Electricity Consumption Prediction for Xinjiang Electric Energy Replacement

Song, Xinfu; Liang, Gang; Li, Changzu; Chen, Weiwei

doi:https://doi.org/10.1155/2019/3262591

Mathematical Problems in Engineering

On this page

Abstract Introduction Results and Discussion Conclusions Data Availability Conflicts of Interest Acknowledgments References Copyright Related Articles

Research Article | Open Access

Volume 2019 | Article ID 3262591 | https://doi.org/10.1155/2019/3262591

Electricity Consumption Prediction for Xinjiang Electric Energy Replacement

Xinfu Song,¹Gang Liang,¹Changzu Li,^2,3and Weiwei Chen¹

Academic Editor: Gaetano Zizzo

Received16 Jul 2018

Revised24 Nov 2018

Accepted13 Jan 2019

Published20 Mar 2019

Abstract

In recent years, the phenomenon of wind and solar energy abandoned in Xinjiang’s new energy has become severe, the contradiction between the supply and demand of the power grid is obvious, and the proportion of power in the energy consumption structure is relatively low, thus hindering the development of Xinjiang’s green power. In this context, the focus of Xinjiang’s power has shifted to promote the development of electric energy replacement. Therefore, using the Xinjiang region as an example, we first select the important indicators such as the terminal energy substitution in Xinjiang, added value of the secondary industry, population, terminal power consumption intensity, and per capita disposable income. Subsequently, eight combined forecasting models based on the grey model (GM), multiple linear regression (MLR), and error back propagation neural network (BP) are constructed to predict and analyse the electricity consumption of the whole society in Xinjiang. The results indicate the optimal weighted combination forecasting model, GM-MLR-BP of the induced ordered weighted harmonic averaging operator (IOWHA operator), exhibits better prediction accuracy, and the effectiveness of the proposed method is proven.

1. Introduction

Xinjiang is located in the northwest of China and contains abundant energy resource. In addition to the three major oil fields, the nine major coal fields, and other fossil resource mining bases, nine wind zones and renewable energy power generation bases such as those for solar energy exist. Among them, the total amount of wind and solar energy of Xinjiang ranks second in China, including wind power of approximately 3 trillion kW·h, and reserve of solar energy of approximately 20 trillion kW·h that provide a solid material foundation for the development of Xinjiang’s national economy and electric energy replacement [1, 2]. In addition, a low-level industrial structure, an unbalanced regional economy, highly extensive energy consumption, and increasingly prominent environmental problems exist. Severe air pollution has restricted the economic development of Xinjiang, but the application of electric energy in the terminal consumption is significantly better than that of fossil fuel. With the introduction of energy-saving and emission-reduction policies, the proportion of clean energy generation is increasing. The replacement of energy in the terminal is critical for reducing urban pollutant emissions. According to statistics, as of March this year, Xinjiang Electric Power Company has implemented 6,408 electric energy replacement projects, achieving a total of 8.272 billion kW·h of electricity, thus effectively reducing the consumption of 2.895 million tons of standard coal, reducing carbon dioxide emissions by 7.238 million tons, thus reducing the use of terminal direct-fired coal and direct fuel significantly, and promoting the “electrification” process of various industries [3]. Therefore, it is critical both theoretically and economically to study the power consumption forecasting method under the background of Xinjiang electric energy replacement to maintain the safe, efficient, and stable operation of the power grid system.

Currently, many methods exist for predicting electric load in China. According to the structure of the model, it can be divided into the single prediction method and combined prediction method. Among them, single prediction methods include the time series method [4–6], linear regression method [7–9], grey forecasting method [10, 11], support vector machine [12–14], and BP neural network [15–17]. However, any single prediction method in practical applications, owing to their own defects, causes insufficient prediction accuracy, and it is difficult to accurately predict the future power consumption level of the region. The combined prediction model can comprehensively utilise the information provided by various methods and assign different weights according to the precision of the single prediction method, which is helpful to improve the scientificity and effectiveness of the prediction. The literature [18] analysed the annual peak load changes in Guangzhou and considered the electricity consumption situation in Guangzhou from the perspective of the annual peak load, total output value of the three major industries, population, and the per capita Gross Domestic Product (GDP). Subsequently, the variance-covariance method based on the combined BP neural network, grey prediction model, and multiple linear regression method is proposed to predict the annual maximum load in 2017–2019. The literature [19] selected the total social electricity consumption data of Anhui province from 2000 to 2014 and constructed multiple linear regression based on the economics such as the population, GDP, total energy consumption, household consumption level, and per capita disposable income of urban households. The combined predictive model of the IOWA operator, IOWHA operator, and IOWGA operator based on the multiple linear regression, double exponential smoothing, and polynomial fitting is constructed to predict the total social power consumption in Anhui province for the next five years. However, it is rare to study the long-term load forecasting method under the electric energy replacement in Xinjiang from the perspective of electric energy replacement and electricity intensity. Therefore, we use the IOWHA preferred combination prediction model based on the grey forecasting method, multiple linear regression, and BP neural network (GM-MLR-BP) to analyse the influence of the terminal electric energy replacement amount, terminal electric energy consumption intensity, added value of the second industry, population, and per capita disposable income of residents in the whole district. The experimental results indicate that the prediction method is effective, and a new idea for regional electricity consumption forecasting under electric energy replacement is provided.

2. Background Theory

2.1. GM (1, 1)

2.1.1. Model Principle

The grey prediction model is a prediction method based on a small amount of incomplete information to establish a mathematical model and predict the future development trend. Through the correlation analysis of system factors, raw data are generated and processed, and the data sequence with strong regularity is generated to obtain the law of system variation to establish the corresponding differential equation model to solve the predicted value. The prediction method is simple to operate and contains a wide range of applications in the field of prediction. Because only a small amount of information is required in the modelling process, high precision requirements can be achieved; therefore, it is also an effective tool for handling small sample prediction problems [20–23].

2.1.2. Modelling Steps

(1) Suppose the raw data sequence of electricity consumption is ,(2) is subjected to an additive preprocessing to generate a new sequence , (3) Establish a corresponding Albino equation:where n is the number of the raw data sequence and and are the parameters.

(4) Suppose . Calculate data Array and data Column , and solve the parameters and ,

where is the pending parameter; and are the predictive values of and , respectively; and are the known quantities that can be obtained by the following two formulas, respectively:(5) Substitute the obtained , into formula (5), which can obtain (6) The generation sequence of GM(1,1) is obtained by the discretisation of equation (9); the grey model is as follows:where .

(7) After the subtractive reduction of (5), the grey prediction model is obtained as follows:where .

2.2. Multiple Linear Regression (MLR)

2.2.1. Model Principle

MLR is based on the optimal combination of multiple independent variables to predict or estimate the dependent variable and subsequently establish a statistical method for the quantitative relationship of linear mathematical models of multiple variables. It reflects the relationship between a phenomenon and a variety of influencing factors. Therefore, it is used widely in all aspects of society [24–26].

Suppose , where i are independent variables without collinearity, Y is a dependent variable; therefore, the MLR model is expressed as follows:where are the unknown parameters; is a constant;

are the regression coefficients of the multiple linear equation;

is a random error.

2.2.2. Modelling Steps

(1)Filter the problem’s independent variables and dependent variable Y. Perform the correlation analysis and collinearity diagnostics of the data using SPSS software to test the model fitness and collinearity between independent variables.(2)Perform the holistic test to test the significance of the overall regression relationship.(3)Test the regression coefficient to discover the significance of each regression coefficient.(4)Define the final independent variables and the corresponding parameters , and establish a multiple linear regression equation.

2.3. Error Back Propagation Neural Network (BP)

2.3.1. Model Principle

BP is an intelligent learning machine trained by an error inverse propagation algorithm. It consists of an input layer, hidden layer, and output layer. One or more nodes exist between each layer, and each node represents a specific one. The excitation function that is the connection between the node and the next node represents a weighting value for the signal passing through the connection, thus reflecting the strength of the connection between the units. The BP neural network has the characteristics of learning and storing a large number of input–output mode mapping relationships. It is suitable for solving any complex nonlinear problems and exhibits good approximation and generalisation ability. It has been used widely in pattern recognition, function approximation, data compression, etc. [27–30].

2.3.2. Modelling Steps

(1) Establish a sample data classification that is used as the training set, test set, and prediction set.

(2) Normalise the data.

(3) Establish a neural network to determine the number of layers in the input layer, hidden layer, output layer, number of nodes between each layer, transfer functions, etc.

(4) Set the network training parameters, including the training times, training accuracy, learning rate, and training network.

(5) Test the model using the test suite.

(6) The prediction results are antinormalized.

(7) Analyse and evaluate the prediction results.

3. Induced Ordered Weighted Harmonic Averaging Operator

3.1. Simple Average Combination [31, 32]

Suppose that the predicted variable Y has m prediction results , and each prediction result is given a weight; subsequently, the predicted simple average combined value is . The predicted simple average combined forecast value is as follows:

3.2. Induced Ordered Weighted Harmonic Averaging Operator (IOWHA)

The traditional prediction method only assigns the weights of different prediction indicators, ignoring the influence of the time point on the prediction accuracy, which leads to the accuracy of one time point prediction in the prediction; however, the prediction error is large. At this time, the combined prediction model IOWHA that overcomes the defect of the traditional combined prediction model’s weight is constructed by the minimum squared error sum of the inverse of the predicted value [33–35].

The model is defined as follows:

Suppose n two-dimensional combinations exist and the n-dimensional induced ordered weighted harmonic mean operator is expressed aswhere the function is generated based on the sequence , the induction value of is , and is the ith data after are arranged by size. The weight vector of the function is that corresponds to . Obviously, the formula indicates that the IOWHA operator is an ordered weighted harmonic average of the data in the corresponding after sorting the . The weight vectors have no concern with position and size of but are related to the position and size of the induced value . Suppose that represents the prediction accuracy of the ith prediction method at the tth time; subsequently, n prediction methods form n two-dimensional arrays at time t, and , , and represent the prediction accuracy, actual observations, and predicted values, respectively. The formula is as follows:where .

Suppose is used as the weighting variable of the prediction model, and the predicted value at time t can be calculated by equation (14):where

The M phase based on the squared sum of the predicted reciprocal errors of the optimal combination prediction model of IOWHA is , which is as follows:Subsequently, the combined prediction model based on the minimum squared error sum of the inverse of the predicted value is as follows:The algorithm steps of the IOWHA optimal weighted prediction model GM-MLR-BP are shown in Figure 1.

4. Results and Discussion

4.1. Data Processing

4.1.1. Definition of Terminal Energy Replacement

To analyse the impact of electric energy substitution policy on Xinjiang’s electric load, this paper quantifies the potential of electric energy replacement and uses terminal electric energy replacement as an influencing factor of the whole society’s electricity consumption under the replacement of Xinjiang electric energy. Suppose the total annual energy consumption is in the base year, electrical energy consumption is in the base year, actual energy consumption is in the tth year, and is the electrical energy consumption in the tth year. Subsequently, the terminal electric energy replacement amount in the tth year is expressed as follows [36]:

4.1.2. Model Application Indicators

In this study, we used four indicators to compare the statistical characteristics between a single prediction model and a combined prediction model.(1)The RMSE error is expressed as follows:(2)The MAPE is expressed as follows:(3)The MSPE is expressed as follows:

4.2. Data Source

Through the correlation analysis of SPSS, the correlation coefficient between the added value of Xinjiang’s secondary industry, per capita disposable income of the whole district, population, terminal energy replacement, power consumption intensity, actual GDP and per capita disposable income of residents in the whole district are both above 0.8, and the P values are less than 0.05. These indicate that the differences between the samples are significant. Therefore, we selected the added value of Xinjiang’s secondary industry, per capita disposable income, population, terminal energy replacement, electricity consumption intensity, and actual GDP from 2000 to 2017 as the influencing factors of the whole society’s electricity consumption in Xinjiang (see Table 1).

The total social power consumption in Xinjiang from 2000–2017 was selected as the research object. The data from 2000 to 2012 was used as the training sample, and the data from the previous five years (2013–2017) were used as the test sample to test the effectiveness of the prediction model. (See Figure 2).

According to the collected data information, the BP neural network structure was constructed, including an input layer with 6 nodes, an intermediate layer with 10 nodes, and an output layer for 1 node. Subsequently, correlation analysis, collinearity diagnostics, and the test of regression coefficient were performed for multiple linear regression equations. Finally, the five indicators of power consumption intensity, such as the added value of Xinjiang’s secondary industry, per capita disposable income of the whole district, population, and terminal energy replacement from 2000 to 2017, were used as independent variables [37].

4.3. Comparison of Prediction Accuracy

Based on the principle of the simple average combined model and the IWOHA operator optimal weighted combination model, the simple average combined weights of GM-MLR, GM-BP, and MLR-BP are , The weight of GM-MLR-BP is . The comparison results of the actual social electricity consumption value and the simple average combined forecast value in Xinjiang in 2013–2017 are shown in Figure 3. The IWOHA optimal weighted combination weight is obtained by Matlab, and the weights of GM-MLR are , . The weights of GM-BP are , ; the weights of MLR-BP are , ; the weights of GM-MLR-BP are , , . The comparison between the actual value of the total social electricity consumption in Xinjiang and the optimal weighted combination forecast of the IWOHA operator from 2013 to 2017 is shown in Figure 4.

As shown in Figure 3, the GM-MLR-BP model exhibits a better fitting effect in the results of the simple combined prediction model.

As shown in Figure 4, the GM-MLR-BP model exhibits a better fitting effect in the IOWHA optimal weighted combination forecasting model.

The comparison between the actual value of total electricity consumption in Xinjiang and the single and combined forecasting model is shown in Table 2. The data indicates that the maximum errors of GM, MLR, and BP are 10.75%, 1.46%, and 4.86%, respectively, and the errors of MLR and BP are relatively small. The error of GM is relatively large, but less than 10%; the maximum phase difference errors of the simple average combination models GM-MLR, GM-BP, MLR-BP, and GM-MLR-BP are 4.86%, 6.44%, 2.68%, and 3.81%, respectively. The maximum phase difference errors of the four IWOHA operator optimal weighted combination prediction models are 0.62%, 4.74%, 1.33%, and 0.33%, respectively. As shown, the IWOHA operator optimal weighted combination prediction model reduces the single prediction. The stability is relatively strong and the prediction accuracy is high.

The comparison of the prediction accuracy of the single evaluation method and the combined prediction method by the application evaluation indexes RMSE, MAPE, and MSPE is shown in Table 3. The study found that the RMSE, MAPE, and MSPE of the single prediction method MLR model demonstrate the lowest index values, indicating that the model is more reasonable for power load forecasting. The prediction accuracy of the simple combination prediction model is smaller than that of the MLR model. The prediction accuracy of the IOWHA optimal weighted combination prediction model is better than that of the corresponding simple combination prediction model and single prediction model. Meanwhile, the RMSE, MAPE, and MSPE values of the GM-MLR-BP model are significantly smaller than those of the MLR model, and the prediction accuracy is higher.

4.4. Analysis of Xinjiang Power Load Forecasting

The GM-MLR-BP model of the optimal weighted combination of IOWHA is used to predict the power load in Xinjiang in the next 10 years, as shown in Table 4.

As shown in Figure 5, the total electricity consumption in Xinjiang is from 15.75 billion kWh in 2000 to 563.791 billion kWh in 2027; the overall trend is increasing, and the upward trend is gradually increasing. In general, it is primarily due to the continuous improvement in relevant plans for electric energy substitution in Xinjiang and the rapid development of the following aspects.

In terms of technology, the Xinjiang region has continuously improved the charging service price mechanism, accelerated the construction of new energy vehicles and charging piles, and promoted the development of transportation electrification. Simultaneously, it has promoted the construction of wind energy storage projects and increased the research and development of energy storage core technologies. The ability is expanded to dissipate the electricity locally.

In terms of policies, the Xinjiang Electric Power Company has continuously accelerated the research and implementation of electric heating price policy to promote the construction of heating facilities such as electric heating and electric boilers. In addition, to fully utilise other provinces’ policy of power transmission and deeply study the scale and development speed of the power market from other provinces, it vigorously promoted the power poverty alleviation policy in southern Xinjiang and established free electricity use files for poor households to improve the rural electrification level. Meanwhile, to accelerate the expansion of Xinjiang’s layout of power silk, it also established long-term cooperative relations with other provinces to improve the power consumption capacity in Xinjiang.

In terms of management, the Xinjiang government guided by market demand has continuously deepened the reform of power supply measurement and scientifically controlled the development pace of new energy projects and the construction scale. Moreover, it continuously improved the market-based economic compensation mechanism to promote the construction of the electricity market, expanded the direct transaction scale of new energy enterprises, and increased the proportion of new energy in peaking alternative trading.

5. Conclusions

In this study, we combined the characteristics of Xinjiang’s current development policy and selected some indicators such as terminal electric energy substitution, power consumption intensity, added value of secondary industry, per capita disposable income of the whole district, and population as the influencing factors of Xinjiang’s electric load. Simultaneously, to fully utilise the useful information of the model and improve the limitations of the single model load forecasting, the optimal weighted combination model of IOWHA was proposed. By comparing the prediction results of the single prediction model and simple combination prediction model, the RMSE, MAPE, and MSPE of the IOWHA optimal weighted combination model GM-MLR-BP were small. Therefore, we effectively verified that the prediction accuracy of the IOWHA optimal weighted combination model GM-MLR-BP was higher than that of the corresponding single model prediction and simple combination prediction model and that it exhibited good practicability for predicting the electricity consumption in the background of electric energy substitution in Xinjiang.

Data Availability

The relevant data involved in this paper are provided by State Grid Xinjiang in China. According to the confidentiality agreement of State Grid Power Company, the relevant data cannot be released completely. However, some data can be found at http://www.xjtj.gov.cn/ from the Bureau of Statistics of Xinjiang, and the data presented herein are summarised in Table 5.

Conflicts of Interest

The authors declare that they have no conflicts of interest.

Acknowledgments

This paper is supported by Natural Science Foundation of China (project no. 71471059). The paper is also supported by the Fundamental Research Funds for the Central Universities (2018ZD14). Finally, we thank the 111 Project (B18021) for their support.

References

H. Y. Ma, “Countermeasures and Suggestions for Promoting Xinjiang's Electric Energy Substitution,” Power Demand Side Management, vol. 18, no. 5, pp. 48–49+51, 2016.
View at: Google Scholar
F. Chen, “Analysis and Suggestions on the Status Quo of Xinjiang Electric Power Development, Modern Industrial Economy and Informationization,” Modern Industrial Economy and Informationization, no. 12, pp. 21–23, 2012.
View at: Google Scholar
B. Ran and Y. F. Li, “Xinjiang has realized a total of 8.272 billion kWh of electricity replacement,” People's Daily Online, 2018, http://k.sina.com.cn/article_2286908003_884f72630200063wb.html.
View at: Google Scholar
G. E. P. Box and G. M. Jenkins, “Time Series Analysis: Forecasting and Control rev. Ed,” Journal of Time, vol. 31, no. 4, pp. 238–242, 1976.
View at: Google Scholar | MathSciNet
D. F. Chen, Z. Zheng, and Q. Wang, “Analysis of Short-term Load Forecasting Problem of Power System Based on Time Series,” Automated Application, no. 11, pp. 99–101, 2017.
View at: Google Scholar
P. Jiang, S. Qin, J. Wu, and B. Sun, “Time Series Analysis and Forecasting for Wind Speeds Using Support Vector Regression Coupled with Artificial Intelligent Algorithms,” Mathematical Problems in Engineering, vol. 2015, Article ID 939305, 14 pages, 2015.
View at: Google Scholar
B. Zhang, S. Y. Yang, and Q. Wang, “Application of Linear Regression Method in Power Multilateral Trading,” Science & Technology of BaoTou Steel, vol. 43, no. 2, pp. 82–84+88, 2017.
View at: Google Scholar
V. Bianco, O. Manca, and S. Nardini, “Electricity consumption forecasting in Italy using linear regression models,” Energy, vol. 34, no. 9, pp. 1413–1421, 2009.
View at: Publisher Site | Google Scholar
K. Panklib, C. Prakasvudhisarn, and D. Khummongkol, “Electricity Consumption Forecasting in Thailand Using an Artificial Neural Network and Multiple Linear Regression,” Energy Sources, Part B: Economics, Planning, and Policy, vol. 10, no. 4, pp. 427–434, 2015.
View at: Publisher Site | Google Scholar
C. Zhao, H. H. Jie, and H. X. Yang, “GM(2, 1) Supply Chain Risk Prediction Based on Genetic Optimization Ridge Regression,” Statistics & Decision, no. 10, pp. 51–53, 2018.
View at: Google Scholar
E. Kayacan, B. Ulutas, and O. Kaynak, “Grey system theory-based models in time series prediction,” Expert Systems with Applications, vol. 37, no. 2, pp. 1784–1789, 2010.
View at: Publisher Site | Google Scholar
B. Y. Chen, C. G. Zhang, M. Ni, and L. Zhang, “Power grid load forecasting based on fuzzy information granulating support vector machine,” Electronic Design Engineering, 2018.
View at: Google Scholar
B. R. Chang and H. F. Tsai, “Forecast approach using neural network adaptation to support vector regression grey model and generalized auto-regressive conditional heteroscedasticity,” Expert Systems with Applications, vol. 34, no. 2, pp. 925–934, 2008.
View at: Publisher Site | Google Scholar
J. Duan, X. Qiu, W. Ma, X. Tian, and D. Shang, “Electricity Consumption Forecasting Scheme via Improved LSSVM with Maximum Correntropy Criterion,” Entropy, vol. 20, no. 2, p. 112, 2018.
View at: Publisher Site | Google Scholar
Y. Long, Z. Y. Su, and Y. Wang, “Monthly load forecasting based on seasonal adjustment and BP neural network,” System Engineering Theory and Practice | System Eng Theor Prac, vol. 38, no. 4, pp. 1052–1060, 2018.
View at: Google Scholar
K. Kandananond, “Forecasting electricity demand in Thailand with an artificial neural network approach,” Energies, vol. 4, no. 8, pp. 1246–1257, 2011.
View at: Publisher Site | Google Scholar
P. Singh, K. K. Mishra, and P. Dwivedi, “Enhanced hybrid model for electricity load forecast through artificial neural network and Jaya algorithm,” in Proceedings of the International Conference on Intelligent Computing and Control Systems (ICICCS '18), pp. 115–120, Madurai, 2018.
View at: Publisher Site | Google Scholar
Y. X. Wu and M. Xie, “Annual Maximum Load Forecasting Based on Grey Regression Combined Model of BP Neural Network,” Southern Energy Construction, vol. 4, no. 2, pp. 46–50+57, 2017.
View at: Google Scholar
C. J. W. Wei, G. Y. Yang, and H. J. Yuan, “Research on the Forecast of the Whole Society's Electricity Consumption Based on the GYOWA Operator,” Journal of Jiaxing University, vol. 29, no. 1, pp. 57–63, 2017.
View at: Google Scholar
X. L. Tang, P. Wang, S. C. Li, and Z. Y. Bai, “Medium and long-term electric load forecasting based on variance-covariance combination forecasting,” Electrical Engineering, no. 1, pp. 15–18, 2015.
View at: Google Scholar
Z. Zhong, C. Yang, W. Cao, and C. Yan, “Short-Term Photovoltaic Power Generation Forecasting Based on Multivariable Grey Theory Model with Parameter Optimization,” Mathematical Problems in Engineering, vol. 2017, Article ID 5812394, 9 pages, 2017.
View at: Publisher Site | Google Scholar
H. Zhao and S. Guo, “An optimized grey model for annual power load forecasting,” Energy, vol. 107, pp. 272–286, 2016.
View at: Publisher Site | Google Scholar
Z. He, Y. Shen, and Q. Wang, “Boundary extension for HilbertHuang transform inspired by gray prediction model,” Signal Processing, vol. 92, no. 3, pp. 685–697, 2012.
View at: Publisher Site | Google Scholar
P. F. Zhou and Z. Y. Lu, “Prediction of urban water consumption based on SPSS multiple linear regression model,” Water Conservancy Science and Technology and Economy, vol. 24, no. 5, pp. 6–10, 2017.
View at: Google Scholar
S. Fei and J. Yang, “Towards Sustainable Cities: Using Multiple Linear Regression Model to Identify Influencing Factors to Promote the Use of Public Transport in Downtown Nanjing,” China City Planning Review, vol. 26, no. 4, pp. 17–24, 2017.
View at: Google Scholar
P. Shine, T. Scully, J. Upton, and M. D. Murphy, “Multiple linear regression modelling of on-farm direct water and electricity consumption on pasture based dairy farms,” Computers and Electronics in Agriculture, vol. 148, pp. 337–346, 2018.
View at: Publisher Site | Google Scholar
Y. E. Shao, “Prediction of Currency Volume Issued in Taiwan Using a Hybrid Artificial Neural Network and Multiple Regression Approach,” Mathematical Problems in Engineering, vol. 2013, no. 3, Article ID 676742, pp. 237–245, 2013.
View at: Publisher Site | Google Scholar
Q. Nie, “Talking about Particle Swarm Optimization Algorithm and BP Neural Network,” Qing Fang Gong Ye Yu Ji Shu, vol. 42, no. 1, pp. 68–70, 2013.
View at: Google Scholar
Q. Zhou, B. Yong, and F. Li, “A novel Monte Carlo based neural network model for electricity load forecasting,” International Journal of High Performance Computing & Networking, vol. 1, no. 1, 2018.
View at: Google Scholar
S. X. Wang, M. Li, and L. Zhao, “Short-term wind power prediction based on improved small-world neural network,” Neural Computing & Applications, vol. 2013, pp. 1–13, 2018.
View at: Google Scholar
A. Laouafi, M. Mordjaoui, S. Haddad, T. E. Boukelia, and A. Ganouche, “Online electricity demand forecasting based on an effective forecast combination methodology,” Electric Power Systems Research, vol. 148, pp. 35–47, 2017.
View at: Publisher Site | Google Scholar
Y. M. Yang, Q. S. Li, and J. Fang, “Coal Consumption Prediction Analysis Based on Optimal Weighted Combination Model,” Coal Engineering, vol. 5, pp. 156–160, 2018.
View at: Google Scholar
N. Wang, “Construction of Logistics Demand Combination Improvement and Forecasting Model Based on IOWHA Method,” Journal of Commercial Economics, vol. 14, pp. 67-68, 2016.
View at: Google Scholar
S. Li, Y. Li, and L. Wang, “Improved IOWHA operator combination forecasting model,” Computer Engineering and Applications, vol. 51, no. 3, pp. 260–264, 2015.
View at: Google Scholar
H. Chen, L. Jin, and X. Li, “The optimal interval combination forecasting model based on closeness degree and IOWHA operator under the uncertain environment,” Grey Systems Theory & Application, vol. 1, no. 3, pp. 250–260, 2013.
View at: Google Scholar
B. G. Shan, J. Zhao, D. X. Jia et al., “Analysis method of electric energy substitution potential based on STIRPAT-ridge regression,” Distribution & Utilization, vol. 1, pp. 68–73, 2018.
View at: Google Scholar
Y. Sun, S. Zhou, B. G. Shan, D. X. Jia, and F. Cao, “Analysis of electric energy substitution potential under multiple scenarios,” Power System Technology, vol. 41, no. 1, pp. 118–123, 2017.
View at: Google Scholar

Copyright

Copyright © 2019 Xinfu Song et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

PDF Download Citation

Download other formats

Order printed copies

Views

1392

Downloads

1040

Citations