China
Africa
Explore chapters and articles related to this topic
Artificial Intelligence Application for HVDC Protection
Published in Almoataz Y. Abdelaziz, Shady Hossam Eldeen Abdel Aleem, Anamika Yadav, Artificial Intelligence Applications in Electrical Transmission and Distribution Systems Protection, 2021
Zahra Moravej, Amir Imani, Mohammad Pazoki
Merlin et al. in [37] applied 13 ANNs for the system to detect a fault condition in the two-terminal HVDC system. The method only uses voltage signal measured at the rectifier side and does not need any communication. In [38], a combination of radial basis function network (RBFN) as an effective feed-forward neural network (FNN) and S transform is proposed to identify internal and external fault in the system. By exploiting S transformed input feature vector of RBFN, some shortcomings of traditional protection e.g. difficult threshold setting has been solved. A novel approach for online fault detection in HVDC converters using adaptive linear neuron (ADALINE) is presented in [39]. An ADALINE is an n-input single output neural network, which outputs a linear combination of its inputs [39]. An ADALINE can be used to follow the harmonic content of a signal online. Then an index is defined to be compared with the extracted harmonic content for fault detection. The threshold value selection of index is not mentioned in the paper.
Artificial Neural Networks
Published in Paresh Chra Deka, A Primer on Machine Learning Applications in Civil Engineering, 2019
Adaptive Linear Neuron or Adaptive Linear Element (ADALINE) is a single-layer neural network. It was developed by Professor Bernard Widrow and Ted Hoff in 1960 and is based on the MP neuron with weights, bias, and summation functions. In the learning phase, the weights are adjusted according to the weighted sum of the inputs. In general, ADALINE can be trained by using the delta rule, which is also known as the least mean square (LMS) or the Widrow–Hoff rule. In ADALINE there will be only one input unit. net=b+∑ixiwi
Groundwater Planning and Management
Published in Mohammad Karamouz, Azadeh Ahmadi, Masih Akhbari, Groundwater Hydrology, 2020
Mohammad Karamouz, Azadeh Ahmadi, Masih Akhbari
The Adaptive Linear (ADALINE) network is similar to the perceptron, but its activation function is linear. The linear activation function allows the ADALINE networks to take on any output value, thus solving the linearly separable problems. The architecture of the ADALINE is shown in Figure 7.13. The ADALINE has a counterpart in statistical modeling, in this case, least square regression. The ADALINE network is a widely used ANN found in the practical applications, especially in adaptive filtering. In the ADALINE, the learning rule is based on the least mean squares method, which minimizes the mean square error. The adaptive filter is produced by combining a tapped delay line and an ADALINE network. At each time step, this filter is adjusted and finds the weights and biases that minimize the network’s sum squared error for the recent input and target vectors. In the ADALINE network, the output, a(t) is obtained as follows: () a(t)=∑k=0Nw(k)P(t−k)+b
Application of Hopfield Neural Network for Harmonic Current Estimation and Shunt Compensation
Published in Electric Power Components and Systems, 2018
Prakash Chittora, Alka Singh, Madhusudan Singh
The performance of the proposed HNN technique is compared with LMS technique which belongs to the class of adaptive filters [14]. They are used to imitate the desired output performance by producing weighted output which is related to least mean square of error signal. The error signal is generated by comparing reference signal and actual signal. Adaptive Linear Neuron (ADALINE) is also similar to the LMS technique, where single layer artificial neural network is implemented and is used to extract the weighted component of a signal by real-time learning. LMS-based control algorithm for fundamental weight extraction is designed as per equations in Appendix B.
Wind speed forecasting techniques for maximum power point tracking control in variable speed wind turbine generator
Published in International Journal of Modelling and Simulation, 2019
Fouzia Achouri, Boubekeur Mendil
The use of the ADALINE neural network has a great advantage in the estimation due to its lower complexity and shorter training time; the results obtained demonstrate its efficiency. The Fuzzy estimator gives satisfied results as the ADALINE estimator, but it is slow and heavy to achieve the algorithm operations in simulation tests. These two advanced techniques become the most powerful tool for estimating parameters and controlling systems. Experimental validation with a real plan using the studied control law will be investigated in the future.
Feature selection based reverse design of doubly reinforced concrete beams
Published in Journal of Asian Architecture and Building Engineering, 2022
Won-Kee Hong, Tien Dat Pham, Van Tien Nguyen
The perceptron training rule works well when the samples to be trained are linearly separable while updating weights based on the backpropagation errors. In contrast, the Adaline rule works well even when the samples to be trained are not linearly separable while converging toward the best-fit approximation of the target output. While the training methods work well for single-layer neural networks or perceptrons, the present paper proposes efficient training schemes that will work for ANNs with both shallow and deep layers.