| In the context of energy production and consumption transformation.China is accelerating construction of large clean energy bases in the wind-rich "Three Norths"region.In order to promote the transformation of clean and low-carbon energy and to address the needs of large-scale wind power consumption and clean heat supply in Three Norths region,a general finite difference domain-hybrid semi-mechanism(FDD-HSM)dynamic modelling and optimal control method is proposed.A refined control-oriented dynamic modelling and simulation method and a distributed model predictive control(DMPC)method are developed and applied to a class of continuous process network systems represented by complex electro-thermal system to achieve the synergistic use of electricity and heat and promote large-scale wind power consumption.The main research elements of this paper are as follows:The introduction of wind power for clean heating on the heating side of CHP units constitutes a complex electro-thermal system with dual stochastic energy flows of electricity and heat.The FDD-HSM dynamic modeling and optimization control method is proposed,which verifies actual operation data of complex electro-thermal system,and realizes the collaborative regulation of complex electric heating system based on hierarchical integrated optimization architecture.Firstly,a generic FDD-HSM modelling approach is proposed for continuous process network systems represented by district heating networks(DHN)to achieve an accurate portrayal of their dynamic performance.As anomalous data usually exist in the system operation,a "random sampling consistent+polynomial least squares fitting" method is proposed to identify and reject anomalous data and extract steady-state operating condition data.Based on this,system identification method is used to determine the delay order and establish a finite difference vector.Kmeans++clustering is used and working domain application conditions are estimated to achieve a tight convex partitioning of finite difference space to form multiple finite difference working domains.In each domain,the parameters of mechanism model are identified and HSM model is obtained by training deviation dynamic compensation model through a neural network algorithm.The performance of the modelling method is verified by DHN real data.Secondly,based on the proposed FDD-HSM modelling approach,a multi-level node dynamic transfer system integration modelling idea is proposed.The key components of the system are divided into multiple nodes.After applying FDD-HSM modelling method to each node and establishing mechanism model of each node,neural network algorithm is applied to establish the deviation compensation model to form FDD-HSM model.The dynamic transfer relationship between the input and output of each node is considered to form an integrated system model architecture with infinite approximation capability.In addition,a finite difference-autoregressive model and a finite difference-long and shortterm memory neural network model are developed to compare with them,and the accuracy and performance of the proposed method are verified based on DHN real data.Finally,a hierarchical integrated intra-day optimal control architecture is proposed,consisting of dynamic rolling optimization of thermal power in the upper layer,taking into account heating economy and flexibility,and DMPC-based temperature control in the lower layer.The DMPC-based control strategy splits the system control problem into multiple subsystem control problems,where optimization output of individual subsystems is governed by estimated output of the state of other subsystems,and optimal output is solved by satisfying the overall objective function.DMPC achieves coordinated control on shorter time scales while satisfying the need for real-time online optimization.The simulations show that FDD-HSM modelling and DMPC control method has important application value for the improvement of regional large-scale electro-thermal co-utilization and promotion of wind power consumption. |
PDF Full Text Request