| Recently, actively developing the smart grid has become a new trend of theworld’s electricity, China’s electric power industry is facing the severe challenge inthis new situation. With the development of the electricity market, the production,consumption and trading patterns in competitive electicity market will undergofundamental changes. In addition to the traditional grid security and reliability, theeconomic and risk indicators researches are becoming increasingly important in theelecticity construction; The smart grid is an inevitable result of economic andtechnological development, which will greatly improve the generation, transmission,and distribution efficiency, promote energy conservation and the sustainabledevelopment of society, however, the smart grid also introduced a large number ofnew technical equipment and concepts, adding more uncertainties to power system aswell as optimizing and improving the development of the traditional power system.Network planning is the key link of the modern power grid construction anddevelopment, the theories and methods in the traditional network planning of ourcountry have been in-depth study, on the basis of that, the purpose of this article willintroduce a variety of factors under the new situation of the electricity markitilizationand the smart grid construction to the actual planning work, the specific work is asfollows:(1) Under the new situation, in addition to the traditional factors, i.e. weather,season, day type, the real-time pricing has also become a very important factor in theaspect of affecting the load forecasting accuracy. So on the basis of analyzing theimpact factors of short-term load characteristics, taking into account the impact ofreal-time pricing, this article made a specific and detailed analysis of the relationshipbetween the electricity price and load in the electricity market environment by theprice elasticity of demand matrix and the grey system theory; the gray model canweaken the randomness of the data, as well as the highly nonlinear characteristics ofneural networks, an improved gray Neural Network (IGNN) were proposed to predictshort-term load in real-time pricing environment. There are three measures forimprovement of prediction accuracy and computing speed of the gray neural networkmodel:1) Using gray double exponential smoothing method for input datapreprocessing;2) Using genetic algorithms to optimize the initial parameters of GNN;3) Optimization of learning and training rate of GNN model. At the same time, the relevance of load and electricity price was proposed to determine whether thecandidate samples used as input variables in terms of the processing of the datasample, largely mitigated the low efficiency learning of GNN caused by excessfulnon-associated samples. A practical example demonstrated the accuracy and validityof the method.(2) Existing substation planning model and methods have only staticcharacteristics (ignoring the time factor), so there are differences with the multi-stagecharacteristics of the actual power grid project, and the dynamic programmingoptimization problem is the most clear and effective method of solving the dynamicprocess of time-division stage. Therefore, this paper has used the dynamicprogramming to establish a multi-stage optimization planning model of the substation,as much as possible to meet the dual demands for time and space of the actualsubstation planning. Firstly, combined with the expers’ advices, optimizing thesubstations’ sites, capacities for the planning target year, which the planning resultswere as the candidates for the various stages later; Then dividing the middle years intoseveral stages, and establishing the dynamic programming model as well as usingsome heuristic rules to reduce the dimension of the model, the optimal substationbuilding programs (including the transformation of the existing substation) throughoutthe planning period was founded. Finally, the real case was used to verify theoperability of the method proposed here.(3) The net present value (NPV) method has always been the main method for theassessment of investment in the power industry, but the NPV method ignores thevalue of uncertainties of the investment projects, normally being suit for theinvestment appraisal with characteristics of short-term, little changes and so on. So,on the basis of taking full account of the new uncertainties bringed by the smart grid,this paper has proposed an economic assessment methods based on fuzzy real option(FROV) by using the features of the real options approach (ROA) for assessing theuncertainties for the future and opportunities to choose. Combined with the Bellmandynamic programming, the FROV method was further expanded to establish amulti-stage fuzzy real option evaluation methods (DFROV), which provided a newperspective for economic decision-making evaluation by bringing the projectevaluation, investment in capital budgeting decisions and investor’s psychology.Finally, the method was tested to have more application reference in actual practice.(4) Lacking of the comprehensive index system of smart grid at this stage, this paper tries to put forward a new Smart Grid assessment index system: the indexsystem consists of four tiers, the logic and the inherent law are as the principles forthe elements at all levels, index selection aimed to achieve an important guidingsignificance of the indicator system on the smart grid construction plan.(5) Considering the dynamic and uncertain characteristecs of the constructionprojects planning, this paper proposed a multi-stage fuzzy comprehensive evaluationmethod based on the group decision-making. Adding the time dimension to thetraditional AHPmethod, and introducing the concept of the group decision-makingand fuzzy preference relations; then establishing a multi-stage decision-makingoptimization model with the results of the comprehensive evaluation as the goalfunction, and proposing a hybrid intelligent optimal algorithm with the flexibleadjustment feature to realize the projects preferred choice under the limitedcircumstances of the investment. Intelligent optimization algorithm consists of threeparts:1) Initialization part, greedy algorithm is used to establish the vialble initial set;2) Multi-stage decision-making optimization part, optimizing the allocation and thetiming of items, an improved genetic algorithm (IGA) was proposed to satisfy thesetwo optimization purposes;3) dynamic adjustment part, aiming to largely save moneyon the basis of meeting the technical requirements. Finally, a specific example wasused to verify the practicality and effectiveness of the method here. |
PDF Full Text Request