Short-term Memory Network Algorithm Research Articles

Hollow village detection method based on non-intrusive power load monitoring

With the development of urbanization, the population and resources converge to the city, leading to different degrees of hollow village phenomenon in rural areas. Therefore, a hollow village detection model based on non-invasive load monitoring method is proposed in this paper. Firstly, the power load data of users is collected in rural areas, and the data is processed and analyzed. A data analysis algorithm based on the improved Attention mechanism is constructed, and the existing public data sets are used for verification and improvement. Six kinds of frequently used electrical appliances among rural users were selected as typical electrical appliances, the power load curves of selected typical electrical appliances were obtained, and the long and short term memory network algorithm was used to identify the types of household electrical appliances of rural users. After algorithm recognition and analysis, the usage characteristics of typical rural electric appliances are obtained, and it is taken as the electricity consumption behavior habits of rural residents. According to their electricity consumption habits, rural residents are divided into four categories: young people living alone, the elderly, two adults and a child, and vacant houses. According to the four groups of people, the index of hollow village is determined. By judging the corresponding household population of four groups of rural residents, the selected village is judged whether it is hollow village or whether there is a tendency of hollow village.

Research on air quality prediction based on improved long short-term memory network algorithm.

Air quality is changing due to the influence of industry, agriculture, people's living activities and other factors. Traditional machine learning methods generally do not consider the time series of the data itself and cannot handle long-range dependencies, thus ignoring information relevant to the predicted items and affecting the accuracy of air quality predictions. Therefore, an attention mechanism is introduced based on the long short term memory network model (LSTM), which attenuates unimportant information by controlling the proportion of the weight distribution. Finally, an integrated lightGBM+LSTM-attention model was constructed based on the light gradient boosting machine (lightGBM), and the prediction results were compared with those of 11 models. The experimental results show that the integrated model constructed in this article performs better, with the coefficient of determination (R2) of prediction accuracy reaching 0.969 and the root mean square error (RMSE) improving by 5.09, 4.94, 4.85 and 4.0 respectively compared to other models, verifying the superiority of the model.

Research on Carbon Emissions Prediction Model of Thermal Power Plant Based on SSA-LSTM Algorithm with Boiler Feed Water Influencing Factors

China’s power industry is a major energy consumer, with the carbon dioxide (CO2) generated by coal consumption making the power industry one of the key emission sectors. Therefore, it is crucial to explore energy conservation and emissions reduction strategies suitable for China’s current situation. Taking a typical cogeneration enterprise in North China as an example, this paper aims to establish a generalized regression prediction model for carbon emissions of coal-fired power plants, which will provide a reference for China to seek strategies for carbon peaking and carbon neutralization in the future. Firstly, in terms of the selection of influencing factors, this paper uses objective index screening methods, simulation means, and the eXtreme Gradient Boosting algorithm (XG-Boost) to analyze the feature importance of various influencing factors. It is concluded that the relevant influencing factors of the boiler feed water system have a strong correlation and characteristic importance with the carbon emissions results of coal-fired power plants. Therefore, this paper proposes to introduce these factors into the regression prediction model as auxiliary variables to more scientifically reflect the carbon emissions results of coal-fired power plants. Secondly, in the aspect of regression prediction model establishment, inspired by the sparrow’s foraging behavior and anti-predation behavior, this paper selects the sparrow search algorithm (SSA) with strong optimization ability and fast convergence speed to optimize the super parameters of the long short-term memory network algorithm (LSTM). It is proposed to use the SSA-LSTM algorithm to establish the carbon emissions regression prediction model of coal-fired power plants. The advantage of the SSA-LSTM algorithm is that it can effectively simplify the super parameter selection process of the LSTM algorithm, effectively solve the global optimization problem, prevent the model from falling into overfitting and local optimization, and make the carbon emissions regression prediction model of coal-fired power plants achieve a better fitting effect. By comparing the performance indicators of the model before and after the improvement, it is found that the regression prediction effect of the SSA-LSTM coal-fired power plant carbon emissions regression prediction model, which introduces boiler feed water influencing factors, has been effectively improved. Therefore, the model proposed in this paper can be used to conduct a comprehensive impact factor analysis and regression prediction analysis on the carbon emissions intensity of China’s coal-fired power plants, formulate targeted carbon emissions reduction countermeasures, and provide a theoretical basis for energy conservation and emissions reduction of China’s coal-fired power plants.

Short-term prediction of the power of a new wind turbine based on IAO-LSTM

Short-term wind power forecasting is of great significance to the real-time dispatching of power systems, but the short-term forecasting accuracy of wind power is not high. To this end, this paper proposes a hybrid prediction model that combines the Isolated Forest algorithm, the Synchronous Squeeze Wavelet Transform (SWT) method, the Aquila Optimizer (AO) and the Long Short-term Memory network (LSTM). Firstly, the Isolated Forest algorithm is used to detect abnormal data. Secondly, the SWT method is used to denoise the original power signal of the new wind turbine. Then, the wind power prediction model is established through the long short-term memory network algorithm. The OA is used to optimize the LSTM structure parameters to solve the influence of random parameters on the prediction accuracy. Finally, perform example verification. The results show that the proposed model is effective in power prediction of new wind turbine.