基于小波网模型的区域供热系统负荷预测

中国的供热系统运行调节中普遍存在由于系统本身的热惯性等因素导致供热量不能实时跟随需求量调整的问题。将小波网络引入热力过程优化运行,通过综合分析系统运行数据间关系,识别供热系统复杂非线性动态特性,建立区域供热系统热负荷小波预测模型。仿真实验表明,该方法在网络规模和预测精度上优于常规多层前馈神经网络。实际运行预测结果表明,该方法辨识精度高,实时性强。

Load predictions for district heating systems based on a WNN model

The energy provided by heating plants in China does not match the energy needed for the residents due to reasons such as the thermal inertia of the heating system. Wavelet neural networks (WNNs) were used to optimize operation of thermal energy plants. A wavelet predictive model was designed based on operating data of a district heating system and recognition of heating system's nonlinear dynamic characteristics. Simulation results show that both the structure and generalization ability of WNN are superior to that of the multilayer feedforward neural network. The predictions for the field operation show that WNN method is precise and feasible.