Search

This paper introduces an innovative framework for wind power prediction that focuses on the future of energy forecasting utilizing intelligent deep learning and strategic feature engineering. This research investigates the application of a state-of-the-art deep learning model for wind energy prediction to make extremely short-term forecasts using real-time data on wind generation from New South Wales, Australia. In contrast with typical approaches to wind energy forecasting, this model relies entirely on historical data and strategic feature engineering to make predictions, rather than relying on meteorological parameters. A hybrid feature engineering strategy that integrates features from several feature generation techniques to obtain the optimal input parameters is a significant contribution to this work. The model's performance is assessed using key metrics, yielding optimal results with a Mean Absolute Error (MAE) of 8.76, Mean Squared Error (MSE) of 139.49, Root Mean Squared Error (RMSE) of 11.81, R-squared score of 0.997, and Mean Absolute Percentage Error (MAPE) of 4.85%. Additionally, the proposed framework outperforms six other deep learning and hybrid deep learning models in terms of wind energy prediction accuracy. These findings highlight the importance of advanced data analysis for feature generation in data processing, pointing to its key role in boosting the precision of forecasting applications. [ABSTRACT FROM AUTHOR]

This paper examines the relationship between locus of control (LOC) and the demand for supplementary health insurance (SUPP). Drawing on longitudinal data from Germany, we document robust evidence that individuals internal LOC increases the take up of supplementary private health insurance (SUPP). We find that the effect of one standard deviation increase in the measure of internal LOC on the probability of SUPP purchase is equivalent to a 14 percent increase in household income. Second, we find that the positive association between self-reported health and SUPP becomes small and insignificant when we control for LOC. These results suggest that LOC might be an unobserved individual trait that can partly explain previously documented evidence of advantageous selection into SUPP. Third, we find comparable results using data from Australia, which enhances the external validity of our results. [ABSTRACT FROM AUTHOR]

• Develops a deep Bayesian survival model that integrates deep neural network and Bayesian inference into the survival analysis for rail useful lifetime modelling. • Demonstrates survival analysis is an effective way to handle censored data problems in rail useful lifetime modelling. • Conducts extensive evaluation to verify the effectiveness of this model, using Concordance index, root mean square error, and a proposed CW-index. • The estimated lifetimes obtained by this model are the closest to real lifetimes compared to other commonly used models. • This model can provide the confidence interval of the estimated lifetime, which is more useful than point estimates in decision-making and maintenance planning of railroad systems. Reliable estimation of rail useful lifetime can provide valuable information for predictive maintenance in railway systems. However, in most cases, lifetime data is incomplete because not all pieces of rail experience failure by the end of the study horizon, a problem known as censoring. Ignoring or otherwise mistreating the censored cases might lead to false conclusions. Survival approach is particularly designed to handle censored data for analysing the expected duration of time until one event occurs, which is rail failure in this paper. This paper proposes a deep Bayesian survival approach named BNN-Surv to properly handle censored data for rail useful lifetime modelling. The proposed BNN-Surv model applies the deep neural network in the survival approach to capture the non-linear relationship between covariates and rail useful lifetime. To consider and quantify uncertainty in the model, Monte Carlo dropout, regarded as the approximate Bayesian inference, is incorporated into the deep neural network to provide the confidence interval of the estimated lifetime. The proposed approach is implemented on a four-year dataset including track geometry monitoring data, track characteristics data, various types of defect data, and maintenance and replacement (M&R) data collected from a section of railway tracks in Australia. Through extensive evaluation, including Concordance index (C-index) and root mean square error (RMSE) for evaluating model performance, as well as a proposed CW-index for evaluating uncertainty estimations, the effectiveness of the proposed approach is confirmed. The results show that, compared with other commonly used models, the proposed approach can achieve the best concordance index (C-index) of 0.80, and the estimated rail useful lifetimes are closer to real lifetimes. In addition, the proposed approach can provide the confidence interval of the estimated lifetime, with a correct coverage of 81% of the actual lifetime when the confidence interval is 1.38, which is more useful than point estimates in decision-making and maintenance planning of railroad systems. [ABSTRACT FROM AUTHOR]

This paper presents a soil moisture retrieval scheme from Cyclone Global Navigation Satellite System (CYGNSS) delay-Doppler maps (DDMs) over land. The proposed inversion method consists of a hybrid global and local optimization method and a physics-based bistatic scattering forward model. The forward model was developed for bare-to-densely vegetated terrains, and it predicts the circularly polarized bistatic radar cross section DDM of the land surface. This method was tested on both simulated DDMs and CYGNSS DDMs over the Soil Moisture Active Passive (SMAP) Yanco core validation site in Australia. About 250 CYGNSS DDMs from 2019 and 2020 over the Yanco site were used for validation. The simulated DDMs were for grassland and forest vegetation types. The vegetation type of the Yanco validation site was grassland. The vegetation water content (VWC) was 0.19   kg m − 2 and 4.89   kg m − 2 for the grassland and forest terrains, respectively. For the case when the surface roughness is known to the algorithm, the unbiased root mean square error (ubRMSE) of soil moisture estimates was less than 0.03   m 3 m − 3 while it was approximately 0.06   m 3 m − 3 and 0.09   m 3 m − 3 for the validation results from 2019 and 2020, respectively. The retrieval algorithm generally had enhanced performance for smaller values of soil moisture. For the case when both the soil moisture and surface roughness are unknown to the algorithm and only a single DDM is used for retrieval, the validation results showed an expected reduced performance, with an an ubRMSE of less than 0.12   m 3 m − 3 . [ABSTRACT FROM AUTHOR]

Mungbean, a grain legume with high nutritional value, is grown widely throughout Asia and increasingly in Australia. Despite growing interest amongst farmers, mungbean remains an inconsistent and thus risky crop to plant in Australia. Cropping system models like the Agricultural Production Systems sIMulator (APSIM) are valuable tools for helping farmers to examine options for improving crop management and assess production risks across potential growing regions for mungbean. This paper outlines the simulation capacity of a new mungbean crop model parameterized using the Plant Modelling Framework in APSIM Next Generation, the newest version of the APSIM framework. The aim of the paper is to document the parameterization and validation processes of the model. The new mungbean model was built using data from 28 field experiments to simulate measured phenology, canopy development, biomass accumulation/partitioning, stress responses, N fixation, root growth, and yield across a wide range of environments. The root mean squared error (RMSE) in predictions for grain weight and aboveground weight were 25.4 g m−2 and 91.4 g m−2, respectively. The model successfully captured the dynamics of crop response to sowing dates, water/irrigation regimes, and climate. The new mungbean model is a robust and accurate tool for use in Australia and tropical/sub-tropical Asia. Researchers can use the new mungbean model to determine best management practices such as the optimal time to sow mungbeans in different environments. The output from model simulations can help farmers assess risks associated with sowing at different times and soil water conditions specific to their region. Such risk analysis can improve farmer decision-making confidence in mungbean, increasing its potential production for Australia. Overall, the new APSIM mungbean model can be used effectively to identify and close the mungbean yield gap, mitigate risk of crop failure, and increase profits for mungbean farmers in Australia and tropical/sub-tropical Asia; it has the capacity to assist with increasing mungbean production globally under changing climate conditions. • The APSIM Next Generation mungbean model was built using data from 28 diverse field experiments. • Detailed phenology, biomass and N partitioning, and canopy data was used to parameterize parameters. • The model captures mungbean's dynamic response to water stress and targeted irrigation. • It is a robust, data-driven crop model that performs well in Australia and tropical/subtropical Asia. [ABSTRACT FROM AUTHOR]

• An innovative framework to derive intensity-frequency-duration curves from satellite data for remote areas. • A variant of the Bartlett-Lewis model calibrated with sub-daily satellite precipitation data to disaggregate daily gauge data. • This new framework was tested for a wide range of climates in Australia. • IFD curves can be derived with reasonable accuracy for data scarce regions. Intensity-frequency-duration (IFD) curves (also known as intensity-duration-frequency (IDF) curves in some regions) are widely used to take into consideration the probability of occurrence of extreme precipitation for the design of water related infrastructure. Derivation of IFD curves is based on statistical analysis of observed extreme precipitation events and thus the reliability of the curves mostly depends on the record length of observations. At sub-hourly timescales however, ground-based precipitation observations are often absent or very limited in terms of spatial coverage and record length, particularly in remote areas around the world. Given the availability of satellite-derived precipitation products (SPPs) with finer spatiotemporal resolution and near-global coverage, there is great potential to use these products in the planning and design of infrastructure. This paper proposes a novel framework, SIFD (Satellite-derived IFD curves) to combine long-term daily gauge data and sub-hourly SPP to estimate IFD curves for time intervals from 0.5 to 24 h. A recent variant of the Random Parameter Bartlett-Lewis Rectangular Pulse model (RPBLRPX) was used to disaggregate daily gauge data to derive IFD curves. In the absence of sub-daily rain gauge data, the statistics required for the RPBLRPX model were obtained using the Integrated Multi-satellitE Retrievals for GPM (IMERG; 30-min, 0.1° resolution) product derived from the Global Precipitation Measurement (GPM) mission. Performance of the SIFD framework was tested against gauge data from seven Australian sites across a wide range of climatologies. The disaggregation scheme coupled with the RPBLRPX model was capable of reproducing the wet-period mean intensity, wet-period standard deviation, and wet-period fraction with reasonable accuracy across the sub-daily timescales. The coefficient of determination ( r 2) values were 0.39 (mean intensity), 0.38 (standard deviation), and 0.87 (wet-period fraction) at 0.5 h timescale. Furthermore, despite some differences in methodology and record length, the IFD estimates computed using the SIFD framework were fairly close to those of the Bureau of Meteorology (BoM). The skill metrics for Sydney were the best with a bias ratio (BR) of 1.01, root mean square error (RMSE) of 13% of the mean of the BoM estimates, and a Nash-Sutcliffe coefficient of efficiency (NSE) of 0.97. The worst case was found at Hobart (BR = 1.43, RMSE = 66%, and NSE = 0.26). Overall, the results indicate that the proposed SIFD framework could provide a viable alternative to estimate unbiased IFD curves at the sub-hourly timescale in the absence of adequate sub-hourly ground-based observations. [ABSTRACT FROM AUTHOR]

This study proposes a new hybrid deep learning (DL) model, the called CSVR, for Global Solar Radiation (GSR) predictions by integrating Convolutional Neural Network (CNN) with Support Vector Regression (SVR) approach. First, the CNN algorithm is used to extract local patterns as well as common features that occur recurrently in time series data at different intervals. Then, the SVR is subsequently adopted to replace the fully connected CNN layers to predict the daily GSR time series data at six solar farms in Queensland, Australia. To develop the hybrid CSVR model, we adopt the most pertinent meteorological variables from Global Climate Model and Scientific Information for Landowners database. From a pool of Global Climate Models variables and ground-based observations, the optimal features are selected through a metaheuristic Feature Selection algorithm, an Atom Search Optimization method. The hyperparameters of the proposed CSVR are optimized by mean of the HyperOpt method, and the overall performance of the objective algorithm is benchmarked against eight alternative DL methods, and some of the other Machine Learning approaches (LSTM, DBN, RBF, BRF, MARS, WKNNR, GPML and M5TREE) methods. The results obtained shows that the proposed CSVR model can offer several predictive advantages over the alternative DL models, as well as the conventional ML models. Specifically, we note that the CSVR model recorded a root mean square error/mean absolute error ranging between ≈ 2.172–3.305 MJ m2/1.624–2.370 MJ m2 over the six tested solar farms compared to ≈ 2.514–3.879 MJ m2/1.939–2.866 MJ m2 from alternative ML and DL algorithms. Consistent with this predicted error, the correlation between the measured and the predicted GSR , including the Willmott's, Nash-Sutcliffe's coefficient and Legates & McCabe's Index was relatively higher for the proposed CSVR model compared to other DL and Machine Learning methods for all of the study sites. Accordingly, this study advocates the merits of CSVR model to provide a viable alternative to accurately predict GSR for renewable energy exploitation, energy demand or other forecasting-based applications. [Display omitted] • This paper presents a hybrid CSVR algorithm for global solar radiation prediction. • The proposed approach combines Convolutional Neural Network with a Support Vector Machine. • An Atom Search-based feature selection is applied to improve the performance of the system. • Experiments are carried out in data at six solar energy farms in Queensland, Australia. • The results show the efficacy of deep learning CSVR for solar energy prediction problems. [ABSTRACT FROM AUTHOR]

Medium resolution satellite-derived fractional cover estimates of bare soil (f BS), photosynthetic vegetation (f PV), and non-photosynthetic vegetation (f NPV) provide a powerful means to study arid ecosystem dynamics. This paper employed remote sensing estimates of f PV and f NPV from five case study sites from Australia's vegetated dunefields to observe (a) vegetation growth response to rainfall 'pulses' and subsequent transition to non-photosynthetic dormancy or senescence; (b) multiple time scales of antecedent climatic influence on vegetation cover; (c) the susceptibility of dunes to wind-blown sand drift during periods of low cover; and (d) the implications of image resolution choice when ground cover is heterogeneous. A spectral unmixing model for Australia's arid zone (termed 'AZN') was first developed by generating endmembers from a dataset of 1405 field surveys; Landsat time series estimates of f PV and f NPV were subject to a Seasonal-Trend decomposition by Loess (STL); Time series components were correlated with rainfall (P) and aridity at various accumulation periods; Fire maps were used to compare the climatic response of unburnt and burnt vegetation; Landform maps were used to isolate dune vegetation cover from the adjacent interdunes; and Landsat estimates of erodible area were compared with Sentinel-2 and WorlView-3 data. The new AZN model yielded Root Mean Square Error (RMSE) estimates of 14.5% (f BS), 6.5% (f PV) and 15.8% (f NPV) during cross validation. The AZN model also compared favourably to an existing continental-scale model when evaluated with independent reference data. Rainfall pulse responses of dune vegetation were detected initially as f PV , and 3–9 months later as a peak in f NPV. Components of f PV responded to P accumulated over 3–9-months (intra-annual), and 12–15-months (trend). The long-term build-up of f NPV , if left unburnt, was influenced by rainfall patterns over the preceding 45–114 months. Fires reduced both the depth and strength of antecedent rainfall's influence on vegetation, and vegetation was often more sensitive to P than to aridity. Erodibility (total cover <14%) and partial erodibility (cover <35%) were more common at the driest sites but did not universally match aridity levels, due to fires and differing vegetation. The targeting of dune crest regions highlighted their enhanced susceptibility to sand drift (in most cases), and, given their occurrence on relatively narrow ridges (~30 m), the importance of estimating cover at Landsat resolutions or better (e.g. Sentinel-2). • Arid fractional vegetation cover was unmixed from Landsat and Sentinel-2 data. • Pulse-reserve behaviour was detected as coupled peaks in green then dry vegetation. • Decomposed time series revealed multiple time scales of rainfall response. • 30 m resolution or less is required to detect wind erosion risk on vegetated dunes. [ABSTRACT FROM AUTHOR]