Wind Power Forecasting by the BP Neural Network with the Support of Machine Learning

Tian, Weihua; Bao, Yan; Liu, Wei

doi:https://doi.org/10.1155/2022/7952860

Mathematical Problems in Engineering

On this page

Abstract Introduction Materials and Methods Results Conclusion Data Availability Conflicts of Interest Acknowledgments References Copyright Related Articles

Special Issue

Advanced Aspects of Computational Intelligence and Applications of Fuzzy Logic and Soft Computing

View this Special Issue

Research Article | Open Access

Volume 2022 | Article ID 7952860 | https://doi.org/10.1155/2022/7952860

Wind Power Forecasting by the BP Neural Network with the Support of Machine Learning

Weihua Tian,¹Yan Bao,¹and Wei Liu¹

Academic Editor: Naeem Jan

Received18 Feb 2022

Revised15 Mar 2022

Accepted18 Mar 2022

Published28 Apr 2022

Abstract

The goal of the research is to increase the accuracy of wind power forecasts while maintaining the power system’s stability and safety. First, the wireless sensor network (WSN) is used to collect the meteorological data of wind power plants in real time. Second, the real-time data collected by WSN are combined with the meteorological forecast data of some meteorological organizations. Then, the fruit fly optimization algorithm (FOA) is improved, and the improved fruit fly optimization algorithm (IFOA) and back propagation neural network (BPNN) are combined to construct the wind power forecast model. Finally, the signal reception of the WSN and the error of wind power forecast under different receiving distances and different antenna heights are tested. The results show that with the increase of receiving and transmitting distance, the signal strength decreases, the packet loss rate increases, and the electromagnetic wave of wind plants will cause some interference to the signal strength. The fly optimization algorithm-back propagation (IFOA-BP) wind power forecast model has a better effect than other models in wind power prediction and can better fit the actual tested wind power. Its root mean square error (RMSE) and mean absolute error (MAE) are 0.16 and 0.11, respectively. The research results provide a reference for improving the forecast accuracy of wind power.

1. Introduction

As an application form of clean energy, wind power generation in the total power generation system increases year by year. However, the wind is unstable energy with significant volatility, which dramatically impacts the stable operation of the power system. A timely and accurate forecast of wind power plants can significantly enhance the security, stability, economy, and controllability of the power system [1].

In recent years, scholars have done a lot of research on the forecast method of wind power plants, which has made some achievements. Studies have shown combined Levenberg–Marquardt algorithm with particle swarm optimization (PSO) to update the weight and bias of the artificial neural network (ANN) for wind speed forecast. The short-term forecast of wind power 1 hour in advance is realized through the ANN [2]. Some researchers studied and analyzed the wind power forecast and evaluation methods from users’ perspectives and discussed the practicability of different wind power evaluation methods in different examples [3]. Some researchers took a wind power plant with an installed capacity of 135 MW in Turkey as the research object, proposed a wind power prediction method based on an extreme learning machine (ELM), and compared it with the traditional wind power forecast method ANN. This shows better performance than conventional methods in power forecast [4]. Some scholars proposed a comprehensive forecast method of signal filtering and selecting the best candidate input based on minimum redundancy. The proposed forecast method comprises a two-dimensional convolutional neural network (CNN), and POS trains the neural network to improve the forecast accuracy [5]. Related research studies introduced artificial intelligence (AI), and the seasonal short-term forecast of hourly wind power in advance is discussed by using the ANN, adaptive network-based fuzzy inference system (ANFIS), and the radial basis function network (RBFN), respectively. Moreover, three different learning algorithms are used for RBFN, including stochastic gradient descent (SGD), the hybrid algorithm, and orthogonal least squares (OLS). The results show that ANFIS performs best, ANN and RBFN-OLS also show strong performance, and RBFN and RBFN-SGD perform poorly [6]. “Multiobjective prediction intervals for wind power forecast based on deep neural networks,” was true according to Zhou et al. [7]. “A wireless communication system based on space-and frequency-division multiplexing employing digital met surfaces,” Zhang et al. [8] suggested.

These research results help improve the forecasting accuracy of wind power generation systems. Nonetheless, these accuracies are primarily based on meteorological forecast values and historical meteorological data collected from wind farms. The forecast of wind power plants depends on the accuracy of meteorological forecast values. However, the numerical meteorological forecast system assists the wind power forecast in China. The size of accumulated data is small, which is easy to lead to significant forecast errors. The wireless communication technology is applied to the wind power forecast system innovatively. The real-time data collected by the wireless sensor network (WSN) is combined with the weather forecast values, and the fruit fly optimization algorithm (FOA) is improved. On this basis, the BP (back propagation) neural network is used to construct the wind power prediction model. The wind power prediction model has superior performance, high accuracy, and pass rate and can be practically applied in power generation. However, in practice, the electromagnetic wave of the wind farm will affect the signal transmission strength of the F in the signal transmission. This may reduce the prediction accuracy.

This study is divided into four parts: Section 1 is the introduction, which mainly introduces the current research status and research hotspots related to the topic. Section 2 is the method, which mainly analyzes and describes the theoretical basis and algorithms related to the research. Section 3 is the result, which mainly analyzes the performance and superiority of the proposed algorithm. Section 4 is the conclusion, which mainly summarizes the obtained results and puts forward the shortcomings and prospects of the research.

2. Materials and Methods

In wind power forecasting, the hottest topic in current research is how to use AI technology to improve the efficiency of wind power forecasting, including deep learning technology, machine learning technology, and wireless communication technology. The focus of the applied method is to combine deep learning neural networks, machine learning algorithms, and wireless communication technology to realize the optimization of the wind power forecasting method.

2.1. Principle and Characteristics of Wind Power Generation

Building a wind power forecast model needs to know about the principle and characteristics of wind power generation. Wind power generation relies on wind energy. Wind turbines convert wind energy into mechanical energy and then convert mechanical energy into electric energy through wind turbine generator sets. The core of wind power generation lies in wind turbines. The absorbed power of wind turbine is calculated by the following:

In equation (1), is the absorbed power of the wind turbine, is the wind energy utilization coefficient, is the pitch angle, is the wind wheel radius, is the function of and , is the air density, and is the wind speed.

However, the power absorbed by the wind turbine does not represent the output power of the wind turbine. It is also necessary to convert the mechanical energy absorbed by the wind turbine into electric energy through the gearbox speed-up generator. The output power of the wind turbine is mainly related to the wind speed, and its relations can be expressed by approximating the following:

In equation (2), is the output power of the fan, is the cut-in wind speed of the fan, is the cut-out wind speed of the fan, is the rated power of the fan, and is the rated wind speed of the fan.

The output power of wind turbines is mainly affected by wind speed. After the distribution law of wind speed and wind power in the dataset is analyzed, wind power has apparent randomness, fluctuation, and intermittence [7].

Wind power forecast falls into the following steps:

2.1.1. Data Collection and Transmission

Wind power is affected by wind [9]. The wind in nature is related to meteorological factors such as temperature, air pressure, and humidity. These data can be learned from the numerical weather forecast released by relevant meteorological departments. Moreover, the data are downloaded directly and input into the wind power forecast model. It is necessary to obtain real-time wind speed, wind direction, temperature, pressure, and other relevant data to predict the wind accurately. Here, the WSN is used to collect real-time data, process these data, and transmit the processed data to the wind power forecast model.

2.1.2. Forecast of Wind Power

The wind power forecast model is implemented and used to complete the forecast, and the power is dispatched according to the forecasted power. The basic architecture of the wind power forecast model is shown in Figure 1.

It is mainly composed of three parts: data acquisition and transmission, wind power prediction, and power dispatching center.

2.2. Wireless Communication Technology and Deep Learning (DL)

2.2.1. Wireless Communication

It refers to the communication mode between multiple nodes through the electromagnetic wave rather than a conductor or a cable [8]. Wireless communication can be carried out by using radios.

2.2.2. WSN

Networks formed by freely organizing and combining tens of thousands of sensor nodes through wireless communication technology. The sensor node units are the data acquisition unit, data transmission unit, data processing unit, and energy supply unit [10].

2.2.3. DL

It is a new research direction in machine learning (ML). Its concept originates from people’s research on the ANN. DL is the internal law and representation level of sample data. Its goal is to enable machines to learn and think like people [11]. At present, it has achieved fruitful results in speech and image recognition.

2.2.4. BPNN

It is the most successful neural network. It is a multilayer feedforward neural network. Its characteristic is that the signal propagates forward, and the error propagates back. It is composed of the input layer, hidden layer, and output layer [12]. The network structure is shown in Figure 2.

Figure 2 shows that the BPNN is divided into two stages: forward signal transmission stage and back error transmission stage. The first stage is the forward propagation of the signal, passing through the hidden layer from the input layer to the output layer. The second stage is the backpropagation of errors, and the weight and bias from the hidden layer to the output layer and from the input layer to the hidden layer are adjusted in turn [13].

2.3. WSN for Wind Power Forecast

The networking mode of the wireless transmission network of wind power plants needs to be selected. Because they are usually set in open places outdoors, wind power plants are greatly affected by the environment. Therefore, in building a WSN, their wide distribution range should be considered. In addition, they will be affected by electromagnetic waves. Cluster-based ad hoc networks are used for networking [14]. The clustered ad hoc network is shown in Figure 3.

In Figure 3, each cluster includes nodes in the cluster, cluster heads, and sink nodes. The nodes in the cluster completely transmit the collected data to the cluster head node, compress the collected data at the cluster head by data fusion, and transmit the compressed data to the sink node. Moreover, cross-communication is made between clusters.

Then, the data acquisition and fusion of WSN wind power plants are studied [15]. Because of the complex environmental conditions of the wind power plant, more data need to be collected, which is easy to produce data redundancy and affect the transmission efficiency. Through the fusion of the collected data, the energy consumption can be reduced, the accuracy and efficiency of the collected data are improved. The data acquisition and fusion technology of WSN are divided into the following steps: containing the original signal, transforming, processing, compressing and encoding the collected signal, and decompressing the collected data [16].

Finally, the signal is transmitted. The strength of signal transmission can be expressed by the packet loss rate [17]. The erection distance and height of transceiver nodes are the main factors affecting the packet loss rate. The structure of WSN for wind power forecast is shown in Figure 4.

It is divided into three parts: original data acquisition, data processing, and signal transmission.

2.4. Construction of the Wind Power Forecast Model

The key to wind power forecast lies in constructing a wind power forecast model. Here, the wind power prediction model is implemented based on the BPNN. The weight and threshold of BPNN are initialized randomly [18], which affects the stability of forecast results. In this case, the fruit fly optimization algorithm (FOA) combines with the BPNN to implement the wind power forecast model because it has a global solid optimization ability [19].

The traditional FOA has inconsistent optimization results and quickly falls into the optimal local value [20]. The abovementioned problems are the uneven distribution of the initialization population, and the judgment value of drosophila taste concentration must be positive. On this basis, the FOA is optimized.

In response to the uneven distribution of the initialization population, the chaotic mapping method is used to initialize the positions of all fruit flies [21] so that the optimized positions of fruit flies have the characteristics of randomness, ergodicity, regularity of chaotic phenomenon, and sensitivity to initial values. The chaotic system is the simplest and has a good effect on logistic mapping [22], as shown below:

In equation (3), is the times of iterations and is the control parameter. When , the system is in a chaotic state.

A transformation model of chaotic variable is shown below:

In equation (4), is the value of mapped chaotic variable after -step chaotic transformation. When and , chaos will appear. Chaotic variable is obtained after transformations, and then the chaotic variable is mapped repeatedly by equations (5) and (6).

In equations (5) and (6), is the optimization variable before chaotic mapping and .

As for the problem that the judgment value of drosophila taste concentration is always positive, the taste concentration is used to solve it [23]. The optimization process is shown in equations (7)–(9).

In equations (7)–(9), represents the position of individual fruit fly from the origin, is the optimized taste concentration, is the optimization factor, and is the optimization coefficient and obeys the uniform distribution.

The improved fruit fly optimization algorithm (IFOA) is combined with the BPNN. To begin, the goal function and population dimension in IFOA are the mean square error (MSE) function, weight, and threshold from the BPNN. Then, whether the IFOA iteration reaches the maximum iteration times or whether the output of the objective function meets the accurate value is judged. The qualified weights and thresholds are substituted into the BPNN. Finally, after the maximum iteration times of BPNN or the target accurate value point are achieved, the model training is completed [24]. The flowchart is shown in Figure 5.

IFOA-BP determines the initialization weight and threshold of the BPNN.

2.5. Evaluation Indexes of Forecast Performance

Evaluating the performance of the wind power forecast model is essential to improve the forecast accuracy and optimize the wind power forecast model. The performance of the wind power prediction model is generally evaluated by root mean square error (RMSE) [25] and mean absolute error (MAE) [26]. RMSE is the square root of the ratio of the square sum of the deviation between the forecast value and the actual value and the observation times n, which can be used to evaluate the stability of the model. It is calculated by the following equation:

In equation (10), is the number of forecast results, is the rated power of the fan, is the actual value of target data, is the predicted value of target data, and is the serial number of the real deal and forecasted value.

MAE evaluates the forecast accuracy of the forecast model from the absolute error of the forecast value, and its calculation equation is as follows:

In equation (11), is the number of forecast results, is the rated power of the fan, is the actual value of target data, is the forecasted value of target data, and is the serial number of the real deal and projected value.

3. Results

This part mainly analyzes the wind power characteristics, signal transmission test, and the test results of the wind power prediction model after the method is proposed.

3.1. Characteristic Analysis of Wind Power

In the dataset, the wind speed and wind power on March 18 are selected for analysis. The changes in wind speed and wind power are shown in Figure 6.

Figure 6 shows that in the same period, the distribution of fan power does not obey a certain known law, and there is no regularity in the change between the power value at the last time and the power value at the next time. Wind power output is large between 22:00 and 23:00, but there is no power output between 0:00-1:00 and 7:00-8:00. After the detection data at other times in the dataset are compared, this power output is widespread. This shows that wind power has obvious randomness, fluctuation, and intermittence.

3.2. Test Results of Signal Transmission at Different Distances and Heights

The influence of the length and height between transceiver nodes are explored, and the signal strengths of varying transceiver distances and antenna heights are tested under a 2.4 GHz channel. The results are shown in Figure 7.

Figure 7 shows that with the increase of the transmitting and receiving distance, the signal strength tends to weaken. When the distance is greater than 30 m, the attenuation trend is more pronounced. After 30 m, the attenuation rate slows down. During the increase of the transmitting and receiving distance, the signal strength occasionally fluctuates, caused by the electromagnetic wave interference of the wind power plants.

Under the 2.4 GHz channel, the packet loss rates of different antenna heights and transceiver distances are tested, and the results are shown in Figure 8.

Figure 8 shows that with the increase of the transmitting and receiving distance, the packet loss rate of the signal also shows an upward trend. When the transmitting and receiving distance is more excellent than 30 m, the packet loss will be more obvious. When the antenna height is 1.5 m, the packet loss rate is higher. When the antenna distance is 0 m, the packet loss rate is lower, and when the antenna distance exceeds 55 m, the packet loss rate is still significant and more than 35%.

3.3. Test Results of the Wind Power Forecast Model

BPNN, FOA-BP, and IFOA-BP are used to implement the wind power forecast model. The model predicts within 5 minutes and records the forecasted value within 1 hour. The results are shown in Figure 9.

Figure 9 shows that the three models can better fit the measured wind power data. However, compared with the wind power forecast models implemented by FOA-BP and IFOA-BP, the wind power forecast model implemented by the BPNN has more errors. In comparison, the errors of the FOA-BP model are fewer, and those of the IFOA-BP model are the most irregular, indicating that the forecast accuracy of the IFOA-BP model is the highest. RMSE and MAE of the wind power forecast model implemented by BPNN, FOA-BP, and IFOA-BP are compared, and the results are shown in Figure 10.

Figure 10 shows that the prediction effects of the three models are suitable. However, compared with the FOA-BP model and IFOA-BP model, the prediction effect of the BPNN wind power forecast model is poor, with RMSE and MAE of 0.42 and 0.25, respectively. Moreover, RMSE and MAE of the FOA-BP wind power forecast model are 0.19 and 0.19, respectively, and RMSE and MAE of the IFOA-BP wind power prediction model are 0.16 and 0.11, respectively. It is concluded that RMSE and MAE of the IFOA-BP wind power prediction model are the smallest, indicating that its prediction accuracy is the highest and more practical.

4. Conclusion

With natural resources becoming increasingly scarce, the use of renewable energy has progressively gained traction. The wind energy research study in power production technologies has made some progress as a sustainable and clean energy source. Here, the principle and characteristics of wind power generation are studied, and the wind power prediction model is implemented. First, the WSN is used to collect the real-time meteorological data of the wind power plant. Wind speed, direction, and other wind power data are obtained in conjunction with appropriate meteorological bureaus’ weather forecast values. Second, the fun optimization technique is enhanced, and the wind power forecast model is implemented using IFOA and BPNN. Finally, the signal strength and packet loss rate of WSN are tested under different antenna heights and receiving and transmitting distances through experiments. The IFOA-BP wind power prediction model is compared with the BPNN and FOA-BP models. The results reveal that when the receiving and sending distances increase, the signal strength falls, the overall packet loss rate rises, and the electromagnetic wave from the wind power plant interferes with the signal strength. The IFOA-BP wind power prediction model has a better effect than other models in wind power forecast, and its RMSE and MAE are 0.16 and 0.11, respectively. The research study makes contribution to improve the forecast accuracy of wind power, but there are still deficiencies in this research study. First, the signal transmission strength of the WSN in signal transmission will be affected by the electromagnetic wave of wind power plants in practice, which could result in a shortage of data and a reduction in forecast accuracy. Second, the wind power forecast model is evaluated by RMSE and MAE, which belong to the mean index, and cannot accurately and comprehensively analyze the system. The research result provides a feasible method for improving the forecast accuracy of wind power and a reference for enhancing the stability and reliability of the power system. The method’s practical usefulness will be further investigated and applied to wind power prediction in functional scenarios in the future. The power system’s stability will be improved much more.

Data Availability

The simulation experiment data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest

The authors declare that there are no conflicts of interest regarding the publication of this paper.

Acknowledgments

This work was funded by Scientific research Funds of Education Department of Liaoning Province in 2021, “Research on short-term wind power prediction based on attention-GRU recurrent neural network,” project number: LJKZ1099.

References

D. H. Barus and R. Dalimi, “Multi-stage Statistical approach to wind power forecast errors evaluation: a Southern Sulawesi case study,” International Journal of Advanced Science, Engineering and Information Technology, vol. 11, no. 2, pp. 633–641, 2021.
View at: Publisher Site | Google Scholar
G. Nazaré, R. Castro, and L. R. Gabriel Filho, “Wind power forecast using neural networks: Tuning with optimization techniques and error analysis,” Wind Energy, vol. 23, no. 3, pp. 810–824, 2020.
View at: Publisher Site | Google Scholar
J. W. Messner, P. Pinson, J. Browell, M. B. Bjerregård, and I. Schicker, “Evaluation of wind power forecasts-An up‐to‐date view,” Wind Energy, vol. 23, no. 6, pp. 1461–1481, 2020.
View at: Publisher Site | Google Scholar
H. Acikgoz, C. Yildiz, and M. Sekkeli, “An extreme learning machine based very short-term wind power forecasting method for complex terrain,” Energy Sources, Part A: Recovery, Utilization, and Environmental Effects, vol. 42, no. 22, pp. 2715–2730, 2020.
View at: Publisher Site | Google Scholar
O. Abedinia, M. Bagheri, M. S. Naderi, and N. Ghadimi, “A new combinatory approach for wind power forecasting,” IEEE Systems Journal, vol. 14, no. 3, pp. 4614–4625, 2020.
View at: Publisher Site | Google Scholar
M. Godinho and R. Castro, “Comparative performance of AI methods for wind power forecast in Portugal,” Wind Energy, vol. 24, no. 1, pp. 39–53, 2021.
View at: Publisher Site | Google Scholar
M. Zhou, B. Wang, S. Guo, and J. Watada, “Multi-objective prediction intervals for wind power forecast based on deep neural networks,” Information Sciences, vol. 550, pp. 207–220, 2021.
View at: Publisher Site | Google Scholar
L. Zhang, M. Z. Chen, W. Tang et al., “A wireless communication scheme based on space- and frequency-division multiplexing using digital metasurfaces,” Nature Electronics, vol. 4, no. 3, pp. 218–227, 2021.
View at: Publisher Site | Google Scholar
A. Das, H. K. Jani, S. S. Kachhwaha, and G. Nagababu, “Assessment of factors affecting Onshore wind power Deployment in India,” Environmental and Climate Technologies, vol. 24, no. 1, pp. 185–208, 2020.
View at: Publisher Site | Google Scholar
S. K. Das and R. Kapelko, “On the range assignment in wireless sensor networks for minimizing the coverage-connectivity cost,” ACM Transactions on Sensor Networks, vol. 17, no. 4, pp. 1–48, 2021.
View at: Publisher Site | Google Scholar
A. Saxe, S. Nelli, and C. Summerfield, “If deep learning is the answer, what is the question?” Nature Reviews Neuroscience, vol. 22, no. 1, pp. 55–67, 2021.
View at: Publisher Site | Google Scholar
Y. G. Zhang, J. Tang, R. P. Liao et al., “Application of an enhanced BP neural network model with water cycle algorithm on landslide prediction,” Stochastic Environmental Research and Risk Assessment, vol. 35, no. 6, pp. 1273–1291, 2021.
View at: Publisher Site | Google Scholar
A. Afzal, K. M. Yashawantha, N. Aslfattahi, R. Abdul Razak, R. K. Subbiah, and R. Subbiah, “Back propagation modeling of shear stress and viscosity of aqueous Ionic-MXene nanofluids,” Journal of Thermal Analysis and Calorimetry, vol. 145, no. 4, pp. 2129–2149, 2021.
View at: Publisher Site | Google Scholar
P. D. Nguyen and L.-W. Kim, “Sensor system: a Survey of sensor Type, ad hoc network Topology and energy Harvesting techniques,” Electronics, vol. 10, no. 2, p. 219, 2021.
View at: Publisher Site | Google Scholar
H. Wang, L. Song, J. Liu, and T. au, “An efficient intelligent data fusion algorithm for wireless sensor network,” Procedia Computer Science, vol. 183, pp. 418–424, 2021.
View at: Publisher Site | Google Scholar
J. Botero-Valencia, L. Castano-Londono, D. Marquez-Viloria, and M. Rico-Garcia, “Data reduction in a low-cost environmental monitoring system based on LoRa for WSN,” IEEE Internet of Things Journal, vol. 6, no. 2, pp. 3024–3030, 2019.
View at: Publisher Site | Google Scholar
B. Sun, C. Gui, Y. Song, and H. Chen, “Network coding-based maximum lifetime algorithm for sliding window in WSNs,” KSII Transactions on Internet and Information Systems, vol. 13, no. 3, pp. 1298–1310, 2019.
View at: Google Scholar
Y. Wu, R. Gao, and J. Yang, “Prediction of coal and gas outburst: a method based on the BP neural network optimized by GASA,” Process Safety and Environmental Protection, vol. 133, pp. 64–72, 2020.
View at: Publisher Site | Google Scholar
L. Jerlin Rubini, E. Perumal, and E. Perumal, “Efficient classification of chronic kidney disease by using multi‐kernel support vector machine and fruit fly optimization algorithm,” International Journal of Imaging Systems and Technology, vol. 30, no. 3, pp. 660–673, 2020.
View at: Publisher Site | Google Scholar
K. Sun, Z. Dou, Y. Zhu et al., “Scheduling model of power system based on forecasting error of wind power plant output,” IEEJ Transactions on Electrical and Electronic Engineering, vol. 16, no. 4, pp. 526–535, 2021.
View at: Publisher Site | Google Scholar
W. Luo, H. Jin, H. Li, X. Fang, and R. Zhou, “Optimal performance and application for firework algorithm using a novel chaotic approach,” IEEE Access, vol. 8, pp. 120798–120817, 2020.
View at: Publisher Site | Google Scholar
A. A. Shah, S. A. Parah, M. Rashid, and M. Elhoseny, “Efficient image encryption scheme based on generalized logistic map for real time image processing,” Journal of Real-Time Image Processing, vol. 17, no. 6, pp. 2139–2151, 2020.
View at: Publisher Site | Google Scholar
C. Xiong and S. Lian, “Structural Damage Identification based on improved fruit fly optimization algorithm,” KSCE Journal of Civil Engineering, vol. 25, no. 3, pp. 985–1007, 2021.
View at: Publisher Site | Google Scholar
N. Zougagh, A. Charkaoui, and A. Echchatbi, “Artificial intelligence hybrid models for improving forecasting accuracy,” Procedia Computer Science, vol. 184, pp. 817–822, 2021.
View at: Publisher Site | Google Scholar
J. Xu, Z. Xiao, Z. Lin, and M. Li, “System bias correction of short-term hub-height wind forecasts using the Kalman filter,” Protection and Control of Modern Power Systems, vol. 6, no. 1, pp. 37–39, 2021.
View at: Publisher Site | Google Scholar
E. Oh and S. Y. Son, “Theoretical energy storage system sizing method and performance analysis for wind power forecast uncertainty management,” Renewable Energy, vol. 155, pp. 1060–1069, 2020.
View at: Publisher Site | Google Scholar

Copyright

Copyright © 2022 Weihua Tian 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