| In the context of the "carbon peak,carbon neutral" national strategic objectives,the construction of new power system requires a higher level of precision in infrastructure investment.The electricity infrastructure investment accounts for about80% of the total investment of power grid companies.The electricity infrastructure projects are characterized by long construction periods,numerous types and large spatial spans,which makes project optimization a discrete combination optimization problem with multi-time coupling,strong constraints and multiple objectives.However,the traditional investment optimization model ignores the key characteristics of a large number of electricity infrastructure projects.It not only neglects the complex correlation characteristics between the projects,but also ignores the dynamic time-series characteristics of new construction,continuing capital construction and expansion projects.Therefore,a multi-objective project decision-making method that takes into account the dynamic time series of investment and the characteristics of project association is needed to provide quantitative decision basis and technical support for the scientific preparation of grid investment plan.The study in this paper is as follows:Firstly,a method of analyzing the characteristics of association between massive electricity infrastructure projects based on relational graph convolution neural network is proposed.Considering multiple dimensions of economic benefits,social benefits,safety benefits and double carbon benefits,the key feature indicator system is constructed for the decision making of electricity infrastructure projects under the dual carbon strategy.The key feature indicator system is used to extract engineering attributes and key features of infrastructure projects to form a massive project portrait.The projects to be selected are classified and matched into mandatory construction,inter-existence,subordination and mutual exclusion based on relational graph convolution neural network and the graph link predication to explore the potential project-related characteristics,so as to gradually improve the optimal correlation constraint data set of massive electricity projects.The algorithm shows that the proposed method can effectively extract the portraits of various types of infrastructure projects and form a large number of project correlation characteristics to provide a correlation constraint set for optimal project selection subsequently.Secondly,a multi-objective electricity infrastructure optimal project selection model considering dynamic investment time sequences and project association characteristics is proposed.Based on the key attributes of electricity infrastructure projects,a multi-objective economic/safety/social efficiency investment optimization model is constructed,which takes into account the new start investment/gross investment ratio and new start investment/continuing ratio investment time sequences constraints,the constraint of summer and winter power supply capacity for facing the peak usage of electricity,the constraints of multi-type projects association,and the constraints of investment scale.The NSGA-III algorithm is adopt to find the pareto front of the multi-objective project combination optimization problem,and then the fuzzy many-criteria decision making method is applied to obtain the optimal project portfolio.Compared with the existing preferred method,the proposed method can solve better economic benefit,social benefit and safety benefit objectives,which can meet the investment timing constraints to achieve the basic investment requirements.Finally,a multi-level coordinated portfolio optimization decision method for high-dimensional many-objective is proposed.A high-dimensional many-objective investment portfolio optimization model considering dual-carbon benefit objectives is established.By analyzing the multi-level coupling coordination degree of construction sequence and characteristic attributes of infrastructure projects under different voltage levels,a multi-attribute coordinated investment portfolio optimization decision-making method based on grey correlation analysis of high-dimensional objectives is constructed.Furthermore,a high-dimensional objective evolutionary algorithm based on hyper cube space projection transformation is proposed to solve the optimal investment portfolio selection of multiple benefit objectives.The example analysis shows that the proposed method is suitable for high-dimensional project optimal investment decision-making and reasonable arrangement of investment plans.In summary,this paper conducts research from three aspects: the analysis of association characteristics of massive investment projects,the multi-objective optimal project selection model with dynamic investment time sequences constraints,and the multi-level coordinated portfolio optimization algorithm for high-dimensional objectives to provide reference for accurate investment strategies. |
