精确高效的多元负荷短期预测对于综合能源系统的运行控制与调度具有重要意义.为了改善负荷预测效果,提出一种量子加权遗忘门与输入门结合的长短期记忆(Quantum weighted coupled input and forget gate long short-term memory,QWCIFGLSTM)神经网络模型.在模型结构方面,将长短期记忆(Long short-term memory,LSTM)神经网络中的遗忘门和输入门结合起来,形成遗忘门与输入门结合长短期记忆(Coupled input and forget gate long short-term memory,CIFGLSTM)神经网络,从而减少了网络参数,优化了网络结构;在模型构成方面,采用量子加权神经元替代传统神经元,构建了QWCIFGLSTM神经网络预测模型.量子加权神经元具有较强的数据处理能力和并行计算能力,可以有效提高负荷预测的精度.通过算例仿真验证,所构建的模型相较于基于反向传播(Back propagation,BP)的神经网络预测模型、传统LSTM神经网络预测模型和遗忘门与输入门结合的长短期记忆神经网络预测模型,具有更好的预测效果.
Research on Multi-load Short-term Forecasting Model of Integrated Energy System Based on QWCIFGLSTM
Accurate and efficient multi-load short-term forecasting is of great significance for the operation control and scheduling of integrated energy system.In order to improve the effect of load forecasting,a quantum weighted coupled input and forget gate long short-term memory(QWCIFGLSTM)neural network model is proposed.In terms of model structure,the forget gate and input gate in the long short-term memory(LSTM)neural network are combined to form a coupled input and forget gate long short-term memory(CIFGLSTM)neural network,so as to reduce the network parameters and optimize the network structure.In terms of model composition,quantum weighted neurons are used to replace traditional neurons to build QWCIFGLSTM neural network forecasting model.Quantum weighted neurons have good data processing ability and parallel computing ability,which can effectively improve the accuracy of load forecasting.The simulation results show that the proposed model has better forecasting effect than back propagation(BP)neural network forecasting model,conventional LSTM neural network forecasting model and the coupled input and forget gate long short-term memory neural network forecasting model.
Integrated energy systemmulti-load forecastinglong short-term memory neural networkquantum weighted neuron
NETL
NSTL
万方数据
