面向高效清洁供热的复杂电热系统综合建模与协同调控研究

After bringing in clean heating by wind power at heating side of combined heat and power (CHP) unit, the complex heating system is formed including stochastic wind power and user heat load, whose collaborative utilization level to electricity and heat need to be raised urgently. Facing efficient and clean heating, this project is devoted to studying the comprehensive modeling and coordinated regulation of the complex electricity-heat system. The research contents are as following. The data preprocessing method based on joint probability space of high-dimensional variables is firstly studied, providing the data basis. Then, the linear state space modeling method of complex network system based on dynamics transfer of multi-stage nodes is studied. The nonlinear dynamic of each node is fully approached via the piecewise affine (PWA) model and hybrid semi-parametric model. Finally, on the basis of high-precision stepwise interval prediction of stochastic variable, the optimization control based on the hierarchical-layers integrated optimization structure is studied. Considering the heating economy and flexibility, description and solution of the receding optimization problem is studied for the upper layer. Output setting values of the upper layer can coordinate the distributed control of the lower layer. Based on linear PWA state space model, the distributed multi-objective robust model predictive control is studied for the lower layer, exploring sufficient conditions for closed-loop stability and iterative feasibility. In summary, implementation of this project not only provides theoretical support for the modeling and control of complex electricity-heat system, but also has important value to raise collaborative utilization level of electricity-heat for heating and to promote wind power absorption.

在热电联产(CHP)机组供热侧引入风电清洁供暖，构成含风电、用户热负荷双随机性的复杂供热系统，其电热协同利用水平亟待增强。本项目面向高效清洁供热，致力于复杂电热系统的综合建模与协同调控研究。研究内容包括：研究基于高维变量联合概率空间的数据预处理方法，奠定数据基础；研究基于多级节点动态传递的复杂网络系统线性状态空间建模方法，单一节点非线性动态采用分段仿射(PWA)模型和混合半参数模型充分逼近；研究随机变量的高精度分步区间预测，基于此，研究基于分层集成优化架构的复杂电热系统优化控制，上层研究考虑供热经济性、灵活性的滚动优化问题描述及求解，通过输出设定值协调下层分布式控制，下层研究基于线性PWA状态空间模型的分布式多目标鲁棒模型预测控制，探索闭环稳定性和迭代可行性充分条件。本项目的实施不仅将为复杂电热系统建模与控制提供理论支撑，而且对提升电热协同利用水平及促进风电消纳具有重要价值。

近年来，中国深入推进清洁、低碳能源转型，我国正加快风光资源丰富的“三北”地区大型清洁能源基地建设。为解决其大规模新能源电力消纳难题，本研究提出了区域电热协同利用解决方案。在热电联产(CHP)机组供热侧引入风电清洁供暖，构成含风电、用户热负荷双随机性的复杂电热系统，其电热协同利用水平亟待增强。本项目面向高效清洁供热，致力于复杂电热系统的综合建模与协同调控研究。首先，针对风电连续变化运行数据治理，提出了基于“运行机理识别+高维变量高斯混合聚类识别+Copula空间多向四分位优化识别”的最大/限功率模式下全工况异常数据优化识别方法、基于“相似工况序列动态筛选+前后向时序动态建模+多重填补”的大量连续缺失数据填补方法；针对供热系统离散稳态工况运行数据，提出了基于“随机抽样一致+多项式最小二乘拟合”的异常数据识别方法、基于“多元特征提取+相似工况序列动态筛选+前后向时序动态建模+多重填补”的大量连续缺失数据填补方法；基于实测数据验证了所提方法的有效性，为后续研究奠定了数据基础。接着，提出复杂电热系统模块化建模方案，发明单节点“有限差分域-混合半机理”建模及多级节点动态传递的电热系统大范围工况动态建模方法，搭建区域供热系统实验台验证了所提方法的有效性，为后续研究奠定了模型基础。然后，提出了基于“多元特征风机选择及滚动预测+多元风况-场级功率有限差分动态建模+条件概率建模”的风电场功率超短期滚动区间预测方法、基于“特征变量选择及预测+多元特征输入-区域负荷有限差分动态建模+条件概率建模”的区域热负荷滚动区间预测方法。在此基础上，提出了复杂电热系统的日内分层集成优化调控架构，包含计及供热经济性和灵活性的上层热功率动态滚动优化调度、基于分布式模型预测控制的下层温度优化控制，通过仿真验证了方法的有效性。本项目研究成果不仅将为复杂电热系统建模与控制提供理论支撑，而且对提升电热协同利用水平及促进风电消纳具有重要价值。