Knowledge tracing (KT) defines the task of predicting whether students can correctly answer questions based on their historical response. Although much research has been devoted to exploiting the question information, plentiful advanced information among questions and skills hasn't been well extracted, making it challenging for previous work to perform adequately. In this paper, we demonstrate that large gains on KT can be realized by pre-training embeddings for each question on abundant side information, followed by training deep KT models on the obtained embeddings. To be specific, the side information includes question difficulty and three kinds of relations contained in a bipartite graph between questions and skills. To pre-train the question embeddings, we propose to use product-based neural networks to recover the side information. As a result, adopting the pre-trained embeddings in existing deep KT models significantly outperforms state-of-the-art baselines on three common KT datasets.
Abstract:In order to maintain the stability and security of the power system, the uncertainty and intermittency of wind power must be taken into account in economic dispatch (ED) problems. In this paper, a dynamic economic dispatch (DED) model based on chance constrained programming is presented and an improved particle swarm optimization (PSO) approach is proposed to solve the problem. Wind power is regarded as a random variable and is included in the chance constraint. New formulation of up and down spinning reserve constraints are presented under expectation meaning. The improved PSO algorithm combines a feasible region adjustment strategy with a hill climbing search operation based on the basic PSO. Simulations are performed under three distinct test systems with different generators. Results show that both the proposed DED model and the improved PSO approach are effective.
An accurate prediction of wind speed is crucial for the economic and resilient operation of power systems with a high penetration level of wind power. Meteorological information such as temperature, humidity, air pressure, and wind level has a significant influence on wind speed, which makes it difficult to predict wind speed accurately. This paper proposes a wind speed prediction method through an effective combination of principal component analysis (PCA) and long short-term memory (LSTM) network. Firstly, PCA is employed to reduce the dimensions of the original multidimensional meteorological data which affect the wind speed. Further, differential evolution (DE) algorithm is presented to optimize the learning rate, number of hidden layer nodes, and batch size of the LSTM network. Finally, the reduced feature data from PCA and the wind speed data are merged together as an input to the LSTM network for wind speed prediction. In order to show the merits of the proposed method, several prevailing prediction methods, such as Gaussian process regression (GPR), support vector regression (SVR), recurrent neural network (RNN), and other forecasting techniques, are introduced for comparative purposes. Numerical results show that the proposed method performs best in prediction accuracy.
Wind Power Producers (WPPs) seek to maximize profit and minimize the imbalance costs when bidding into the day-ahead market, but uncertainties in the hourly available wind and forecasting errors make the bidding risky. This paper assumes that hourly wind power output given by the forecast follows a normal distribution, and proposes three different bidding strategies, i.e., the expected profit-maximization strategy (EPS), the chance-constrained programming-based strategy (CPS) and the multi-objective bidding strategy (ECPS). Analytical solutions under the three strategies are obtained. Comparisons among the three strategies are conducted on a hypothetical wind farm which follows the Spanish market rules. Results show that bid under the EPS is highly dependent on market clearing price, imbalance prices, and also the mean value and standard deviation of wind forecast, and that bid under the CPS is largely driven by risk parameters and the mean value and standard deviation of the wind forecast. The ECPS combining both EPS and CPS tends to choose a compromise bid. Furthermore, the ECPS can effectively control the tradeoff between expected profit and target profit for WPPs operating in volatile electricity markets.Keywords: wind power; bidding; day-ahead electricity market; risk; chance-constrained programming; multi-objective optimization
Data centres (DCs) have rapidly expanded in recent years due to increasing demand for cloud services. The costs incurred by electricity and water consumption contribute to a major portion of DC operating costs. This article addresses the day-ahead operation of a DC in the context of a virtual power plant (VPP) under the grid-connected mode. The day-ahead operation problem is formulated as a two-stage stochastic mixed integer linear programming (MILP) model that considers workload schedules among server clusters, water consumption, and uncertainties of onsite renewable energy and electricity prices. In addition, the conditional value at risk (CVaR) is utilised to manage the risk caused by various uncertainties that challenge DCs, especially electricity prices. The impacts of the risk level, the water price, and uncertainty on the day-ahead operation of the DC are studied. The numerical test results show that the proposed model can efficiently generate optimal dayahead schedules.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.