Conference PaperPDF Available

Forecasting day, week and month ahead electricity load consumption of a building using empirical mode decomposition and extreme learning machine

Authors:

Abstract and Figures

Forecasting of building energy consumption plays a key role in the energy management of the modern power system. However, the noise and randomness in the electricity load data makes it difficult to forecast accurate electricity load. In this paper, a novel scheme namely Empirical Mode Decomposition based Extreme Learning Machine (EMD-ELM) is proposed to forecast the electricity load consumption of a building. Randomness in the electric load data is removed using EMD, whereas, ELM is used to forecast the day, week and month ahead electricity load. To illustrate the usefulness of EMD-ELM, the performance is compared with the renowned neural networks namely Convolution Neural Network (CNN), Long Short Term Memory (LSTM) and ELM. The simulation results clearly indicate that EMD-ELM outperforms CNN, LSTM and ELM in forecasting the day, week and month ahead electricity load consumption of a building.
Content may be subject to copyright.
Forecasting day, week and month ahead electricity
load consumption of a building using empirical
mode decomposition and extreme learning machine
Sajjad Khan1, Nadeem Javaid1, Annas Chand2, Raza Abid Abbasi1
Abdul Basit Majeed Khan3, Hafiz Muhammad Faisal1
1COMSATS University Islamabad, Islamabad 44000, Pakistan
2COMSATS University Islamabad, Abbotabad Campus 22010, Pakistan
3Abasyn University Islamabad Campus, Islamabad 44000, Pakistan
* Correspondence: nadeemjavaidqau@gmail.com, www.njavaid.com
Abstract—Forecasting of building energy consumption plays
a key role in the energy management of the modern power
system. However, the noise and randomness in the electricity
load data makes it difficult to forecast accurate electricity
load. In this paper, a novel scheme namely Empirical Mode
Decomposition based Extreme Learning Machine (EMD-ELM)
is proposed to forecast the electricity load consumption of a
building. Randomness in the electric load data is removed using
EMD, whereas, ELM is used to forecast the day, week and
month ahead electricity load. To illustrate the usefulness of EMD-
ELM, the performance is compared with the renowned neural
networks namely Convolution Neural Network (CNN), Long
Short Term Memory (LSTM) and ELM. The simulation results
clearly indicate that EMD-ELM outperforms CNN, LSTM and
ELM in forecasting the day, week and month ahead electricity
load consumption of a building.
Index Terms—Smart grid, forecasting, empirical mode decom-
position, extreme learning machine, neural network
I. INTRODUCTION
Nowadays, electricity load consumption is increased due
to increase in the world’s population. The electricity load
consumed in buildings plays a vital role in the total energy
consumed. According to a survey conducted by the United
States (department of energy), about 39% of the total electric-
ity is consumed in buildings [1]. Similarly, the study in [2]
shows that, in China, 28% of the total energy is consumed
by the buildings. Therefore, forecasting the electricity load
consumption of buildings can significantly improve the energy
utilization rate.
Accurate forecast of electricity load consumption of a build-
ing is not only beneficial for building managers and owners but
also for the utility companies. From the perspective of building
owners and managers, it is helpful in building the electricity
load profile of the building. Building the electricity load
profile of building can be helpful in reducing the electricity
consumption cost. Furthermore, it is also useful in taking
design decision and optimizing the heating, ventilation and air
conditioning system of the buildings. Whereas, from the view
point of the utility, forecasting the accurate electricity load
consumption of buildings is helpful in developing effective
demand side management strategies.
Forecasting the accurate electricity load consumption of a
building is quite challenging [3]. There are many factors that
affect the electricity load consumption of a building, e.g., large
amount of heating, ventilation as well as air-conditioning loads
and variations in the weather conditions, etc.,. Furthermore,
the irregular electricity consumption pattern of the inhabitants
adds randomness in the electricity load consumption of the
buildings.
For the last few decades, many techniques have been
presented by researchers to forecast electricity load. These
forecasting techniques are mainly categorized in to two sub-
categories namely artificial intelligence methods (i.e., Support
Vector Machine (SVM), Artificial Neural Networks (ANN),
etc.,) and statistical methods (i.e., Auto Regressive Integrated
Moving Average (ARIMA), Seasonal ARIMA etc.,) [4]. These
techniques mainly focus on load forecasting and less attention
has been paid to tackle the randomness in the electricity load
data. Therefore, in this paper, a novel technique is proposed
to deal with the randomness in the electricity load data and
forecast the electricity load consumption of a building. The
main contribution of our work can be summarized as follows.
A novel scheme namely Empirical Mode Decomposition
based Extreme Learning Machine (EMD-ELM) is devel-
oped to tackle the randomness and noise in the electricity
load data and forecast the day, week as well as month
ahead electricity load consumption of a building.
A comparison analysis of the renowned forecasting
schemes is performed.
The rest of this paper is organized as follows. Section II dis-
cusses the electricity load forecasting techniques. Motivation
and problem statement is discussed in section III. Section IV
describes the proposed forecasting model. The existing and
proposed forecasting techniques are discussed in section V.
Section VI is the simulation results and discussion. Finally,
the paper is concluded in section VII along with the future
directions of our work.
978-1-5386-7747-6/19/$31.00 ©2019 IEEE 1600
II. RE LATE D WO RK
In this section, we discussed the well known techniques
used for electricity load forecasting.
Chen et al. in [5] forecast the electricity load of three
states in Australia using mixed kernel based ELM. In this
work, the authors combined the functionality of Radial Basis
Function (RBF) kernel and KMOD kernel to improve the
forecasting accuracy. EMD is used to remove randomness
and noise from the original load series. The authors in [6]
proposed two models to forecast the electricity load and
price. Grid search and cross validation techniques are used
to tune the parameters of Convolution Neural Network (CNN)
and Support Vector Regressor (SVR). Whereas, for features
selection and extraction this work used XG-Boost, recursive
feature elimination, decision tree and Random Forest (RF).
The performance comparison of these models with the bench-
mark schemes show that CNN and SVR outperforms the others
schemes in terms of forecasting accuracy.
In [7], Qiu et al. forecast the electricity load by decomposing
the original load series into several Intrinsic Mode Functions
(IMFs) and a residue. In this work, the authors used a Deep
Belief Network (DBN) with two restricted Boltzmann ma-
chines to forecast the electricity load using each IMF. The final
results were obtained by combining the forecasting result of
each IMF. The effectiveness of this scheme was demonstrated
by comparing the performance with nine benchmark schemes
namely SVR, ANN, Persistence, RF, EMD-ANN, EMD-RF,
EMD-SVR, DBN and ensemble DBN. This work ignored
the non-linear features of the electricity load data. In [9],
an extreme deep learning approach is presented to forecast
the electricity load consumption of a building. The authors in
this work extract the energy consumption features of a build-
ing using stacked auto encoders. The input parameters were
determined using partial auto correlation analysis, whereas,
ELM forecast the energy consumption of the building. To
evaluate the effectiveness of this scheme, the performance
is compared with back propagation Neuarl Network (NN),
SVR, multiple linear regression and Generalized RBF NN
(GRBFNN). Experimental results show that this work out-
performs the benchmark schemes. However, in this work, the
training performance of ELM is poor as compare to GRBFNN.
A hybrid model to forecast short term electricity load de-
mand using an improved feature selection and a novel data
framework strategy is developed in [8]. In this work, genetic
algorithm and improved binary cuckoo search algorithm are
used to select the important features. To forecast the electricity
load, ELM is used. The authors in this work ignored the impact
of environmental factors on the load data.
Using a priority index, Mahnoor et al. [10] developed a
knowledge based short term electricity load forecasting model
for demand response management. In the knowledge base,
binary firefly algorithm and affinity propagation are used.
Whereas, for efficient energy management between consumers
and the utility, the authors used Stackelberg game plan. To
evaluate the effectiveness of their proposed scheme, compari-
son is performed with DBN and fuzzy local linear model tree.
Experimental results indicate that the proposed knowledge
based scheme outperforms the benchmark schemes in fore-
casting the electricity load. Guo et al. [11] develop a deep feed
forward NN model to forecast electricity load. In this work, the
authors first identified several environmental factors that affect
the electricity load consumption. Secondly, important features
are identified using deep learning, gradient boosting and RF.
Lastly, electricity load is forecasted using probability density
forecasting method based on quantile regression, deep learning
and kernel density estimation. The performance comparison
of this scheme illustrate that the deep learning technique
outperforms the other benchmark schemes.
In [12], Grzegorz Dudek compared the performance of
Multi-layer perceptron, RBFNN, Generalized Regression NN
(GRNN), fuzzy counter propagation NN and self organizing
map in short term electricity load forecasting. The performance
comparison of the aforementioned schemes show that GRNN
outperforms all other schemes. The authors in [13] forecast the
electricity load for each day of the week using a deep CNN. In
this work, the authors enhanced the forecasting performance
of CNN by increasing the number of layers in CNN. [14]
forecast the electricity load of a building. In this work, the
authors combined Elman NN (ENN) and novel Shark smell
optimization method to develop a hybrid forecast engine.
Furthermore, sliding window EMD is used to decompose the
original electricity load signal and the best input candidates
are selected on the basis of Pearson’s correlation. To evaluate
the forecasting performance of this scheme, comparison is
made with nineteen methods. Simulation results affirm that
this scheme has the lowest performance metric values.
Qiu et al. [15] develop an ensemble incremental learning
approach to forecast short term electricity load. The proposed
approach comprises of EMD, Discrete Wavelet Transformation
(DWT) and Random Vector Functional Link (RVFL). The
forecasting performance of this scheme is compared with
eight benchmark schemes. Experimental results show that the
proposed DWT-EMD based learning approach outperforms the
EMD and non-EMD based forecasting schemes. Sana Mujeeb
et al. [16] present a deep Long Short Term Memory (LSTM)
model to forecast the electricity load and price. In this work,
electricity load of the first day and week of every month
is forecasted. To compare the effectiveness of deep LSTM,
the performance is compared with the ANN, non linear auto
regressive network with exogenous variables and ELM in
terms of forecasting accuracy. Experimental results validate
the effectiveness of this scheme.
III. MOTI VATION AND P RO BLEM STATEMENT
The authors in [5] developed a mixed kernel based ELM
to forecast the electricity load of three states. Qiu et al. in
[6] developed a DBN to forecast the future electricity load by
decomposing the original electricity signal into several IMFs.
The final forecasting results are obtained by combing the
forecasting result of each IMF. The authors in [14] determined
the input variables using EMD and forecast the electricity load
1601
Fig. 1. Historic load values consumed
using ENN. A forecasting technique using EMD, DWT and
RVFL is developed in [15]. The aforementioned techniques
motivate us to forecast the electricity load.
Randomness in the electricity load data has a huge impact
on the forecasting accuracy. The work in [5] developed an
EMD based mixed kernel ELM model to remove randomness
and noise from the electricity load data. In this work, the
authors forecast the electricity load of three different states.
The work in [9] and [14] forecast the electricity load of a
building. However, none of the work considered randomness
and noise in the electricity load consumption data of buildings.
IV. SYS TE M MO DE L
In this paper, the aim is to develop a simple, effective and
accurate model to forecast the electricity load consumed in a
building. This section is further divided into three subsections.
Subsection A discusses the dataset used in this paper. The
detailed description of the proposed system model is discussed
in subsection B, whereas, subsection C presents the perfor-
mance evaluation metrics used to assess the effectiveness of
our proposed forecasting model.
A. Dataset description
In this paper, we use the building energy consumption
dataset which is publicly available at [17]. The dataset contains
the electricity load consumed in building 74 on the Lawrence
Berkeley National Lab campus for every fifteen minutes. For
training and testing of our proposed model, we have only used
the load data consumed from January 1, 2014 to June 30, 2014.
The training dataset contains the load values for the first five
months, whereas, testing is performed on the last thirty days.
Figure 1 presents the electricity load consumption graph. In
order to better visualize the electricity load data, we have only
plotted the load consumed in one week.
B. Proposed forecasting model
Figure 2 shows the proposed forecasting model. The elec-
tricity load data is affected by external factors, e.g., electricity
price, holidays and management policies of the utility. Fur-
thermore, the hourly, daily, weekly and seasonal variations
also cause randomness in the electricity load data. Due to
these factors the forecasting accuracy is affected. Therefore,
in the paper, we have first removed the randomness and noise
using EMD. After removing the randomness and noise from
the electric load data, the de-noised load series is converted
into a supervised learning problem. To convert the de-noised
load series into supervised learning problem, the input and
Start
Load series
De-noised load series
Convert into a supervised learning problem
Train / test data
ELM (RBF)
Performance metrics: MAE, MSE and MFE
End
. . .
IMF1IMF2Residue
EMD
IMFn
. . . .
Fig. 2. Proposed system model
output variables are formed by shifting the load series data
one step forward. Afterwards, the data is divided into train
and test dataset.
C. Performance metrics
To evaluate the forecasting performance of our proposed
model, we use three performance metrics namely Mean Fore-
casting Error (MFE), Mean Absolute Error (MAE) and Mean
Square Error (MSE).
M F E =1
N
N
X
t=1
(yi¯yi)(1)
MAE =1
N
N
X
t=1
|yi¯yi|(2)
MSE =1
N
N
X
t=1
(yi¯yi)2(3)
1602
In the above equations yipresents the actual load value,
whereas, ¯yishows the forecasted load value.
V. EXISTING AND PROPOSED FORECASTING SCHEMES
This section discusses the existing and proposed schemes
used in this paper to forecast the load consumed in a building.
This section is further divided into four subsections.
A. LSTM
Hochreiter and Schmidhuber first proposed LSTM [19]
which is one of the most commonly used RNN. The difference
between LSTM and other NNs is that LSTM can store
information for a longer period of time. It is mainly due to
the three types of gates namely input gate, forget gate and
output gate. The input gate decides which information must be
added or updated in one cell state. Forget gate decides which
information from the previous state must be kept or discarded.
Whereas, output gate decides how much information must be
preserved as output.
B. CNN
CNN is one of the most common and successful NN used
in computer vision and medical imaging [20]. It has the
capability to handle large amount of input data, Furthermore,
it can also extract the hidden features from the input data. A
typical CNN consist of an input, convolution, max pool, fully
connected, dense, dropout, flatten and output layers. The CNN
used in this paper comprises of three hidden layers.
C. ELM
Huang et al. first proposed ELM [21] which is a single layer
feed forward NN. Due to its simplicity and learning capability
without iteratively tunning the hidden layers parameters, ELM
has gained enormous popularity in various fields. Furthermore,
the biasness and hidden layer parameters in ELM can be ran-
domly assigned. Similarly, in ELM, less human intervention is
required, i.e., if the user is not aware of the features mapping,
kernels are applied to ELM.
D. EMD-ELM
EMD is a signal decomposition method. It decomposes the
original time series load data into several IMFs as well as a
residue. The residue presents the trend. EMD is mainly used
to extract the instantaneous frequency data from the datasets
which possess non-linear and non-stationary characteristics.
To remove the randomness and noise from the load series,
we dropped the first IMF and reconstruct a new load series
named as de-noised series. The de-noised load series is then
input to the ELM to forecast the day, week and month ahead
load consumption of the building.
VI. SIMULATION RESULTS AND DISCUSSION
This section discusses the forecasting results. A brief sum-
mary of the forecasting models used in this paper is given in
table I. To better understand the day, week and month ahead
load forecasting performance of each model, this section is
further divided into four subsections.
TABLE I
PERFORMANCE METRIC VALUES
LSTM CNN ELM EMD-ELM
Day
MSE 3.65 2.19 1.95 0.84
MAE 1.57 1.07 1.19 0.71
MFE -0.91 0.07 0.07 0.01
Week
MSE 7.26 4.18 4.23 1.36
MAE 2.03 1.52 1.56 0.84
MFE -1.005 -0.002 -0.08 -0.020
Month
MSE 9.03 5.04 4.72 1.94
MAE 2.25 1.59 1.63 0.95
MFE -0.48 0.009 -0.31 -0.019
Fig. 3. Load forecasted by all schemes for one day
A. Forecasting performance of LSTM
The three performance metric parameters of LSTM in
forecasting the electricity load of one day are shown in figure
4. From this figure it can be clearly seen that LSTM has
the worst performance metrics values. As shown in figure3,
the load forecasted by LSTM for all time slots lies far away
from the actual load. Similarly, LSTM has poor performance
in forecasting the week and month ahead consumption load
as shown in figure 5 and 7. As shown in table I as well as
figures 6 and 8 LSTM as a load forecasting model to forecast
the electricity load consumption of a building is not a good
choice.
B. Forecasting performance of CNN
The performance of CNN in forecasting building energy
consumption is better as compare to LSTM. However, the load
forecasted by CNN is not acceptable. As shown in figure 4,
6 and 8, CNN achieves average values. Furthermore, the load
values forecasted by CNN have a similar trend with the actual
load data. However, the difference in the actual and forecasted
load is vey large as shown in figure 3. According to table I,
CNN minimized the MSE by 44.18% as compare to LSTM in
forecasting the month ahead electricity load consumption of
the building.
1603
Fig. 4. Performance metrics for one day
Fig. 5. Load forecasted by all schemes for one week
Fig. 6. Performance metrics for one day
Fig. 7. Load forecasted by all schemes for one month
Fig. 8. Performance metrics for one month
C. Forecasting performance of ELM
In forecasting the electricity load consumption of a building,
ELM outperforms CNN and LSTM as shown in figures 4, 6
and 8. Furthermore, it can be seen in figure 3, the load values
forecasted by ELM overlaps the actual load at several time
slots. Similarly, in the week and month ahead load forecasting,
ELM overlaps the actual load as shown in figures 5 and
7. Here, it is worth mentioning that ELM outperforms all
the existing schemes in forecasting the load consumption of
the building. However, on its own, ELM cannot tackle the
randomness in the electricity load data. In forecasting the
month ahead electricity load consumption of the building,
ELM minimized the MSE by 47.72% and 6.34% as compare
to LSTM and CNN respectively.
D. Forecasting performance of EMD-ELM
According to figures 4, 6 and 8 EMD-ELM outperforms
all the benchmark schemes in forecasting the day, week and
month ahead electricity load consumption of the building.
From these figures, it can be clearly seen that EMD-ELM
has the lowest performance metric values in forecasting the
day, week and month ahead load consumption of the building.
1604
Furthermore, the electricity load forecasted by EMD-ELM and
the actual load have a similar pattern as shown in figures
3, 5 and 7. According to table I EMD-ELM minimized the
MSE by 78.51%, 61.50% and 58.89% in forecasting the month
ahead load consumption of the building as compare to LSTM,
CNN and ELM. Similarly, in forecasting the week ahead
load consumption of the building, EMD-ELM is 81.28%,
64.46% and 67.84% efficient in minimizing MSE as compare
to LSTM, CNN and ELM respectively. Whereas, in the day
ahead forecasting of building energy consumption, EMD-ELM
reduced the MSE by 76.98%, 61.64% and 56.92% as compare
to LSTM, CNN and ELM.
VII. CONCLUSION AND FUTURE WORK
In this paper, day, week and month ahead electricity load
consumption of a building is forecasted using ELM. To
improve the forecasting accuracy, randomness in the electricity
load data is removed using EMD. To evaluate the forecasting
performance of the proposed scheme comparison with the well
known forecasting scheme is performed. Simulation results
validate that, the proposed scheme outperforms LSTM by
78.51%, CNN by 61.5% and ELM by 58.5% in term of
minimizing MSE. In future, we plan to develop a deep learning
technique to forecast long term electricity load consumption of
residential buildings. Furthermore, the proposed work can be
extended to forecast the long term electricity load consumption
of the building.
REFERENCES
[1] https://catalog.data.gov/dataset/buildings-energy-data-book-6d4d2. Last
accessed: February 15, 2019
[2] Chen, Hua, W. L. Lee, and Xiaolin Wang. “Energy assessment of office
buildings in China using China building energy codes and LEED 2.2.
Energy and buildings 86 (2015): 514-524.
[3] Ahmad, Ashfaq, Nadeem Javaid, Mohsen Guizani, Nabil Alrajeh, and
Zahoor Ali Khan. “An accurate and fast converging short-term load
forecasting model for industrial applications in a smart grid.” IEEE
Transactions on Industrial Informatics 13, no. 5 (2017): 2587-2596.
[4] Ahmad, Ashfaq, Nadeem Javaid, Abdul Mateen, Muhammad Awais,
and Zahoor Khan. “Short-Term Load Forecasting in Smart Grids: An
Intelligent Modular Approach.” Energies 12, no. 1 (2019): 164.
[5] Chen, Yanhua, Marius Kloft, Yi Yang, Caihong Li, and Lian Li. “Mixed
kernel based extreme learning machine for electric load forecasting.
Neurocomputing (2018).
[6] Zahid, Maheen, Fahad Ahmed, Nadeem Javaid, Raza Abid Abbasi, Zainab
Kazmi, Hafiza Syeda, Atia Javaid, Muhammad Bilal, Mariam Akbar, and
Manzoor Ilahi. “Electricity Price and Load Forecasting using Enhanced
Convolutional Neural Network and Enhanced Support Vector Regression
in Smart Grids.” Electronics 8, no. 2 (2019): 122.
[7] Qiu, Xueheng, Ye Ren, Ponnuthurai Nagaratnam Suganthan, and Gehan
AJ Amaratunga. “Empirical mode decomposition based ensemble deep
learning for load demand time series forecasting.” Applied Soft Comput-
ing 54 (2017): 246-255.
[8] Jiang, Ping, Feng Liu, and Yiliao Song. “A hybrid forecasting model
based on date-framework strategy and improved feature selection tech-
nology for short-term load forecasting.” Energy 119 (2017): 694-709.
[9] Li, Chengdong, Zixiang Ding, Dongbin Zhao, Jianqiang Yi, and Guiqing
Zhang. “Building energy consumption prediction: An extreme deep
learning approach.” Energies 10, no. 10 (2017): 1525.
[10] Khan, Mahnoor, Nadeem Javaid, Adnan Naseem, Salman Ahmed,
Muhammad Riaz, Mariam Akbar, and Manzoor Ilahi. “Game Theoretical
Demand Response Management and Short-Term Load Forecasting by
Knowledge Based Systems on the basis of Priority Index.” Electronics
7, no. 12 (2018): 431.
[11] Guo, Zhifeng, Kaile Zhou, Xiaoling Zhang, and Shanlin Yang. “A
deep learning model for short-term power load and probability density
forecasting.” Energy 160 (2018): 1186-1200.
[12] Dudek, Grzegorz. “Neural networks for pattern-based short-term load
forecasting: A comparative study.” Neurocomputing 205 (2016): 64-74.
[13] Sajjad Khan, Nadeem Javaid, Annas Chand, Abdul Basit Majeed Khan,
Fahad Rashid and Imran Uddin Afridi, “Electricity load forecasting for
each day of week using deep CNN”, in 33rd International Conference on
Advanced Information Networking and Applications (AINA), 2019
[14] Liu, Yang, Wei Wang, and Noradin Ghadimi. “Electricity load forecast-
ing by an improved forecast engine for building level consumers.” Energy
139 (2017): 18-30.
[15] Qiu, Xueheng, Ponnuthurai Nagaratnam Suganthan, and Gehan AJ
Amaratunga. “Ensemble incremental learning Random Vector Functional
Link network for short-term electric load forecasting.” Knowledge-Based
Systems 145 (2018): 182-196.
[16] Sana Mujeeb, Nadeem Javaid, Manzoor Ilahi, Zahid Wadud and Farruh
Ishmanov. “Deep Long Short-Term Memory: A New Price and Load
Forecasting Scheme for Big Data in Smart Cities.” Sustainability 2019,
11(4), 987
and Muhammad Khalil Afzal 4
[17] https://trynthink.github.io/buildingsdatasets/ last accessed: February 12,
2019
[18] Singh, Shailendra, and Abdulsalam Yassine. “Big data mining of energy
time series for behavioral analytics and energy consumption forecasting.
Energies 11, no. 2 (2018): 452.
[19] Hochreiter, Sepp, and Jrgen Schmidhuber. ”Long short-term memory.”
Neural computation 9, no. 8 (1997): 1735-1780.
[20] Krizhevsky, Alex, Ilya Sutskever, and Geoffrey E. Hinton. ”Imagenet
classification with deep convolutional neural networks.” In Advances in
neural information processing systems, pp. 1097-1105. 2012.
[21] Huang, Guang-Bin, Lei Chen, and Chee Kheong Siew. ”Universal
approximation using incremental constructive feedforward networks with
random hidden nodes.” IEEE Transactions. Neural Networks 17, no. 4
(2006): 879-892.
1605
... Recently, many techniques and methodologies have been applied to forecast electricity load. These forecasting techniques are mainly classified into two classes: artificial intelligence methods (Support Vector Machine, Artificial Neural Networks, etc.) and statistical methods (Multiple Regression, Exponential Smoothing, ARIMA and Seasonal ARIMA, etc.) [5,6]. Recent developments in artificial neural networks, especially Deep Learn-0 2. Research method ...
Article
Full-text available
The Convolutional Neural Network (CNN) model is one of the most effective models for load forecasting with hyperparameters which can be used not only to determine the CNN structure and but also to train the CNN model. This paper proposes a framework for Grid Search hyperparameters of the CNN model. In a training process, the optimal models will specify conditions that satisfy requirement for minimum of accuracy scores of Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE) and Mean Absolute Error (MAE). In the testing process, these optimal models will be used to evaluate the results along with all other ones. The results indicated that the optimal models have accuracy scores near the minimum values. Load demand data of Queensland (Australia) and Ho Chi Minh City (Vietnam) were utilized to verify the accuracy and reliability of the Grid Search framework. © 2021. The Author(s). This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License (CC BY-NC-ND 4.0, https://creativecommons.org/licenses/by-nc-nd/4.0/), which permits use, distribution, and reproduction in any medium, provided that the Article is properly cited, the use is non-commercial, and no modifications or adaptations are made.
Article
This study investigates data standardization methods based on the grid search (GS) algorithm for energy load forecasting, including zero-mean, min-max, max, decimal, sigmoid, softmax, median, and robust, to determine the hyperparameters of deep learning (DL) models. The considered DL models are the convolutional neural network (CNN) and long short-term memory network (LSTMN). The procedure is made over (i) setting the configuration for CNN and LSTMN, (ii) establishing the hyperparameter values of CNN and LSTMN models based on epoch, batch, optimizer, dropout, filters, and kernel, (iii) using eight data standardization methods to standardize the input data, and (iv) using the GS algorithm to search the optimal hyperparameters based on the mean absolute error (MAE) and mean absolute percent error (MAPE) indexes. The effectiveness of the proposed method is verified on the power load data of the Australian state of Queensland and Vietnamese Ho Chi Minh city. The simulation results show that the proposed data standardization methods are appropriate, except for the zero-mean and min-max methods.
Article
Accurate forecasting of residential electricity consumption plays an important role in formulating energy plans and ensuring the safety of power system operations. In order to improve forecasting accuracy, we propose a novel hybrid model with online search data for residential electricity consumption forecasting. Two main steps are involved: (1) Time difference correlation analysis, cointegration test, and Granger causality test are employed to investigate the relationship between online search data and residential electricity consumption. Qualified search keywords are selected to serve as predictors. (2) An extreme learning machine model optimized by Jaya algorithm, together with the selected search keywords from the first step, is proposed to predict residential electricity consumption. Furthermore, monthly residential electricity consumption data from China are used to validate the effectiveness of the proposed model. The experimental results show that the incorporation of online search data into the model can significantly improve forecasting accuracy. After incorporating online search data, improvement rates of all the forecasting models exceed 10%. In addition, the proposed model has the best forecasting performance compared with seasonal autoregressive integrated moving average (SARIMA(X)), support vector regression (SVR), back propagation neural network (BPNN) and extreme learning model (ELM). Root mean squared error (RMSE), mean absolute percentage error (MAPE), and mean absolute error (MAE) of the proposed model with online search data decrease by 34%-51.2%, 43.03%-53.92%, and 41.35%-54.85% relative to other benchmark models, respectively.
Chapter
Full-text available
In smart grid, precise and accurate electricity load forecasting is one of the most challenging tasks. It is due to the high volatile, non-stationary and non-linear behavior of electricity load data. In this paper, a Deep Convolution Neural Network (DCNN) model is proposed to forecast the electricity load for each day of the week of Victoria (Australia). To forecast the electricity load for one day of the week, we analyzed the electricity load data consumed on the same day for the previous three months. To show the usefulness of our proposed scheme, comparison is made with the state of the art forecasting models namely recurrent neural network, extreme learning machine, CNN and auto regressive integrated moving average. Results show that the proposed DCNN has the lowest mean absolute percentage error, mean absolute error and root mean square error of 2.1%, 138.771 and 116.417.
Article
Full-text available
This paper focuses on analytics of an extremely large dataset of smart grid electricity price and load, which is difficult to process with conventional computational models. These data are known as energy big data. The analysis of big data divulges the deeper insights that help experts in the improvement of smart grid’s (SG) operations. Processing and extracting of meaningful information from data is a challenging task. Electricity load and price are the most influential factors in the electricity market. For improving reliability, control and management of electricity market operations, an exact estimate of the day ahead load is a substantial requirement. Energy market trade is based on price. Accurate price forecast enables energy market participants to make effective and most profitable bidding strategies. This paper proposes a deep learning-based model for the forecast of price and demand for big data using Deep Long Short-Term Memory (DLSTM). Due to the adaptive and automatic feature learning mechanism of Deep Neural Network (DNN), the processing of big data is easier with LSTM as compared to the purely data-driven methods. The proposed model was evaluated using well-known real electricity markets’ data. In this study, day and week ahead forecasting experiments were conducted for all months. Forecast performance was assessed using Mean Absolute Error (MAE) and Normalized Root Mean Square Error (NRMSE). The proposed Deep LSTM (DLSTM) method was compared to traditional Artificial Neural Network (ANN) time series forecasting methods, i.e., Nonlinear Autoregressive network with Exogenous variables (NARX) and Extreme Learning Machine (ELM). DLSTM outperformed the compared forecasting methods in terms of accuracy. Experimental results prove the efficiency of the proposed method for electricity price and load forecasting.
Article
Full-text available
Short-Term Electricity Load Forecasting (STELF) through Data Analytics (DA) is an emerging and active research area. Forecasting about electricity load and price provides future trends and patterns of consumption. There is a loss in generation and use of electricity. So, multiple strategies are used to solve the aforementioned problems. Day-ahead electricity price and load forecasting are beneficial for both suppliers and consumers. In this paper, Deep Learning (DL) and data mining techniques are used for electricity load and price forecasting. XG-Boost (XGB), Decision Tree (DT), Recursive Feature Elimination (RFE) and Random Forest (RF) are used for feature selection and feature extraction. Enhanced Convolutional Neural Network (ECNN) and Enhanced Support Vector Regression (ESVR) are used as classifiers. Grid Search (GS) is used for tuning of the parameters of classifiers to increase their performance. The risk of over-fitting is mitigated by adding multiple layers in ECNN. Finally, the proposed models are compared with different benchmark schemes for stability analysis. The performance metrics MSE, RMSE, MAE, and MAPE are used to evaluate the performance of the proposed models. The experimental results show that the proposed models outperformed other benchmark schemes. ECNN performed well with threshold 0.08 for load forecasting. While ESVR performed better with threshold value 0.15 for price forecasting. ECNN achieved almost 2% better accuracy than CNN. Furthermore, ESVR achieved almost 1% better accuracy than the existing scheme (SVR).
Article
Full-text available
Daily operations and planning in a smart grid require a day-ahead load forecasting of its customers. The accuracy of day-ahead load-forecasting models has a significant impact on many decisions such as scheduling of fuel purchases, system security assessment, economic scheduling of generating capacity, and planning for energy transactions. However, day-ahead load forecasting is a challenging task due to its dependence on external factors such as meteorological and exogenous variables. Furthermore, the existing day-ahead load-forecasting models enhance forecast accuracy by paying the cost of increased execution time. Aiming at improving the forecast accuracy while not paying the increased executions time cost, a hybrid artificial neural network-based day-ahead load-forecasting model for smart grids is proposed in this paper. The proposed forecasting model comprises three modules: (i) a pre-processing module; (ii) a forecast module; and (iii) an optimization module. In the first module, correlated lagged load data along with influential meteorological and exogenous variables are fed as inputs to a feature selection technique which removes irrelevant and/or redundant samples from the inputs. In the second module, a sigmoid function (activation) and a multivariate auto regressive algorithm (training) in the artificial neural network are used. The third module uses a heuristics-based optimization technique to minimize the forecast error. In the third module, our modified version of an enhanced differential evolution algorithm is used. The proposed method is validated via simulations where it is tested on the datasets of DAYTOWN (Ohio, USA) and EKPC (Kentucky, USA). In comparison to two existing day-ahead load-forecasting models, results show improved performance of the proposed model in terms of accuracy, execution time, and scalability.
Article
Full-text available
Demand Response Management (DRM) is considered one of the crucial aspects of the smart grid as it helps to lessen the production cost of electricity and utility bills. DRM becomes a fascinating research area when numerous utility companies are involved and their announced prices reflect consumer’s behavior. This paper discusses a Stackelberg game plan between consumers and utility companies for efficient energy management. For this purpose, analytical consequences (unique solution) for the Stackelberg equilibrium are derived. Besides this, this paper presents a distributed algorithm which converges for consumers and utilities. Moreover, different power consumption activities on the basis of time series are becoming a basic need for load prediction in smart grid. Load forecasting is taken as the significant concerns in the power systems and energy management with growing technology. The better precision of load forecasting minimizes the operational costs and enhances the scheduling of the power system. The literature has discussed different techniques for demand load forecasting like neural networks, fuzzy methods, Naïve Bayes, and regression based techniques. This paper presents a novel knowledge based system for short-term load forecasting. The algorithms of Affinity Propagation and Binary Firefly Algorithm are integrated in knowledge based system. Besides, the proposed system has minimum operational time as compared to other techniques used in the paper. Moreover, the precision of the proposed model is improved by a different priority index to select similar days. The similarity in climate and date proximity are considered all together in this index. Furthermore, the whole system is distributed in sub-systems (regions) to measure the consequences of temperature. Additionally, the predicted load of the entire system is evaluated by the combination of all predicted outcomes from all regions. The paper employs the proposed knowledge based system on real time data. The proposed scheme is compared with Deep Belief Network and Fuzzy Local Linear Model Tree in terms of accuracy and operational cost. In addition, the presented system outperforms other techniques used in the paper and also decreases the Mean Absolute Percentage Error (MAPE) on a yearly basis. Furthermore, the novel knowledge based system gives more efficient outcomes for demand load forecasting.
Article
Full-text available
Responsible, efficient and environmentally aware energy consumption behavior is becoming a necessity for the reliable modern electricity grid. In this paper, we present an intelligent data mining model to analyze, forecast and visualize energy time series to uncover various temporal energy consumption patterns. These patterns define the appliance usage in terms of association with time such as hour of the day, period of the day, weekday, week, month and season of the year as well as appliance-appliance associations in a household, which are key factors to infer and analyze the impact of consumers’ energy consumption behavior and energy forecasting trend. This is challenging since it is not trivial to determine the multiple relationships among different appliances usage from concurrent streams of data. Also, it is difficult to derive accurate relationships between interval-based events where multiple appliance usages persist for some duration. To overcome these challenges, we propose unsupervised data clustering and frequent pattern mining analysis on energy time series, and Bayesian network prediction for energy usage forecasting. We perform extensive experiments using real-world context-rich smart meter datasets. The accuracy results of identifying appliance usage patterns using the proposed model outperformed Support Vector Machine (SVM) and Multi-Layer Perceptron (MLP) at each stage while attaining a combined accuracy of 81.82%, 85.90%, 89.58% for 25%, 50% and 75% of the training data size respectively. Moreover, we achieved energy consumption forecast accuracies of 81.89% for short-term (hourly) and 75.88%, 79.23%, 74.74%, and 72.81% for the long-term; i.e., day, week, month, and season respectively.
Article
Full-text available
Building energy consumption prediction plays an important role in improving the energy utilization rate through helping building managers to make better decisions. However, as a result of randomness and noisy disturbance, it is not an easy task to realize accurate prediction of the building energy consumption. In order to obtain better building energy consumption prediction accuracy, an extreme deep learning approach is presented in this paper. The proposed approach combines stacked autoencoders (SAEs) with the extreme learning machine (ELM) to take advantage of their respective characteristics. In this proposed approach, the SAE is used to extract the building energy consumption features, while the ELM is utilized as a predictor to obtain accurate prediction results. To determine the input variables of the extreme deep learning model, the partial autocorrelation analysis method is adopted. Additionally, in order to examine the performances of the proposed approach, it is compared with some popular machine learning methods, such as the backward propagation neural network (BPNN), support vector regression (SVR), the generalized radial basis function neural network (GRBFNN) and multiple linear regression (MLR). Experimental results demonstrate that the proposed method has the best prediction performance in different cases of the building energy consumption.
Article
Accurate load forecasting is critical for power system planning and operational decision making. In this study, we are the first to utilize a deep feedforward network for short-term electricity load forecasting. Our results are compared to those of popular machine learning models such as random forest and gradient boosting machine models. Then, electricity consumption patterns are explored based on monthly, weekly and temperature-based patterns in terms of feature importance. Also, a probability density forecasting method based on deep learning, quantile regression and kernel density estimation is proposed. To verify the efficiency of the proposed methods, three case studies based on daily electricity consumption data for three Chinese cities for 2014 are conducted. The empirical results demonstrate that (1) the proposed deep learning-based approach exhibits better forecasting accuracy in terms of measuring electricity consumption relative to the random forest and gradient boosting model; (2) monthly, weekly and weather-related variables are key factors that have a great influence on household electricity consumption; and (3) the proposed probability density forecasting method is capable of forecasting high-quality prediction intervals via probability density forecasting.
Article
Short term electric load forecasting, as an important tool in the electricity market, plays a critical role in the management of electric systems. Proposing an accuracy and optimization method is not only a challenging task but also an indispensable part of the energy system. More and more accurate forecasting methods are needed by different people in different areas. This paper proposes a novel short-term electric load forecasting method EMD-Mixed-ELM which based on empirical mode decomposition (EMD) and extreme learning machine (ELM). EMD-Mixed-ELM first uses the empirical mode decomposition to decompose the load series for capturing the complicated features of the electric load and de-noising the data. Considering that the performance of extreme learning machine (ELM) is greatly influenced by the choice of kernel, the mixed kernel method is proposed for ELM. The mixed kernel combines the RBF kernel and the UKF kernel. The forecasting results of the EMD-Mixed-ELM are proved to be better than all the other three methods (RBF-ELM, UKF-ELM and Mixed-ELM) and other existing methods (MFES, ESPLSSVM and Combined method). To verify the forecasting ability of the EMD-Mixed-ELM, half-hourly electric load data from the state of New South Wales, Victoria and Queensland in Australia are used in this paper as a case study. The experimental results clearly indicate that for this three datasets, the forecasting accuracy of the proposed method is superior to other methods.
Article
Short-term electric load forecasting plays an important role in the management of modern power systems. Improving the accuracy and efficiency of electric load forecasting can help power utilities design reasonable operational planning which will lead to the improvement of economic and social benefits of the systems. A hybrid incremental learning approach composed of Discrete Wavelet Transform (DWT), Empirical Mode Decomposition (EMD) and Random Vector Functional Link network (RVFL) is presented in this work. RVFL network is a universal approximator with good efficiency because of the randomly generated weights between input and hidden layers and the close form solution for parameter computation. By introducing incremental learning, along with ensemble approach via DWT and EMD into RVFL network, the forecasting performance can be significantly improved with respect to both efficiency and accuracy. The electric load datasets from Australian Energy Market Operator (AEMO) were used to evaluate the effectiveness of the proposed incremental DWT-EMD based RVFL network. Moreover, the attractiveness of the proposed method can be demonstrated by the comparison with eight benchmark forecasting methods.