scholarly journals Predicting Performance of a District Heat Powered Adsorption Chiller by Means of an Artificial Neural Network

Energies ◽  
2019 ◽  
Vol 12 (17) ◽  
pp. 3328 ◽  
Author(s):  
Tomasz Halon ◽  
Ewa Pelinska-Olko ◽  
Malgorzata Szyc ◽  
Bartosz Zajaczkowski
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
District Heating ◽  
Cooling Capacity ◽  
Optimal Number ◽  
Normal Operation ◽  
Adsorption Chiller ◽  
Capacity Curves ◽  
Predicting Performance ◽  
Artificial Neural

In this paper, the feasibility of a multi-layer artificial neural network to predict both the cooling capacity and the COP of an adsorption chiller working in a real pilot plant is presented. The ANN was trained to accurately predict the performance of the device using data acquired over several years of operation. The number of neurons used by the ANN should be selected individually depending on the size of the training base. The optimal number of datasets in a training base is suggested to be 35. The predicted cooling capacity curves for a given adsorption chiller driven by the district heating are presented. Predictions of the artificial neural network used show good correlation with experimental results, with the mean relative deviation as low as 1.36%. The character of the cooling capacity curve is physically accurate, and during normal operation for cooling capacities ≥8 kW, the errors rarely exceed 1%.

scholarly journals Partial discharge detection for stator winding insulation of motors using artificial neural network

2018 ◽  
Vol 10 (7) ◽  
pp. 168781401878612 ◽  
Author(s):  
Yu-Tung Chen ◽  
Jui-Chien Lai ◽  
Yu-Ming Jheng ◽  
Cheng-Chien Kuo ◽  
Hong-Chan Chang
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
High Voltage ◽  
Partial Discharge ◽  
Recognition Rate ◽  
Gradient Algorithm ◽  
Normal Operation ◽  
Test Models ◽  
Artificial Neural ◽  
Stator Fault

In this article, the insulation fault detection of high-voltage motors by the artificial neural network algorithm is used. The proposed method can evaluate the status of operating motor without interrupting the normal operation. According to the measurement of partial discharge information, this research establishes the relationship of stator failures and pattern features. This study uses common high-voltage motor stator fault types to experimentally produce four types of stator test models with insulation defects; these models are compared with a healthy motor model. Through the learning of the artificial neural network, the experimental results show that the artificial neural network–based stator fault diagnosis system proposed in this article has a recognition rate as high as 90% when the conjugate gradient algorithm is used, and there are 20 neurons in the hidden layer.

An Artificial Neural Network Recognition Method for Single-Phase Earth Fault of Ultra-High Voltage Power Grid

2013 ◽  
Vol 333-335 ◽  
pp. 1659-1662
Author(s):  
Hai Wei Lu ◽  
Gang Wu ◽  
Chao Xiong
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Single Phase ◽  
Training Sample ◽  
Normal Operation ◽  
Recognition Method ◽  
Time Motion ◽  
Training Samples ◽  
Artificial Neural ◽  
Network Diagnosis

Fault diagnosis is very important to make the system return to normal operation quickly after an accident. This paper diagnoses the specific component failure and failure area when the real-time motion information of inputting protection and switch transferred to a trained artificial neural network model by building an artificial neural network diagnosis model of components such as transmission line, bus bar and transformer, training the artificial neural network through taking the failure rule which is found by the historic fault data as a training sample. This method has obvious advantages in the accuracy and speed of diagnosis compared with the previous artificial neural network and overcomes the shortcomings of the incompletion of training samples and not well dealing with the heuristic knowledge.

scholarly journals HETEROTIC GROUP FORMATION IN PSIDIUM GUAJAVA L. BY ARTIFICIAL NEURAL NETWORK AND DISCRIMINANT ANALYSIS

2016 ◽  
Vol 38 (1) ◽  
pp. 151-157 ◽  
Author(s):  
BIANCA MACHADO CAMPOS ◽  
ALEXANDRE PIO VIANA ◽  
SILVANA SILVA RED QUINTAL ◽  
CIBELLE DEGEL BARBOSA ◽  
ROGÉRIO FIGUEIREDO DAHER
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Discriminant Analysis ◽  
Group Formation ◽  
Psidium Guajava ◽  
Optimal Number ◽  
Heterotic Group ◽  
Linear Discriminant ◽  
Artificial Neural ◽  
Artificial Neural Network Ann

ABSTRACT The present study aimed at evaluating the heterotic group formation in guava based on quantitative descriptors and using artificial neural network (ANN). For such, we evaluated eight quantitative descriptors. Large genetic variability was found for the eight quantitative traits in the 138 genotypes of guava. The artificial neural network technique determined that the optimal number of groups was three. The grouping consistency was determined by linear discriminant analysis, which obtained classification percentage of the groups, with a value of 86 %. It was concluded that the artificial neural network method is effective to detect genetic divergence and heterotic group formation.

Predictability of the Loop Current Variation and Eddy Shedding Process in the Gulf of Mexico Using an Artificial Neural Network Approach

2015 ◽  
Vol 32 (5) ◽  
pp. 1098-1111 ◽  
Author(s):  
Xiangming Zeng ◽  
Yizhen Li ◽  
Ruoying He
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Gulf Of Mexico ◽  
Function Analysis ◽  
Optimal Number ◽  
Loop Current ◽  
Neural Network Approach ◽  
Current Variation ◽  
Artificial Neural ◽  
Eddy Shedding

AbstractA novel approach based on an artificial neural network was used to forecast sea surface height (SSH) in the Gulf of Mexico (GoM) in order to predict Loop Current variation and its eddy shedding process. The empirical orthogonal function analysis method was applied to decompose long-term satellite-observed SSH into spatial patterns (EOFs) and time-dependent principal components (PCs). The nonlinear autoregressive network was then developed to predict major PCs of the GoM SSH in the future. The prediction of SSH in the GoM was constructed by multiplying the EOFs and predicted PCs. Model sensitivity experiments were conducted to determine the optimal number of PCs. Validations against independent satellite observations indicate that the neural network–based model can reliably predict Loop Current variations and its eddy shedding process for a 4-week period. In some cases, an accurate forecast for 5–6 weeks is possible.

scholarly journals The Water Level and Outflow Prediction Using the Artificial Neural Network (ANN) for the Management of the Reservoir Flooding

2019 ◽  
Vol 8 (10) ◽  
pp. 4699-4706
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Water Level ◽  
Discharge Rate ◽  
Reservoir Management ◽  
Normal Operation ◽  
Ann Model ◽  
Reservoir Water ◽  
Artificial Neural ◽  
Artificial Neural Network Ann

The aim of operation reservoir during flood is to prevent overflow that endangers the dams. It is also to prevent flooding in the downstream of the dam, which leads to loss of life and property. This aim can be achieved with optimal reservoir management which is influenced by the reservoir’s condition during flooding such as: rain, reservoir storage, inflow, water level, and discharge of reservoir water released to the downstream. The successfully of the reservoir management depends on the accuracy of the estimated a). water level (due to the inflow of the reservoir) and b). outflow from the reservoir. One of the models which can be used to predict the water level and reservoir water released during flooding is the Artificial Neural Network (ANN). ANN can simulates flood events that are similar in fact to the previous occurence In this study a backpropagation ANN model was applied to the Wonogiri Reservoir in Central Java, Indonesia. The optimal ANN architecture produced in this study are the Input Pattern of 5-3-4 (which has a rain input recorded 1 – 5 hours earlier, a water level input recorded 1 – 3 hours earlier and a release input recorded 1 – 4 hours earlier). 27 pieces hidden layer, total epoch which is 200 and the learning rate of 0.01. The output is predicting the water level, the Outflow and Gate Opening of Reservoir. The current flood data was applied to the above model and it was concluded that the network can follow the flood management pattern adequately. In addition, the network is extra flexible with a lower flood discharge rate; and has the final elevation of the reservoir slightly lower than the normal operation.

scholarly journals Multi-variant differential evolution algorithm for feature selection

2020 ◽  
Vol 10 (1) ◽  
Author(s):  
Somaia Hassan ◽  
Ashraf M. Hemeida ◽  
Salem Alkhalaf ◽  
Al-Attar Mohamed ◽  
Tomonobu Senjyu
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Feature Selection ◽  
Differential Evolution ◽  
Optimization Technique ◽  
Differential Evolution Algorithm ◽  
Optimal Number ◽  
Scaling Factor ◽  
Redundant Data ◽  
Artificial Neural

Abstract This work introduces a new population-based stochastic search technique, named multi-variant differential evolution (MVDE) algorithm for solving fifteen well-known real world problems from UCI repository and compared to four popular optimization methods. The MVDE proposes a new self-adaptive scaling factor based on cosine and logistic distributions as an almost factor-free optimization technique. For more updated chances, this factor is binary-mapped by incorporating an adaptive crossover operator. During the evolution, both greedy and less-greedy variants are managed by adjusting and incorporating the binary scaling factor and elite identification mechanism into a new multi-mutation crossover process through a number of sequentially evolutionary phases. Feature selection decreases the number of features by eliminating irrelevant or misleading, noisy and redundant data which can accelerate the process of classification. In this paper, a new feature selection algorithm based on the MVDE method and artificial neural network is presented which enabled MVDE to get a combination features’ set, accelerate the accuracy of the classification, and optimize both the structure and weights of Artificial Neural Network (ANN) simultaneously. The experimental results show the encouraging behavior of the proposed algorithm in terms of the classification accuracies and optimal number of feature selection.

Sign in / Sign up

Export Citation Format

Share Document