Dst Index Forecast Based on Ground‐Level Data Aided by Bio‐Inspired Algorithms

Abstract

In this study, different hybridized techniques that combine an artificial neural network (ANN) with bio‐inspired optimization algorithms such as particle swarm optimization (PSO), genetic algorithm (GA), and a hybridized PSO+GA were applied to update the ANN connection weights and so forecast the disturbance storm time (Dst) index. The past values of Dst index time series were used as input parameters to forecast its variation from 1 to 6 hours ahead. The database collected considers 233,760 hourly data from 01 January 1990 to 31 August 2016, containing storms and quiet period, grouped into three data sets: learning set (116,880 hourly data points), validation set (58,440 data points), and testing set (58,440 data points). Several ANN configurations were studied and optimized during the training process by evaluating the root mean square error (RMSE) and the correlation coefficient (R). An analysis of the predictive capability of each method was made year per year, and according to the levels of the geomagnetic storm. Also, an additional test was applied to the proposed method using 17 intense geomagnetic storms reported during solar cycle 24, including the St. Patrick's Day storm of 2015. Results show that the hybridized ANN+PSO method can forecast the Dst index quite accurately from 1 to 3 h in advance (with RMSE≤5 nT and R≥0.9), while the ANN+PSO+GA method can forecast the Dst index quite accurately from 4 to 6 h ahead (RMSE≤7 nT and R≥0.8)

Bio-Inspired Hybridization of Artificial Neural Networks: An Application for Mapping the Spatial Distribution of Soil Texture Fractions

“…The local optima, convergence rate, and training time of the neural networks are greatly affected by their weights and biases, thus resulting in the need for optimizing the ANN parameters during the iteration process [57]. The hybridization of ANN with bio-inspired, optimization algorithms has the potential to be a powerful tool, which may overcome these challenges [28,57]. Hence, four bio-inspired, metaheuristic algorithms, namely, GA, BAT, PSO, and MBO, were trained to predict the PSFs as the target variables.…”

“…This process is repeated until the stopping criteria are met. By changing the particle's position and velocity from their solution space, PSO optimizes the weights and biases of ANN [57]. The cost function (fitness function) of the ith particle can be defined in the terms of RMSE in Equation 1:…”

“…The performances of the five ANN models (i.e., BP-ANN, GA-ANN, PSO-ANN, Bat-ANN, and MBO-ANN) for predicting three components of PSFs were compared in terms of MAE, RMSE, R 2 , and CCC. Once the ANN models were trained using the aforementioned algorithms, the optimal weights and biases of each input covariates were assigned to predict the PSFs [57,70]. Table 3 indicates the mean values of the accuracy metrics, calculated using the testing data from the repeated, nested 10-fold cross-validation procedure.…”

Deep Neural Networks With Convolutional and LSTM Layers for SYM‐H and ASY‐H Forecasting

“…However, despite the overall performance, it is unable to obtain accurate predictions of intense storms where the Dst reaches values lower than −250 nT. Lazzús et al (2019) explored and compared several ML techniques for the Dst index forecast problem. In his work, several Artificial Neural Networks (ANNs) are studied, as well as its combination with bio-inspired algorithms, such as particle swarm optimization, genetic algorithms, and a hybridization of both, to improve the system's accuracy.…”

Prominence of the training data preparation in geomagnetic storm prediction using deep neural networks

“…For such a comparison, we consider both 25 , and 26 . In the first, a similar way of splitting the data is used, and their test set has a consistent overlap with ours.…”