The tremendous rise of electrical energy demand worldwide has led to many problems related to efficient use of electrical energy, consequently posing difficult challenges to electricity consumers of all levels—from households to large companies’ facilities. Most of these challenges could be overcome by the accurate prediction of electricity demand. Additionally, balance responsibility includes the penalty-based financial mechanism causing extra expense for badly estimated consumption, above the allowed imbalance limits. In this paper, a method for electricity consumption prediction based on artificial neural networks is proposed. The electricity consumption dataset is obtained from a cold storage facility, which generates data in hourly intervals. The data obtained are measured for a period of over 2 years and then separated to four seasons, so different models are developed for each season. Five different network structures (ordinary RNN, LSTM, GRU, bidirectional LSTM, bidirectional GRU) for five different values of horizon, i.e., input data (one day, two days, four days, one week, two weeks) are examined. Performance indices, such as mean absolute percentage error (MAPE), root mean square error (RMSE), mean absolute error (MAE) and mean square error (MSE), are used in order to obtain qualitative and quantitative comparisons among the obtained models. The results show that the modifications of recurrent neural networks perform much better than ordinary recurrent neural networks. GRU and LSTMB structures with horizons of 168h and 336h are found to have the best performances.

Regular surveillance, data acquisition, and visual observation of high embankment dams are extremely important for the stability analysis of these structures. The stability issues that could occur during a dam's lifetime are mainly related to slope i...

The increasing integration of photovoltaic systems into modern power grids requires forecasting models that not only provide accurate predictions but also reliable uncertainty quantification under evolving operating conditions. In this paper, we propose an Out-of-distribution-aware time series conformal prediction framework with adaptive retraining, designed to address key limitations of standard conformal prediction methods in temporally dependent and dynamically changing environments. The framework is built upon the Ensemble batch prediction intervals method, which enables distribution-free uncertainty quantification without relying on a fixed calibration set, making it particularly suitable for time series applications. To ensure robustness to distribution shifts, a conformal out-of-distribution detection module is incorporated, where out-of-distribution detection is formulated as a hypothesis testing problem and enhanced through calibration-conditional p-values obtained via the Simes correction, providing conservative false-positive control intended to limit unnecessary model retraining. The proposed framework demonstrates superior performance compared to state-of-the-art approaches in uncertainty quantification, while conformal out-of-distribution detection reduces false positives and the adaptive retraining mechanism ensures effective adaptation to evolving data distributions in real-world scenarios.

This paper discusses the process of training a neural network intended for predicting power generation in a solar power plant based on meteorological data and the historical data of power measurement readings. Initially, the neural network was trained on a wide array of different solar and weather quantities as inputs (26 in total), but since the prediction errors of this model were considerably large, the model was retrained, but on a much smaller subset of critical input variables, which resulted in an increase of accuracy where MAE dropped significantly.

In this paper, we firstly investigate the relationships between weather conditions and solar plant power output during winter and near-winter months (March and October) and afterwards, we create multiple neural network models for predicting the solar power output based on weather parameters which are prominent during such seasons. The motivation for this work comes from the fact that successful prosumer management necessitates accurate predictions of solar outputs, especially during winter due to effects of snow soiling of the PV cells and lower air temperature on the output. Moreover, we imply that a neural network model trained on data collected for a particular season will be more accurate than a model trained on data collected all year-round, due to significant differences in data distribution between different year seasons. To confirm this implication, we create several groups of such models and compare their prediction accuracy to a more complex model from our previous work which was trained on an all-year round dataset. The results show that the mean-absolute error (MAE) of the most accurate winter-specialized model is 76.85 kW, which is around 5% lower than the MAE of the full dataset model (80.08 kW). This accuracy improvement sets path for further research in creating segmented neural network models for predicting power output under different weather conditions.

Anticipating water levels in vast riverbeds is crucial for preventing and mitigating floods or droughts, assessing power plant capacity, and facilitating navigation management. This study introduces an innovative water level prediction model utilizing an Extreme Learning Machine developed to solve the issues of low performance of existing forecasting methods. Development of such a system is of extreme importance when talking about the largest European river – the Danube River. Experimental findings reveal the model's satisfactory performance across various accuracy metrics, complexity considerations, and calculation speed. The prediction with the highest error rate based on MAPE criteria was for Prahovo water level prediction over a 365-day period at 2.02%, whilst the most accurate predictions were for Novi Sad and Banatska Palanka over 30 days and 180 days horizons, respectively, at 0.0550%. The highest coefficient of determination (R2) was achieved with the Novi Sad data at 0.9968, whilst the lowest was observed with the Prahovo data at 0.7353. The ELM model achieved high precision by adjusting the activation functions of the hidden layer neurons, which involved using different combinations of sigmoid and radial-basis activation functions.