Researches On Distributed Coordinated Optimal Decision Methods For Energy Internet

As the deep mix of advanced energy and information technologies,the Energy Internet(EI)is regarded as an effective way to solve the energy crisis and environ-mental pollution in the international energy field.It is also the most likely energy supply/demand fashion for the future human society.The distributed method fo-cuses on making the complex and large-scale optimization problems be divided into the individual node to achieve distributed computation,which is very suitable for the development of EI.Thus,the distributed method is seen as an indispensable and basis technology for EI.Many research problems in EI,especially the cooperative energy management problems,can be modeled as distributed optimization problem-s,which can be solved by employing distributed algorithms.Therefore,the design of distributed approaches is of great significance in both theory and practice for the development of EI.In the context of EI,this thesis researches distributed theory to solve the unconstrained and constrained distributed optimization problems,and the distributed energy management problems for EI under different system models in final.Firstly,the issue of unconstrained distributed optimization problem over networked multi-agent systems is investigated;meanwhile,several distributed algo-rithms with faster convergence rate are proposed to address this issue.Secondly,the thesis investigates and presents a distributed energy management strategy to solve the issue of combined heat and power energy management with quadratic for-m cost function and many constraint limits.On the basis,the energy management framework of EI and its mathematical models axe further investigated,which is formulated as a kind of distributed coupled optimization problem with global equal-ity constraints,local inequality constraints and generalized convex cost function.To address this issue,the corresponding distributed energy management strategies are proposed within the context of basic model and improved model.The main contributions and salient features are summarised as follows:(1)For the issue of distributed convex optimization in multi-agent networks,a distributed indirect continuous-time Newton-Raphson algorithm is proposed,which features reduced communication,fast convergence and distributed exe-cution.To avoid continuous control signal updating,an asynchronous event-triggered scheme is proposed for each agent.As a consequence,the continuous-time Newton-Raphson can be implemented with discrete-time control law.Under an undirected and connected graph,it is proved that each agent can exponentially converge to the global optimal point and the proposed event-triggered scheme is naturally free of Zeno behavior.(2)In order to overcome the shortcomings of the indirect continuous-time Newton-Raphson algorithm in the aspects of information utilization,convergence rate estimation and communication topology,an event-triggered based distributed direct Newton-Raphson is further proposed.Under strongly connected and weight-balanced directed graphs,it is proved that the proposed algorithm can convergence to the global optimal point;meanwhile,we give the estimation of exponential convergence rate in theory.In addition,by designing event-triggered broadcasting strategy,the proposed algorithm can be executed with discrete-time communication when necessary only.As a result,the communi-cation expenditure can be greatly reduced.(3)For the issue of the combined heat and power energy management,a consensus-based distributed alternating iterative optimization method is proposed to address this issue.This method is built upon two modified consensus protocols and makes use of the alternating iterative fashion to handle the heat-electrical coupling problem existed in the objective function and the feasible operating regions,which can solve the combined heat and power energy management problem in a fully distributed fashion.In addition,by making use of eigenvalue perturbation approach,it is proved that each variable can converge to its corresponding optimal solution.(4)For the issue of the energy management of EI,a novel energy management framework is presented for the first time to describe the features of multi-coupling of different energy forms,diversified energy roles and peer-to-peer energy supply/demand,etc.The energy body is defined and formulated for the system model development.Further,a novel distributed-consensus al-ternating direction method of multipliers algorithm is presented to solve the energy management problem of EI in a fully distributed manner.The theo-retical analysis shows that,by using the proposed algorithm,both the optimal energy market clearing price and the optimal energy outputs/demands can be obtained through local communication and computation.(5)In order to improve the model of EI and overcome the shortcoming of the distributed-consensus alternating direction method of multipliers algorithm in the aspect of information utilization,this thesis employs the Hub structure to improve the energy body model and proposes an asynchronous event-triggered based distributed algorithm to solve the issues of day-ahead and real-time co-operative energy management for EI.The improved model can reflect the internal conversion among different energy flows in a better way and also take the different timescale characteristics between electricity and heat power into consideration.Meanwhile,an event-triggered communication strategy is em-bedded into the execution of the distributed algorithm,which can dramatically reduce the communication among energy bodies.The convergence and opti-mization of the proposed algorithm can be verified by the theoretical analysis.