Accurate solar generation forecasting is critical for ensuring power system reliability, economics, and effectiveness and controlling the supply-demand balance. This research offers novel multi-branch spatio-temporal forecasting models to improve forecasting accuracy and minimize forecasting errors. The first step is to build temporal models employing advanced deep learning architectures, such as Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and GRU with Attention (AttGRU). Next, spatio-temporal solar forecasting models are constructed. A novel multi-branch Attentive Gated Recurrent Residual network (ResAttGRU) consisting of multiple branches of residual networks (ResNet), GRU, and the attention mechanism is introduced. The proposed multi-branch ResAttGRU is capable of modeling data at various resolutions, extracting hierarchical features, and capturing short- and long-term dependencies. Moreover, this network also presents a strong multi-time-scale representative, while GRUs can exploit temporal information at less computational cost than the popular LSTM. The novelty of the developed architecture is in the utilization of multiple convolutional-based branches to learn multi-time-scale features jointly, accelerate the learning process, and reduce overfitting. This dissertation also compares the multi-branch ResAttGRU networks with state-of-the-art deep learning methods using 18 years of NSRDB data at 12 solar sites. The proposed multi-branch ResAttGRU requires 7.1% fewer parameters than multi-branch residual LSTM (ResLSTM) while achieving similar average RMSE, MAE, and R-squared values. Finally, to effectively model spatial correlation among neighboring solar sites as well as to alleviate performance degradation due to overfitting of conventional neural networks, a spatio-temporal framework comprised of concatenated multi-branch Residual network and Transformer (ResTrans) is developed. Numerical results indicate that the multi-branch ResTrans structure achieves the highest forecasting accuracy, with an average RMSE of 0.049 ( W/m^2 ), an average MAE of 0.031 (W/m^2 ), and a R^2 coefficient of 97%.

Accurate solar irradiance forecasting is essential for minimizing operational costs of solar photovoltaic (PV) generation as it is commonly used to predict the power output. This thesis presents and compares three different machine learning approaches of solar irradiance forecasting: Random Forest (RF), Feedforward Neural Networks (FNNs) and Long Short-Term Memory (LSTM) networks. Each model was tested on two different forecasts: the next hour average and the hourly day-ahead averages. The machine learning algorithms were trained and tested on data from a weather station located at Tampere University (TAU) in Tampere, Finland. Data were preprocessed before training the algorithms and the relevant features were selected. Moreover, Grid Search and Random Search techniques were used along with multiple train and validation splits to find the optimal hyperparameters for each machine learning algorithm. Persistence model is set as a baseline model for comparison while RMSE and MAE are used to quantify the prediction error. For the next hour forecast, LSTM achieved the highest accuracy in terms of RMSE (76.14 W/m2 ), 2.1% and 1.1% better than RF and FNN respectively. Instead, FNN generally produced the best results in the day-ahead forecast. In all models, the prediction error increases as the forecast horizon increases until it stabilizes at 10 hours approximately. Further, the error keeps increasing but slower. Besides, the next hour forecast models were able to predict considerably better the next hour solar irradiance than the day-ahead forecast models.

Solar Energy Forecasting and Resource Assessment is a vital text for solar energy professionals, addressing a critical gap in the core literature of the field. As major barriers to solar energy implementation, such as materials cost and low conversion efficiency, continue to fall, issues of intermittency and reliability have come to the fore. Scrutiny from solar project developers and their financiers on the accuracy of long-term resource projections and grid operators' concerns about variable short-term power generation have made the field of solar forecasting and resource assessment pivotally important. This volume provides an authoritative voice on the topic, incorporating contributions from an internationally recognized group of top authors from both industry and academia, focused on providing information from underlying scientific fundamentals to practical applications and emphasizing the latest technological developments driving this discipline forward. - The only reference dedicated to forecasting and assessing solar resources enables a complete understanding of the state of the art from the world's most renowned experts. - Demonstrates how to derive reliable data on solar resource availability and variability at specific locations to support accurate prediction of solar plant performance and attendant financial analysis. - Provides cutting-edge information on recent advances in solar forecasting through monitoring, satellite and ground remote sensing, and numerical weather prediction.

Solar generation is an increasing part of the energy portfolio in the United States. An accurate forecast of the available solar resource and power is essential to managing the electric grid, market operations, and reducing the cost of solar energy. High-frequency forecasts of solar radiation in intra-hour horizons is important for real-time electric power system energy management, especially at the distribution level. Conventional Numerical Weather Prediction models perform poorly in intra-hour, high-frequency forecasts because of the limits on real-time computing, spatial resolution, and infrequent availability of observations. Although a number of alternative technologies, e.g., time-series analysis and machine learning, have been used to fill this gap, the smart persistence model is among the top-performing models in short-term forecasting and therefore often serves as the baseline to evaluate other forecasting models. Although the smart persistence model often serves as the baseline model in these intra-hour forecasts, obvious uncertainties exist in the current smart persistence model: (1) clear-sky index does not respond to the variation of the solar incident angle when cloud conditions are persistent within the forecast horizon, and (2) cloud coverage is inherently persistent though it is constrained by cloud advection. In this study, we developed a Physics-Based Smart Persistence Model for Intra-Hour Solar Forecasting (PSPI) that integrates cloudy property estimation, a radiative transfer model, and cloud fraction forecasts to improve the performance of the smart persistence model. Compared to the smart persistence model, PSPI does not require additional observations of various atmospheric parameters past global horizontal irradiance, but it is customizable because additional observations, if available, can be ingested to further improve the forecast. Our results show that the PSPI outperforms the persistence and smart persistence model on 5-minute, 15-minute, and 30-minute forecast horizons. The software package of PSPI is flexible to users' needs and provides low computational time to run at site-specific locations across the continental United States.

This book presents an overview of modeling and simulation of environmental systems via diverse research problems and pertinent case studies. It is divided into four parts covering sustainable water resources modeling, air pollution modeling, Internet of Things (IoT) based applications in environmental systems, and future algorithms and conceptual frameworks in environmental systems. Each of the chapters demonstrate how the models, indicators, and ecological processes could be applied directly in the environmental sub-disciplines. It includes range of concepts and case studies focusing on a holistic management approach at the global level for environmental practitioners. Features: Covers computational approaches as applied to problems of air and water pollution domain. Delivers generic methods of modeling with spatio-temporal analyses using soft computation and programming paradigms. Includes theoretical aspects of environmental processes with their complexity and programmable mathematical approaches. Adopts a realistic approach involving formulas, algorithms, and techniques to establish mathematical models/computations. Provides a pathway for real-time implementation of complex modeling problem formulations including case studies. This book is aimed at researchers, professionals and graduate students in Environmental Engineering, Computational Engineering/Computer Science, Modeling/Simulation, Environmental Management, Environmental Modeling and Operations Research.

This book discusses control and optimization techniques in the broadest sense, covering new theoretical results and the applications of newly developed methods for PV systems. Going beyond classical control techniques, it promotes the use of more efficient control and optimization strategies based on linearized models and purely continuous (or discrete) models. These new strategies not only enhance the performance of the PV systems, but also decrease the cost per kilowatt-hour generated.