Conference PaperPDF Available

Short Term Electricity Price Forecasting Through Convolutional Neural Network (CNN)


Abstract and Figures

High price fluctuations have a direct impact on electricity market. Thus, accurate and plausible price forecasts have been implemented to mitigate the consequences of price dynamics. This paper proposes two techniques to deal with the Electricity Price Forecasting (EPF) problem. Firstly, Convolutional Neural Network (CNN) model is used to predict the EPF. Secondly, a principle component analysis model is used for the feature extraction. We have conducted simulations to prove the effectiveness of the proposed approach, which show that CNN based approach outperforms the multilayer perceptron model.
Content may be subject to copyright.
Short Term Electricity Price Forecasting
Through Convolutional Neural Network (CNN)
Zahoor Ali Khan, Sahiba Fareed, Mubbashra Anwar, Afrah Naeem, Hira Gul,
Arooj Arif and Nadeem Javaid
Abstract High price fluctuations have a direct impact on electricity market. Thus,
accurate and plausible price forecasts have been implemented to mitigate the con-
sequences of price dynamics. This paper proposes two techniques to deal with the
Electricity Price Forecasting (EPF) problem. Firstly, Convolutional Neural Network
(CNN) model is used to predict the EPF. Secondly, a principle component analysis
model is used for the feature extraction. We have conducted simulations to prove
the effectiveness of the proposed approach, which show that CNN based approach
outperforms the multilayer perceptron model.
1 Introduction
Short term price forecasting has become the crucial issue in electricity market [1].
Electricity Price Forecasting (EPF) has been classified into five categories: (1) self-
organized system, (2) fundamental models, (3) non-structured methods, (4) statis-
tical models, and (5) computational intelligence approaches. Computational intelli-
gence models are also referred to as artificial intelligence models [2]. The Artificial
Intelligence (AI) model is able to deal with complex systems including nonlinear
and statistical systems. Convolution Neural Network (CNN), Artificial Neural Net-
work (ANN) and Recurrent Neural Network (RNN) have been extensively used as
classifiers in literature and real-time applications. ANN is the widely used method
for forecasting short term electricity load and price forecasting. However, it uses
Zahoor Ali Khan
Faculty of Computer Information Science,
Higher Colleges of Technology, Fujairah 4114, United Arab Emirates
Sahiba Fareed, Mubbashra Anwar, Afrah Naeem, Hira Gul, Arooj Arif and
Nadeem Javaid (corresponding author), e-mail:
COMSATS University Islamabad, Islamabad 44000, Pakistan
2 Sahiba
back propagation algorithm, which increases the complexity of algorithm. ANN
takes too many features which cause the problem of overfitting. ANN can easily
trap in local minima. It takes greater time in testing the data. According to the his-
torical study, the lack of feature selection, data extraction, and efficient training pro-
cedures have become a major problem for accurate price forecasting [1]. However,
advancement in neural networks have a major impact on reducing such problems.
The development in neural networks progresses to make an accurate measurement
with less computational time. Therefore, the main purpose of this literature is to
propose the finest model for Electricity Price Forecasting (EPF). CNN is the intel-
ligent classifier for the prediction. Some external and internal features can easily
affect the EPF. Internal factors such as balance in supply, demand and EPF have an
immense effect on price forecasting. Market design, weather conditions, humidity
and temperature are the external factors affecting the price forecasting.These factors
have an important affect on the EPF shown in Fig. 1.
Fig. 1: Factors Affecting EPF
2 Related Work
In [1], authors use fundamental models for forecasting purpose in electricity market.
These models accurately forecast the electricity price as model depicts its demand
Short Term Electricity Price Forecasting 3
and supply. Recently, econometric time series models and machine learning meth-
ods, including Artificial Neural Networks (ANN) have been developed. However,
ANN model gets trapped in local minima, which is considered the biggest drawback
of this model. In [3], owing to complex drivers and sharp changes of electricity price,
forecasting electricity price accurately becomes a difficult task. Two main problems
are found in existing relevant models. Firstly, a hybridized model combined with
Empirical Mode Decomposition (EMD) suffers from its limitation, results in re-
ducing the accuracy of the model. Secondly, the aforementioned model is not able
to classify the features of electricity price, when it combined with linear and non-
linear models . However, improved EMD has a mode mixing problem. In [4], as
the increase in integration of Renewable Energy Resources (RER), it increases the
volatility of price, hence; there is unpredictable change in the market agent’s behav-
ior, so sudden drops in production and consumption occur more likely. However, this
imbalance increases the electrical grids unstability. In [6], the ANN model is used
for the prediction of electric load forecasting. Some optimization technique can be
applied to increase efficiency of deep learning model. In [7], authors have proposed
a framework for the prediction of over voltage. Sparse auto-encoder is used for the
forecasting. Single layer sparse auto-encoder is used for the dimensionality reduc-
tion and automate feature extraction. In [8], EMD is used to denoise input electrical
load signal. The processed data is transferred to mixed-Extreme Learning Machine
(ELM) for forecasting. Mixed kernel combines Radial Basis Function (RBF) ker-
nel and Unscented Kalman Filter (UKF) kernel. However, EMD has mode mixing
problem. ELM performs forecasting in one go as it is single feed forward neural
network, though, it cannot tune parameters effectively. So, accuracy of such models
is questionable. In [9], they proposed two stage model, the first is Simple Moving
Average (SMA) and second one is Random Forest [RF]. Although, the number of
decision trees increase in RF, it causes over fitting problem. RF does not perform
well in this work, when the number of trees are greater than 128. SMA predicts load
on previous trends. Although, there is more volatility in short time frames as com-
pared to long time frames. So, it does not show good results in short time frame.
In [10], they deal with the problem of feature selection. Feature selection is the key
concept for the accurate results. In many cases, load and price forecast methods face
the problem of appropriate feature selection techniques. In [12], the traditional short
term load forecasting is conducted by deterministic theories, i.e., Gray Model (GM),
time series method, Support Vector Regression (SVR), and Back-Propagation Neu-
ral Network (BPNN) model. However, all of these techniques have few limitations,
which are enlisted below.
The simulations of the interrelated parameters of loads and mathematical model’s
parameters become complex.
Enhancements are required in the forecasting accuracy.
There are not satisfactory forecasting effects.
These models are not able to reflect the actual load in real time environment.
4 Sahiba
The techniques [13]-[19] are also developed for the energy management in different
perspectives. By following all the aforementioned techniques, we have proposed a
new technique in order to overcome their trade-offs.
3 Proposed Methodology
In this paper, the proposed model is used for electricity price forecasting. In the pro-
posed model, three modules are used. In the first module, Mutual Information (MI)
is used for feature selection. In the second module, Principal Component Analysis
(PCA) is used for feature extraction. In the third module, CNN is used for electricity
price forecasting.
Fig. 2: Overview of System Model
3.1 Data Processing
CNN requires the best selection of input data. The dataset includes the feature of
weather conditions which require normalization and standardization. These pro-
cesses provide a common scale to input data. Seasonality is removed through nor-
3.2 Feature Selection
Feature selection is the process, which reduces the complexity of model and makes
it easier to interpret. It provides more accuracy to the model, if input data is correct
according to the situation. In paper [1], MI provides the best detection over the non-
linear relationship.
Short Term Electricity Price Forecasting 5
3.3 Feature Extraction
In the proposed model, PCA is used for feature extraction. This provides the less
computational time and more accuracy. PCA reduces the over fitting problem and
dimensionality by denoising the data. Data normalization is important for perform-
ing PCA.
3.4 Price Forecasting
The appropriate methods for electricity prices have considered the deterministic pat-
ters. CNN is used for electricity price forecasting. CNN consists of the multiple hid-
den layers. The first layer is called the convolution layer. Convolution layer contains
the filters used for mapping input data and prioritizing them. Relu function applies
the kernel filter to the matrices and assigns them weights in the first layer. The
neurons in this layer connect to other neurons and pass the information to hidden
layers. The second layer is pooling, which takes information from the convolution
layer, calculates the average and the maximum value of a class. It also reduces the
dimensionality. There are two types of pooling, one is large and other is small. Small
pooling covers little area of the field. Secondly, the pooling layer is used to extract
these features with high priority, which also processes the noise suppression. The
third layer is dense layer, in which each neuron is connected to the neuron of the
other layers. The structure of dense layer is similar to multiple layer perceptron. The
dense layer uses the matrix form of data and converts it into the vector form. The
fourth layer is dropout layer, which is a regularization technique aims to reduce the
complexity of the model with the goal to prevent over fitting.
4 Simulation Results
The final results of the prediction have been discussed in this section. Importance of
selected features is calculated by MI. A threshold value is defined for the selection
of the best features. Features that have importance smaller than or equal to a thresh-
old value as shown in Fig. 4, are selected for the prediction results. Some features
are highly important for the prediction, i.e., RT-LMP, RT-EC. Some features have
importance greater than 0.15. However, the value of the threshold is taken as 0.15.
The features that have greater values than the threshold, are dropped. PCA is used
for the extraction, which deals with dimensionality reduction, as shown in Fig. 5.
The appropriate methods for electricity prices have considered the deterministic
patters as well as stochastic components. CNN is used for EPF.
6 Sahiba
Fig. 3: Flow Chart of CNN
Fig. 4: Feature Selection using MI
Short Term Electricity Price Forecasting 7
Fig. 5: Feature Extraction Using PCA
Fig. 6: Electricity Price Forecasting Using CNN
5 Conclusion
CNN is the widely used method for forecasting short term EPF. ANN is used for
classification; however, it uses back propagation algorithm, which increases the
complexity of algorithm. ANN takes too many features, which cause the problem of
overfitting. ANN can easily trap in local minima. In this paper, we have proposed a
model for short term EPF using CNN with PCA. The proposed model is compared
with benchmark methods from the results and it is concluded that proposed model
performs better than the benchmark method.
8 Sahiba
1. Keles, D., Scelle, J., Paraschiv, F., and Fichtner, W. (2016). Extended forecast methods for
day-ahead electricity spot prices applying artificial neural networks. Applied energy, 162,
2. Wang, J., Liu, F., Song, Y., and Zhao, J. (2016). A novel model: Dynamic choice artificial neu-
ral network (DCANN) for an electricity price forecasting system. Applied Soft Computing,
48, 281-297.
3. Zhang, J. L., Zhang, Y. J., Li, D. Z., Tan, Z. F., and Ji, J. F. (2019). Forecasting day-ahead elec-
tricity prices using a new integrated model. International Journal of Electrical Power Energy
Systems, 105, 541-548.
4. Gao, W., Darvishan, A., Toghani, M., Mohammadi, M., Abedinia, O., and Ghadimi, N.
(2019). Different states of multi-block based forecast engine for price and load prediction.
International Journal of Electrical Power Energy Systems, 104, 423-435.
5. Wang, K., Xu, C., Zhang, Y., Guo, S., and Zomaya, A. Y. (2017). Robust big data analytics
for electricity price forecasting in the smart grid. IEEE Transactions on Big Data, 5(1), 34-45.
6. Qiu, X., Ren, Y., Suganthan, P. N., and Amaratunga, G. A. (2017). Empirical mode decom-
position based ensemble deep learning for load demand time series forecasting. Applied Soft
Computing, 54, 246-255.
7. Chinnathambi, R. A., Mukherjee, A., Campion, M., Salehfar, H., Hansen, T. M., Lin, J., and
Ranganathan, P. (2018). A Multi-Stage Price Forecasting Model for Day-Ahead Electricity
Markets. Energies, 1(1), 1-21.
8. Fan, G. F., Guo, Y. H., Zheng, J. M., and Hong, W. C. (2019). Application of the Weighted
K-Nearest Neighbor Algorithm for Short-Term Load Forecasting. Energies, 12(5), 1-19.
9. Chen, Y., Kloft, M., Yang, Y., Li, C., and Li, L. (2018). Mixed kernel based extreme learning
machine for electric load forecasting. Neurocomputing, 312, 90-106.
10. Qiu, X., Suganthan, P. N., and Amaratunga, G. A. (2018). Ensemble incremental learning
random vector functional link network for short-term electric load forecasting. Knowledge-
Based Systems, 145, 182-196.
11. Alanis, A. Y. (2018). Electricity prices forecasting using artificial neural networks. IEEE Latin
America Transactions, 16(1), 105-111.
12. Guo, Y., Han, S., Shen, C., Li, Y., Yin, X., and Bai, Y. (2018). An adaptive SVR for high-
frequency stock price forecasting. IEEE Access, 6, 11397-11404.
13. Hameed, A. R., Javaid, N., Islam, S. U., Ahmed, G., Qasim, U., and Khan, Z. A. (2016,
September). BEEC: Balanced energy efficient circular routing protocol for underwater wire-
less sensor networks. In 2016 International Conference on Intelligent Networking and Col-
laborative Systems (INCoS) (pp. 20-26). IEEE.
14. Zain-ul-Abidin, M., Khan, M. A., Javaid, N., Khizar, M., Khan, Z. A., and Qasim, U. (2016,
March). Enhanced single chain-based scheme in cylindrical underwater wireless sensor net-
works. In 2016 30th International Conference on Advanced Information Networking and Ap-
plications Workshops (WAINA) (pp. 343-348). IEEE.
15. Hafeez, T., Javaid, N., Hameed, A. R., Sher, A., Khan, Z. A., and Qasim, U. (2016, July).
AVN-AHH-VBF: Avoiding void node with adaptive hop-by-hop vector based forwarding for
underwater wireless sensor networks. In 2016 10th International Conference on Innovative
Mobile and Internet Services in Ubiquitous Computing (IMIS) (pp. 49-56). IEEE.
16. Shah, M., Javaid, N., Tariq, S., Imran, M., and Alnuem, M. (2015, September). A balanced
energy consumption protocol for underwater ASNs. In 18th IEEE International Conference
on Network-Based Information Systems (NBiS-2015), Taipei, Taiwan.
17. Fahim, H., Javaid, N., Qasim, U., Khan, Z. A., Javed, S., Hayat, A., ... and Rehman, G.
(2015, July). Interference and bandwidth aware depth based routing protocols in underwater
WSNs. In 2015 9th International Conference on Innovative Mobile and Internet Services in
Ubiquitous Computing (pp. 78-85). IEEE.
18. Awais, M., Javaid, N., Shaheen, N., Iqbal, Z., Rehman, G., Muhammad, K., and Ahmad, I.
(2015, September). An efficient genetic algorithm based demand side management scheme
Short Term Electricity Price Forecasting 9
for smart grid. In 2015 18th International Conference on Network-Based Information Systems
(pp. 351-356). IEEE.
19. Ashraf, H., Hassan, A., Khurshid, U., Mahmood, A., Shaheen, N., Khan, Z. A., ... and Javaid,
N. (2015, July). Peak load shaving model based on individual’s habit. In 2015 Ninth Inter-
national Conference on Complex, Intelligent, and Software Intensive Systems (pp. 276-282).
... After 2021, some researchers have also proposed a kernel-based extreme learning machine power price prediction model [20,21]. Recently, with the development of deep learning, electricity price prediction models based on deep learning have been continuously proposed, including BP, CNN, RNN, etc. [22][23][24]. Recently, the LSTM-based electricity price prediction method has received widespread attention because of its excellent performance. ...
Full-text available
The price of electricity is an important factor in the electricity market. Accurate electricity price forecasting (EPF) is very important to all competing electricity market parties. Decision-making in the electricity market is highly dependent on electricity prices, making an EPF model an important part of the orderly and efficient operation of the electricity market. Especially during the COVID-19 pandemic, the prices of raw materials for electricity production, such as coal, have risen sharply. Forecasting electricity prices has become particularly important. Currently, existing EPF prediction models face two main challenges: First, how to integrate multiscale electricity price data to obtain a higher prediction accuracy. Second, how to solve the problem of data noise caused by the fusion of EPF samples and multiscale data. To solve the above problems, we innovatively propose a tensor decomposition method to integrate multiscale electricity price data and use L1 regularization and wavelet transform to remove data noise. In general, this paper proposes a deep learning EPF prediction model, named the WT_TDLSTM model, based on tensor decomposition, a wavelet transform, and long short-term memory (LSTM). Among them, the LSTM method is used to predict electricity prices. We conducted experiments on three datasets. The experimental results of three data prove that the WT_TDLSTM model is better than the compared EPF model. The indicators of MSE and RMSE are 33.65%–99.97% better than the comparison model. We believe that the WT_TDLSTM model is a good supplement to the EPF model.
... CNN based models have gained attention in speech recognition, object detection, video processing and image classification due to their state-of-the-art performance [131,132,133]. Researchers have been using a variant of CNN model, known as the TCN model, for sequence modeling activities. ...
Electricity theft (ET) is a major problem in developing countries. It a�ects the economy that causes revenue loss. It also decreases the reliability and stability of electricity utilities. Due to these losses, the quality of supply e�ects and tari � imposed on legitimate consumers. ET is an essential part of Non-technical loss (NTL) and it is challenging for electricity utilities to �nd the responsible people. Several methodologies have developed to identify ET behaviors automatically. However, these approaches mainly assess records of consumers' electricity usage, may prove inadequate in detecting ET due to a variety of theft attacks and irregularity of consumers' behavior. Moreover, some important challenges are needed to be addressed. (i) The number of normal consumers has been wrongly identi�ed as fraudulent. This leads to high False-positive rate (FPR). After the detection of theft, on-site inspection is needed to validate the detected person, either is it fraudulent or not and it is costly. (ii) The imbalanced nature of datasets which negatively a�ect on the model's performance. (iii) The problem of over�tting and generalization error is often faced in deep learning models, predicts unseen data inaccurately. So, the motivation for this work to detect illegal consumers accurately. We have proposed four Arti�cial intelligence (AI) models in this thesis. In system model 1, we have proposed Enhanced arti�cial neural network blocks with skip connections (EANNBS). It makes training easier, reduces over�tting, FPR and generalization error, as well as execution time. Temporal convolutional network with enhanced multi-layer perceptron (TCN-EMLP) is proposed in system model 2. It analyzes the sequential data based on daily electricity-usage records, obtained from smart meters. At the same time, EMLP integrates the non-sequential auxiliary data, such as data related to electrical connection type, property area, electrical appliances usage, etc. System model 3 based on Residual network (RN) that is used to automate feature extraction while three tree-based classi�ers such as Decision tree (DT), Random forest (RF) and Adaptive boosting (AdaBoost) are trained on the obtained features for classi�cation. Hyperparameter tuning toolkit is presented in this system model, named as Hyperactive optimization toolkit. Bayesian is used as an optimizer in this toolkit that aims to simplify the tuning process of DT, RF and AdaBoost. In system model 4, input is forwarded to three di�erent and well-known Machine learning (ML) techniques, i.e., Support vector machine (SVM), as an input. At this stage, a meta-heuristic algorithm named Simulated annealing (SA) is employed to acquire optimal values for ML models' hyperparameters. Finally, ML models' outputs are used as features for meta-classi�ers to achieve �nal classi�cation with Light Gradient boosting machine (LGBM) and Multi-layer perceptron (MLP). Furthermore, Pakistan residential electricity consumption dataset (PRECON1), State grid corporation of china (SGCC2) and Commission for energy regulation (CER3) datasets is used in this thesis. SGCC dataset contains 9% fraudulent consumers, which are extremely less than non-fraudulent consumers, due to the imbalance nature of data. Furthermore, many classi�cation techniques have poor predictive class accuracy for the positive class. These techniques mainly focus on minimizing the error rate while ignoring the minority class. Many re-sampling techniques are used in literature to adjust the class ratio; however, sometimes, these techniques remove the important information that is necessary to learn the model and cause over�tting. By using six previous theft attacks, we generate theft cases to mimic the real world theft attacks in original data. We have proposed the combination of oversampling and under-sampling techniques that is Near miss borderline synthetic minority oversampling technique (NMB-SMOTE), Tomek link borderline synthetic minority oversampling technique with support vector machine (TBSSVM) and Synthetic minority oversampling technique with near miss (SMOTE-NM) to handle imbalanced classi�cation problem. We have conducted a comprehensive experiment using SGCC, CER and PRECON datasets. The performance of suggested model is validated using di�erent performance metrics that are derived from Confusion matrix (CM).
Recently, forecasting is an important task in various domains. Forecasting has applications in various domains where future estimates are very useful such as economics, weather, transportation, environment, sales and production, finance, sports, and health care. It helps to predict future values based on present and past data. It needs to process a large volume of data to analyze the trend and identify the behavior of the historical data to forecast the future. In a day, a large volume of data in terms of petabytes is generated from various devices, sensors, and social media which helps to develop forecasting applications. To perform forecasting, different approaches like qualitative and quantitative methods are used. Traditional approaches are not suitable to process the huge volume of data results of the inefficient forecasting models. Hence, a deep architecture of neural networks like deep learning techniques (DLT) is more suitable to process, analyze, and forecast models. Also, DLT supports the nonlinear data, multivariate data, and multistep data in developing efficient forecasting models. The main motivation of this chapter is to brief the use of deep learning techniques to develop forecasting models in various domains. This chapter mainly focuses on the different deep learning architectures such as fully connected neural networks (FCNN), recursive neural network, convolutional neural network (CNN), recurrent neural network (RNN), generative adversarial networks (GAN), and deep reinforcement learning to develop forecasting models with various parameters and reveal the applications of these methods.
Full-text available
In this paper, the historical power load data from the National Electricity Market (Australia) is used to analyze the characteristics and regulations of electricity (the average value of every eight hours). Then, considering the inverse of Euclidean distance as the weight, this paper proposes a novel short-term load forecasting model based on the weighted k-nearest neighbor algorithm to receive higher satisfied accuracy. In addition, the forecasting errors are compared with the back-propagation neural network model and the autoregressive moving average model. The comparison results demonstrate that the proposed forecasting model could reflect variation trend and has good fitting ability in short-term load forecasting.
Full-text available
Forecasting hourly spot prices for real-time electricity markets is a key activity in economic and energy trading operations. This paper proposes a novel two-stage approach that uses a combination of Auto-Regressive Integrated Moving Average (ARIMA) with other forecasting models to improve residual errors in predicting the hourly spot prices. In Stage-1, the day-ahead price is forecasted using ARIMA and then the resulting residuals are fed to another forecasting method in Stage-2. This approach was successfully tested using datasets from the Iberian electricity market with duration periods ranging from one-week to ninety days for variables such as price, load and temperature. A comprehensive set of 17 variables were included in the proposed model to predict the day-ahead electricity price. The Mean Absolute Percentage Error (MAPE) results indicate that ARIMA-GLM combination performs better for longer duration periods, while ARIMA-SVM combination performs better for shorter duration periods.
Full-text available
This paper presents the results of the use of training algorithms for recurrent neural networks based on the extended Kalman filter and its use in electric energy price prediction, for both cases: one-step ahead and n-step ahead. In addition, it is included the stability proof using the well-known Lyapunov methodology, for the proposed artificial neural network trained with an algorithm based on the extended Kalman filter. Finally, the applicability of the proposed prediction scheme is shown by mean of the one-step ahead and n-step ahead prediction using data from the European power system.
Full-text available
Load demand forecasting is a critical process in the planning of electric utilities. An ensemble method composed of Empirical Mode Decomposition (EMD) algorithm and deep learning approach is presented in this work. For this purpose, the load demand series were first decomposed into several intrinsic mode functions (IMFs). Then a Deep Belief Network (DBN) including two Restricted Boltzmann Machines (RBMs) was used to model each of the extracted IMFs, so that the tendencies of these IMFs can be accurately predicted. Finally, the prediction results of all IMFs can be combined by either unbiased or weighted summation to obtain an aggregated output for load demand. The electricity load demand data sets from Australian Energy Market Operator (AEMO) are used to test the effectiveness of the proposed EMD-based DBN approach. Simulation results demonstrated attractiveness of the proposed method compared with nine forecasting methods.
Electricity price forecasting proves useful for power producers and consumers to make proper decisions in a market-oriented environment. However, due to the complex drivers and sharp fluctuation of electricity prices, accurate electricity price forecasting turns to be very difficult. To better capture the characteristics of day-ahead electricity prices, a new integrated model based on the improved empirical mode decomposition (IEMD), autoregressive moving average with exogenous terms (ARMAX), exponential generalized autoregressive conditional heteroscedasticity (EGARCH) and adaptive network-based fuzzy inference system (ANFIS) is proposed in this paper. Then it is validated by using the data from Spanish and Australian electricity markets. The results indicate that the forecasting accuracy of the new integrated model proves higher than that of some well-recognized models in the literature.
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.
In order to mitigate investmentsk, stock price forecasting has attracted more attention in recent years. Aiming at the discreteness, non-normality, high-noise in high-frequency data, a support vector machine regression(SVR) algorithm is introduced in this paper. However, the characteristics in different periods of the same stock, or the same periods of different stocks are significantly different. So, SVR with fixed parameters is difficult to satisfy with the constantly changing data flow. To tackle this problem, an adaptive SVR was proposed for stock data at three different time scales, including daily data, 30-minute data, and 5-minute data. Experiments show that the improved SVR with dynamic optimization of learning parameters by Particle Swarm Optimization(PSO) can get a better result than compared methods including SVR and Back-propagation neural network(BPNN).
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.
Electricity price forecasting is a significant part of smart grid because it makes smart grid cost efficient. Nevertheless, existing methods for price forecasting may be difficult to handle with huge price data in the grid, since the redundancy from feature selection cannot be averted and an integrated infrastructure is also lacked for coordinating the procedures in electricity price forecasting. To solve such a problem, a novel electricity price forecasting model is developed. Specifically, three modules are integrated in the proposed model. First, by merging of Random Forest (RF) and Relief-F algorithm, we propose a hybrid feature selector based on Grey Correlation Analysis (GCA) to eliminate the feature redundancy. Second, an integration of Kernel function and Principle Component Analysis (KPCA) is used in feature extraction process to realize the dimensionality reduction. Finally, to forecast price classification, we put forward a differential evolution (DE) based Support Vector Machine (SVM) classifier. Our proposed electricity price forecasting model is realized via these three parts. Numerical results show that our proposal has superior performance than other methods.