Scholarlink

<p>In response to the problem of insufficient multi-source heterogeneous data fusion and uncertainty quantification mechanism in load forecasting of integrated energy systems, a convolutional neural network (CNN) is used to extract local features of load data, and a bidirectional long short-term memory network (BiLSTM) is used to capture bidirectional temporal dependencies. Random forest (RF) is used to process high-dimensional nonlinear relationships, and kernel density estimation (KDE) is used to quantify prediction uncertainty, thus establishing a CNN BiLSTM RF-KDE hybrid model; At the same time, a multi energy flow coupling model of electricity heat gas is constructed to analyze the impact of different carbon price ranges on scheduling strategies. The example analysis shows that the determination coefficients for predicting power and heat loads on the training set are 0.93 and 0.97, respectively; The determination coefficients for predicting power and heat loads on the test set are 0.79 and 0.85, respectively. The predicted power generation and heat generation of each device are highly consistent with the average trend, indicating that the use of this model can obtain a load that is closer to the accurate value. Based on this data, more reliable analysis and scheduling of the integrated energy system can be carried out.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Research on Optimization Methods for Management of Electric Thermal Gas Multi Energy Systems

  • 期刊论文

  • 发表日期：

    2026年3月1日

查看全文
  • Lin Liao,
  • Yong Chen,
  • Mengyang Xu,
  • Jian Dai,
  • Xiangjun Peng

摘要

In response to the problem of insufficient multi-source heterogeneous data fusion and uncertainty quantification mechanism in load forecasting of integrated energy systems, a convolutional neural network (CNN) is used to extract local features of load data, and a bidirectional long short-term memory network (BiLSTM) is used to capture bidirectional temporal dependencies. Random forest (RF) is used to process high-dimensional nonlinear relationships, and kernel density estimation (KDE) is used to quantify prediction uncertainty, thus establishing a CNN BiLSTM RF-KDE hybrid model; At the same time, a multi energy flow coupling model of electricity heat gas is constructed to analyze the impact of different carbon price ranges on scheduling strategies. The example analysis shows that the determination coefficients for predicting power and heat loads on the training set are 0.93 and 0.97, respectively; The determination coefficients for predicting power and heat loads on the test set are 0.79 and 0.85, respectively. The predicted power generation and heat generation of each device are highly consistent with the average trend, indicating that the use of this model can obtain a load that is closer to the accurate value. Based on this data, more reliable analysis and scheduling of the integrated energy system can be carried out.