What is the network analysis method
The Network analysis Method (ANP) is a decisionmaking method of Pittsburg University of America, T.l.saaty, which is adapted to the nonindependent hierarchical substructure in 1996, it is in the analytic hierarchy process (Analytic Hierarchy process, or AHP). Based on the development of a new practical decisionmaking method.
As a decisionmaking process, AHP provides a basic method to represent the measure of decision factors. This method uses the form of relative scale and takes full advantage of human experience and judgment. Under the hierarchical substructure, according to the relative scaleproportional scale, it relies on decisionmaker's judgment, compares the relative importance of the elements of the same level 22, and measures the decisionmaking target according to the hierarchy from top to bottom. This hierarchical substructure is convenient for handling system problems, and it also restricts its application in complex decisionmaking problems. In many practical problems, the internal elements of each level tend to be dependent on the C lowlevel elements that have a dominant effect on the upper elements, i.e. there is feedback. At this point the structure of the system is more similar to the network structure. The network analysis method is the system decisionmaking method which adapts to this need and is developed by the extension of AHP.
ANP first divides the system elements into two parts: the first part is called the control factor layer, including the problem target and decision criterion. All decisionmaking criteria are considered independent of each other and are governed only by the target elements. There can be no decisionmaking criteria in the control factor, but there is at least one goal. The weights of each criterion in the control layer can be obtained by AHP method. The second part is the network layer, which is a group of elements dominated by the control layer C whose interior is the network structure of mutual influence, it is composed of all the elements controlled by the control layer, the elements are interdependent, each other, the elements and the hierarchy is not independent, Each guideline in the hierarchical hierarchy is not a simple internal independent element, but an interdependent, feedback network structure. The control layer and the network layer are composed of typical ANP hierarchies, see.
Characteristics of network analysis method
AHP analyzes a series of factors affecting the target, compares its relative importance, and finally chooses the highest scoring scheme as the optimal scheme. Harker and Vargas once evaluated AHP: "AHP is a complex evaluation system, when we are multiobjective, multicriteria and multijury decisionmaking, faced with a number of options, AHP can be used to solve a variety of quantitative and nonquantitative, rational and irrational decisionmaking problems. "AHP is simple and easy to use, its rigorous theoretical basis determines that it can solve a variety of practical problems." The AHP model makes the various decisionmaking interrelated, and can introduce the interlevel relationship. The top level of the model is the overall goal of the enterprise, then decomposed into specific criteria, subcriteria and so on, until the manager can quantify the relative weights of the subcriteria.
Analytic hierarchy process can solve various complex system problems for decisionmakers, but it also has some shortcomings. For example, AHP does not take into account the interaction between different decisionmaking or the same level, the AHP model only emphasizes the oneway hierarchical relationship between decisionmaking, that is, the impact of the next layer on the previous layer. However, in the actual work of the total target layer of the layer decomposition, often encounter the crosseffect of various factors. If the different research stages of a project have different weights for the judges, the judges will also change the grading of each evaluation index at different stages of the project. At this point, the AHP model seems to be a little powerless.
The characteristic of network analysis method is that, on the basis of analytic hierarchy process, taking into account the interaction between each factor or adjacent level, the "Super Matrix" is used to analyze the influence factors of each interaction and obtain the mixed weights. The ANP model does not require the same hierarchical relationship as the AHP model, there is interaction between the decisionmaking or the same level, and the interaction relationship between the levels is represented by double arrows. If the interaction in the same layer is represented by a doubleloop arrow. The arrows point to factors that influence the decisionmaking factors of the tail. Based on this feature, ANP is becoming more and more popular with policymakers, and is an effective tool for enterprises to make decisions on many complex issues. The determination of relative importance index of each factor in ANP is basically the same as that of AHP. The relative importance index (scale) of each factor is obtained by questionnaire survey to the decisionmaker, but there are some inconsistent phenomena (such as I and H ratio, the scale is 3;j and k ratio, the scale is 5, and I and k ratio, the scale is 6).
Case Analysis of network analysis method

Case study: The risk analysis model of hydropower project based on ANP

1. Identification of risk factors for hydropower projects
Because of the large number of projects in hydropower project, and the general long construction period, the risk of each subproject will be varied, the risk of hydropower project risk identification will be difficult, and it is likely to omit the more important risk factors, Therefore, before identifying the risk, it is necessary to divide the whole project into appropriate subprojects, and then carry out the risk identification for each subproject. At the same time, due to the diversification of risk factors, it is necessary to decompose the risk according to certain risk principle. So this paper uses the method of project decomposition structure (WBS) and risk decomposition structure (RBS) to identify the risk. In addition, using this method to identify risk factors will be beneficial to the establishment and solution of ANP structure model.

2. The hierarchical structure model of the project
In establishing the network analysis model structure of the whole project risk factor, the work structure model of the project should be established first. Because each subproject has its corresponding engineering control objectives: Cost, schedule, quality, safety, and each subproject to the overall project objectives must have different important degree of influence. Therefore, in the construction of the project hierarchy, the engineering objectives should be judged as the criterion of the importance of each subproject. The importance degree model of each subproject established on the basis of WBS is the AHP structure, as the example.

3. Network structure model of risk factors
The risk factors are divided into five categories: natural risk, technical risk, economic risk, organization management risk and social policy risk according to the source of risk. The specific risk factors are then divided according to the five categories of risk.
The traditional risk analysis considers that risk has two properties of occurrence probability and loss, but this definition obviously cannot reflect the nature of risk comprehensively, so Zhang Construction will take predictability, controllable and transferable into risk attribute, and describe risk as a multidimensional attribute object with these five attributes. Inductive multidimensional attributes describe risk in a more comprehensive way to reflect the characteristics of risk factors, but in which transferability and predictability can be reflected in the controllable, so only the control can be estimated to more comprehensively reflect the characteristics of the risk, according to the needs of research, It is concluded that risk estimation is generally an estimation of negative impact, and the risk of hydropower project is evaluated by "uncontrollable" on the basis of traditional twodimensional attribute.
In the process of risk identification, only the risk factors are identified, and the interaction between the risk factors must be studied to establish ANP model. Risk factors can ultimately be influenced by expert or panel discussion, as shown in the table below.
Based on the influence relation table, the ANP structure model is established according to the risk factors ' occurrence probability, loss and uncontrollable nature, as shown in Figure 3.
To establish the corresponding risk factors ANP structure model for each subproject risk factor, we can get the whole structure model of the project, and establish the whole risk factor ANP structure model as multicriteria and multilevel model.

4. Analysis of the risk model of hydropower project based on ANP
(1) The determination of the subproject selfimportance degree.
The importance degree of each subproject is calculated, because the structure model based on WBS is the form of AHP, so it is easy to solve the model by using the traditional method.
(2) Weight vectors and sequencing of risk factors for subengineering.
The weight vector of the interrelated risk factors under each subproject is the ANP structure model of the risk factors of the subproject in Fig. 3.
The key step of the whole project risk factor sequencing is also the core of the risk analysis using ANP. According to the ANP structure model in Figure 3 and the risk factors in table 1, the weight calculation of risk factors under the subproject is carried out as follows:
1) Calculate risk attribute weights. It is important to compare the probability, loss and uncontrollable nature of the risk size. These three attributes are regarded as the criteria for judging the risk factors, so the traditional AHP method can be used to determine the weight.
2) Calculate the weight of each risk factor under the single criterion. Since this model is a multicriteria problem, the interrelated risk factors should be judged separately under three criteria, and the risk factors are studied by the probability criterion, which can be divided into the following steps:
A. Create a hypermatrix. In this paper, the relative importance of other risk factors is compared by the probabilitybased criterion and one of the risk factors, that is, the influence degree of other risk factors on the probability of the risk factor is compared in an important degree. Since not every other risk factor has an impact on it, not all elements are compared under this guideline, and other risk factors that affect this risk factor can be found in the impact relationship table. Then, the feature vectors are calculated separately for each risk category group, that is, the corresponding local weight vectors. After the comparison of each element as a subcriterion and the calculation of the formula (1) to establish a supermatrix.
(1)
where (;)) represents the risk factor category R j Risk factors under R i category The vector matrix of factors affecting. W i j The column vectors are made up of R i each factor in R j a factor is a subcriterion, which is compared and judged to get the eigenvector of the judging matrix.
B. Establishment of a rights matrix. Based on the probability principle, the risk category rI is the subcriterion, the judgment matrix of all categories is compared, that is, the Ri in each risk category The degree of influence of risk category occurrence probability is compared. The comparison of these judgments includes the comparison of the influence of Ri and other categories on the self. Because each risk factor is affected by the degree of comparison in each risk category, the columns of the matrix consisting of multiple matrices are not normalized, that is, the column vector and not 1, can not compare the elements that exist in different categories to a subcriterion of the impact of the degree of the extent of factors; The nonweighted supermatrix cannot solve the limit relative weight vector by the power method, so we should make a comparative judgment on the importance of each risk category. After comparing and judging by the subcriteria of each category, five judgment matrices are obtained, and the eigenvector is calculated, and finally the weight matrix of the formula (2) is available.
(2)
C. Create a weighted supermatrix and solve it. The weighted supermatrix is weighted by the supermatrix by the formula (3), and the size of the column vector element in the weighted supermatrix is the size of the factors of each risk factor on this column, and if a risk factor has no effect on this factor, the corresponding value is zero. At this point, the weighted supermatrix can be solved by the method of power or other methods, and the final relative ordering vector is the weight of each risk factor under the probability criterion.
(3)
3) Calculate the weight of multicriteria risk factors. In order to calculate the weight vector of each risk factor according to step (2), and then synthesize the risk factor weights of each single criterion by the weight obtained in step (1), the risk factors in the subproject can be sorted by the risk factor.
(3) Sequencing of risk factors for the overall project
In order to solve the risk factors of each subproject, we can calculate the weight synthesis and the total sort of the risk factors of the whole project.
The weight of the risk factors of each subproject is corresponding to all risk factors of the whole project, and the weight is set to zero if the risk factors of the subproject are not affected. The weight of each subproject in the whole project is obtained by the calculation of the importance degree of the project, so the rank of each risk factor in the previous project can be obtained by weighting the weighting of the risk factors under each subproject. In the end, total risk factors of the whole project can be sorted.
It is easy to find the biggest and most critical risk factors in the project, and the final result will reflect the actual situation more objectively and realistically because of the mutual influence of the risk factors.
According to the above research, the flow chart of the risk analysis of hydropower project based on ANP is obtained.
Reference documents
 Tangxiaoli, Feng Junwen, Wang Sherong. Project risk management based on network analysis method, statistics and decisionmaking, 2005 issue 16
 DV laugh, Peng Yumei. Selection of R&D items by means of network analysis (J). Science and Science and Technology Management, 2003 2
 Zhong Denghua Cai Shaoquan Li Yuchin. The risk analysis and application of hydropower engineering based on network analysis (ANP). Journal of Hydroelectric Power. 27th February 2008, Issue 1th
Web Analytics (Analytic network PROCESS,ANP)