The ability of artificial intelligence and machine learning techniques in classification and detection of the types of data in large datasets lead to their popularity among scientists and researchers. Because of the presence of different load at different times in power systems, it is hard to provide an accurate mathematical model for such systems. On the other hand, most of the available protection devices in power grids work based on the estimated mathematical models of the grid. For this reason, power system utilizers usually suffer from the low accuracy of the available protection systems in fault detection and diagnosis. In this paper, a reliable machine learning technique is proposed to detect and classify different faults of smart grids. The proposed technique benefits from the principal component analysis (PCA) and linear discriminant analysis (LDA). The PCA is used to reduce the size of the dataset matrixes. The applied PCA reduces the dataset sizes and eliminates the possible singularity of the datasets. The LDA method is applied to the outputs data of the PCA to minimize the within class distance of the dataset and maximize the distance between classes. Finally, the well-known K-nearest neighbor technique is applied to detect the fault and determine its classes. The paper results demonstrate the effectiveness and robustness of the proposed algorithm in the determination of the fault class in smart grids.
Novel computational methods such as artificial neural networks (ANNs) can facilitate modeling and predicting results of tissue culture experiments and thereby decrease the number of experimental treatments and combinations. The objective of the current study is modeling and predicting in vitro shoot proliferation of Erysimum cheiri (L.) Crantz, which is an important bedding flower and medicinal plant. Its micropropagation has not been investigated before and as a case study multilayer perceptron- non-dominated sorting genetic algorithm-II (MLP-NSGAII) can be applied. MLP was used for modeling three outputs including shoots number (SN), shoots length (SL), and callus weight (CW) based on four variables including 6-benzylaminopurine (BAP), kinetin (Kin), 1-naphthalene acetic acid (NAA) and gibberellic acid (GA3). The R2 correlation values of 0.84, 0.99 and 0.93 between experimental and predicted data were obtained for SN, SL, and CW, respectively. These results proved the high accuracy of MLP model. Afterwards the model connected to Non-dominated Sorting Genetic Algorithm-II (NSGA-II) was used to optimize input variables for obtaining the best predicted outputs. The results of sensitivity analysis indicated that SN and CW were more sensitive to BA, followed by Kin, NAA and GA. For SL, more sensitivity was obtained for GA3 than NAA. The validation experiment indicated that the difference between the validation data and MLP-NSGAII predicted data were negligible. Generally, MLP-NSGAII can be considered as a powerful method for modeling and optimizing in vitro studies.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.