Daily electricity consumption forecasting of home appliances can improve the accuracy and efficiency of the operation of home energy management systems. In this paper, an improved bidirectional long short memory network (BILSTM) model for predicting daily electricity consumption of household air conditioning is proposed. Firstly, and the “mutual information” is used to analyze the correlation between the daily electricity consumption of air conditioning and some environmental factors. Second, the environmental factors with strong correlation with the daily electricity consumption of air conditioning are selected as the influence factors, and these influence factors and the electricity data are taken as the characteristic input of the network. Finally, the improved bidirectional LSTM load prediction model which has been trained is used to forecast the daily electricity consumption of air conditioning. The experimental results show that the improved bidirectional LSTM network proposed in this paper can predict the daily electricity consumption of air conditioning in short term, and the maximum relative error of the predicted result is less than 5%.
Synchronization signals are requisite for calibrating electrical measurement devices with digital output when using conventional calibration methods. However, since the signal sampling process of the analogue merging units (MUs) operating in an intelligent substation does not rely on external synchronization signals, accuracy calibration without the use of synchronization signals is of particular importance in order to guarantee the measurement accuracy in practical situations. So far, very little research on calibration systems independent of synchronization signals has been performed. This paper presents a design of the calibration system without dependence on synchronization signals. To verify the feasibility of the proposed design, the designed system and a conventional calibration system have been employed in testing the accuracy of the same analogue MU of a 0.2 accuracy class. The comparison of the test results shows that the differences of ratio errors are below 0.02%, and the maximum difference of phase errors is about 4 ′ . This paper also provides a new efficient and significant calibration method which does not require any external synchronization signals.
Mining of valuable information from user energy consumption data efficiently and accurately has always been a research hotspot in the power industry. Bayesian classification method is one of the important data processing methods in the field of machine learning and data mining research. It has the advantages of simplicity, high efficiency and stable classification effect, and it provides an effective solution to the user’s comprehensive energy consumption feature identification. A model describing energy consumption data to predict trends is built, training data set is analyzed, a classification model is constructed, and the data records in the database to a given category are maped, which can be applied to data prediction. By studying incomplete information systems, expanding rough sets, constructing extended models, Bayesian classification algorithms based on rough sets theory is designed, and user comprehensive energy consumption feature identification is realized. Experiments show that the user comprehensive energy consumption feature identification algorithm based on rough set Bayesian classification can greatly improve the classification accuracy.
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.