CN107239905A - Onboard networks safety risk estimating method based on advanced AHP GCM - Google Patents

Abstract

The invention discloses an airborne network security risk assessment method based on the improved AHP-GCM, comprising the following steps: 1. Analyzing the risk source of the new airborne network architecture; 2. Establishing an evaluation index system for the risk source and its threat state; 3. Assign the weight of the index and check the consistency; 4. Find the weight of each layer of indicators and normalize it; 5. Divide the risk level to determine the gray class; 6. Establish a whitening weight function; 7. Calculate the gray clustering coefficient and judge It belongs to the gray category; 8. Based on the above steps, conduct a second assessment of the threat status of the risk source. The patent determines its risk sources and risk status on the basis of analyzing the new airborne network architecture, then evaluates the risk sources based on tomographic analysis and gray clustering methods, and finally conducts a second assessment of its threat status, and synthesizes it twice The assessment optimizes the risk assessment results to prevent the deviation and inaccuracy of a single assessment result.

Description

Airborne Network Security Risk Assessment Method Based on Improved AHP-GCM

technical field

The invention is applied to the field of new airborne network security risk assessment, and in particular relates to an airborne network security risk assessment method based on the improved AHP-GCM.

Background technique

Flight safety is an eternal theme in the aviation industry. With the widespread implementation of onboard Wi-Fi plus modification and the important application of portable airborne equipment such as electronic flight bag (Electronic Flight Bag System-EFB) in aviation, the risk of hacker access attacks Higher and higher; the new aircraft adopts an advanced integrated modular avionics system network architecture, which breaks the physical isolation between the traditional flight control network domain, airline information service domain, and in-flight entertainment domain, and even increases Internet access related functions of the in-flight passenger entertainment domain.

Traditional approaches do not fully counteract the risk of impacting the airborne control domain through external networks or devices. Therefore, it is an urgent problem to design an effective risk assessment method for new airborne network security. Such an assessment method can intuitively reflect the security risks of the new airborne network, allowing managers to understand the security of the network based on the assessment results, and formulate appropriate defense or repair strategies in advance, which is of great significance to the security of the new airborne network.

Scholars at home and abroad have discussed many risk assessment methods in view of the characteristics of the network. Although these methods are different in specific implementation means and calculation methods, they all follow the basic risk assessment process. According to different calculation methods, risk assessment methods are generally divided into three categories: qualitative, quantitative, and a combination of qualitative and quantitative.

Analytic Hierarchy Process (AHP) is a systematic hierarchical weight decision-making analysis method that combines qualitative and quantitative analysis formally proposed by American operations researcher Thomas Setty (T.L.Satty) in the mid-1970s. . This method is a systematic analysis method, which is simple and practical, requires less quantitative information, but cannot provide new solutions, has many qualitative factors, and the statistics are large when there are many indicators.

Gray Theory was first proposed by Professor Deng Julong in the 1980s. The gray system concept of gray clustering method is extended according to the concept of "gray box". It is a clustering method based on the generation of whitening function of gray number. Poor, less linkages between indicators.

Gray clustering methods can be divided into gray relational clustering and gray whitening weight clustering according to different division objects, among which gray whitening weight clustering includes three evaluations: gray fixed weight clustering, gray variable weight clustering and triangular whitening weight based clustering. function evaluation.

Both methods are commonly used network risk assessment methods, and combined use can overcome the shortcomings of the separation of qualitative analysis and quantitative assessment. However, the AHP relies on the experience of technical engineers and is highly subjective; and it may not be possible to evaluate only risk points. Comprehensive analysis of the risk posture of the network.

Contents of the invention

The technical problem to be solved by the present invention is: the purpose of the present invention is to provide an airborne network security risk assessment method based on the improved AHP-GCM, and this patent determines its risk source and risk status on the basis of analyzing the new airborne network architecture , then evaluate the risk sources based on the tomographic analysis method and the gray clustering method, and finally conduct a second evaluation of its threat status, and optimize the risk evaluation results by combining the two evaluations to prevent the deviation and inaccuracy of the single evaluation results.

The technical scheme that the present invention takes for solving the technical problem existing in known technology is:

An airborne network security risk assessment method based on improved AHP-GCM, comprising the following steps:

Step 101, analyzing the risk source of the airborne network architecture; the specific method is:

The airborne network architecture includes an aircraft control domain, an information trusted domain, an information open domain, a cabin network domain, and a public network domain; the interface bus-oriented data transmission structure of the airborne network architecture includes ARINC429, ARINC825, and ARINC664;

The network interface points exposed by the airborne network architecture include the gateway and each network domain, and between each network domain; the network data interaction points between the airborne information systems include the passenger interface, the software platform provided by the refitting company, and the onboard entertainment system. The network data interaction mode between the airborne information system and the ground information system includes electronic flight bag, flight attendant operation PAD; access authority includes illegal users entering the airborne network; the threat state of the airborne network architecture includes: obtaining information, tampering Information, use of services, denial of services and illegal operations of elevated privileges;

Step 102, establishing evaluation index systems for risk sources and their threat states; specific methods are:

First, establish an evaluation index system for the risk source: the target layer is the risk assessment target to be achieved; the attribute layer is the risk probability, risk impact, and uncontrollability; the program layer is the threat caused by the risk source;

Secondly, establish an evaluation index system for the threat state of the risk source: the target layer is the risk assessment goal to be achieved; the attribute layer is confidentiality, integrity, and availability; the third layer is the specific performance of the threat state;

Step 103, assigning index weights and checking consistency; the specific method is:

Two technical engineers assign values to the indicators, and the assignment rules are as follows: compare the importance of each indicator in the lower layer relative to the element based on a certain element in the upper layer, and select the appropriate one from the specified scale. value, establish a judgment matrix: B ₁ ={b _ij }, B ₂ ={b _ij };

In order to obtain the importance of the comparison between each index, take two comparisons each time and use the scaling method proposed by Saaty for numericalization, where: b _ij = 1, b _ij = 1/b _ij (i, j = 1 ,2,…,n)

Weighted average comprehensive technical engineer opinion, construct B ^* = {b _ij }, where:

Carry out a consistency test on the constructed matrix, that is, calculate the corrected consistency index: CR=CI/RI, where CI is the consistency index: RI is the average random consistency index obtained by comparison search, and the judgment condition is:

Any revalidation by changing some comparison values when the matrix does not meet the consistency requirements;

When the matrix meets the consistency requirement, then perform step 104;

Step 104, calculating and normalizing the index weights of each layer; the specific steps are:

Calculation of relative weight: that is, the importance of the compared index relative to the elements of the previous layer. This sorting method is called hierarchical single sorting;

When the judgment matrix B ^* passes the consistency check, calculate the product M _i of the elements in each row of the judgment matrix:

Use the square root method to get the weight before normalization

where the vector

Will Perform normalization processing to obtain the weight W, then:

W=(W ₁ ,W ₂ ,...,W _n ) ^T ,

The weight W is the required feature vector, where:

Calculate the composite weight of the third layer: the composite weight refers to the weight of an index relative to the top layer element, which is calculated layer by layer through the sum product method from bottom to top; for the index directly related to the top layer, its composite weight is equal to the relative Weight; otherwise, it is equal to the product of the relative weight of the indicator and the composite weight of its upper-level associated elements;

Step 105, dividing the risk level is to determine the gray class; the specific steps are:

Divide the network risk level into 5 risk levels, that is, 5 gray classes, which are scored by two technical engineers and recorded as the evaluation value E _i (c ₁ , c ₂ , c ₃ , c ₄ , c ₅ , c ₆ ) , the evaluation matrix D=(d _ij ) _n*m is calculated, where:

Step 106, establishing a whitening weight function; the specific method of constructing a whitening weight function is:

There are m indicators, s gray classes, and n objects, and x _ij (i=1,2,…,n; j=1,2,…,m) is the sample value of object i with respect to index j, namely actual value d _ij ;

Assume The whitening weight function called k subclass j index:

① if only Two turning points, the function represents an upper bound measure:

② if only Two turning points, the function represents the lower bound measure;

③If The second and third turning points of coincide, the function represents a moderate measure:

Step 107, calculating the gray clustering coefficient and judging the gray class to which it belongs; the specific steps are:

Suppose η _j (j=1,2,...,m) is the weight of each clustering index, where η _j is W;

Assume is the gray fixed weight clustering coefficient of the object i belonging to the k gray class, then for:

According to the cluster analysis based on the principle of maximization, the object i belongs to the gray class k:

Step 108 , based on the above steps, conduct a second assessment of the threat status of the risk source; the specific method of the second assessment is the same as that of steps 101 to 107 .

Further: third layer weights It is equal to the product of its relative weight W _c and the composite weight of the associated element W _i of layer B:

The advantages and positive effects that the present invention has are:

The patent determines its risk sources and risk status on the basis of analyzing the new airborne network architecture, then evaluates the risk sources based on tomographic analysis and gray clustering methods, and finally conducts a second assessment of its threat status, and synthesizes it twice The assessment optimizes the risk assessment results to prevent the deviation and inaccuracy of a single assessment result.

Description of drawings:

Figure 1 Schematic diagram of the new airborne network architecture;

Fig. 2 Improved AHP-GCM method design process;

Figure 3 New airborne network security risk assessment system 1;

Figure 4 New airborne network security risk assessment system II;

Figure 5 is a schematic diagram of the whitening weight function of the upper measurement limit;

Figure 6 is a schematic diagram of the whitening weight function of the lower measurement limit;

Fig. 7 Schematic diagram of moderate measure whitening weight function.

detailed description

In order to further understand the invention content, characteristics and effects of the present invention, the following examples are given, and detailed descriptions are as follows in conjunction with the accompanying drawings:

Please refer to Figure 1, Figure 1 shows the design pattern of the new airborne network architecture: the aircraft control domain refers to the airborne data network related to flight safety and target systems that require security protection, which connects the core system of avionics, flight control , navigation, power, display, communication and other systems are mainly airborne systems that control the normal flight of the aircraft and provide safety measures.

The information trusted domain is a transitional network between the low security level network and the high security level network. Its main function is to isolate the internal and external networks and ensure the security of network communication from the information open domain to the aircraft control domain.

The information open domain ensures the communication security between the airborne network and the ground public network, and mainly refers to the connection network between the information system and the cabin network domain of the off-board network domain.

The cabin network domain connects the cabin core, in-flight entertainment and external communication systems.

Public network domains are untrusted networks.

Please refer to Figure 2 to Figure 7, an airborne network security risk assessment method based on the improved AHP-GCM, including the following steps:

Step 101, analyzing the risk sources of the new airborne network architecture; the specific method is:

Due to the different security levels of aircraft on-board data network and ground public network, general aircraft is mainly divided into aircraft control domain, airline information service domain, passenger information and in-flight entertainment domain; domain, information open domain, cabin network domain and public network domain.

Since the new airborne network architecture supports the interaction of more network domains, from the previous single interface-oriented simplex bus ARINC429 data transmission structure to the coexistence of ARINC429, ARINC825, and ARINC664, the source of exposed risks has also greatly increased:

Exposed network interface points; network data interaction points between airborne information systems; network data interaction methods between airborne information systems and ground information systems. The threat states also have different specific manifestations;

Step 102, establishing evaluation index systems for risk sources and their threat states; the specific method for establishing the evaluation index system is as follows:

First, establish an evaluation index system for the risk source: the target layer, the highest layer, is the risk assessment target to be achieved; the attribute layer, the second layer, is the risk probability, risk impact, and uncontrollability; the third layer, the program layer, is caused by the risk source Threats: interface points between gateways and network domains, interface points between network domains, data interaction points between software platforms and in-flight entertainment systems, use of PADs and electronic flight bags (EFBs) by flight attendants, and illegal modification of user codes;

Secondly, establish an evaluation index system for the threat state of the risk source: the target layer is the risk assessment goal to be achieved; the attribute layer, that is, the second layer is confidentiality, integrity, and availability; the third layer is the specific performance of the threat state: obtaining information, Tampering with information, using services, refusing services, and escalating privileges for illegal operations.

Step 103, assigning values to index weights and checking consistency; the specific method of assigning values to index weights and checking consistency is as follows:

Invite two technical engineers to assign values to the indicators. The assignment rules are as follows: compare the importance of each indicator in the lower layer relative to the element, and select the appropriate one from the specified scale. value, establish a judgment matrix: B ₁ ={b _ij }, B ₂ ={b _ij }.

In order to obtain the importance of the comparison between each index, take two comparisons each time and use the scaling method proposed by Saaty for numericalization, where: b _ij = 1, b _ij = 1/b _ij (i, j = 1 ,2,…,n)

Weighted average comprehensive technical engineer opinion, construct B ^* = {b _ij }, where:

Consistency check is performed on the constructed matrix, that is, to calculate the corrected consistency index: CR=CI/RI, where CI is the consistency index: RI is the average random consistency index obtained by comparison search, and the judgment condition is:

When the matrix does not meet the consistency requirements, some comparison values need to be changed for re-inspection;

Step 104, calculating the index weights of each layer and performing normalization processing; the specific steps for obtaining the index weights of each layer and performing normalization processing are:

Calculation of relative weight: that is, the importance of the compared index relative to the elements of the previous layer. This sorting method is called hierarchical single sorting;

The calculation problem of hierarchical single ranking can be reduced to the problem of calculating the largest eigenvalue and its eigenvector of the judgment matrix. But in general, the largest eigenvalues and corresponding eigenvectors of the judgment matrix do not require high precision. This paper adopts a simple method of calculating the largest eigenvalue of the judgment matrix and the corresponding eigenvector:

When the judgment matrix B ^* passes the consistency check, calculate the product M _i of the elements in each row of the judgment matrix:

Use the square root method to get the weight before normalization

where the vector

Will Perform normalization processing to obtain the weight W, then:

W=(W ₁ ,W ₂ ,...,W _n ) ^T ,

The weight W is the required feature vector, where:

Calculate the composite weight of the third layer: the composite weight refers to the weight of an indicator relative to the topmost element, which is calculated layer by layer through the sum-product method from bottom to top. For the indicator directly associated with the top layer, its composite weight is equal to the relative weight; otherwise, it is equal to the product of the relative weight of the indicator and the composite weight of its upper layer associated element.

In this paper, the third layer weight It is equal to the product of its relative weight W _c and the composite weight of the associated element W _i of layer B:

Step 105, dividing the risk level is to determine the gray class; the specific steps are:

The network risk level is divided into 5 risk levels, that is, 5 gray categories, which are scored by two technical engineers and recorded as the evaluation value E _i (c ₁ , c ₂ , c ₃ , c ₄ , c ₅ , c ₆ ), calculate the evaluation matrix D=(d _ij ) _n*m , where:

Step 106, establishing a whitening weight function; the specific method of constructing a whitening weight function is:

There are m indicators, s gray classes, and n objects, and x _ij (i=1,2,…,n; j=1,2,…,m) is the sample value of object i with respect to index j, namely actual value d _ij ;

Assume The whitening weight function called k subclass j index:

④If only Two turning points, the function represents an upper bound measure:

⑤ if only Two turning points, the function represents the lower bound measure;

⑥ if The second and third turning points of coincide, the function represents a moderate measure:

Step 107, calculating the gray clustering coefficient and judging the gray class to which it belongs; the specific steps for calculating the gray clustering coefficient and judging the gray class to which it belongs are:

Suppose η _j (j=1,2,...,m) is the weight of each clustering index, in this paper η _j is W;

Assume is the gray fixed weight clustering coefficient of the object i belonging to the k gray class, then for:

According to the cluster analysis based on the principle of maximization, the object i belongs to the gray class k:

Step 108 , based on the above steps, conduct a second assessment of the threat status of the risk source; the specific method of the second assessment is the same as that of steps 101 to 107 .

The specific implementation of the improved AHP-GCM method is as follows:

Step1 Analyze the risk sources of the new airborne network architecture

Analyze and classify the risk sources of the new airborne network architecture, which are mainly divided into: exposed network interface points—the gateway and each network domain, and between each network domain (each network domain does not include the aircraft control domain); Network data interaction points between information systems—passenger interface, software platform provided by modification companies, and in-flight entertainment systems; network data interaction methods between airborne information systems and ground information systems—electronic flight bag, flight attendant operation PAD ; Access rights - illegal users enter the airborne network (modify executable code). Its threat status is mainly divided into five types: obtaining information, tampering with information, using services, denying services, and illegal operations with elevated privileges; the flow chart of the method design for improving AHP-GCM is shown in Figure 2;

Step 2 Establish evaluation index systems for risk sources and their threat status

First, establish an evaluation index system for the risk source: the target layer, the highest layer, is the risk assessment target to be achieved; the attribute layer, the second layer, is the risk probability, risk impact, and uncontrollability; the third layer, the program layer, is caused by the risk source Threats: interface points between gateways and network domains, interface points between network domains, data interaction points between software platforms and in-flight entertainment systems, flight attendants use PADs, electronic flight bags (EFBs), and illegally modify user codes; as shown in Figure 3

Secondly, establish an evaluation index system for the threat state of the risk source: the target layer is the risk assessment goal to be achieved; the attribute layer, that is, the second layer is confidentiality, integrity, and availability; the third layer is the specific performance of the threat state: obtaining information, Illegal operations of tampering with information, using services, refusing services, and escalating privileges; as shown in Figure 2;

Step3 Assign the index weight and check the consistency

The judgment matrix constructed from the opinions of technical engineers is as follows:

The matrix after weighted average is:

Consistency check on matrix B ₁ :

RI＝0.52, CR＝0.05154＜0.1

Then the consistency requirement is satisfied, and similarly, both B ₂ and B ^* are satisfied.

Step4 Calculate the index weights of each layer and normalize them

The product of elements in each row of matrix B ^* is: M ₁ =11.3136, M ₂ =0.0791, M ₃ =1.1150; the weight obtained by the square root method is:

but

Will After normalization, the weight W of the second layer is obtained as:

W=(0.6048, 0.1156, 0.2796) ^T , that is, the weights of b ₁ , b ₂ , and b ₃ are 0.6048, 0.1156, and 0.2796, respectively.

third layer weight is equal to the product of its relative weight W _c and the combined weight of the associated element W _i of layer B, where the relative weight W _c is:

Wc = ( _0.3686 , 0.1469, 0.0778, 0.0778, 0.1420, 0.1869) ^T ;

Then the composite weight of layer C for:

Step5 Divide the risk level to determine the gray class

Divide the network risk level into 5 risk levels, that is, 5 gray classes, the risk levels are low, low, medium, high, high, and the corresponding quantitative values are 1, 2, 3, 4, 5 .

According to the evaluation value scored by the technical engineer, the evaluation matrix D=(d _ij ) _n*m is calculated:

Step6 Establish whitening weight function

There are m indicators, s gray classes, and n objects, that is, in this article, m=6, s=5, n=2; set x _ij (i=1,2,...,n; j=1, 2,...,m) is the sample value of object i with respect to index j, that is, the actual value d _ij ;

Assume It is called the whitening weight function of k subclass j index, and the three whitening weight functions are shown in Figure 5, Figure 6, and Figure 7, and the whitening weight function is established as follows:

Step7 Calculate the gray clustering coefficient and judge the gray class to which it belongs

The weight η _j of each cluster index in this paper is W, then η _j = (0.6048, 0.1156, 0.2796), then the gray fixed weight clustering coefficient for:

Then the clustering vector of technical engineer 1 to the network is: σ ₁ =(0.3952 0.5596 0.0452 0 0), and similarly, the clustering vector of technical engineer 2 to the network is: σ ₂ =(0.5922 0.4078 0.0632 0 0);

According to the principle of maximization for cluster analysis, which is:

k=1, so that it is determined that the gray category, that is, the risk level, of the new airborne network security risk is the first category, that is, "low" risk.

Step8 Based on the above steps, conduct a second assessment of the threat status of the risk source

Perform the analysis and calculation steps from Step 1 to Step 7 in Figure 4 to obtain the composite weight for:

The obtained gray fixed weight clustering vector is: σ ₁ =(0 0.1002 0.2786 0.6212 0),

σ ₂ =(0 0.1002 0.0632 0.5675 0.2732).

According to the principle of maximization for cluster analysis, which is:

Therefore, it is determined that the new type of airborne network security risk belongs to the gray category, that is, the risk level is the fourth category, that is, "higher" risk.

From the above analysis and evaluation, it can be known that the index system is established based on the risk probability, impact, uncontrollability and threats caused by the risk source, and the security risk level of the new airborne network is "low". However, after analyzing the threat status, it is found that the risk level to "higher".

If risk assessors give risk assessment conclusions based only on the former, staff are likely to ignore the serious consequences of these risk factors. Only by combining the threats of risk sources and their threat status to construct two assessment index systems can a more reasonable and comprehensive risk assessment result be given.

The embodiments of the present invention have been described in detail above, but the content described is only a preferred embodiment of the present invention, and cannot be considered as limiting the implementation scope of the present invention. All equivalent changes and improvements made according to the application scope of the present invention shall still belong to the scope covered by the patent of the present invention.

Claims (2)

1. a kind of onboard networks safety risk estimating method based on advanced AHP-GCM, it is characterised in that：Comprise the following steps：

Step 101, analysis onboard networks framework risk source；Specific method is：

The onboard networks framework includes aircraft control domain, information recipient domain, information opening domain, cabin network domain and public network Domain；The data transmission structure towards interface type bus of the onboard networks framework include ARINC429, ARINC825, ARINC664；

The network interface point of the onboard networks framework exposure is included between gateway and each network domains, each network domains；Airborne information Network data exchange point between system includes passenger interface, the software platform that repacking company provides, airborne entertainment system；It is airborne Network data exchange mode between information system and Ground Information System includes EFB, service on buses or trains operation PAD；Access right Limit includes disabled user and enters onboard networks；The threatened status of the onboard networks framework include：Obtain information, distort information, Utilize service, refusal service and lifting authority illegal operation；

Step 102, evaluation index system is established respectively to risk source and its threatened status；Specific method is：

Evaluation index system is set up to risk source first：Destination layer is the risk assessment target to be reached；Attribute layer is that risk is general Rate, venture influence, uncontrollability；Solution layer is the threat that risk source is caused；

Secondly evaluation index system is set up to risk source threatened status：Destination layer is the risk assessment target to be reached；Attribute layer It is confidentiality, integrality, availability；Third layer is the specific manifestation in threatened status；

Step 103, to index weights assignment and check consistency；Specific method is：

Assignment is carried out to index by two technicians, the rule of wherein assignment is：Using a certain element in upper strata as standard, under Each index of layer is compared two-by-two relative to the importance of the element, and suitable value is selected in defined scale, and foundation is sentenced Disconnected matrix：B₁={ b_ij, B₂={ b_ij}；

In order to obtain the significance level compared between each index, two marks for being compared and being proposed using Saaty are taken every time Degree method is quantized, wherein：b_ij=1, b_ij=1/b_ij(i, j=1,2 ..., n)

Weighted average complex art engineer's opinion, constructs B^*={ b_ij, wherein：

Consistency check is carried out to the matrix of construction, that is, calculates the coincident indicator of amendment：CR=CI/RI, wherein CI are consistent Property index：RI searches obtained Aver-age Random Consistency Index for contrast, and Rule of judgment is：

It is any to examine again by changing some fiducial values when matrix does not meet coherence request；

When matrix meets coherence request, then step 104 is performed；

Step 104, ask each layer index weights and normalized；Concretely comprise the following steps：

Calculate relative weighting：Significance level i.e. by Comparative indices relative to last layer element, this sortord is referred to as level Single sequence；

As judgment matrix B^*After consistency check, the product M of each row element of judgment matrix is calculated_i：

<mrow> <msub> <mi>M</mi> <mn>1</mn> </msub> <mo>=</mo> <msubsup> <mo>&amp;Pi;</mo> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>n</mi> </msubsup> <msub> <mi>b</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> </msub> <mo>,</mo> <mi>i</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mn>2</mn> <mo>,</mo> <mn>3</mn> <mo>,</mo> <mo>...</mo> <mo>,</mo> <mi>n</mi> <mo>;</mo> </mrow>

Weight before being normalized with root method

<mrow> <mover> <msub> <mi>W</mi> <mi>i</mi> </msub> <mo>&amp;OverBar;</mo> </mover> <mo>=</mo> <mroot> <msub> <mi>M</mi> <mi>i</mi> </msub> <mi>n</mi> </mroot> <mo>;</mo> </mrow>

It is wherein vectorial

<mrow> <mover> <mi>W</mi> <mo>&amp;OverBar;</mo> </mover> <mo>=</mo> <msup> <mrow> <mo>(</mo> <mover> <msub> <mi>W</mi> <mi>i</mi> </msub> <mo>&amp;OverBar;</mo> </mover> <mo>,</mo> <mover> <msub> <mi>W</mi> <mn>2</mn> </msub> <mo>&amp;OverBar;</mo> </mover> <mo>,</mo> <mn>...</mn> <mo>,</mo> <mover> <msub> <mi>W</mi> <mi>n</mi> </msub> <mo>&amp;OverBar;</mo> </mover> <mo>)</mo> </mrow> <mi>T</mi> </msup> <mo>;</mo> </mrow>

WillIt is normalized, obtains weight W, then：

W=(W₁,W₂,…,W_n)^T,

Weight W is required characteristic vector, wherein：

<mrow> <msub> <mi>W</mi> <mi>i</mi> </msub> <mo>=</mo> <mfrac> <mover> <msub> <mi>W</mi> <mi>i</mi> </msub> <mo>&amp;OverBar;</mo> </mover> <mrow> <msubsup> <mi>&amp;Sigma;</mi> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>n</mi> </msubsup> <mover> <msub> <mi>W</mi> <mi>i</mi> </msub> <mo>&amp;OverBar;</mo> </mover> </mrow> </mfrac> <mo>;</mo> </mrow>

Calculate third layer synthetic weight：Synthetic weight refers to weight of certain index relative to the superiors' element, is logical from bottom to top Cross and area method successively calculates what is obtained；For the index with the superiors direct correlation, its synthetic weight is equal to relative weighting；Otherwise Equal to the product of the relative weighting and the synthetic weight of its upper strata associated element of the index；

Step 105, division risk class are to determine grey class；It is concretely comprised the following steps：

It is 5 risk class by network risks grade classification, i.e., 5 grey classes are given a mark by two technicians, are denoted as commenting Valuation E_i(c₁, c₂, c₃, c₄, c₅, c₆), calculating obtains evaluating matrix D=(d_ij)_n*m, wherein：

<mrow> <msub> <mi>d</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> </msub> <mo>=</mo> <msub> <mi>E</mi> <mi>i</mi> </msub> <mo>*</mo> <msubsup> <mi>W</mi> <msub> <mi>b</mi> <mi>i</mi> </msub> <mo>*</mo> </msubsup> <mo>;</mo> </mrow>

Step 106, set up whitened weight function；Construction whitened weight function specific method be：

Provided with m index, s grey class, n object, if x_ij(i=1,2 ..., n；J=1,2 ..., m) for object i on index j Sample value, i.e. actual value d_ij；

IfThe referred to as whitened weight function of k subclasses j indexs：

If 1.OnlyTwo turning points, then function representation upper measure：

<mrow> <msubsup> <mi>f</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mrow> <mo>(</mo> <msub> <mi>x</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> </msub> <mo>)</mo> </mrow> <mo>=</mo> <mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mn>0</mn> </mtd> <mtd> <mrow> <mi>x</mi> <mo>&amp;Element;</mo> <mo>&amp;lsqb;</mo> <mn>0</mn> <mo>,</mo> <msubsup> <mi>x</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mrow> <mo>(</mo> <mn>1</mn> <mo>)</mo> </mrow> <mo>&amp;rsqb;</mo> </mrow> </mtd> </mtr> <mtr> <mtd> <mfrac> <mrow> <msub> <mi>x</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> </msub> <mo>-</mo> <msubsup> <mi>x</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mrow> <mo>(</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow> <mrow> <msubsup> <mi>x</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mrow> <mo>(</mo> <mn>2</mn> <mo>)</mo> </mrow> <mo>-</mo> <msubsup> <mi>x</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mrow> <mo>(</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow> </mfrac> </mtd> <mtd> <mrow> <mi>x</mi> <mo>&amp;Element;</mo> <mrow> <mo>(</mo> <msubsup> <mi>x</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mo>(</mo> <mn>1</mn> <mo>)</mo> </mrow> <mo>,</mo> <msubsup> <mi>x</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mrow> <mo>(</mo> <mn>2</mn> <mo>)</mo> </mrow> <mo>&amp;rsqb;</mo> </mrow> </mtd> </mtr> <mtr> <mtd> <mn>1</mn> </mtd> <mtd> <mrow> <mi>x</mi> <mo>&amp;Element;</mo> <mrow> <mo>(</mo> <msubsup> <mi>x</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mo>(</mo> <mn>2</mn> <mo>)</mo> <mo>,</mo> <mo>+</mo> <mi>&amp;infin;</mi> <mo>)</mo> </mrow> </mrow> </mtd> </mtr> </mtable> </mfenced> <mo>;</mo> </mrow>

If 2.OnlyTwo turning points, then function representation lower limit estimate；

<mrow> <msubsup> <mi>f</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mrow> <mo>(</mo> <msub> <mi>x</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> </msub> <mo>)</mo> </mrow> <mo>=</mo> <mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mn>1</mn> </mtd> <mtd> <mrow> <mi>x</mi> <mo>&amp;Element;</mo> <mo>&amp;lsqb;</mo> <mn>0</mn> <mo>,</mo> <msubsup> <mi>x</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mrow> <mo>(</mo> <mn>3</mn> <mo>)</mo> </mrow> <mo>&amp;rsqb;</mo> </mrow> </mtd> </mtr> <mtr> <mtd> <mfrac> <mrow> <msubsup> <mi>x</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mrow> <mo>(</mo> <mn>4</mn> <mo>)</mo> </mrow> <mo>-</mo> <msub> <mi>x</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> </msub> </mrow> <mrow> <msubsup> <mi>x</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mrow> <mo>(</mo> <mn>4</mn> <mo>)</mo> </mrow> <mo>-</mo> <msubsup> <mi>x</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mrow> <mo>(</mo> <mn>3</mn> <mo>)</mo> </mrow> </mrow> </mfrac> </mtd> <mtd> <mrow> <mi>x</mi> <mo>&amp;Element;</mo> <mo>(</mo> <mn>1</mn> <mo>,</mo> <mn>2</mn> <mo>&amp;rsqb;</mo> </mrow> </mtd> </mtr> <mtr> <mtd> <mn>0</mn> </mtd> <mtd> <mrow> <mi>x</mi> <mo>&amp;Element;</mo> <mrow> <mo>(</mo> <msubsup> <mi>x</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mo>(</mo> <mn>4</mn> <mo>)</mo> <mo>,</mo> <mo>+</mo> <mi>&amp;infin;</mi> <mo>)</mo> </mrow> </mrow> </mtd> </mtr> </mtable> </mfenced> <mo>;</mo> </mrow>

If 3.Second and the 3rd turning point overlap, then function representation is moderate estimates：

<mrow> <msubsup> <mi>f</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mrow> <mo>(</mo> <msub> <mi>x</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> </msub> <mo>)</mo> </mrow> <mo>=</mo> <mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mfrac> <mrow> <msub> <mi>x</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> </msub> <mo>-</mo> <msubsup> <mi>x</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mrow> <mo>(</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow> <mrow> <msubsup> <mi>x</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mrow> <mo>(</mo> <mn>2</mn> <mo>)</mo> </mrow> <mo>-</mo> <msubsup> <mi>x</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mrow> <mo>(</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow> </mfrac> </mtd> <mtd> <mrow> <mi>x</mi> <mo>&amp;Element;</mo> <mo>&amp;lsqb;</mo> <msubsup> <mi>x</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mrow> <mo>(</mo> <mn>1</mn> <mo>)</mo> </mrow> <mo>,</mo> <msubsup> <mi>x</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mrow> <mo>(</mo> <mn>2</mn> <mo>)</mo> </mrow> <mo>&amp;rsqb;</mo> </mrow> </mtd> </mtr> <mtr> <mtd> <mfrac> <mrow> <msubsup> <mi>x</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mrow> <mo>(</mo> <mn>4</mn> <mo>)</mo> </mrow> <mo>-</mo> <msub> <mi>x</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> </msub> </mrow> <mrow> <msubsup> <mi>x</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mrow> <mo>(</mo> <mn>4</mn> <mo>)</mo> </mrow> <mo>-</mo> <msubsup> <mi>x</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mrow> <mo>(</mo> <mn>2</mn> <mo>)</mo> </mrow> </mrow> </mfrac> </mtd> <mtd> <mrow> <mi>x</mi> <mo>&amp;Element;</mo> <mrow> <mo>(</mo> <msubsup> <mi>x</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mo>(</mo> <mn>2</mn> <mo>)</mo> </mrow> <mo>,</mo> <msubsup> <mi>x</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mrow> <mo>(</mo> <mn>4</mn> <mo>)</mo> </mrow> <mo>&amp;rsqb;</mo> </mrow> </mtd> </mtr> <mtr> <mtd> <mn>0</mn> </mtd> <mtd> <mrow> <mi>x</mi> <mo>&amp;NotElement;</mo> <mo>&amp;lsqb;</mo> <msubsup> <mi>x</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mrow> <mo>(</mo> <mn>1</mn> <mo>)</mo> </mrow> <mo>,</mo> <msubsup> <mi>x</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mrow> <mo>(</mo> <mn>4</mn> <mo>)</mo> </mrow> <mo>&amp;rsqb;</mo> </mrow> </mtd> </mtr> </mtable> </mfenced> <mo>;</mo> </mrow>

Step 107, calculating grey cluster coefficient simultaneously judge affiliated grey class；Concretely comprise the following steps：

If η_j(j=1,2 ..., m) be each clustering target power, wherein η_jAs W；

IfBelong to the grey fixed weight cluster coefficient of the grey classes of k for object i, thenFor：

<mrow> <msubsup> <mi>&amp;sigma;</mi> <mi>i</mi> <mi>k</mi> </msubsup> <mo>=</mo> <msubsup> <mi>&amp;Sigma;</mi> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>m</mi> </msubsup> <msubsup> <mi>f</mi> <mi>j</mi> <mi>k</mi> </msubsup> <mrow> <mo>(</mo> <msub> <mi>x</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> </msub> <mo>)</mo> </mrow> <mo>*</mo> <msub> <mi>&amp;eta;</mi> <mi>j</mi> </msub> <mo>;</mo> </mrow>

Object i is obtained as clustering according to maximization principle and belongs to grey class k：

<mrow> <munder> <mrow> <msubsup> <mi>&amp;sigma;</mi> <mi>i</mi> <mi>k</mi> </msubsup> <mo>=</mo> <mi>m</mi> <mi>a</mi> <mi>x</mi> </mrow> <mrow> <mn>1</mn> <mo>&amp;le;</mo> <mi>k</mi> <mo>&amp;le;</mo> <mi>s</mi> </mrow> </munder> <mo>{</mo> <msubsup> <mi>&amp;sigma;</mi> <mi>i</mi> <mi>k</mi> </msubsup> <mo>}</mo> <mo>;</mo> </mrow>

Step 108, based on above step to risk source threatened status carry out secondary evaluation；The specific method and step of secondary evaluation 101 is identical to step 107.

2. the onboard networks safety risk estimating method based on advanced AHP-GCM according to claim 1, it is characterised in that： Third layer weightEqual to its relative weighting W_cWith B layers of associated element W_iSynthetic weight product：

<mrow> <msubsup> <mi>W</mi> <msub> <mi>b</mi> <mi>i</mi> </msub> <mo>*</mo> </msubsup> <mo>=</mo> <msub> <mi>W</mi> <mi>c</mi> </msub> <mo>*</mo> <msub> <mi>W</mi> <mi>i</mi> </msub> <mo>.</mo> </mrow> 3