Estimation of wind speed by artificial intelligence method: A case study

Omer Ahmed

Makale detayları

Estimation of wind speed by artificial intelligence method: A case study

Yazarlar:

Omer Ahmed

Yayıncı:

Yildiz Technical University

Ana Başlıklar:

Thermodynamics and Statistical Physics

Dergi:

Journal of Thermal Engineering Vol.10 Issue 5, pp.1347-1361

Belge Tipi:

Makale

Yayın Tarihi:

10/2024

Belge Dili:

İngilizce

Document ID (Org No.):

2001326784

pISSN:

2148-7847

Anahtar kelimeler:

Özet:

 Wind speed changes from one region to another due to several influencing variables. In this article, a software method has been proposed to determine the future wind speed at any time and under any conditions. Neural Networks were used with engineering data regarding the method of education, training algorithms, and different activation functions between the input and output layers, each according to the nature of the data that would be generated. Back-propagation Neural was used with three variables chosen to be the inputs for the learning and training network wind speed, humidity, and time, which are considered the most important in determining the proposed or expected speed at the relevant time and place. The hidden layer consists of 10 neurons, which are determined according to the precision of output. After comparing the measurements from the weather system with the expected values, a very tiny percentage of error was found since these readings are regarded as acceptable and aid in the problem-solving process for running companies and researchers. The error rate recorded in this work ranged between 3  10-3 and 3  10-5, and the average number of attempts for the training and examination process reached 33 attempts, as it is known that neural networks carry out the training process based on specific mathematical functions and closed loops that depend on the lowest possible error rate.

 Wind speed changes from one region to another due to several influencing variables. In this article, a software method has been proposed to determine the future wind speed at any time and under any conditions. Neural Networks were used with engineering data regarding the method of education, training algorithms, and different activation functions between the input and output layers, each according to the nature of the data that would be generated. Back-propagation Neural was used with three variables chosen to be the inputs for the learning and training network wind speed, humidity, and time, which are considered the most important in determining the proposed or expected speed at the relevant time and place. The hidden layer consists of 10 neurons, which are determined according to the precision of output. After comparing the measurements from the weather system with the expected values, a very tiny percentage of error was found since these readings are regarded as acceptable and aid in the problem-solving process for running companies and researchers. The error rate recorded in this work ranged between 3  10-3 and 3  10-5, and the average number of attempts for the training and examination process reached 33 attempts, as it is known that neural networks carry out the training process based on specific mathematical functions and closed loops that depend on the lowest possible error rate.

 Wind speed changes from one region to another due to several influencing variables. In this article, a software method has been proposed to determine the future wind speed at any time and under any conditions. Neural Networks were used with engineering data regarding the method of education, training algorithms, and different activation functions between the input and output layers, each according to the nature of the data that would be generated. Back-propagation Neural was used with three variables chosen to be the inputs for the learning and training network wind speed, humidity, and time, which are considered the most important in determining the proposed or expected speed at the relevant time and place. The hidden layer consists of 10 neurons, which are determined according to the precision of output. After comparing the measurements from the weather system with the expected values, a very tiny percentage of error was found since these readings are regarded as acceptable and aid in the problem-solving process for running companies and researchers. The error rate recorded in this work ranged between 3  10-3 and 3  10-5, and the average number of attempts for the training and examination process reached 33 attempts, as it is known that neural networks carry out the training process based on specific mathematical functions and closed loops that depend on the lowest possible error rate.

 Wind speed changes from one region to another due to several influencing variables. In this article, a software method has been proposed to determine the future wind speed at any time and under any conditions. Neural Networks were used with engineering data regarding the method of education, training algorithms, and different activation functions between the input and output layers, each according to the nature of the data that would be generated. Back-propagation Neural was used with three variables chosen to be the inputs for the learning and training network wind speed, humidity, and time, which are considered the most important in determining the proposed or expected speed at the relevant time and place. The hidden layer consists of 10 neurons, which are determined according to the precision of output. After comparing the measurements from the weather system with the expected values, a very tiny percentage of error was found since these readings are regarded as acceptable and aid in the problem-solving process for running companies and researchers. The error rate recorded in this work ranged between 3  10-3 and 3  10-5, and the average number of attempts for the training and examination process reached 33 attempts, as it is known that neural networks carry out the training process based on specific mathematical functions and closed loops that depend on the lowest possible error rate.

Makale detayları

Estimation of wind speed by artificial intelligence method: A case study

Yazarlar:

Omer Ahmed

Yayıncı:

Yildiz Technical University

Ana Başlıklar:

Thermodynamics and Statistical Physics

Dergi:

Journal of Thermal Engineering Vol.10 Issue 5, pp.1347-1361

Belge Tipi:

Makale

Yayın Tarihi:

10/2024

Belge Dili:

İngilizce

Document ID (Org No.):

2001326784

pISSN:

2148-7847

Artificial Intelligence

Assessment

Neural Network

Wind Energy

Sayı

Makale detayları

Estimation of wind speed by artificial intelligence method: A case study

Yazarlar: Omer Ahmed

Yayıncı: Yildiz Technical University

Ana Başlıklar: Thermodynamics and Statistical Physics

Dergi: Journal of Thermal Engineering Vol.10 Issue 5, pp.1347-1361

Belge Tipi: Makale

Yayın Tarihi: 10/2024

Belge Dili: İngilizce

Document ID (Org No.): 2001326784

pISSN: 2148-7847

Artificial Intelligence

Assessment

Neural Network

Wind Energy

Sayı

Yazarlar:

Omer Ahmed

Yayıncı:

Yildiz Technical University

Ana Başlıklar:

Thermodynamics and Statistical Physics

Dergi:

Journal of Thermal Engineering Vol.10 Issue 5, pp.1347-1361

Belge Tipi:

Makale

Yayın Tarihi:

10/2024

Belge Dili:

İngilizce

Document ID (Org No.):

2001326784

pISSN:

2148-7847