基于自适应生成式对抗神经网络的综合能源优化主体调度策略  

Principal Scheduling Strategy for Comprehensive Energy Optimization Based on the Self-adaptive Generative Adversarial Neural Network

作　　者：刘洪广 崔双喜[1] 宋江涛 樊小朝 胡衡 Liu Hongguang;Cui Shuangxi;Song Jiangtao;Fan Xiaochao;Hu Heng(School of Electrical Engineering,Xinjiang University,Urumqi Xinjiang 830017,China;School of Energy Engineering,Xinjiang University of Engineering,Urumqi Xinjiang 830023,China)

机构地区：[1]新疆大学电气工程学院,新疆乌鲁木齐830017 [2]新疆工程学院能源工程学院,新疆乌鲁木齐830023

出　　处：《电气自动化》2025年第1期105-108,共4页Electrical Automation

基　　金：国家自然科学基金项目(52266018);新疆工程学院博士启动金项目(2023XGYBQJ01)。

摘　　要：针对传统调度策略在进行综合能源调度过程中存在的调度效率较低、异常调度数据识别准确率不高等问题,设计了一种基于注意力机制和反向传播(back propagation,BP)神经网络相结合的自适应生成式对抗神经网络调度策略。根据综合能源系统结构,构建以BP神经网络为主体算法的调度优化模型,分别对神经网络中的输入层、隐藏层和输出层进行数据优化模拟,提升综合能源系统的调度效率;同时引入注意力机制技术,与BP神经网络构成自适应生成式对抗神经网络,利用注意力机制技术对BP神经网络中的数据处理环节进行特征提取与特征分析,提升能源系统对调度过程中异常数据的识别能力。最后,在MATLAB中模拟综合能源系统结构,并加入能源数据进行调度模拟试验。试验结果表明,自适应生成式对抗神经网络提升了综合能源调度过程中的调度效率与异常调度数据识别准确率。A self-adaptive generative adversarial neural network scheduling strategy based on attention mechanism and back propagation(BP)neural network was designed to address the problems of low scheduling efficiency and low accuracy in identifying abnormal scheduling data in traditional scheduling strategies during comprehensive energy scheduling.Based on the structure of the integrated energy system,a scheduling optimization model with BP neural network as the principal algorithm was constructed.Data optimization simulations were conducted on the input layer,hidden layer,and output layer of the neural network to improve the scheduling efficiency of the integrated energy system;at the same time,the attention mechanism technology was introduced to form an self-adaptive generative adversarial neural network with BP neural network.The attention mechanism technology was used to extract and analyze features in the data processing stage of BP neural network,improving the energy system’s ability to identify abnormal data in the scheduling process.Finally,the comprehensive energy system structure was simulated in MATLAB and the energy data was added for scheduling experiment simulation.The experimental results show that the self-adaptive generative adversarial neural network improves the scheduling efficiency and accuracy of identifying abnormal scheduling data in the comprehensive energy scheduling process.

关 键 词：综合能源 调度主体 BP神经网络 注意力机制 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]