The smart grid (SG) has emerged as an important form of the Internet of Things. Despite the high promises of renewable energy in the SG, it brings about great challenges to the existing power grid due to its nature of intermittent and uncontrollable generation. In order to fully harvest its potential, accurate forecasting of renewable power generation is indispensable for effective power management. In this paper, we propose a least absolute shrinkage and selection operator (LASSO)-based forecasting model and algorithm for solar power generation forecasting. We compare the proposed scheme with two representative schemes with three real world datasets. We find that the LASSO-based algorithm achieves a considerably higher accuracy comparing to the existing methods, using fewer training data, and being robust to anomaly data points in the training data, and its variable selection capability also offers a convenient tradeoff between complexity and accuracy, which all make the proposed LASSO-based approach a highly competitive solution to forecasting of solar power generation.

Energy management is indispensable in the smart grid, which integrates more renewable energy resources, such as solar and wind. Because of the intermittent power generation from these resources, precise power forecasting has become very crucial to achieve efficient energy management. In this paper, we propose a novel adaptive learning hybrid model (ALHM) for precise solar intensity forecasting based on meteorological data. We first present a time-varying multiple linear model (TMLM) to capture the linear and dynamic property of the data, We then construct simultaneous confidence bands (SCB) for variable selection. Next we apply the genetic algorithm back propagation neural network (GABP) to learn the nonlinear relationships in the data. We further propose ALHM by integrating TMLM, GABP and the adaptive learning online hybrid algorithm (ALOHA). The proposed ALHM captures the linear, temporal and nonlinear relationships in the data, and keeps improving the predicting performance adaptively online as more data collected. Simulation results show that ALHM outperforms several benchmarks in both short-term and long-term solar intensity forecasting.

In this paper, an ensemble learning approach is proposed for load forecasting in urban power systems. The proposed framework consists of two levels of learners that integrate clustering, Long Short-Term Memory (LSTM), and a Fully Connected Cascade (FCC) neural network. Historical load data is first partitioned by a clustering algorithm to train multiple LSTM models in the level-one learner, and then the FCC model in the second level is used to fuse the multiple level-one models. A modified Levenberg-Marquardt (LM) algorithm is used to train the FCC model for fast and stable convergence. The proposed framework is tested with two public datasets for short-term and mid-term forecasting at the system, zone and client levels. The evaluation using real-world datasets demonstrates the superior performance of the proposed model over several state-of-the-art schemes. For the ISO-NE Dataset for Years 2010 and 2011, an average reduction in mean absolute percentage error (MAPE) of 10.17% and 11.67% are achieved over the four baseline schemes, respectively.

Quaternary pumped storage hydropower (Q-PSH) technology, as one of the new advanced-PSH technology, has been developed by taking advantage of Conventional-PSH (C-PSH) and Adjustable Speed-PSH (AS-PSH). By combining adjustable-speed pump unit and conventional hydropower turbine unit in the quaternary configuration, Q-PSH has the more competitive capability of providing fast power support in the future high renewable penetrated power system. Acting as energy storage (ES), Q-PSH provides promising power supply to deal with the uncertainty and variability from renewable energy generation. This paper focuses on the dynamic modeling of Q-PSH technology employing full-converter machine and the impact of Q-PSH on the frequency response in a system. The detailed model of Q-PSH is developed and implemented in the IEEE 14-bus system based on GE Positive Sequence Load Flow (PSLF) platform, which captures the dynamic of multiple operation modes, especially hydraulic short-circuit (HSC) operation mode. Several cases are set up to reveal the advantages of Q-PSH technology when power electronic based renewable energy generation is deployed in the system. Sensitivity studies of the controller in pump governor show the impact of parameters in pump response performance. The comparison case illustrates the impact of frequency response provided by the Q-PSH in the system.

Nonintrusive load monitoring (NILM) is to obtain individual appliance’s electricity consumption from aggregated smart meter data. In this article, we propose a middle window transformer model, termed Midformer , for NILM. Existing models are limited by high computational complexity, dependency on data, and poor transferability. In Midformer, we first exploit patchwise embedding to shorten the input length, and then reduce the size of queries in the attention layer by only using global attention on a few selected input locations at the center of the window to capture the global context. The cyclically shifted window technique is used to preserve connection across patches. We also follow the pretraining and fine-tuning paradigm to relieve the dependency on data, reduce the computation in modeling training, and enhance transferability of the model to unknown tasks and domains. Our experimental study using two real-world data sets demonstrates the superior performance and transferability of Midformer over three baseline models.

Non-intrusive load monitoring (NILM) is to estimate individual appliance’s power consumption from aggregated smart meter data, which is useful for optimized energy management and provisioning of customized services. While deep learning (DL) has achieved state-of-the-art NILM performance, it is still constrained by the dependency on large amounts of data and intensive computations on training. In this paper, we propose a pre-training approach to address the generalization of DL models for NILM. We develop a meta-learning based approach and an ensemble learning based approach, which pre-train a base model and then fine-tune it with few-short learning when applied to an unknown dataset. The models are validated with two real-world datasets and shown to achieve a superior transferability performance compared with traditional DL and transfer learning methods.