Research on the optimal model for the evaluation of new power system investment projects based on the cloud model–DS evidence theory–TOPSIS method

Considering that the traditional investment project evaluation system can no longer fully adapt to the characteristics of the new power system under the investment environment of the new power system, this study constructed a new power system investment project evaluation and optimization method based on the combination of the cloud model, evidence theory and the TOPSIS method. First, 16 evaluation indexes were considered based on four dimensions, namely green and low‐carbon, safe and reliable, flexible and intelligent, and economical and efficient, and the optimal evaluation index system of new power system investment projects was designed. Second, the normal cloud model was used to obtain the evaluation membership distribution of each index and turn it into evidence. Considering the conflicts between indicators and the combinational contradictions of traditional evidence theories, an improved DS evidence theory based on a game weight coefficient discount to reduce conflicting evidence was used to fuse the membership information of multiple indicators. On the basis of the evaluation results of each scheme, this study used the TOPSIS method to sort the advantages and disadvantages of each scheme. Finally, the project of planning and putting this model into operation in the 14th Five‐Year project database of a power grid enterprise was selected as an example to carry out empirical analysis, and the effectiveness and superiority of the proposed model were verified.

To achieve the goal of "reaching the peak of carbon and carbon neutrality," the construction of a new power system with new energy as the main body is an important task in the future stages of the development of China's energy and power field. 1,2New power system investment projects have many types, with high requirements and long payback periods, and are closely related to national development, human welfare, and so on.Relying solely on the construction of traditional investment project access systems cannot solve the problems brought about by large-scale wind power, photovoltaic, and other new energy power generation systems connected to the grid. 3,4At the same time, according to the data released by the China Electricity Council (2022), with the comprehensive promotion of the construction of new power systems in China, the annual investment plan of the State Grid Corporation exceeded CNY 500 billion, an increase of 8.84%, and the planned investment in fixed assets of the Southern Power Grid Corporation reached CNY 125 billion, more than twice the value in previous years. 5Therefore, combined with the regional development level and grid development demands, there is an urgent need to construct an investment optimization framework that will consider the new power system projects with volatile and intermittent new energy generation characteristics.
The optimization and evaluation system of new power system investment projects is a complicated problem, which needs to consider the multidimensional optimization index of investment projects.At present, the research on power grid project investment optimization is mainly divided into two categories-one relates to investment decision and project optimization, and the other covers the comprehensive evaluation of project investment programs.In terms of investment decision and project optimization, Qiao et al. 6 extracted and evaluated the risk indicators affecting wind power investment projects based on four dimensions: technology, market, management, and external environment.To improve the system sufficiency brought about by largescale renewable energy promotion, Zhao et al. 7 evaluated transnational investment projects based on four dimensions: cost, society, environment, and technology.Yang et al. 8 evaluated photovoltaic investment projects based on two aspects: evaluating the return on investment after removing central subsidies based on uncertain investment environment, and determining the best investment time for photovoltaic projects.To improve the investment efficiency of power grid enterprises, Wu et al. 9 evaluated eight investment indicators based on five dimensions, namely the improvement effect of the distribution network, the power supply level of the distribution network, the scale of the distribution network, the operation level of the distribution network and profitability.To improve the ability to perform the quantitative evaluation and prediction of power grid investment benefits, Guo et al. 10 evaluated the investment benefit indicators such as enterprise asset economic growth level, price index, asset-liability ratio, return on investment, debt financing level, and benefit index.To solve the problem that the investment allocation of provincial companies is prone to deviation from the actual investment demand, Li et al. 11 evaluated 18 indicators based on five dimensions: power supply quality, power grid structure, equipment level, power supply capacity, and intelligence level.Han et al. 12 pointed out problems such as an investment imbalance caused by power grid enterprises in the process of allocating investment funds and the inability to reasonably strengthen the weak links of power grid construction, and evaluated 24 investment benefit indicators based on four dimensions: economic benefit, technical level, coordinated development and social benefit.Xu et al. 13 promoted supply-side structural adjustment and demand-side response through investment, and conducted investment evaluation on 14 indicators based on three dimensions: power grid investment benefit, power grid operation and maintenance effect, and sustainable development value.For example, considering the verification rules of transmission and distribution price, input indicators such as network investment, number of employees, and substation capacity, and output indicators such as market trading power, main business income, and power supply reliability rate, this study evaluated the investment efficiency of provincial power grid enterprises under new power transformation.In terms of the comprehensive evaluation of project investment plans, Nurmamet Ruze et al. 14 performed a macroeconomic evaluation of the investment project of power grid foundation engineering in Germany.The above research results solved certain practical problems, but their deficiency is that when selecting new power system investment projects, the uncertainty and other characteristics brought about by the large-scale grid connection of new energy such as wind power and photovoltaic should be considered, as well as the security impact on the original system.
Because of the different investment environments of each power grid project, the optimal selection method is not the same.In general, the subjective and objective combination method is used to determine the weight of investment project optimization indicators, which mainly covers the analytic hierarchy process, entropy weight method, coefficient of variation method, and so on. 15The decision-making method is mostly the improvement of the multiattribute decision-making method and the application of related new theories, such as prospect theory, matter-element theory, the envelopment analysis method, and so on.Hasan Di ̇ncer et al., 16 based on Pythagorean fuzzy DEMATEL, TOPSIS, and Shapley values, identified renewable energy investment projects.Pawe Ziemba et al. 17 used the fuzzy multicriteria decision analysis (FMCDA) method, sensitivity analysis, and robustness analysis based on Monte-Carlo simulations to evaluate offshore wind investment projects.According to Guo et al., 18 to improve power quality and increase the absorption capacity of renewable energy generation, by introducing probabilistic hesitation-fuzzy elements to describe quantitative indicators, decision-makers can use hesitation-fuzzy language term sets to evaluate qualitative indicators and evaluate offshore wind storage investment projects.Wu et al. 19 designed an investment decision index system for engineering projects by monitoring the investment cost of large-scale basic scientific research projects, adopting the entropy method and the optimal difference method to calculate the objective weight and subjective weight of investment project indicators, and adopting the ROMETHEE method to prioritize investment projects.To improve the investment accuracy of power grid investment projects and effectively prevent investment risks, Yuan et al. 20 designed a multidimensional and multiangle risk evaluation index system for power grid investment projects, determined the index weights by means of the combination weighting method and proposed an investment optimization decisionmaking method for multiple power grid construction projects under a certain investment scale.In the above research results, the weight setting method of fixed strategy was mostly adopted, and new power system investment projects in dynamic development environments were not considered.Although the above method can evaluate the optimal situation for power grid investment projects, it is slightly insufficient to deal with uncertain environments.
Therefore, the innovation of this paper is mainly reflected in three aspects, namely, first, combining the current development status of China's power system, deeply integrating the performance characteristics of the traditional power system and new power system investment environments, and combining the characteristics of the development of the new power system.Considering 16 evaluation indexes based on four dimensions, namely green and low-carbon, safe and reliable, flexible and intelligent, economical and efficient, the optimal index system of new power system investment projects is designed.The second innovation is proposing a method based on the combination of the cloud model, evidence theory, and the TOPSIS method.While solving the problems of randomness, fuzziness, and uncertainty in the process of investment project optimization evaluation, the importance difference between the evidence bodies is fully considered, the conflicting evidence is corrected, the correct focal element is identified more efficiently, and the accuracy of evidence fusion is improved.A new power system investment project optimization evaluation model is established, and a typical new power system demonstration project is selected to carry out empirical analysis, verifying the effectiveness and feasibility of the proposed model, which will promote the new power system in China.
This paper is organized as follows.In Section 2, the design of an optimal evaluation index system for new power system investment projects is described.In Section 3, a new power system investment project optimization evaluation model based on the cloud model-evidence theory-TOPSIS method is described.In Section 4, the optimization evaluation calculation process for power grid investment projects based on the cloud model-evidence theory-TOPSIS method is described.In Section 5, the selection of a typical new power system demonstration investment project for empirical analysis is described.In Section 6, some conclusions of this paper are discussed.

| New energy electricity penetration rate
The penetration rate of new energy refers to the proportion of new energy generation to total power generation within a certain time period within a certain power grid control area, which reflects the power support capacity of new energy for the system. 21The higher the penetration rate of new energy, the stronger the acceptance rate of grid investment projects for green power resources such as distributed energy, and the greater its contribution to the construction of a new power system with new energy as the main body.The calculation formula of new energy electricity permeability R supply is as follows: where S new indicates new energy generation and S total indicates the total power output.consumption The proportion of new energy consumption refers to the proportion of the power generation actually consumed by wind and solar energy to the available power generation. 22The higher the proportion of new energy consumption, the stronger the support capacity of investment projects for the consumption of new energy on the user side.The formula for calculating the R Consumption of new energy consumption is as follows: where C new indicates the actual consumption of new energy.

| Reduction in CO 2 emissions
The reduction in CO 2 emissions refers to the amount of CO 2 reduction generated by investment projects such as power grid renovation and upgrading. 23The higher the reduction in CO 2 emissions, the greater the investment project's contribution to achieving the zero-carbon goal.
The formula for calculating carbon dioxide emission reduction Q co 2 is: where q i co , 2 represents the CO 2 emissions reduced after the i power grid investment node is put into operation and I represents the number of nodes included in a power grid investment project.

| Income from pollutant emission reduction
For investment projects where new energy gains access to the power grid, it replaces traditional thermal power generation, effectively reducing the emissions of sulfur dioxide, nitrogen oxides and other pollutants, thereby reducing the cost of pollutant treatment, and indirectly translates into the new energy emission reduction benefit B reduction .The calculation formula is as follows: where A reduction represents the pollutant emission reduction benefit per k.Wh of electricity; l represents the number of types of clean energy generation; P k rated, denotes the rated power of k clean energy generation at a certain voltage level; T k denotes the maximum utilization hours of k clean energy power generation; λ SO 2 and λ NO X represent sulfur dioxide emissions and nitrogen oxide emissions in the clean generation of 1 k.Wh of electricity, respectively; V SO 2 and V NO X represent the environmental treatment cost per ton of sulfur dioxide and environmental treatment cost of nitrogen oxides, respectively.

| Substation load coordination degree
The substation load coordination degree reflects the load balance degree between the project and the adjacent substations after the project is put into operation.The more balanced the load between the planned project and the adjacent substations, the stronger the matching between the spatial layout of the substations and the load distribution, the smaller the difference between the operating conditions of the substations, and the greater the safety and robustness of the power grid's operation.The calculation formula of substation load coordination degree R trans is as follows: trans, trans 2 trans trans (6)   where N trans represents the substation load rate of the planned project; L i trans, represents the average load rate of adjacent substations; and L ¯trans indicates the total number of adjacent substations.

| Line power flow balance
The power flow coordination degree reflects the power flow balance between the project and the adjacent power lines after the project is put into operation. 24Under the same load level, the more balanced the planned project load is with the power flow of adjacent lines, the higher the safety and economy of power grid operation.The line power flow coordination degree R line is calculated as follows: line line (7)   where L i line, indicates the planned project line load ratio.L ¯line indicates the average load ratio of the adjacent lines.M line indicates the total number of adjacent lines.

| Power supply capacity
Power supply capacity refers to the capacity of the distribution network to supply electricity to users in a certain area under the constraints of line capacity and main transformer capacity, reflecting the scale of the distribution network's available load.The objective function and constraint conditions for the calculation of power supply capacity S capacity are as follows: is the rated capacity of the feeder m.

| Overload/low voltage risk
In power system safety analysis, risk is generally defined as the product of accident probability and accident consequence.The power system risk includes overload risk and low voltage risk, and is measured by the comprehensive risk of both.The formula for calculating the comprehensive risk where ω l and ω v represent line the overload risk weight factor and bus low-voltage risk weight factor, respectively; indicate the severity of the overload risk of a line and the severity of the low voltage of a bus caused by an accident, respectively.

| Distribution line connection rate
If the two wires are connected directly by the contact switch, it is said that the two lines are connected, and the line connection rate is the proportion of the connected lines in the total line.The higher the distribution line connection rate, the more flexible the network structure is.The distribution line contact rate N contact is calculated as follows: where R contact indicates the number of connected lines and N total indicates the number of bus routes in the distribution network.

| Proportion of installed flexible resources
Flexible resources usually refer to flexible regulation resources on the power side, including hydropower with regulation capacity, pumped storage, new energy storage power stations, gas turbines, and coal power with flexible regulation capacity.The higher the proportion of installed flexible power supply, the stronger the acceptance of flexible resources for the power grid investment project is.The R Flexible calculation formula is as follows: where P Flexible represents the flexible resource installed capacity and P total represents the total installed capacity.

| Peak-valley difference
The peak-valley difference is the difference between the user's maximum load and the minimum load, and the peak-valley difference rate is the ratio of the peak-valley difference to the maximum load.The higher the peak-valley difference ratio, the more the power network investment project can guide the user to use electricity flexibly, and the load side adjustment and power system flexibility will also improve.Its calculation formula is as follows: where R L Δ is the daily peak-valley difference rate, while L k max, and L k min, are the maximum and minimum load of the power system on day k, respectively.

| Market electricity proportion
Electricity market transactions undoubtedly require power grid enterprises to invest capital in different dimensions such as the power supply side, the power grid side and the load side to provide strong support for market transactions.The higher the proportion of field electricity, the more power grid investment projects can support and promote the construction of the power market.The market electricity ratio R trading calculation formula is: where Q trading represents the amount of traded electricity included in the electricity market and Q total represents the total transaction power.

| Net present value (NPV) ratio
The NPV refers to the algebraic sum of the net cash inflow value of each year during the life of the power grid investment project, discounted to the base period according to the base discount rate. 25The NPV reflects the investment profitability of the planning scheme, but when the NPV of different schemes is the same, the investment amount of the project should be considered at the same time-that is, the relative profitability of the investment project should be judged by the NPV rate (NPVR).The higher the NPV ratio, the stronger the relative profitability of the investment scheme.The formula for calculating NPVR is as follows: where PVI represents the present value of the project investment; CI expresses the cash inflow; CO expresses the cash outflow; and n indicates the number of counting periods.

| Internal rate of return(IRR)
The IRR refers to the discount rate when the present value of the net cash flow of each year is equal to 0 during the construction period and the production period of the investment plan, reflecting the return rate desired by the investor.A higher IRR indicates that the cost of project investment is relatively small, but the benefit is relatively large.The IRR is calculated as follows:

| Payback period
The dynamic payback period (DPP) refers to the time required for the net income of the project to cover the entire investment based on the benchmark rate of return of the power industry, taking into account the time value of the capital.The DPP measures the fund recovery ability of the project from a dynamic point of view.The shorter the DPP, the stronger the fund recovery ability of the project is.  where I c represents the benchmark rate of return of the power industry and DPP represents the project's dynamic investment payback period.

| Expected maximum gains
The expected maximum electricity income refers to the expected maximum net income after the investment project is put into operation, considering the reliability and line loss conditions, reflecting the economy of power grid enterprises.The expected maximum power return E max is calculated as follows: where Q out,max indicates the maximum expected sales of electricity; P out represents the average selling price; Q capacity indicates the power supply capacity of the grid; Q in,max indicates the maximum expected power purchase; η ply sup indicates the average power supply availability; P in represents the average power purchase price; η loss represents the line loss ratio.
To sum up, the evaluation index system constructed in this paper is shown in Table 1.

| A NEW POWER SYSTEM INVESTMENT PROJECT OPTIMIZATION EVALUATION MODEL BASED ON THE CLOUD MODEL-EVIDENCE THEORY-TOPSIS METHOD IS ESTABLISHED
The optimization of new power system investment projects is essentially a typical multiattribute evaluation and decision problem.To take into account the optimization of new power system investment projects, by combining the cloud model and evidence theory, the randomness, fuzziness, and uncertainty in the optimization process of investment projects are solved, making the evaluation results more reliable.

| Cloud model
In the 1990s, to solve the problem of the uncertainty of research objects, Li Deyi proposed the concept of the cloud model. 26The cloud model is used to calculate the expected value, entropy, and hyper entropy of the new power system investment project, reflecting the randomness and fuzziness of the investment project optimization problem, and realizing the mutual mapping relationship between qualitative and quantitative evaluation indicators.
Here, the evaluation set of the new power system investment project optimization evaluation is determined-that is, grade I, grade II, grade III, grade IV, and grade V-and the corresponding cloud model of each investment project optimization index is established.The X-conditional cloud generator is used to obtain the membership of each optimization evaluation index value under different levels-that is, the membership matrix.The X-conditional cloud generator algorithm is as follows Input: Digital characteristics of the target to be evaluated E E H ( , , ) x n e , index evaluation value x 0 and the number of cloud droplets to be generated N. Output: The quantitative value of N cloud droplets corresponding to a specific value x μ i N , ( = 1, 2, …, ) i 0 .

| Basic principle
DS evidence theory is a method proposed by Dempster and Shafer to deal with uncertain and incomplete information.The identification frame composed of all of the objects to be studied is denoted as Θ, and the elements in the frame are mutually exclusive, while 2 Θ is a subset of Θ, and is in the interval [0,1]. 27et the function mass satisfy the following conditions, namely: where mass is the basic probability distribution function of the identification frame Θ; K is the focal element of the basic probability distribution function; and K mass( ) is the degree of support of relevant evidence for focal element K , also known as the membership degree.
Multiple groups of basic probability distribution are denoted by the BPA expression formula, namely:

| Combination weight determination based on game thought
Taking full account of the importance difference between the evidence bodies, we revise the conflicting evidence, identify the correct focal element more efficiently, and improve the accuracy of evidence fusion. 28The steps are as follows.
(1) Determine the dynamic weight coefficient According to the basic probability distribution matrix U′ m n ×( +1) obtained from the cloud model, the recognition framework Θ has a total of n + 1 propositions-that is, a n + 1 preference level-and each proposition has a total of group n evidence and n preference indicators.The dynamic weight coefficient is calculated as follows: 1.The average membership degree μ mass( ¯′) k of n preference indicators in each preference level μ′ k is calculated to reflect the average support degree of all indicators for grade μ′ k .
2. The distance d i between the membership degree of the i-th index and the average membership degree is calculated, namely: 3. The weight ω i of each index is assigned according to the distance, and it is believed that the index with the larger gap between the average membership degree and most indicators will have greater conflict, so the assigned weight is inversely proportional to the distance, that is: (2) Determine the static weight coefficient The fuzzy consistency matrix is introduced to solve the problem of struggling to pass the consistency test and the subjectivity being too strong in the analytic hierarchy process.The steps are as follows.

Construct fuzzy complementary matrix
Experts in fields related to new power system investment projects are invited to make pairwise comparisons according to the importance of scale judgment on indicators, and a fuzzy complementary matrix is obtained, namely: where r ij is the relative importance degree of indicator i compared with indicator j, and r ij meets:

Solve the fuzzy uniform matrix
To ensure the consistency of the judgment results, the fuzzy complementary matrix is transformed into a fuzzy consistent matrix, which naturally satisfies the consistency.

Calculate index weight
The weight of each factor in the fuzzy consistent matrix relative to the previous layer is calculated by the formula, which belongs to the hierarchical single ranking weight, namely: where the parameter α satisfies α n − 1 2  .
(3) Combined weight coefficient The game combination weight synthesizes the information of dynamic weight and static weight, and provides a scientific tool for the identification and correction of conflict evidence.
The weighting method of m indicators is linearly weighted to obtain the combined weight-that is: where w ij is the weight of index j under i weighting methods and μ t represents the weight distribution factor for each weighting method.

| Conflict evidence identification and correction
Suppose there is m group of evidence, and the average weight m 1/ assigned to each piece of evidence is taken as the threshold value. 29When the weight W m 1/ i  is combined, the evidence is considered as nonconflicting evidence and the evidence is retained.When the weight W m < 1/ i is combined, the evidence is considered as conflicting evidence and corrected.The modified basic probability distribution function is: where discount is the correction factor.

| Based on the TOPSIS method, the average proximity degree is calculated for project comparison
The TOPSIS method is used to calculate the average closeness between the basic probability distribution function formed after evidence fusion and the basic probability distribution function corresponding to the positive and negative ideal solution of each new power system investment project, and can determine the optimal order of the new power system investment project.The steps are as follows.
1. Calculate the probability assignment function of the virtual best/worst scheme Based on the optimal value and the worst value of each optimal evaluation index of the new power system investment project, the membership matrix of the positive ideal cloud and negative ideal cloud is obtained by using the cloud model.Then, the basic probability distribution function of the virtual worst/ best scheme after the evidence fusion is obtainedthat is, mass i − and mass i + , which are the optimal solution and the least ideal solution, respectively.2. Calculate the degree of closeness between each alternative new power system investment project and the virtual optimal/column scheme Let r ij + be the degree of fit between the investment project i and the virtual optimal scheme at the level j, and the degree of proximity matrix R p n × + is as follows: Similarly, the fit degree matrix R − between the investment project of the new power system i and the virtual worst scheme under grade j is obtained.

Calculate the average fit difference
According to the average fit degree, we calculate the average fit degree difference between the basic probability distribution of each new power system investment project and the basic probability distribution of the virtual optimal/worst scheme r Δ¯i + and r Δ¯i − , as follows:

| DESIGN OF THE OPTIMIZATION EVALUATION CALCULATION FLOW OF POWER GRID INVESTMENT PROJECT BASED ON THE CLOUD MODEL-EVIDENCE THEORY-TOPSIS METHOD
To solve a series of problems such as randomness, fuzziness and uncertainty in the optimization process of new power system investment projects, this paper presents an optimization method for new power system investment projects based on the cloud model and improved evidence theory.First, the new power system investment project optimization evaluation index performs system classification; secondly, the cloud model is used to calculate the membership matrix of the preferred index of each new power system investment project at different levels, and the uncertainty focal element is transformed into the basic probability distribution function in evidence theory.Then, based on the deficiency of traditional evidence theory in solving the conflicting evidence problem, an improved method of identifying and correcting conflicting evidence based on game combination weight is proposed. 29inally, to overcome the situation in which several new power system investment projects have the same rating and the pros and cons cannot be judged, TOPSIS thought is used to calculate the basic probability distribution function of the virtual optimal/inferior scheme, and then to determine the priority investment order of the project according to the average proximity degree between the investment projects and the basic probability distribution function of the virtual optimal/ inferior scheme.The specific calculation process is shown in Figure 1.

| Data basis
Based on the 10 kV distribution network frame formed by a power grid enterprise at the end of 2022, and Optimization calculation process of a new power system investment project.
taking the planned load in 2023 as the boundary, the planned production projects in 2023 in the "14th Five-Year Plan" project library were selected for optimal ordering, and the scientificity and rationality of the proposed power grid investment project evaluation and optimization method were verified.The data for each indicator of each of the five projects are shown in Table 2.

| Cloud model affiliation distribution evidence
Taking the best/worst values of the indicators in the five items as the boundary, they are evenly divided into four sub-intervals, which respectively indicate the initial grading criteria of each index for the rating (excellent, good, medium, and poor), as shown in Table 3.
Based on the grading standard of the index, the digital characteristics of the benchmark cloud are calculated using the aforementioned numerical characteristics calculation method, and the numerical characteristics of the benchmark cloud are obtained as shown in Table 4.
Taking the new energy electricity penetration index as an example, the corresponding evaluation benchmark cloud is shown in Figure 2: Based on the evaluation index data, the X-conditional cloud generator is used to calculate the membership degree of each index at each comment level in the benchmark cloud, and the upper formula is transformed into a basic probability distribution in line with evidence theory, and the result is shown in Figure 3.

| Improve the theory of evidence and multiindex subordination information fusion
Considering the conflict between evidence, the discount coefficient based on the combined weights is used to correct the original basic probability distribution.Dynamic weights are calculated by means of the Manhattan distance method, static weights are calculated by means of the fuzzy analytic hierarchy method, and combined weights are calculated by means of game ideas.| 33 F I G U R E 3 Basic probability distribution plot.
For the fuzzy analytic hierarchy method, five senior industry experts (three senior engineers, one professor-level engineer, and one dean of the Electric Power Research Institute) were invited to compare the importance of the evaluation indicators presented in Table 1, and the experts unanimously agreed that the importance of the four firstlevel indicators and their respective second-level indicators was the same, and the fuzzy judgment matrix of the threelevel indicators formed was:  The static weight of the indicator was calculated, that is: (0.079, 0.063, 0.063, 0.046, 0.050, 0.050, 0.067, 0.083, 0.050, 0.083, 0.050, 0.067, 0.067, 0.050, 0.067, 0.067).
Taking the first scheme as an example, the static weights and dynamic weights of each scheme were combined, and the comparison results of the three weights are shown in Figure 4.
It can be seen from the figure that the combination weighting method based on game ideas can effectively combine dynamic weights and static weights, fully take into account the information of the two, and obtain more accurate weight results, providing a strong reference for the identification of conflict evidence.
Based on the combined weights, the discount coefficient of the conflicting evidence of each scheme is calculated as shown in Figure 5.When the discount coefficient is 1, it indicates that there is no conflict with the corresponding evidence, and when the discount coefficient is less than 1, it indicates that the corresponding evidence is conflicting, and it is necessary to reduce the discount, and the revised evidence is synthesized.

| Optimal decision-making of investment solutions based on TOPSIS thinking
We selected the optimal and worst values of each index in the scheme set to form the corresponding virtual optimal scheme ( S Best ) and the virtual worst scheme Worst ).Using the same method described above, the basic probability distribution of the two was calculated and the evidence was synthesized.The results of the evidence synthesis of five investment solutions, the virtual optimal solution, and the virtual worst scenario are shown in Figure 6.
According to the TOPSIS idea, the closeness between the basic probability distribution of each investment scheme and F I G U R E 4 Three weights corresponding to scheme 1.

MA ET AL.
| 35 the basic probability of the virtual best/worst scenario is calculated, and the results are shown in Figure 7: According to the degree of proximity, the preferred order of the five investment projects is as follows: project 4 ≻ project 1 ≻ project 5 ≻ project 3 ≻ project 2, where the ≻ operator means "better than."

| Verification of model validity and superiority
To verify the effectiveness and superiority of the proposed method, other literature methods are introduced for comparison with it, and the comparison results are shown in Table 5.
The comparative results show that the project optimization results of the proposed method are consistent with the results of method 1, which verifies the effectiveness of the model, while compared with method 1, the uncertainty of the evaluation results can be measured by the uncertainty focal element combined with evidence theory.In addition, method 2 and method 3 are inconsistent with the preferred results in this paper, because method 2 does not deal with the conflicting nature of the evidence, resulting in the problem of combinatorial contradiction in the synthesis of evidence, while method 3 deals with the conflicting evidence, but only uses a single dynamic weight to identify and correct the conflicting evidence, and on the one hand, the weight is not considered comprehensively, while on the other hand, there are also defects in the identification performance of conflict.Therefore, the evidence F I G U R E 6 Evidence synthesis results of five investment schemes, the virtual optimal scheme, and the virtual worst scenario.
F I G U R E 7 Proximity map between the basic probability distribution of each investment scenario and the basic probability of the virtual best/worst scenario.
synthesis method based on the cloud model and game combination weight modification of conflicting evidence can obtain comprehensive evaluation results for the project more accurately, and provide decision-making support for the optimization of new power system investment projects.

| CONCLUSION
Considering the randomness, ambiguity, and uncertainty of the optimization problem of new power system investment projects, this paper proposes a new power system investment project optimization method based on cloud model-evidence theory-TOPSIS method, and the research conclusions are as follows: 1. Considering the characteristics of the new power system, the optimal evaluation index system of new power system investment projects, including dimensions such as green and low-carbon, safe and reliable, flexible and intelligent, and cost-effective is designed.2. In the evaluation optimization, the membership value between the measured sample and the benchmark cloud is calculated by the cloud model, which realizes the mutual transformation between the quantitative value and the qualitative concept, and overcomes the ambiguity and randomness of the traditional membership function.3. The results show that compared with cloud model + fuzzy comprehensive evaluation, cloud model + traditional evidence theory, and cloud model + single weight improvement evidence theory, the method proposed in this paper can not only measure the uncertainty of evaluation results, but also effectively overcome the combination contradiction defects of traditional evidence theory in dealing with highconflict evidence, and can clearly determine the priority of projects in combination with TOPSIS.
the load carried by the main transformer i; L m feeder, is the load carried by the feeder m; L S i feeder ∈ is the corresponding bus of the autonomous transformer i for the feeder m; L m n feeder, . is the load carried by the m paragraph n of the feeder; R i Substation, is the rated capacity of the main transformer i; and R m feeder,

1 F I G U R E 5
Scheme 1 Scheme 1 Scheme 1 Scheme 1 Scheme 1

T A B L E 5 4 ≻ 1 ≻ 4 ≻ 5 ≻ 4 ≻
Comparison of results.Project 1 ≻ Project 5 ≻ Project 3 ≻ item 2 Method2 Cloud model + traditional evidence theory Project 1 ≻ Project 4 ≻ Project 5 ≻ Project 2 ≻ Project 3 Method 3 Cloud model + single weight improves evidence theory Project Project Project Project Project 1 ≻ Project 5 ≻ Project 3 ≻ Project 2 the total risk of overloading of the entire grid line caused by a component failure and the total risk of low voltage caused by a component failure, respectively; Rating index system.
T A B L E 1 Initial rating of the indicator.
T A B L E 4 Benchmark cloud digital characteristics.
MA ET AL.