Electrical Load Forecasting

—Load forecasting, particularly short-term load forecasting (STLF) plays a vital role in the economy streaming and tracking of power system. Many stochastic and artificial intelligence techniques haven been used in order to come up with an accurate (less error) short-term load forecast. Here, we introduce a new approach to short-term load forecasting (STLF) using the conventional Hidden Markov Model (HMM) then compare it with Autoregressive Integrated Moving Average (ARIMA) models. Three-dimensional continuous multivariate Gaussian emission probabilities are used in this experiment for HMM. Meanwhile for ARIMA models, different parameters are used for different kinds of dataset. Comparison is done afterwards to the actual load value using MAPE and RMSE.

Applications of time series typically involve an analysis of the existing series
of data values and predicting the future values. Energy load forecasting is
considered one of the interesting types of time series applications and plays
an important role in a power system operation and control. Power companies
have used different conventional methods to forecast energy loads. Yet,
since the relationship between the energy load and the factors that affect the
energy load are not linear, it is difficult to capture the nonlinearity using conventional
methods. Some forecasting systems predict the energy load based on
exogenous variables only, such as the absolute temperature. However, since a
change in temperature can cause dramatic changes in the energy load, that approach
suffers from significant forecasting errors. Other companies use data
from ’similar’ days to learn the trend of similarity in the energy load and use
that for prediction. This approach is very complex and it does not work for
neural network learning. Neural networks deviate from other models by their
ability to map inputs to the outputs and build complex relationships among
variables without specifying them explicitly.
In this work we provide an extensive literature survey of the related problems
and study several approaches, including conventional predictive methods. As
a result of our analysis we propose two new methods, the multi–context recurrent
networks and the hybrid networks, i.e., the auto–regressive multi–context
recurrent neural networks. We consider them in context of the forecasting system
design and development.
We developed a system of adaptive and dynamic networks for predicting the
daily peak load of electric energy. The system maps the exogenous and endogenous
inputs to the endogenous outputs. It can be used for characterizing
the seasonal and daily loads of electric power as well as unseasonably hot
or cold days, holidays and other exceptional situations. We use an adaptive
mechanism to train the network, deal with old data, and reduce the growth in
energy load. The system is characterized by maintaining the memory of past
inputs and outputs and being able to predict the future behavior in a compact
and robust way, using natural representation. The system is tested on two very
different data sets in two countries. The results show that our models are more
reliable and more suitable for energy load forecasting than the existing methods.
Compared to others, the performance of our system with other models
(statistical methods and machine learning) is very encouraging and produces
the mean absolute percentage error of (MAPE) ±1.5%.

Accurate electricity load estimation is an essential issue for the operation of the power system, and it is one of the essential works of future power planning for large cities. Every power prediction model has its benefits and drawbacks and has its particular application range. Researchers categorized the load energy forecasting as Short-Term Load Forecasting (STLF), Medium-Term Load Forecasting (MTLF), and Long-Term Load Forecasting (LTLF), which entirely depends on the time in which estimation is scheduled. As electricity load forecasting can be seen as a machine learning problem, several automated methodologies and models are included in the literature review. In this work, we aim to explore and implement state-of-the-art machine learning techniques like Polynomial Regression, Support Vector Regression (SVR), and Artificial Neural Networks (ANN) to predict the short-term and medium-term load consumption for the historical data accurately. For this study, hourly load data of an Urban 11 KV Feeder was collected from the 220 KV grid station. Weather parameters like temperature, pressure, and humidity data for the particular region were taken from Lahore Meteorological Department. Data were divided into several datasets (daily, weekly, and monthly) to achieve short-term and medium-term electrical load prediction using the techniques mentioned earlier. Input parameters used in this study were temperature (both dry and wet), humidity, and pressure, while predicted hourly load demand was used as output. The final results tables show that the performance of the SVR predictor is much better than other techniques both in the short term and medium load forecasting.

Cite this as: Khaliq, A., Khosa, I., Muneeb, M. (2019). Short term electrical load forecasting for an urban 11 KV feeder using machine learning techniques. J. Multidiscip. Sci. 1(1), 1-11.

Forecasting of electrical load is vital in electrical sector; it helps in the process of planning periodical operations and facility expansion. In the deregulated economy, it has many applications which include energy purchasing, and generation, load switching, contract evaluation and infrastructure development. Therefore finding the most appropriate forecasting method for a specific electricity network is not an easy task. Although many forecasting methods were developed none can be generalized for all load patterns. This paper presents a framework of various approaches to optimize the forecasting of electrical loads.

This paper shows a methodology for short term load forecasting. The forecasting methodology proposed is based on the combined use of a fuzzy logic model for predicting the load profile up to 48 hours ahead, a polynomial neural network to predict daily values of minimum demand and peak demand and cubic splines interpolation for disaggregation of the hourly load forecasts in different time resolutions (10, 15 and 30 minutes). The performance of the proposed methodology is illustrated by the results from a computational experiment with the load records of the National Interconnected System (SIN).

Multi-step forecasting is very challenging and there are a lack of studies available that consist of machine learning algorithms and methodologies for multi-step forecasting. It has also been found that lack of collaborations between these different fields is creating a barrier to further developments. In this paper, multi-step time series forecasting are performed on three nonlinear electric load datasets extracted from Open-Power-System-Data.org using two machine learning models. Multi-step forecasting performance of Auto-Regressive Integrated Moving Average (ARIMA) and Long-Short-Term-Memory (LSTM) based Recurrent Neu-ral Networks (RNN) models are compared. Comparative analysis of forecasting performance of the two models reveals that the LSTM model has superior performance in comparison to the ARIMA model for multi-step electric load forecasting.

Load forecasting is a foundational requirement for an electricity distribution business (EDB).
Load forecasts estimate where and how much load will grow on the network and allow
effective planning of network augmentation projects. Unison Networks Limited (Unison) has
spent considerable effort developing a robust decision support tool (DST) for forecasting
changes in demand, known as the Load Forecast Tool (LFT).
The purpose of this paper is to share Unison’s experiences in developing the LFT and to
examine how load forecasting tools can be developed to achieve maximum benefit.
The paper begins by introducing Unison, and details the structure and the objectives of the
LFT. It then introduces challenges that were encountered during the development of the LFT
and the steps that were taken to meet these challenges to produce a tool of real value.

Load forecasting of a rustic area where grid connection does not exist is a process of calculating the required electric load using a method named Matrix Analysis. In our neighbor country India meet their clients need but in Bangladesh which is most densely populated country in the world. Most of the people of Bangladesh live in below poverty level. In Bangladesh almost 66 percent people access the grid connection but more than 45 percent people of rural area in Bangladesh does not get the grid connection. It can be seen from the calculated amount the number of people who does not access grid power supply yet now is more than 34 percent. This is not expected actually. According to blessing of electric supply we can get better life style.

Healthcare buildings exhibit a different electrical load predictability depending on their size and nature. Large hospitals behave similarly to small cities, whereas primary care centers are expected to have different consumption dynamics. In this work, we jointly analyze the electrical load predictability of a large hospital and that of its associated primary care center. An unsupervised load forecasting scheme using combined classic methods of principal component analysis (PCA) and autoregressive (AR) modeling, as well as a supervised scheme using orthonormal partial least squares (OPLS), are proposed. Both methods reduce the dimensionality of the data to create an efficient and low-complexity data representation and eliminate noise subspaces. Because the former method tended to underestimate the load and the latter tended to overestimate it in the large hospital, we also propose a convex combination of both to further reduce the forecasting error. The analysis of data from 7 years in the hospital and 3 years in the primary care center shows that the proposed low-complexity dynamic models are flexible enough to predict both types of consumption at practical accuracy levels.

The typical load profiles are used in formulating the electricity distribution tariffs. Usually, clustering
algorithms are used in identification of typical load profiles from a sample with load curves
measurements. These methods group together load curves having similar shapes, from the which are
extracted the typical profiles. The various clustering algorithms can be applied to identify load profiles
but only the Self-Organising Map also known as Kohonen Map, a kind of neural network, can
organize a map where the clusters with similar load diagrams appears together in the map. The map
shows simultaneously all clusters and allows the visual identification of the load profiles. This paper
shows some results of a case study performed with Kohonen map applied on a sample with 125 load
curves.

This paper focuses on the study of short term load forecasting (STELF) using interval Type-2 Fuzzy Logic (IT2FL) and feed-forward Neural Network with back-propagation (NN-BP) tuning algorithm to improve their approximation capability, flexibility and adaptiveness. IT2FL for STELF is presented which provides additional degrees of freedom for handling more uncertainties for improving prediction accuracy and reducing cost. The IT2FL comprises five components which include; the fuzzification unit, the knowledge base, the inference engine, the type reducer and the defuzzification unit. Gaussian membership function is used to show the degree of membership of the input variables. The lower and upper membership functions (fired rules) as well as their consequent coefficients of IT2FL are fed into a (NN) which produces a crisp value coresponding to the optimal defuzzified output of IT2FLSs. The NN type reducer is trained to optimize parameters of membership function (MF) so as to produce an output with minimum error function with the purpose of improving forecasting performance of IT2FLS models. The IT2FNN system has the ability to overcome the limitations of individual technique and enhances their strengths to handle electric load forecasting. The IT2FNN is applied for STELF in Akwa Ibom State-Nigeria. The result of performance of IT2FNN is compared with IT2FLS and T1FLS methods for short term load forecasting with MSE of 0.00123, 0.00185 and 0.00247 respectively. Also, the results of forecasting are compared using RMSE of 0.035, 0.043 and 0.035 respectively, indicating a best accurate forecasting with IT2FNN. In addition, the result of performance of IT2FNN is compared with IT2FLS and T1FLS methods for short term load forecasting with MAPE of 1.5%, 3% and 4.5% respectively. Simulation results show that the IT2FNN approach takes advantages of accuracy and efficiency and performs better in prediction than IT2FL and T1FL methods in power load forecasting task.. KEYWORDS Interval type-2 fuzzy logic; feed-forward neural network; back propagation; electric load; Interval type-2 fuzzy neural networks; type-1 fuzzy logic.

Neste trabalho são descritas as linhas gerais de uma metodologia para previsão de carga capaz
de fornecer previsões até 48 horas à frente, com resolução temporal de 10, 15, 30 e 60 minutos. A
metodologia de previsão baseia-se no uso combinado de um modelo de lógica nebulosa para a previsão
do perfil horário da curva de carga até 48 horas á frente, de uma rede neural polinomial para prever
valores diários da demanda mínima e da demanda de ponta e da interpolação spline para obter previsões
com diferentes resoluções de tempo. O desempenho da metodologia proposta é ilustrado pelos resultados
obtidos em um experimento computacional com os registros de carga do Sistema Interligado Nacional
(SIN).

Cloud computing is an emerging and rapid growing field of Infrastructure as a Service (IaaS), it has to deal with resource allocation and power management issues. This paper proposes CGPANN to accurately forecast the client's requests for a very short term duration of 1 second. A forecasting accuracy as high as 99.81% has been attained that verifies the accuracy of the proposed model. The experimental results show that the model outperforms all the contemporary models proposed in past.

— This paper proposes a new objective function and quantile regression (QR) algorithm for load forecasting (LF). In LF, the positive forecasting errors often have different economic impact from the negative forecasting errors. Considering this difference, a new objective function is proposed to put different prices on the positive and negative forecasting errors. QR is used to find the optimal solution of the proposed objective function. Using normalized net energy load of New England network, the proposed method is compared with a time series method, the artificial neural network method, and the support vector machine method. The simulation results show that the proposed method is more effective in reducing the economic cost of the LF errors than the other three methods. Index Terms—Economic objective function, load forecast, power system planning, quantile regression, weighted objective function.

In these days we are more focused on good planning for load demand that’s why we all are highly interested in the load forecasting techniques. Precise load forecasting clutches an excellent saving strength for electric utility in the organization, industries and institutes [1]. In this paper I show a comparative study of two soft computing models namely artificial neural network and generalized neural network for the hourly electricity demand forecast of DEI 33 KV substation. The soft computing models were trained and tested using the actual hourly weather and load data which is obtained from DEI 33 KV substation and DRDO, Agra .these models comparisons of the proposed schemes is presented for 3 month demands for electricity. My work result shows the generalized neural network is much better to artificial neural network.

This work deals with the residential electricity projection based on a
macroeconomic scenario. The main issue is incorporate an energy efficiency framework
based on the future evolution of household income and their household budgets, because
this budgets are the most important factor that affects the residential average
consumption. The PNADs microdata together with the 10 years macroeconomic
projection and households forecast permitted to model the household income. These
results guide the projection of the residential average consumption and consequently
allow the formulation of different energy conservation scenarios.

In this report our point is to figure the power costs as precisely as conceivable by planning a half and half model and utilizing different blunder capacities to really take a look at the exactness of our outcome. Before liberation come to presence years and years back, the electric power enterprises have been overwhelmed by utilities that had full command over movements of every kind nearby. In any case, after its first endeavor in Latin America, the business has been on the move in many nations all over the planet. In a liberated market, end-use clients have the decision to choose their power provider.

Este artigo apresenta um estudo de caso onde a modelagem estrutural de séries temporais é aplicada na série mensal do consumo de eletricidade. Os modelos estruturais fazem uso da decomposição de uma série temporal em suas componentes não observáveis de tendência, sazonalidade, cíclica e irregular, oferecendo uma descrição dos traços estilizados da série o que possibilita analisar mudanças de tendência e alterações no padrão sazonal ou no comportamento cíclico.

This paper focuses on the study of short term load forecasting (STELF) using interval Type-2 Fuzzy Logic (IT2FL) and feed-forward Neural Network with back-propagation (NN-BP) tuning algorithm to improve their approximation capability, flexibility and adaptiveness. IT2FL for STELF is presented which provides additional degrees of freedom for handling more uncertainties for improving prediction accuracy and reducing cost. The IT2FL comprises five components which include; the fuzzification unit, the knowledge base, the inference engine, the type reducer and the defuzzification unit. Gaussian membership function is used to show the degree of membership of the input variables. The lower and upper membership functions (fired rules) as well as their consequent coefficients of IT2FL are fed into a (NN) which produces a crisp value coresponding to the optimal defuzzified output of IT2FLSs. The NN type reducer is trained to optimize parameters of membership function (MF) so as to produce an ...

This article proposes a hybrid technique for very short-term load forecasting in microgrids. The proposed technique integrates genetic algorithm, particle swarm optimization, and adaptive neuro fuzzy inference systems. An accurate electricity demand forecasting in microgrids is crucial task for efficient operation of power systems and to ensure system reliability in general. In this article, binary genetic algorithm is used to select important predictors that significantly influence the load pattern among a number of candidate input variables; and particle swarm optimization algorithm is used to optimize an adaptive neuro fuzzy inference systems-based model for very short-term forecasting of load. The effectiveness of the proposed technique is demonstrated using a historical load demand dataset of a case study microgrid system in Beijing. Performance of the proposed approach is compared with backpropagation neural network, auto regressive integrated moving average, and persistence models. The proposed approach outperformed all three methods for 10 min ahead forecasting of load, demonstrating its effectiveness and applicability for very short-term load forecasting in microgrids. The average execution time for one-step-ahead forecasting is reduced by over 40% compared to a backpropagation neural network-based approach.

Recent advances in integrated electronic devices motivated the use of Wireless Sensor Networks (WSNs) in many applications including target surveillance and tracking. A number of sensor nodes are scattered within a sensitive region to detect the presence of intruders and forward subsequent events to the analysis center(s). Obviously, the sensor deployment should guarantee an optimal event detection rate. This paper proposes a tracking framework based on Voronoi tessellations. Two mobility models are proposed to control the coverage degree according to target presence. The objective is to set a non-uniform coverage within the monitored zone to allow detecting the target(s) by multiple sensor nodes. We show how the proposed algorithm adapts to the situation where multiple targets move in the monitored zone. Moreover, we introduce an algorithm to discover redundant nodes (which do not provide additional information about target position). This algorithm is shown to be effective in reducing the energy consumption using an activity scheduling approach. Simulations are carried out to underline the efficiency of the proposed models.

http://ojs.academypublisher.com/index.php/jnw/article/viewArticle/0407589597