Classical Artificial Neural Network Model Research Articles

Development of Artificial Intelligence Based Safety Performance Measures for Urban Roundabouts

A reliable model for predicting crash frequency at roundabouts is an essential tool for evaluating the safety measures of a roundabout. This study developed a hybrid PSO-ANN model by optimizing the modeling parameters of the classical artificial neural network (ANN) model with the particle swarm optimization algorithm (PSO). The performance accuracy of the models was evaluated using the mean absolute error (MAE), root mean square error (RMSE), and determination coefficients (DC). The PSO-ANN model predicted the crash frequency with very good accuracy at the testing stage (DC = 0.7935). The hybrid model could improve the performance of the classical ANN model by up to 23.3% in the training stage and 16.9% in the testing stage. In addition to the statistical measures, graphical approaches (scatter and violin plots) were also used for evaluating the models’ accuracy. Both statistical and graphical evaluation techniques prove the reliability and accuracy of the proposed hybrid model in predicting the crash frequency at roundabouts.

Efficient seismic risk analysis of gravity dams via screening of intensity measures and simulated non-parametric fragility curves

The computational cost of nonlinear numerical simulation of gravity dams brings resistance to the popularization of seismic risk analysis under the framework of Performance-based Earthquake Engineering. Insufficient seismic records and inappropriate parametric fragility models will mislead the assessment of structural performance and total risk. A computationally efficient methodology suitable for a large number of seismic waves is proposed based on screening for intensity measures and a surrogate model to obtain the non-parametric fragility curves. It specifically integrates the rigorous numerical technique of seismic damage to the dam-reservoir-foundation system, the comprehensive comparison of intensity measures, the surrogate model for classification of limit states, and the seismic risk analysis of gravity dams. A typical high gravity dam and a classical artificial neural network model are employed to illustrate the effectiveness of the proposed methodology through a step-by-step scheme. The result shows that the simulated non-parametric fragility curves are more competent to consider the uncertainty of ground motions. The obtained structural fragility and risk is more accurate than the conventional method.

Improving the Spatial Prediction of Soil Organic Carbon Content in Two Contrasting Climatic Regions by Stacking Machine Learning Models and Rescanning Covariate Space

Understanding the spatial distribution of soil organic carbon (SOC) content over different climatic regions will enhance our knowledge of carbon gains and losses due to climatic change. However, little is known about the SOC content in the contrasting arid and sub-humid regions of Iran, whose complex SOC–landscape relationships pose a challenge to spatial analysis. Machine learning (ML) models with a digital soil mapping framework can solve such complex relationships. Current research focusses on ensemble ML models to increase the accuracy of prediction. The usual ensemble method is boosting or weighted averaging. This study proposes a novel ensemble technique: the stacking of multiple ML models through a meta-learning model. In addition, we tested the ensemble through rescanning the covariate space to maximize the prediction accuracy. We first applied six state-of-the-art ML models (i.e., Cubist, random forests (RF), extreme gradient boosting (XGBoost), classical artificial neural network models (ANN), neural network ensemble based on model averaging (AvNNet), and deep learning neural networks (DNN)) to predict and map the spatial distribution of SOC content at six soil depth intervals for both regions. In addition, the stacking of multiple ML models through a meta-learning model with/without rescanning the covariate space were tested and applied to maximize the prediction accuracy. Out of six ML models, the DNN resulted in the best modeling accuracies, followed by RF, XGBoost, AvNNet, ANN, and Cubist. Importantly, the stacking of models indicated a significant improvement in the prediction of SOC content, especially when combined with rescanning the covariate space. For instance, the RMSE values for SOC content prediction of the upper 0–5 cm of the soil profiles of the arid site and the sub-humid site by the proposed stacking approaches were 17% and 9% respectively, less than that obtained by the DNN models—the best individual model. This indicates that rescanning the original covariate space by a meta-learning model can extract more information and improve the SOC content prediction accuracy. Overall, our results suggest that the stacking of diverse sets of models could be used to more accurately estimate the spatial distribution of SOC content in different climatic regions.

Monthly streamflow forecasting using neuro-wavelet techniques and input analysis

ABSTRACTCombinations of low-frequency components (also known as approximations) resulting from the wavelet decomposition are tested as inputs to an artificial neural network (ANN) in a hybrid approach, and compared to classical ANN models for flow forecasting for 1, 3, 6 and 12 months ahead. In addition, the inputs are rewritten in terms of the flow, revealing what type of information was being provided to the network, in order to understand the effect of the approximations on the forecasting performance. The results show that the hybrid approach improved the accuracy of all tested models, especially for 1, 3 and 6 months ahead. The input analyses show that high-frequency components are more important for shorter forecast horizons, while for longer horizons, they may worsen the model accuracy.

Short term electricity price forecast based on environmentally adapted generalized neuron

The liberalization of the power markets gained a remarkable momentum in the context of trading electricity as a commodity. With the upsurge in restructuring of the power markets, electricity price plays a dominant role in the current deregulated market scenario which is majorly influenced by the economics being governed. In the deregulated environment price forecasting is an important aspect for the power system planning. The problem of price forecasting can be entirely viewed as a signal processing problem with proper estimation of model parameters, modeling of uncertainties, etc. Among the different existing models the artificial neural network based models have gained wide popularity due their black box structure but it too has its own limitations. In the proposed work in order to overcome the limitations of the classical artificial neural network model, generalized neuron model is used for forecasting the short term electricity price of Australian electricity market. The pre-processing of the input parameters is accomplished using wavelet transform for better representation of the low and high frequency components. The free parameters of the generalized neuron model are tuned using environment adaptation method algorithm for increasing the generalization ability and efficacy of the model.

Estimates of hydroelectric generation using neural networks with the artificial bee colony algorithm for Turkey

The primary objective of this study was to apply the ANN (artificial neural network) model with the ABC (artificial bee colony) algorithm to estimate annual hydraulic energy production of Turkey. GEED (gross electricity energy demand), population, AYT (average yearly temperature), and energy consumption were selected as independent variables in the model. The first part of the study compared ANN-ABC model performance with results of classical ANN models trained with the BP (back propagation) algorithm. Mean square and relative error were applied to evaluate model accuracy. The test set errors emphasized positive differences between the ANN-ABC and classical ANN models. After determining optimal configurations, three different scenarios were developed to predict future hydropower generation values for Turkey. Results showed the ANN-ABC method predicted hydroelectric generation better than the classical ANN trained with the BP algorithm. Furthermore, results indicated future hydroelectric generation in Turkey will range from 69.1 to 76.5 TWh in 2021, and the total annual electricity demand represented by hydropower supply rates will range from 14.8% to 18.0%. However, according to Vision 2023 agenda goals, the country plans to produce 30% of its electricity demand from renewable energy sources by 2023, and use 20% less energy than in 2010. This percentage renewable energy provision cannot be accomplished unless changes in energy policy and investments are not addressed and implemented. In order to achieve this goal, the Turkish government must reconsider and raise its own investments in hydropower, wind, solar, and geothermal energy, particularly hydropower.

Daily streamflow forecasting using a wavelet transform and artificial neural network hybrid models

New wavelet and artificial neural network (WA) hybrid models are proposed for daily streamflow forecasting at 1, 3, 5 and 7 days ahead, based on the low-frequency components of the original signal (approximations). The results show that the proposed hybrid models give significantly better results than the classical artificial neural network (ANN) model for all tested situations. For short-term (1-day ahead) forecasts, information on higher-frequency signal components was essential to ensure good model performance. However, for forecasting more days ahead, lower-frequency components are needed as input to the proposed hybrid models. The WA models also proved to be effective for eliminating the lags often seen in daily streamflow forecasts obtained by classical ANN models. Editor D. Koutsoyiannis; Associate editor L. SeeCitation Santos, C.A.G. and Silva, G.B.L., 2013. Daily streamflow forecasting using a wavelet transform and artificial neural network hybrid models. Hydrological Sciences Journal, 59 (2), 312–324.

Multi-lead ahead prediction model of reference evapotranspiration utilizing ANN with ensemble procedure

Obtaining an accurate estimate of the reference evapotranspiration (ETo) can be difficult, especially when there is insufficient data to utilize the Penman–Monteith method. Artificial intelligence–based methods may provide reliable prediction models for several applications in engineering. However, time-series prediction based on artificial neural network (ANN) learning algorithms is fundamentally problematic. For example, the ANN model can experience over-fitting during training and, in consequence, lose its generalization. In this research, several over-fitting procedures have been augmented with the classical ANN model, are proposed. This model was applied to the prediction of the daily ETo at Rasht city, located in the north part of Iran, by using the minimum and maximum daily temperature of the region collected from 1975–1988. In addition, three different scenarios have been developed in order to achieve better prediction accuracy. The results showed that the proposed ENN model successfully predicted the daily ETo with a significant level of accuracy using only the maximum and minimum temperatures. The model also outperformed the classical ANN method. In addition, the proposed ENN compared with Hargreaves and Samani (Appl Eng Agric 1:96–99, 1985) (HGS) model and showed the ENN provides more accurate prediction for ETo. Furthermore, the proposed model could provide relatively good level of accuracy when examined for multi-lead predictions, which could not be afford by HGS model.

Discrete wavelet neural network approach in significant wave height forecasting for multistep lead time

Recently Artificial Neural network (ANN) was extensively used as non-linear inter-extrapolator for ocean wave forecasting as well as other application in ocean engineering. In this current study, the Wavelet transform was hybridised with ANN naming Wavelet Neural Network (WLNN) for significant wave height forecasting near Mangalore, west coast of India, upto 48h lead time. The main time series of significant wave height data were decomposed to multiresolution time series using discrete wavelet transformations. Then, the multiresolution time series data were used as input of the ANN to forecast the significant wave height at different multistep lead time. It was shown how the proposed model, WLNN, that makes use of multiresolution time series as input, allows for more accurate and consistent predictions with respect to classical ANN models. The proposed wavelet model (WLNN) results revealed that it was better forecasted and consistent than single ANN model because of using multiresolution time series data as inputs.

Regularized versus non-regularized neural network model for prediction of saturated soil-water content on weathered granite soil formation

Modeling unsaturated water flow in soil requires knowledge of the hydraulic properties of soil. However, correlation between soil hydraulic properties such as the relationship between saturated soil-water content θs and saturated soil hydraulic conductivity k s as function of soil depth is in stochastic pattern. On the other hand, soil-water profile process is also believed to be highly non-linear, time varying, spatially distributed, and not easily described by simple models. In this study, the potential of implementing artificial neural network (ANN) model was proposed and investigated to map the soil-water profile in terms of k s and θs with respect to the soil depth d. A regularized neural network (NN) model is developed to overcome the drawbacks of conventional prediction techniques. The use of regularized NN advantaged avoid over-fitting of training data, which was observed as a limitation of classical ANN models. Site experimental data sets on the hydraulic properties of weathered granite soils were collected. These data sets include the observed values of saturated and unsaturated hydraulic conductivities, saturated water contents, and retention curves. The proposed ANN model was examined utilizing 49 records of data collected from field experiments. The results showed that the regularized ANN model has the ability to detect and extract the stochastic behavior of saturated soil-water content with relatively high accuracy.