Impact of Noise on a Dynamical System: Prediction and Uncertainties from a Swarm-Optimized Neural Network

López-Caraballo, C. H.; Lazzús, Juan A.; Salfate, Ignacio; Rojas, Pedro Díaz; Rivera, Marco; Palma-Chilla, L.

doi:10.1155/2015/145874

Cited by 3 publications

(7 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recent studies have shown that PSO has been successfully applied to ANN training [e.g., Grimaldi et al , ; Lazzús et al , ; López‐Caraballo et al , ]. So once the ANN output values are calculated, the ANN weights and biases are adjusted

w_{i, j} \to w_{i, j}^{'}

by the PSO algorithm via equations and [ Lazzús et al , ; López‐Caraballo et al , ]. Here PSO modifies each ANN weight using the training data set.…”

Section: Methodsmentioning

confidence: 99%

“…No specific methods for determining the optimum number of neurons in the hidden layer (NHL) exist, giving rise to many potential alternate combinations [ Lazzús et al , ]. We opted to determine this number of neurons by adding neurons in a systematic manner from 1 to 40 units and by evaluating the RMSE, MAE, and R for each neural network architecture during the training phase process [ Lazzús et al , ; López‐Caraballo et al , ]. Figure S4 shows the correlation coefficient ( R ) found in the prediction of Dst ( t + d ) as a function of the number of neurons in the hidden layer NHL (see supporting information).…”

Section: Forecasting With Ann+psomentioning

confidence: 99%

“…Here the designed ANN was optimized using a particle swarm algorithm (PSO) [ Kennedy and Eberhart , ] to update ANN weights. Note that PSO has been recently applied in the optimization of ANN weights with promising results and performance [ Lazzús et al , ; López‐Caraballo et al , ]. For ANN, commonly applied optimization algorithms such as backpropagation learning algorithm [ Rumelhart and McClelland , ] can determine the network connection weights but do not have the capacity of controlling the parameter optimization in the absence of gradient information, while PSO can solve this problem [ Lazzús et al , ].…”

Section: Introductionmentioning

confidence: 99%

“…For ANN, commonly applied optimization algorithms such as backpropagation learning algorithm [ Rumelhart and McClelland , ] can determine the network connection weights but do not have the capacity of controlling the parameter optimization in the absence of gradient information, while PSO can solve this problem [ Lazzús et al , ]. The task of PSO in ANN is to obtain the optimum weights (i.e., best particle position) where several particles (i.e., possible problem solutions) move in a search space to obtain the best solution to the problem [ López‐Caraballo et al , ]. Note that the application of PSO to solve optimization problems in geosciences is scarce [ Yuan et al , ].…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Forecasting the Dst index using a swarm‐optimized neural network

et al. 2017

Self Cite

View full text Add to dashboard Cite

A hybrid technique that combines an artificial neural network with a particle swarm optimization (ANN+PSO) was used to forecast the disturbance storm time (Dst) index from 1 to 6 h ahead. Our ANN was optimized by PSO to update ANN weights and to predict the short‐term Dst index using past values as input parameters. The database used contains 233,760 hourly data from 1 January 1990 to 31 August 2016, considering storms and quiet period, grouped into three data sets: learning set (with 116,880 hourly data points), validation set (with 58,440 data points), and testing set (with 58,440 data points). Several ANN topologies were studied, and the best architecture was determined by systematically adding neurons and evaluating the root‐mean‐square error (RMSE) and the correlation coefficient (R) during the training process. These results show that the hybrid algorithm is a powerful technique for forecasting the Dst index a short time in advance like t + 1 to t + 3, with RMSE from 3.5 nT to 7.5 nT, and R from 0.98 to 0.90. However, t + 4 to t + 6 predictions become slightly more uncertain, with RMSE from 8.8 nT to 10.9 nT, and R from 0.86 to 0.79. Additionally, an exhaustive analysis according to geomagnetic storm magnitude was conducted. In general, the results show that our hybrid algorithm can be correctly trained to forecast the Dst index with appropriate precision and that Dst past behavior significantly affects adequate training and predicting capabilities of the implemented ANN.

show abstract

w_{i, j} \to w_{i, j}^{'}

by the PSO algorithm via equations and [ Lazzús et al , ; López‐Caraballo et al , ]. Here PSO modifies each ANN weight using the training data set.…”

Section: Methodsmentioning

confidence: 99%

Section: Forecasting With Ann+psomentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Forecasting the Dst index using a swarm‐optimized neural network

et al. 2017

Self Cite

View full text Add to dashboard Cite

show abstract

“…ANNs are accepted as the most powerful nonlinear technique in mathematical modeling (Lazzús, 2011; López-Caraballo et al, 2015), and many models of neural networks have been used in the wind speed estimation. ANNs are accepted as the most powerful nonlinear technique in mathematical modeling (Haykin, 1999; Lazzús, 2011; López-Caraballo et al, 2015), and many models of neural networks have been used for wind speed estimation with several methodologies (see section “Introduction”). In this work, a BPNN with Levenberg‒Marquardt algorithm (Hagan and Menhaj, 1994) was used to represent nonlinear relationships among parameters (Haykin, 1999).…”

Section: Methodsmentioning

confidence: 99%

24-hours wind speed forecasting and wind power generation in La Serena (Chile)

et al. 2018

View full text Add to dashboard Cite

This article presents 24-h wind speed forecasting for the city of La Serena in Chile and a methodology to explore forecasting effects on the production of wind turbine power. To that end, we used meteorological data from a weather station located in the southern zone of the hyper-arid Atacama Desert. In this area, energy resources are economically and environmentally important, and wind speed forecasting plays a vital role in the management and marketing processes of wind potential via wind farms. To contribute to the development of this energy, we propose carrying out the short-term prediction of 12 and 24 h ahead (identified as Ws(t + 12) and Ws(t + 24), respectively) using an artificial neural network with backpropagation approach. Hourly time series of wind speed, temperature, and relative humidity (from 2003 to 2006) were considered to characterize the artificial neural network in the training phase, while we used data from the year 2007 to check the efficiency of our prediction. For artificial neural network Ws(t + 12) and Ws(t + 24) models, we obtained similar performance of wind speed prediction with root mean square error of around 0.7 m s −1 and with maximum and minimum residuals of +4 and-4 m s −1 , respectively. Based on the results, we gain a reliable tool to characterize wind speed properties in the range of 1 day within 20% of uncertainty. Moreover, this tool becomes useful to study the effects of our artificial neural network Ws(t + 12) and Ws(t + 24) models on the generation of wind energy from a wind power turbine parametrization.

show abstract

K-means++ for Critical Component Identification: Power Grid Case Study Using Measurement-Based Analysis

SaeedKandezy,

Jiang

2024

Preprint

View full text Add to dashboard Cite

The inherent capabilities of the K-means++ algorithm to approximate system dynamics within complex systems are subjected by constructing a network structure that captures interconnections among identified components, extending its original purpose as a data clustering method and transforming it into a tool for systems analysis. Leveraging advanced measurement technologies and sophisticated data collection systems, the K-means++ algorithm unveils hidden relationships among components and identifies critical elements. This study explored the algorithm's potential, facilitating the identification of critical components to enhance system operation, control, optimization, and decision-making and examining the practicality and resiliency of the method in real-world application with noisy and limited data. A case study conducted on power systems (IEEE 39-bus and IEEE 300-bus systems) exemplifies K-means++'s capacity to accurately identify critical components and approximate system dynamics, supported by performance metrics affirming its effectiveness and robustness in system analysis through measurements of bus similarity within clusters based on standard deviation and comparison of net tie-line flows in equivalent networks with the original network across scenarios. Performance metrics, including the Silhouette Score, Davies-Bouldin Index, and Variation of Information (VI) score, further validated K-means++'s performance, yielding reliable and consistent results.

show abstract

Impact of Noise on a Dynamical System: Prediction and Uncertainties from a Swarm-Optimized Neural Network

Cited by 3 publications

References 39 publications

Forecasting the Dst index using a swarm‐optimized neural network

Forecasting the Dst index using a swarm‐optimized neural network

24-hours wind speed forecasting and wind power generation in La Serena (Chile)

K-means++ for Critical Component Identification: Power Grid Case Study Using Measurement-Based Analysis

Contact Info

Product

Resources

About