Short-term solar irradiance forecasting (STSIF) is of great significance for the optimal operation and power predication of grid-connected photovoltaic (PV) plants. However, STSIF is very complex to handle due to the random and nonlinear characteristics of solar irradiance under changeable weather conditions. Artificial Neural Network (ANN) is suitable for STSIF modeling and many research works on this topic are presented, but the conciseness and robustness of the existing models still need to be improved. After discussing the relation between weather variations and irradiance, the characteristics of the statistical feature parameters of irradiance under different weather conditions are figured out. A novel ANN model using statistical feature parameters (ANN-SFP) for STSIF is proposed in this paper. The input vector is reconstructed with several statistical feature parameters of irradiance and ambient temperature. Thus sufficient information can be effectively extracted from relatively few inputs and the model complexity is reduced. The model structure is determined by cross-validation (CV), and the Levenberg-Marquardt algorithm (LMA) is used for the network training. Simulations are carried out to validate and compare the proposed model with the conventional ANN model using historical data series (ANN-HDS), and the results indicated that the forecast accuracy is obviously improved under variable weather conditions.

A Real-Time Price (RTP) based Automatic Demand Response (ADR) strategy for PV-assisted Electric Vehicle (EV) Charging Station (PVCS) without Vehicle to Grid (V2G) is proposed. The charging process is modeled as a Dynamic Linear Program (DLP) instead of the normal day-ahead and real-time regulation strategy, to capture the advantages of both global and real-time optimization. Different from conventional price forecasting algorithms, a Dynamic Price Vector Formation Model (DPVFM) is proposed based on a clustering algorithm to form a RTP vector for a particular day. A Dynamic Feasible Energy Demand Region (DFEDR) model considering grid voltage profiles is designed to calculate the lower and upper bounds. A deduction method is proposed to deal with the unknown information of future intervals, such as the actual stochastic arrival and departure times of EVs, which make the DFEDR model suitable for global optimization. Finally, both the comparative cases articulate the advantages of the developed methods and the validity in reducing electricity costs, mitigating peak charging demand, and improving PV self-consumption of the proposed strategy are verified through simulation scenarios.

The optimized operation of building energy management system (BEMS) is of great significance to its operation security, economy and efficiency. This paper proposed a day-ahead multi-objective optimization model for a BEMS under time-of-use (TOU) price based demand response (DR), which integrates building integrated photovoltaic (BIPV) with other generations to optimize the economy and occupants’ comfort by the synergetic dispatch of source-load-storage. The occupants' comfort contains three aspects of the indoor environment: visual comfort, thermal comfort, and indoor air quality comfort. With the consideration of controllable load that could participate in DR programs, the balances among different energy styles, electric, thermal and cooling loads are guaranteed during the optimized operation. YALMIP toolbox in MATLAB was applied to solve the optimization problem. Finally, a case study was conducted to validate the effectiveness and adaptability of the proposed model.

Accurate solar photovoltaic (PV) power forecasting is an essential tool for mitigating the negative effects caused by the uncertainty of PV output power in systems with high penetration levels of solar PV generation. Weather classification based modeling is an effective way to increase the accuracy of day-ahead short-term (DAST) solar PV power forecasting because PV output power is strongly dependent on the specific weather conditions in a given time period. However, the accuracy of daily weather classification relies on both the applied classifiers and the training data. This paper aims to reveal how these two factors impact the classification performance and to delineate the relation between classification accuracy and sample dataset scale. Two commonly used classification methods, K-nearest neighbors (KNN) and support vector machines (SVM) are applied to classify the daily local weather types for DAST solar PV power forecasting using the operation data from a grid-connected PV plant in Hohhot, Inner Mongolia, China. We assessed the performance of SVM and KNN approaches, and then investigated the influences of sample scale, the number of categories, and the data distribution in different categories on the daily weather classification results. The simulation results illustrate that SVM performs well with small sample scale, while KNN is more sensitive to the length of the training dataset and can achieve higher accuracy than SVM with sufficient samples.