 99%

Deep-Based Conditional Probability Density Function Forecasting of Residential Loads

Afrasiabi

Mohammadi

Rastegar

et al. 2020

IEEE Trans. Smart Grid

111

View full text Add to dashboard Cite

show abstract

“…In load forecasting, both terms showed high relevance; see, e.g., [10][11][12][13]. Sometimes, models with many seasonal and autoregressive components performed even very well in short-term forecasting; see, e.g., [14].…”

Section: Proposed Nowcasting Modelmentioning

confidence: 99%

Load Nowcasting: Predicting Actuals with Limited Data

Ziel

2020

Energies

Self Cite

View full text Add to dashboard Cite

We introduce the problem of load nowcasting to the energy forecasting literature. The recent load of the objective area is predicted based on limited available metering data within this area. Thus, slightly different from load forecasting, we are predicting the recent past using limited available metering data from the supply side of the system. Next, to an industry benchmark model, we introduce multiple high-dimensional models for providing more accurate predictions. They evaluate metered interconnector and generation unit data of different types like wind and solar power, storages, and nuclear and fossil power plants. Additionally, we augment the model by seasonal and autoregressive components to improve the nowcasting performance. We consider multiple estimation techniques based on the lassoand ridge and study the impact of the choice of the training/calibration period. The methodology is applied to a European TSO dataset from 2014 to 2019. The overall results show that in comparison to the industry benchmark, an accuracy improvement in terms of MAE and RMSE of about 60% is achieved. The best model is based on the ridge estimator and uses a specific non-standard shrinkage target. Due to the linear model structure, we can easily interpret the model output.

show abstract

“…Point load forecasts refer to forecasts which give a single, usually mean, value for the future load estimate. Regardless, the need to integrate LCT, market competition and electricity trading have brought about the need for probabilistic load forecasts which may include intervals, quantiles, or densities as noted by Hong and Fan [4] and Haben et al [1]. In either point load forecasting or probabilistic load forecasting, many approaches exist and increasingly mixed approaches are being used to create hybrid profiles to better represent load with irregular peaks.…”

Section: Forecastsmentioning

confidence: 99%

“…Each addition resulted in reduction in errors. 1 Dry bulb temperature is the temperature as measured when the thermometer is exposed to air but not to sunlight or moisture. It is associated with air temperature that is most often reported.…”

Section: Linear Regressionmentioning

confidence: 99%

See 1 more Smart Citation

Short Term Load Forecasting

Jacob

Neves

Greetham

2019

Mathematics of Planet Earth

View full text Add to dashboard Cite

Electrification of transport and heating, and the integration of low carbon technologies (LCT) is driving the need to know when and how much electricity is being consumed and generated by consumers. It is also important to know what external factors influence individual electricity demand. Low voltage networks connect the end users through feeders and substations, and thus encompass diverse strata of society. Some feeders may be small with only a handful of households, while others may have over a hundred customers. Some low voltage networks include small-to-medium enterprises (SMES), or hospitals and schools, but others may be entirely residential. Furthermore, local feeders will also likely register usage from lighting in common areas of apartments or flats, street lighting and other street furniture such as traffic lights. Moreover, the way that different households on the same feeder or substation use electricity may be drastically different. For example, load profiles of residential households will vary significantly depending on the size of their houses, occupancy, socio-demographic characteristics and lifestyle. Profiles will also depend on whether households have solar panels, overnight storage heating (OSH) or electric vehicles [1]. Thus, knowing how and when people use electricity in their homes and communities is a fundamental part of understanding how to effectively generate and distribute electrical energy. In short term load forecasting, the aim is to estimate the load for the next half hour up to the next two weeks. For aggregated household demand, many different methods are proposed and tested (see e.g. Alfares and Nazeeruddin [2], Taylor and Espasa [3], Hong and Fan [4], etc.). Aggregating the data smooths it, therefore makes it easier to forecast. The individual level demand forecasting is more challenging and comes with higher errors, as shown in Singh et al. [5], Haben et al. [1]. The growth of literature on short term load forecasting at the individual level has started with the wider access to higher resolution data in the last two decades, and is still developing.

show abstract

Short term load forecasting and the effect of temperature at the low voltage level

Cited by 72 publications

References 47 publications

Deep-Based Conditional Probability Density Function Forecasting of Residential Loads

Deep-Based Conditional Probability Density Function Forecasting of Residential Loads

Load Nowcasting: Predicting Actuals with Limited Data

Short Term Load Forecasting

Contact Info

Product

Resources

About