High precision and reliable wind speed forecasting have become a challenge for meteorologists. Convective events, namely, strong winds, thunderstorms, and tornadoes, along with large hail, are natural calamities that disturb daily life. For accurate prediction of wind speed and overcoming its uncertainty of change, several prediction approaches have been presented over the last few decades. As wind speed series have higher volatility and nonlinearity, it is urgent to present cutting-edge artificial intelligence (AI) technology. In this aspect, this paper presents an intelligent wind speed prediction using chicken swarm optimization with the hybrid deep learning (IWSP-CSODL) method. The presented IWSP-CSODL model estimates the wind speed using a hybrid deep learning and hyperparameter optimizer. In the presented IWSP-CSODL model, the prediction process is performed via a convolutional neural network (CNN) based long short-term memory with autoencoder (CBLSTMAE) model. To optimally modify the hyperparameters related to the CBLSTMAE model, the chicken swarm optimization (CSO) algorithm is utilized and thereby reduces the mean square error (MSE). The experimental validation of the IWSP-CSODL model is tested using wind series data under three distinct scenarios. The comparative study pointed out the better outcomes of the IWSP-CSODL model over other recent wind speed prediction models.
Weather conditions certainly affect various aspects of life in modern society. Unfavorable weather conditions and events have direct as well as indirect impacts on a large number of businesses and economic fields, like agriculture, transport, and logistics [
Through technological development and fortifying advocacy for ecological protection, wind energy production is more commercially competitive than coal-fired power production [
This paper presents an Intelligent Wind Speed Prediction using Chicken Swarm Optimization with Hybrid Deep Learning (IWSP-CSODL) model. The presented IWSP-CSODL model estimates the wind speed using a hybrid deep learning and hyperparameter optimizer. In the presented IWSP-CSODL model, the prediction process is performed via CNN-based long short-term memory with autoencoder (CBLSTMAE) model. To optimally modify the hyperparameters related to the CBLSTMAE model, the CSO algorithm is utilized and thereby reduces the mean square error (MSE). The experimental validation of the IWSP-CSODL model is tested using wind series data under three distinct scenarios.
The remainder of the paper is organized as follows, Section 2 analysis the related works involved in Wind Speed Prediction. Section 3 describes the Proposed Wind Speed Prediction Model and estimates the wind speed using a hybrid deep learning and hyperparameter optimizer. Section 4 then analyses the experimental data and results, including a performance comparison with alternative methodologies. Finally, Section 5 concludes the critical results of the proposed research.
In Trebing and Mehrkanoon [
Lin and Zhang [
Geng et al. [
An enhanced deep learning (DL)-based hybrid model for forecasting wind speed is projected in [
In this study, a new IWSP-CSODL model has been developed for effectual and precise wind speed forecasting. The presented IWSP-CSODL model estimates the wind speed using a hybrid deep learning and hyperparameter optimizer.
In the presented IWSP-CSODL model, the prediction process is performed by the CBLSTMAE model. The architecture encompasses different frameworks comprising 2 CNN layers and an autoencoder (AE) framework composed of a single LSTM layer as a decoder and a CBLSTM as an encoder [
Especially, CNN is skilled in extracting complicated features and might save many different unequal trends [
Assume the input vector
In
The pooling layer of the convolution layer samples the activation from mapping features to decrease the parameter count and computational network cost. The
The output from the maximal pooling layer was given to the input of the BLSTM layer via the gate components. BLSTM comprises dissimilar gates (forget, input, and output gates) in forward and backward directions, and every gate is activated. At the same time, memory cells upgrade the state, characterized as follows in
In the following equation,
The cell and hidden states are defined by the gate function of CBLSTM for the hidden and cell states, correspondingly in
The output of the BLSTM layer was concatenated with backward and forward directions, formulated in
The output of BLSTM
The CNN layer extracts spatial characteristics from the input dataset, and CBLSTM-AE accepts the feature from a CNN for learning temporal dependency from an estimated output.
To optimally modify the hyperparameters related to the CBLSTMAE model, the CSO algorithm is utilized and thereby reduces the mean square error (MSE). The CSO approach is preferred over other optimization techniques due to its high parallelism and simplicity [
where:
where
In which,
Then replace the
whereas
Now
This section investigates the wind speed prediction outcomes of the IWSP-CSODL model under three distinct scenarios. This proposed work measured the wind speed data of a wind farm, starting from February 1, 2022, to July 30, 2022, with an interval of 3 h, each containing 100 data points in the Kotdwara location.
The major hardware requirements are a cloud-based supercomputer, server, virtual machine (VM), several sensors, and meet towers for wind speed prediction. The various sensors are wind speed and direction sensors, humidity sensors, radiation sensors, precipitation sensors, modems, and data loggers. It implied that the IWSP-CSODL model has accurately forecasted the wind speed under all data points. The current work uses climate data information that has been downloaded from the website of
Scenario-1 | ||||
---|---|---|---|---|
Methods | MSE | MAE | MAPE (%) | R2 |
IWSP-CSODL | 0.5318 | 0.4967 | 10.1240 | 0.9687 |
Jaya-SVM | 0.6492 | 0.5899 | 11.6891 | 0.9468 |
LASSO | 0.6508 | 0.6089 | 12.3410 | 0.9428 |
XGBoost | 0.6699 | 0.6203 | 12.7580 | 0.9413 |
MLPR | 0.6769 | 0.6215 | 12.8411 | 0.9204 |
DBN | 0.7032 | 0.6236 | 12.8487 | 0.9116 |
GPR | 0.7094 | 0.6385 | 13.0254 | 0.9064 |
SSAE | 0.7159 | 0.6505 | 13.5843 | 0.8927 |
GrC | 0.8310 | 0.6938 | 14.3853 | 0.8863 |
A detailed MAPE inspection of the IWSP-CSODL model with recent models for scenario 1 is given in
Scenario-2 | ||||
---|---|---|---|---|
Methods | MSE | MAE | MAPE (%) | R2 |
IWSP-CSODL | 1.0114 | 0.6834 | 12.5490 | 0.9312 |
Jaya-SVM | 1.1309 | 0.7961 | 15.1714 | 0.8955 |
LASSO | 1.1523 | 0.8125 | 16.2675 | 0.8904 |
XGBoost | 1.1550 | 0.8412 | 17.3406 | 0.8791 |
MLPR | 1.1704 | 0.8756 | 15.8917 | 0.8708 |
DBN | 1.1784 | 0.8562 | 16.3558 | 0.8646 |
GPR | 1.2027 | 0.8209 | 15.3223 | 0.8511 |
SSAE | 1.2036 | 0.8618 | 15.7527 | 0.8486 |
GrC | 1.4720 | 0.9527 | 19.1790 | 0.8246 |
A brief MAPE inspection of the IWSP-CSODL approach with current methods on scenario 2 is given in
Scenario-3 | ||||
---|---|---|---|---|
Methods | MSE | MAE | MAPE (%) | R2 |
IWSP-CSODL | 1.2654 | 0.9185 | 16.3842 | 0.8432 |
Jaya-SVM | 1.6437 | 1.0179 | 19.2791 | 0.8086 |
LASSO | 1.7420 | 1.0124 | 18.7221 | 0.7837 |
XGBoost | 1.7547 | 1.0714 | 20.2979 | 0.8120 |
MLPR | 1.7670 | 1.0584 | 18.9103 | 0.7760 |
DBN | 1.8035 | 1.0431 | 18.7332 | 0.8086 |
GPR | 1.8051 | 1.0214 | 19.9954 | 0.7813 |
SSAE | 1.8115 | 1.0667 | 20.3918 | 0.8166 |
GrC | 2.4062 | 1.2386 | 23.5756 | 0.7025 |
A detailed MAPE review of the IWSP-CSODL technique with recent models on scenario 3 is given in
In this study, a new IWSP-CSODL model has been developed for effectual and precise wind speed forecasting. The presented IWSP-CSODL model estimates the wind speed using a hybrid deep learning and hyperparameter optimizer. In the presented IWSP-CSODL model, the CBLSTMAE model performs the prediction process. The CSO algorithm is utilized to optimize the hyperparameters related to the CBLSTMAE model, and thereby reducing MSE. The experimental validation of the IWSP-CSODL model is tested using wind series data under three distinct scenarios. The comparative study pointed out the better outcomes of the IWSP-CSODL model over other recent wind speed prediction models. In the future, the predictive performance of the IWSP-CSODL method could be boosted by the use of hybrid metaheuristic algorithms. The IWSP-CSODL method will be applied for atmospheric pressure prediction.
The authors would like to thank the Deanship of Scientific Research at Umm Al-Qura University for supporting this work by Grant Code: (22UQU4281755DSR01).
This research is funded by
Available Based on Request. The datasets generated and/or analyzed during the current study are not publicly available due to the extension of the submitted research work. Still, they are available from the corresponding author on reasonable request.
The authors declare that they have no conflicts of interest to report regarding the present study.