Steel frame structure safety evaluation method based on boundary adaptive cloud network model

Abstract

The invention discloses a steel frame structure safety evaluation method based on a boundary adaptive cloud network model. The method comprises the following steps: 1) constructing a steel frame structure safety evaluation system model; 2) determining an interval threshold value of each damage performance index in different structure safety levels, and calculating a characteristic parameter of a forepart-'3En' rule cloud and a characteristic parameter of a forepart-variable fuzzy cloud; 3) carrying out information fusion on the membership degree and the membership degree, and constructing a boundary adaptive cloud network model preceding cloud; 4) calculating consequent cloud feature parameters of the boundary adaptive cloud network model, and establishing a boundary adaptive cloud network model based on weighted D-S fusion; and 5) obtaining a structure monitoring index measured value under the current load step, and inputting the measured value into the boundary adaptive cloud network model based on weighted D-S fusion to obtain a safety evaluation quantitative value of the steel frame structure under the current load step. According to the method, the problem of accurate quantification of the safety degree of the steel frame structure under the condition of multi-state indexes and the problem of uncertainty caused by measurement noise are solved.

Description

Technical Field

[0001] The invention relates to the field of steel frame structure safety evaluation, and in particular to a steel frame structure safety evaluation method based on a boundary adaptive cloud network model. Background Art

[0002] Steel frame structures have the characteristics of high strength, light weight, good seismic performance, recyclability, suitability for modular construction, and ease of industrial production. In recent years, steel frame structures have been widely used and developed rapidly in the construction field. Under the action of reciprocating loads, steel frame structures may suffer fatigue and other damages. The continuous accumulation of structural damage will lead to a gradual decrease in bearing capacity, which will seriously affect the overall safety of the structure. In addition, most steel frame structures are supported by supporting steel columns. The loosening of the constraints of the supporting steel columns may also reduce the stability of the structure, which may easily cause the structure to become unstable and damaged, thereby reducing the safety of the structure, and may even cause serious accidents such as the collapse of the steel frame structure as a whole. Since steel frame structures may age and reduce the bearing capacity of the structure and the stability of the support during service, real-time operation and maintenance monitoring and safety status assessment of building structures are particularly important.

[0003] Structural safety evaluation methods include analytic hierarchy process, grey correlation analysis, fuzzy comprehensive evaluation, reliability analysis, artificial neural network and other methods. The analytic hierarchy process is a multi-criteria comprehensive evaluation method for quantitative analysis of qualitative problems, which can reduce the influence of subjective factors to a certain extent. By logically decomposing the object to be evaluated into multiple levels such as target layer, criterion layer, and scheme layer, the qualitative factors are quantified, and finally a decision is made, so that the evaluation results tend to be more scientific. The grey correlation analysis method analyzes the correlation between a small amount of known data and uses the correlation coefficient matrix as the basis for judgment to judge the similarity between several evaluation object data sequences. The fuzzy comprehensive evaluation method is a specific application method of fuzzy mathematics. Using the membership principle of fuzzy mathematics, the influencing factors of the evaluation object are comprehensively considered, and the fuzzy result set is obtained through fuzzy symbol operation, so that the uncertain information is expressed by quantitative evaluation methods. Reliability method refers to the probability that a structure completes the pre-set functions and requirements under specified time and conditions. For the research on reliability calculation methods, the first-order second-moment method was the earliest and most widely used. With the in-depth research on structural reliability analysis methods, reliability calculation methods such as the second-order second-moment method, the higher-order moment method, and the response surface method were subsequently proposed. Currently, reliability analysis methods are also widely used in the field of structural safety evaluation. Artificial neural network is an advanced intelligent algorithm. Through data training, it can have excellent memory and self-learning abilities. At the same time, it performs well in processing massive data, making up for the problem of insufficient computing power of traditional calculation methods, reducing the influence of human subjective factors, and is suitable for the safety status evaluation of structures. However, since neural networks require a large amount of data for training, if there is not enough training sample data, it is easy to cause it to fail to generalize. In addition, since neural networks activate their behavior through the weights of neurons, the model lacks transparency, making it difficult to determine whether the neural network makes evaluations based on wrong factors.

[0004] In actual engineering applications, there are often factors such as measurement noise, initial defects, and uncertainty in the standard values ​​of evaluation indicators, which makes it necessary to deal with uncertainty in the safety evaluation of structures. The cloud model has the advantage of reducing uncertainty by using its three digital features, and has a good application prospect in the evaluation of the safety status of building structures. However, although the cloud model has certain advantages in dealing with uncertainty problems, its evaluation of the safety status of the structure is usually based on the maximum membership principle to determine the current safety level of the research object, and it is difficult to evaluate the actual safety level of the structure. Summary of the invention

[0005] The purpose of the present invention is to provide a steel frame structure safety assessment method based on a boundary adaptive cloud network model, comprising the following steps:

[0006] 1) Select the safety evaluation index of steel frame structure and construct the damage index system U = {U1, U2, ..., U i}, and determine the fusion weight of each damage index;

[0007] 2) Construct the steel frame structure safety evaluation system model M;

[0008] 3) Determine the interval thresholds of each damage state index at different structural safety levels, and calculate the characteristic parameters of the antecedent-"3En" rule cloud and the characteristic parameters of the antecedent-variable fuzzy cloud; the characteristic parameters of the antecedent-"3En" rule cloud include the expected Ex a,3En , Entropy a,3En , super entropy He a,3En , antecedent-"3En" rule cloud membership μ a,3En ; Antecedent - the characteristic parameters of the variable fuzzy cloud include the variable fuzzy cloud expectation Ex a,VFC 、Entropy a,VFC 、Super entropy He a,VFC , membership degree μ of the antecedent-variable fuzzy cloud a,VFC ;

[0009] 4) Using the weighted DS information fusion method to calculate the membership μ a,3En and membership μ a,VFC Perform information fusion to obtain the boundary adaptive antecedent cloud certainty μ a,BACN , and then construct the boundary adaptive cloud network model frontier cloud;

[0010] 5) Calculate the post-cloud characteristic parameters of the boundary adaptive cloud network model, thereby establishing a qualitative rule base including multiple "IF-THEN" rules, and establishing a boundary adaptive cloud network model based on weighted DS fusion;

[0011] 6) Obtain the measured values ​​of the structural monitoring indicators under the current load step (x1, x2, ···, x i ), and input it into the boundary adaptive cloud network model based on weighted DS fusion to obtain the quantitative value of safety evaluation of the steel frame structure under the current load step.

[0012] Furthermore, the entropy En of the “3En” rule cloud model a,3En =(C R -C L ) / 6, where C L , C R are the left and right boundary values ​​of the threshold interval; for a specific value x i The cloud membership μ a,3En (x i )=exp[-(x i -Ex a,3En ) 2 / (2En i' 2 )], where En i ′ is En a,3En is the expected value and He a,3En A normal random number generated for the variance.

[0013] Furthermore, the structural safety evaluation indicators include strain, deflection, rotation angle, stress, and layer displacement.

[0014] Furthermore, the structural safety assessment system model M is as follows:

[0015]

[0016] Where U i represents the i-th damage behavior index in the structural safety evaluation index system; L j It represents the jth safety level divided by the structural safety assessment model M;

[0017] C j =[C 1j C 2j …C ij ] T , where C ij =(a ij ,b ij ) represents the jth structural safety level threshold interval in the i-th damage behavior index. ij ,b ij It is the lower and upper limits of the safety level.

[0018] Furthermore, the variable fuzzy cloud expects Ex a,VFC As shown below:

[0019]

[0020] In the formula, x t,norm (i,j) represents the standardized training data of the i-th monitoring indicator at the j-th safety level, t represents the training data sequence number, μ t ′(i,j) is the standardized training data x of the ith monitoring indicator at the jth safety level t,norm For the relative membership of the jth security level threshold interval, n t is the number of standardized training data for the i-th monitoring indicator at the j-th safety level;

[0021] Entropy a,VFC As shown below:

[0022]

[0023] Where η R(i,j) is the right limit value of the threshold interval of the i-th monitoring indicator at the j-th safety level obtained by appropriately expanding the softening factor, η L (i,j) is the left limit value of the indicator threshold interval obtained by appropriately expanding the i-th monitoring indicator at the j-th safety level through the softening factor.

[0024] Super entropy He a,VFC As shown below:

[0025] He a,VFC =0.1En a,VFC (i,j) (4)

[0026] Antecedent - Membership degree μ of variable fuzzy cloud a,VFC As shown below:

[0027]

[0028] In the formula, x norm (i,j) is the standardized measured value of the i-th monitoring indicator at the j-th safety level. a ' ,VFC (i,j) is En a,VFC (i,j) is the expectation, He a,VFC is a normal random distribution with a standard deviation of .

[0029] Furthermore, the boundary adaptive antecedent cloud certainty μ a,BACN As shown below:

[0030]

[0031] In the formula, m1(A i ) is the evidence source from the antecedent-“3En” rule cloud, which is the i-th piece of evidence data in the evidence source, m2(B j ) comes from the evidence source in the antecedent-variable fuzzy cloud, which is the jth piece of evidence data in the evidence source;

[0032] Among them, the basic probability assignment m1(A i )’s kth monitoring indicator item m 1k (A i ) is as follows:

[0033]

[0034] In the formula, μ ki,3En Assign m1(A) to the basic probability i )’s cloud membership corresponding to the kth monitoring indicator item; n j Indicates the number of security levels;

[0035] Basic probability assignment m2(B j)’s kth monitoring indicator item m 2k (B j ) is as follows:

[0036]

[0037] Where n j Indicates the number of security levels; μ kj,VFC Assign m2(B j ) corresponds to the cloud membership of the kth monitoring indicator item.

[0038] Furthermore, the boundary adaptive cloud network model post-processing cloud characteristic parameters include the boundary adaptive cloud network model post-processing cloud expectation Ex b,BACN , Boundary Adaptive Cloud Network Model Post-Cloud Entropy En b,BACN 、Super entropy He b,BACN .

[0039] Furthermore, the boundary adaptive cloud network model post-cloud expectation Ex b,BACN As shown below:

[0040]

[0041] Where, Ex b,BACN (j) is the expected cloud of the subsequent component of the j-th safety level score of the structure, s R (j) is the right boundary of the jth safety level scoring interval of the structure, s L (j) is the left boundary of the jth safety level scoring interval of the structure;

[0042] Boundary Adaptive Cloud Network Model Post-processing Cloud Entropy En b,BACN As shown below:

[0043]

[0044] In the formula, En b,BACN (j) is the subsequent cloud entropy of the j-th safety level score of the structure;

[0045] Super entropy He b,BACN As shown below:

[0046] He b,BACN =0.1En b,BACN (11)

[0047] Furthermore, in the boundary adaptive cloud network model based on weighted DS fusion, the output of the post-processing cloud is the safety evaluation quantitative value x b,BACN (i,j), that is:

[0048]

[0049] Where Nb ' ,BACN (i,j) is En b,BACN For the expectation, He b,BACN is a normal random distribution with standard deviation; Ex b,BACN (j) is the expected cloud of the subsequent component of the j-th safety level score of the structure; μ a,BACN

[0050] is the membership degree of the subsequent cloud; norm Ex is the standardized measured value. a,3En is the characteristic parameter of the antecedent-“3En” rule cloud.

[0051] The technical effect of the present invention is undoubted. The present invention combines variable fuzzy cloud and "3En"

[0052] The degree of certainty obtained by the rule normal cloud is fused with DS information, and then the boundary adaptive antecedent cloud in the normal cloud reasoning system is constructed, and a mapping relationship is formed with the "3En" rule consequent cloud based on the safety level score, so as to establish a boundary adaptive cloud network model based on DS fusion. The corresponding multidimensional performance index measured value data is input into the input layer, the boundary adaptive cloud network model is activated, and after reasoning and fusion, the overall fusion safety score output value of the structure in the current health state is finally obtained.

[0053] The present invention utilizes a boundary adaptive cloud network model based on multiple performance indicators to perform health identification of steel frame structures, thereby solving the problem of inaccurate health status safety assessment of steel frame structures when considering multiple performance indicators.

[0054] Under the interference of uncertain factors such as measurement noise, the boundary adaptive cloud network model of the present invention can more accurately determine the safety level of the steel frame structure and has a higher accuracy in identifying the health and safety status. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 is a method flow chart;

[0056] Figure 2 (a)-(c) are the size information, finite element model and measurement point locations of the four-story steel frame;

[0057] Figure 3 Extract overall features of a four-story steel frame structure; Figure 3 (a) is the hysteresis curve; Figure 3 (b) is the skeleton curve; Figure 3 (c) is the secant stiffness degradation curve;

[0058] Figure 4 Safety level classification for four-story steel frame structures.

[0059] Figure 5 It is the overall security evaluation result of the structure based on the boundary adaptive cloud network model. DETAILED DESCRIPTION

[0060] The present invention is further described below in conjunction with the embodiments, but it should not be understood that the above subject matter of the present invention is limited to the following embodiments. Without departing from the above technical ideas of the present invention, various substitutions and changes are made according to the common technical knowledge and customary means in the art, which should all be included in the protection scope of the present invention.

[0061] Embodiment 1:

[0062] See also Figures 1 to 5 , a steel frame structure safety evaluation method based on a boundary adaptive cloud network model, comprising the following steps:

[0063] 1) Select the safety evaluation index of steel frame structure and construct the damage index system U = {U1, U2, ..., U i}, and determine the fusion weight of each damage index;

[0064] 2) Construct the steel frame structure safety evaluation system model M;

[0065] 3) Determine the interval thresholds of each damage state index at different structural safety levels, and calculate the characteristic parameters of the antecedent-"3En" rule cloud and the characteristic parameters of the antecedent-variable fuzzy cloud; the characteristic parameters of the antecedent-"3En" rule cloud include the expected Ex a,3En , Entropy a,3En , super entropy He a,3En , antecedent-"3En" rule cloud membership μ a,3En ; Antecedent - the characteristic parameters of the variable fuzzy cloud include the variable fuzzy cloud expectation Ex a,VFC 、Entropy a,VFC 、Super entropy He a,VFC , membership degree μ of the antecedent-variable fuzzy cloud a,VFC ;

[0066] 4) Using the weighted DS information fusion method to calculate the membership μ a,3En and membership μ a,VFC Perform information fusion to obtain the boundary adaptive antecedent cloud certainty μ a,BACN , and then construct the boundary adaptive cloud network model frontier cloud;

[0067] 5) Calculate the post-cloud characteristic parameters of the boundary adaptive cloud network model, thereby establishing a qualitative rule base including multiple "IF-THEN" rules, and establishing a boundary adaptive cloud network model based on weighted DS fusion;

[0068] 6) Obtain the measured values ​​of the structural monitoring indicators under the current load step (x1, x2, ···, xi ), and input it into the boundary adaptive cloud network model based on weighted DS fusion to obtain the quantitative value of safety evaluation of the steel frame structure under the current load step.

[0069] The structural safety evaluation indicators include strain, deflection, rotation angle, stress, and layer displacement.

[0070] The structural safety assessment system model M is as follows:

[0071]

[0072] Where U i represents the i-th damage behavior index in the structural safety evaluation index system; L j It represents the jth safety level divided by the structural safety assessment model M;

[0073] C j =[C 1j C 2j … C ij ] T , where C ij =(a ij ,b ij ) represents the jth structural safety level threshold interval in the i-th damage behavior index. ij ,b ij It is the lower and upper limits of the safety level.

[0074] Variable Fuzzy Cloud Expectation Ex a,VFC As shown below:

[0075]

[0076] In the formula, x t,norm (i,j) represents the standardized training data of the i-th monitoring indicator at the j-th safety level, t represents the training data sequence number, μ t ′(i,j) is the standardized training data x of the ith monitoring indicator at the jth safety level t,norm For the relative membership of the jth security level threshold interval, n t is the number of standardized training data for the i-th monitoring indicator at the j-th safety level;

[0077] Entropy a,VFC As shown below:

[0078]

[0079] Where η R (i,j) is the right limit value of the threshold interval of the i-th monitoring indicator at the j-th safety level obtained by appropriately expanding the softening factor, η L(i,j) is the left limit value of the indicator threshold interval obtained by appropriately expanding the i-th monitoring indicator at the j-th safety level through the softening factor.

[0080] Super entropy He a,VFC As shown below:

[0081] He a,VFC =0.1En a,VFC (i,j) (4)

[0082] Antecedent - Membership degree μ of variable fuzzy cloud a,VFC As shown below:

[0083]

[0084] In the formula, x norm (i,j) is the standardized measured value of the i-th monitoring indicator at the j-th safety level. a ' ,VFC (i,j) is En a,VFC (i,j) is the expectation, He a,VFC is a normal random distribution with a standard deviation of .

[0085] Boundary adaptive antecedent cloud certainty μ a,BACN As shown below:

[0086]

[0087] In the formula, m1(A i ) is the evidence source from the antecedent-“3En” rule cloud, which is the i-th piece of evidence data in the evidence source, m2(B j ) comes from the evidence source in the antecedent-variable fuzzy cloud, which is the jth piece of evidence data in the evidence source;

[0088] Among them, the basic probability assignment m1(A i )’s kth monitoring indicator item m 1k (A i ) is as follows:

[0089]

[0090] In the formula, μ ki,3En Assign m1(A) to the basic probability i )’s cloud membership corresponding to the kth monitoring indicator item; n j Indicates the number of security levels;

[0091] Basic probability assignment m2(B j )’s kth monitoring indicator item m 2k (B j ) is as follows:

[0092]

[0093] Where n j Indicates the number of security levels; μ kj,VFC Assign m2(B j ) corresponds to the cloud membership of the kth monitoring indicator item.

[0094] The characteristic parameters of the post-process cloud of the boundary adaptive cloud network model include the expected value of the post-process cloud of the boundary adaptive cloud network model Ex b,BACN , Boundary Adaptive Cloud Network Model Post-Cloud Entropy En b,BACN 、Super entropy He b,BACN .

[0095] Boundary Adaptive Cloud Network Model Post-Cloud Expectation Ex b,BACN As shown below:

[0096]

[0097] Where, Ex b,BACN (j) is the expected cloud of the subsequent component of the j-th safety level score of the structure, s R (j) is the right boundary of the jth safety level scoring interval of the structure, s L (j) is the left boundary of the jth safety level scoring interval of the structure;

[0098] Boundary Adaptive Cloud Network Model Post-processing Cloud Entropy En b,BACN As shown below:

[0099]

[0100] In the formula, En b,BACN (j) is the subsequent cloud entropy of the j-th safety level score of the structure;

[0101] Super entropy He b,BACN As shown below:

[0102] He b,BACN =0.1En b,BACN (11)

[0103] In the boundary adaptive cloud network model based on weighted DS fusion, the output of the post-processing cloud is the safety evaluation quantitative value x b,BACN (i,j), that is:

[0104]

[0105] Where N b ' ,BACN (i,j) is En b,BACN For the expectation, He b,BACNis a normal random distribution with standard deviation; Ex b,BACN (j) is the expected cloud of the subsequent component of the j-th safety level score of the structure; μ a,BACN

[0106] is the membership degree of the subsequent cloud; norm Ex is the standardized measured value. a,3En is the characteristic parameter of the antecedent-“3En” rule cloud.

[0107] Embodiment 2:

[0108] A steel frame structure safety assessment method based on a boundary adaptive cloud network model comprises the following steps:

[0109] 1) Select the safety evaluation index of steel frame structure and construct the damage index system U = {U1, U2, ..., U i}, and determine the fusion weight of each damage index;

[0110] 2) Construct the steel frame structure safety evaluation system model M;

[0111] 3) Determine the interval thresholds of each damage state index at different structural safety levels, and calculate the characteristic parameters of the antecedent-"3En" rule cloud and the characteristic parameters of the antecedent-variable fuzzy cloud; the characteristic parameters of the antecedent-"3En" rule cloud include the expected Ex a,3En , Entropy a,3En , super entropy He a,3En , antecedent-"3En" rule cloud membership μ a,3En ; Antecedent - the characteristic parameters of the variable fuzzy cloud include the variable fuzzy cloud expectation Ex a,VFC 、Entropy a,VFC 、Super entropy He a,VFC , membership degree μ of the antecedent-variable fuzzy cloud a,VFC ;

[0112] 4) Using the weighted DS information fusion method to calculate the membership μ a,3En and membership μ a,VFC Perform information fusion to obtain the boundary adaptive antecedent cloud certainty μ a,BACN , and then construct the boundary adaptive cloud network model frontier cloud;

[0113] 5) Calculate the post-cloud characteristic parameters of the boundary adaptive cloud network model, thereby establishing a qualitative rule base including multiple "IF-THEN" rules, and establishing a boundary adaptive cloud network model based on weighted DS fusion;

[0114] 6) Obtain the measured values ​​of the structural monitoring indicators under the current load step (x1, x2, ···, x i), and input it into the boundary adaptive cloud network model based on weighted DS fusion to obtain the quantitative value of safety evaluation of the steel frame structure under the current load step.

[0115] Embodiment 3:

[0116] A steel frame structure safety evaluation method based on a boundary adaptive cloud network model, the technical content of which is the same as that of Example 2, and further, the structural safety evaluation indicators include strain, deflection, rotation angle, stress, and layer displacement.

[0117] Embodiment 4:

[0118] A steel frame structure safety evaluation method based on a boundary adaptive cloud network model, the technical content is the same as any one of embodiments 2-3, and further, the structural safety evaluation system model M is as follows:

[0119]

[0120] Where U i represents the i-th damage behavior index in the structural safety evaluation index system; L j represents the jth safety level divided by the structural safety assessment model M; C j =[C 1j C 2j … C ij ] T , where C ij =(a ij ,b ij ) represents the jth structural safety level threshold interval in the i-th damage behavior index.

[0121] Embodiment 5:

[0122] A steel frame structure safety evaluation method based on a boundary adaptive cloud network model, the technical content is the same as any one of embodiments 2-4, further, the variable fuzzy cloud expectation Ex a,VFC As shown below:

[0123]

[0124] In the formula, x t,norm (i,j) represents the standardized training data of the i-th monitoring indicator at the j-th safety level, t represents the training data sequence number, μ′ t (i,j) is the standardized training data x of the i-th monitoring indicator at the j-th safety level t,norm For the relative membership of the jth security level threshold interval, n t is the number of standardized training data for the i-th monitoring indicator at the j-th safety level;

[0125] Entropy a,VFCAs shown below:

[0126]

[0127] Where η R (i,j) is the right limit value of the threshold interval of the i-th monitoring indicator at the j-th safety level obtained by appropriately expanding the softening factor, η L (i,j) is the left limit value of the indicator threshold interval obtained by appropriately expanding the i-th monitoring indicator at the j-th safety level through the softening factor.

[0128] Super entropy He a,VFC As shown below:

[0129] He a,VFC =0.1En a,VFC (i,j) (4)

[0130] Antecedent - Membership degree μ of variable fuzzy cloud a,VFC As shown below:

[0131]

[0132] In the formula, x norm (i,j) is the standardized measured value of the i-th monitoring indicator at the j-th safety level.

[0133] Embodiment 6:

[0134] A steel frame structure safety evaluation method based on a boundary adaptive cloud network model, the technical content is the same as any one of embodiments 2-5, further, the boundary adaptive antecedent cloud certainty μ a,BACN As shown below:

[0135]

[0136] In the formula, m1(A i ) represents the evidence source from the antecedent-“3En” rule cloud, which is the i-th piece of evidence data in the evidence source, m2(B j ) represents the evidence source from the antecedent-variable fuzzy cloud, which is the jth piece of evidence data in the evidence source;

[0137] Among them, the basic probability assignment m1(A i ) is

[0138]

[0139] Basic probability assignment m2(B j ) is

[0140]

[0141] Where n j Indicates the number of security levels.

[0142] Embodiment 7:

[0143] A steel frame structure safety evaluation method based on a boundary adaptive cloud network model, the technical content is the same as any one of embodiments 2-6, further, the boundary adaptive cloud network model post-cloud feature parameters include the boundary adaptive cloud network model post-cloud expected Ex b,BACN , Boundary Adaptive Cloud Network Model Post-Cloud Entropy En b,BACN 、Super entropy He b,BACN .

[0144] Embodiment 8:

[0145] A steel frame structure safety evaluation method based on a boundary adaptive cloud network model, the technical content is the same as any one of embodiments 2-7, further, the boundary adaptive cloud network model post-cloud expected Ex b,BACN As shown below:

[0146]

[0147] Where, Ex b,BACN (j) is the expected cloud of the subsequent component of the j-th safety level score of the structure, s R (j) is the right boundary of the jth safety level scoring interval of the structure, s L (j) is the left boundary of the jth safety level scoring interval of the structure;

[0148] Boundary Adaptive Cloud Network Model Post-processing Cloud Entropy En b,BACN As shown below:

[0149]

[0150] In the formula, En b,BACN (j) is the subsequent cloud entropy of the j-th safety level score of the structure;

[0151] Super entropy He b,BACN As shown below:

[0152] He b,BACN =0.1En b,BACN (11)

[0153] Embodiment 9:

[0154] A steel frame structure safety assessment method based on a boundary adaptive cloud network model, the technical content is the same as any one of embodiments 2-8, further, in the boundary adaptive cloud network model based on weighted DS fusion, the output of the post-cloud is the safety assessment quantitative value x b,BACN (i,j), that is:

[0155]

[0156] Where N b ' ,BACN (i,j) is En b,BACN For the expectation, He b,BACN is a normal random distribution with standard deviation; Ex b,BACN (j) is the expected cloud of the subsequent component of the j-th safety level score of the structure; μ a,BACN

[0157] is the membership degree of the subsequent cloud; norm is the standardized measured value.

[0158] Embodiment 10:

[0159] A steel frame structure safety evaluation method based on a boundary adaptive cloud network model, the technical content is the same as any one of embodiments 2-9, further, the entropy En of the "3En" rule cloud model a,3En =(C R -C L ) / 6, where C L , C R is the left and right boundary value of the threshold interval, and the super entropy He a,3En =En / 10; for a particular value x i Cloud membership

[0160] μ a,3En (x i )=exp[-(x i -Ex a,3En ) 2 / (2En i ' 2 )], where En i ′ is En a,3En is the expected value and He a,3En A normal random number generated for the variance.

[0161] Embodiment 11:

[0162] A safety assessment method for steel frame structures based on a boundary adaptive cloud network model, the steps are as follows:

[0163] (1) Select structural safety evaluation indicators and construct a damage index system U = {U1, U2, ..., U i}, and determine the fusion weight of each damage index, U represents the damage behavior parameter of structural safety assessment, U1,U2,...,U i Characterize specific evaluation indicators in safety assessment, such as strain, deflection, rotation angle, stress, layer displacement, etc.

[0164] (2) Divide the structural safety level, regard the threshold of each level of the structural safety index as a double constraint space, fully consider the uncertainty of the boundary value of the constraint space and appropriately expand it to become a normal cloud. For the structural safety evaluation model M, its safety state is divided into j levels, i structural damage multi-state indicators are selected, and the relevant structural safety evaluation system model M can be expressed as:

[0165]

[0166] Where U i represents the i-th damage behavior index in the structural safety evaluation index system; L j represents the jth safety level divided by the structural safety assessment model M; C j =[C 1j C 2j …C ij ] T , where C ij =(a ij ,b ij ) represents the jth structural safety level threshold interval in the i-th damage behavior index.

[0167] (3) Determine the interval thresholds of each damage state index at different structural safety levels, and calculate the characteristic parameters of the antecedent-“3En” rule cloud and the antecedent-variable fuzzy cloud: Ex a,3En ,En a,3En , He a,3En , μ a,3En and Ex a,VFC ,En a,VFC , He a,VFC , μ a,VFC .

[0168] The characteristic parameters of the “3En” regular cloud are the traditional eigenvalues, while the variable fuzzy cloud expects Ex a,VFC The regulations are as follows:

[0169]

[0170] In the formula, x t,norm (i,j) represents the standardized training data of the i-th monitoring indicator at the j-th safety level, t represents the training data sequence number, μ t ′(i,j) is the standardized training data x of the ith monitoring indicator at the jth safety level t,norm For the relative membership of the jth security level threshold interval, n t is the number of standardized training data for the i-th monitoring indicator at the j-th security level.

[0171] Entropy a,VFC The calculation is as follows:

[0172]

[0173] Where η R (i,j) is the right limit value of the threshold interval of the i-th monitoring indicator at the j-th safety level obtained by appropriately expanding the softening factor, η L (i,j) is the left limit value of the indicator threshold interval obtained by appropriately expanding the i-th monitoring indicator at the j-th safety level through the softening factor.

[0174] Super entropy He a,VFC Generally, it is entropy En a,VFC 0.1 times of.

[0175] Generate En a,VFC For the expectation, He a,VFC N is a normal random number with standard deviation a ' ,VFC (i, j), and then calculate the membership μ of the antecedent-variable fuzzy cloud a,VFC :

[0176]

[0177] In the formula, x norm (i,j) is the standardized measured value of the i-th monitoring indicator at the j-th safety level.

[0178] (4) Using the weighted DS information fusion method to a,3En and μ a,VFC Perform information fusion to obtain the boundary adaptive antecedent cloud certainty μ a,BACN , construct the boundary adaptive cloud network model frontier cloud.

[0179] Need to be the membership degree μ of the antecedent-“3En” rule cloud a,3En and antecedent-variable fuzzy cloud membership μ a,VFC Perform normalization to determine the basic probability assignment

[0180]

[0181] In the formula, m i (A j ) represents the membership degree μ of the i-th monitoring indicator in the j-th security level antecedent cloud ij The basic probability assignment of

[0182] Perform DS information fusion to obtain a more accurate source of membership evidence.

[0183]

[0184] In the formula, m(μ a,BACN) is the basic probability assignment obtained after the fusion of the evidence sources, that is, the antecedent cloud membership of the boundary adaptive cloud network model μ a,BACN .

[0185] (5) Calculate the post-cloud characteristic parameters of the boundary adaptive cloud network model: Ex b,BACN ,En b,BACN and He b,BACN , thereby establishing a qualitative rule base, using the boundary adaptive antecedent cloud and the 3En rule consequent cloud to establish a qualitative rule base consisting of several "IF-THEN" rules containing qualitative concepts. On this basis, a boundary adaptive cloud network model based on weighted DS fusion is established.

[0186] Boundary Adaptive Cloud Network Model Post-Cloud Expectation Ex b,BACN The calculation is as follows:

[0187]

[0188] Where, Ex b,BACN (j) is the expected cloud of the subsequent component of the j-th safety level score of the structure, s R (j) is the right boundary of the jth safety level scoring interval of the structure, s L (j) is the left boundary of the jth safety level scoring interval of the structure.

[0189] Boundary Adaptive Cloud Network Model Post-processing Cloud Entropy En b,BACN for:

[0190]

[0191] In the formula, En b,BACN (j) is the subsequent cloud entropy of the j-th safety level score of the structure.

[0192] Super entropy He b,BACN Generally take entropy En b,BACN One tenth of.

[0193] Calculate the quantitative value x of the subsequent cloud of the boundary adaptive cloud network model b,BACN When En b,BACN For the expectation, He b,BACN N is a normal random distribution with standard deviation b ' ,BACN (i, j), the safety evaluation quantitative value x output by the post-cloud of the boundary adaptive cloud network model is calculated by the following formula b,BACN (i,j):

[0194]

[0195] (6) Input the measured values ​​of the structural monitoring indicators under a certain load step (x1, x2, ···, xi ), activate the boundary adaptive cloud network model, and after reasoning fusion, finally obtain the overall fusion score of the structure under the current load step.

[0196] Embodiment 12:

[0197] A validation of a safety assessment method for steel frame structures based on a boundary adaptive cloud network model is as follows:

[0198] The verification was conducted by low-cycle reciprocating loading of a single-span four-story steel frame structure with a span of 6m, a first floor height of 4.2m, and a second to fourth floor height of 3.9m. The beams and columns are made of Q235 steel, and the beam cross-section size is

[0199] H300mm×300mm×10mm×15mm, column section size is

[0200] H400mm×300mm×10mm×16mm. In order to improve the calculation efficiency, the finite element beams and columns are all two-dimensional beam units. The grid size of the beam and column units of the finite element model of the four-story steel frame structure is divided into 300mm. Figure 2 As shown. In order to simulate the vertical loads that the steel frame structure is subjected to during actual use, a constant vertical concentrated force is applied to the nodes of the finite element model, and the vertical concentrated force applied to each node of each floor is 400kN. It is calculated that the axial compression ratio of the fourth-floor frame column is 0.15, the axial compression ratio of the third-floor frame column is 0.3, the axial compression ratio of the second-floor frame column is 0.45, and the axial compression ratio of the first-floor frame column is 0.6. In order to simulate the connection between the steel frame structure and the ground, the bottom of the four-story steel frame finite element model is set as a fixed constraint.

[0201] The four-story steel frame structure division and monitoring index measurement point locations are as follows Figure 2 As shown in (c), each substructure includes: beam end strain, column upper end strain, column lower end strain, node rotation angle and layer displacement monitoring indicators.

[0202] The four-story steel frame structure was subjected to low-cycle reciprocating loading, and its hysteresis curve, skeleton curve and secant stiffness curve were extracted, as shown in the figure. Figure 3 As shown. By analyzing the hysteresis curve, it can be seen that the hysteresis curve of the four-story steel frame structure is full and has no obvious pinching phenomenon, indicating that the structure has a good energy dissipation capacity. Figure 3It can be seen that before loading to a displacement of 60mm, the hysteresis curve of the four-story steel frame structure is almost a straight line, and the stiffness has not changed significantly. The skeleton curve is almost a straight line, that is, the structural force is almost proportional to the displacement, and the structure can be considered to be in an elastic stage. When the four-story steel frame structure is loaded with a displacement of more than 60mm, its hysteresis curve is no longer a straight line, but a hysteresis loop containing a certain area. The skeleton curve gradually bends, and the structural stiffness gradually decreases. As the loading progresses, the structural stiffness decreases faster and faster. At the end of the loading, the structural stiffness drops to about 60% of the initial stiffness. It can be seen from the skeleton curve that when the loading displacement is 250mm, the structural bearing capacity reaches the maximum, and then begins to gradually decrease as the loading progresses. When the loading reaches the last load step, the structural bearing capacity drops to about 60% of the maximum structural bearing capacity.

[0203] Before establishing the structural safety assessment model, the structural safety level is divided according to the structural stiffness degradation curve, skeleton curve and structural damage characteristics. Figure 4 As shown in the figure, the four-story steel frame structure is divided into five safety levels: intact state, slight damage, moderate damage, severe damage, and failure, corresponding to level 1, level 2, level 3, level 4, and level 5 respectively. In the figure, the stiffness ratio represents the ratio of the structural secant stiffness of the current load step to the initial (first load step) structural secant stiffness, and the peak load ratio represents the ratio of the peak load of the current loading step of the positive segment of the skeleton curve to the maximum bearing capacity of the structure.

[0204] Using the measured data and the boundary adaptive cloud network model, the safety evaluation results of the steel frame structure under low-cycle reciprocating loading based on the boundary adaptive cloud network model are shown in the figure. Figure 4 shown.

[0205] When the structure is actually at safety level 1, the overall safety score of the structure should be in the range of (80,100]. In the safety evaluation results based on the boundary adaptive cloud network model: the load step 6 score is less than 80, and the evaluation accuracy of the four-story steel frame structure at this safety level is 83%; when the structure is actually at safety level 2, the overall safety score of the structure should be in the range of (60,80]. In the safety evaluation results based on the boundary adaptive cloud network model: the load step 18 score is less than 60, and the evaluation accuracy of the four-story steel frame structure at this safety level is 92%; when the structure is actually at safety level 3, the overall safety score of the structure should be in the range of (40,60]. In the safety evaluation results based on the boundary adaptive cloud network model: The load step scores are all in the range of (40,60], and the evaluation accuracy of the four-story steel frame structure at this safety level is 100%; when the structure is actually at safety level 4, the overall safety score of the structure should be in the range of (20,40]. In the safety evaluation results based on the boundary adaptive cloud network model: the load step scores are all in the range of (20,40], and the evaluation accuracy of the four-story steel frame structure at this safety level is 100%; when the structure is actually at safety level 5, the overall safety score of the structure should be in the range of (0,20]. In the safety evaluation results based on the boundary adaptive cloud network model: the load step scores are all in the range of (0,20], and the evaluation accuracy of the four-story steel frame structure at this safety level is 100%.

[0206] The overall safety evaluation accuracy of the four-story steel frame structure under low-cycle reciprocating loading based on the boundary adaptive cloud network model is shown in Table 1, which also lists the calculation results of the “50% certainty” regular normal cloud network model and the “3En” regular normal cloud network model.

[0207] Table 1 Overall safety evaluation accuracy of four-story steel frame structure

[0208]

[0209]

[0210] It can be seen from Table 1 that the boundary adaptive cloud network model has a strong structural safety evaluation capability, and the evaluation accuracy of each safety level and the evaluation accuracy of the structure as a whole are higher than those of the "3En" rule normal cloud network model and the "50% certainty" rule normal cloud network model, and can effectively evaluate the structural health status. In addition, it can be seen from the overall score of the four-story steel frame structure that compared with the "50% certainty" rule normal cloud network model and the "3En" rule normal cloud network model, which have a high error rate in evaluating the load step at the boundary of the safety level, the boundary adaptive cloud network model proposed in the present invention can better evaluate the safety status of the boundary load step, and has a higher recognition reliability.

Claims

1. A steel frame structure safety assessment method based on a boundary adaptive cloud network model, characterized in that: The following steps are involved: 1) Select the safety evaluation index of steel frame structure and construct the damage index system U = {U1, U2, ..., U i }, and determine the fusion weight of each damage index; 2) Construct the steel frame structure safety evaluation system model M; 3) Determine the interval thresholds of each damage state index at different structural safety levels, and calculate the characteristic parameters of the antecedent-"3En" rule cloud and the characteristic parameters of the antecedent-variable fuzzy cloud; the characteristic parameters of the antecedent-"3En" rule cloud include the expected Ex a,3En , Entropy a,3En , super entropy He a,3En , the antecedent-"3En" rule cloud membership μ a,3En ; Antecedent - the characteristic parameters of the variable fuzzy cloud include the variable fuzzy cloud expectation Ex a,VFC 、Entropy a,VFC 、Super entropy He a,VFC , membership degree μ of the antecedent-variable fuzzy cloud a,VFC ; 4) Using the weighted DS information fusion method to calculate the membership μ a,3En and membership μ a,VFC Perform information fusion to obtain the boundary adaptive antecedent cloud certainty μ a,BACN , and then construct the boundary adaptive cloud network model frontier cloud; 5) Calculate the post-cloud characteristic parameters of the boundary adaptive cloud network model, thereby establishing a qualitative rule base including multiple "IF-THEN" rules, and establishing a boundary adaptive cloud network model based on weighted DS fusion; 6) Obtain the measured values ​​of the structural monitoring indicators under the current load step (x1, x2, ···, x i ), and input it into the boundary adaptive cloud network model based on weighted DS fusion to obtain the quantitative value of safety evaluation of the steel frame structure under the current load step.

2. According to claim 1, a steel frame structure safety assessment method based on a boundary adaptive cloud network model is characterized by: The structural safety evaluation indicators include strain, deflection, rotation angle, stress, and layer displacement.

3. The steel frame structure safety assessment method based on the boundary adaptive cloud network model according to claim 1 is characterized in that: Entropy En of the "3En" rule cloud model a,3En =(C R -C L ) / 6, where C L , C R are the left and right boundary values ​​of the threshold interval; for a specific value x i The cloud membership μ a,3En (x i )=exp[-(x i -Ex a,3En ) 2 / (2En i ' 2 )], where En i ′ is En a,3En is the expected value and He a,3En A normal random number generated for the variance.

4. According to claim 1, a steel frame structure safety assessment method based on a boundary adaptive cloud network model is characterized in that: The structural safety assessment system model M is as follows: Where U i represents the i-th damage behavior index in the structural safety evaluation index system; L j It represents the jth safety level divided by the structural safety assessment model M; C j =[C 1j C 2j …C ij ] T , where C ij =(a ij ,b ij ) represents the jth structural safety level threshold interval in the i-th damage behavior index; a ij ,b ij It is the lower and upper limits of the safety level.

5. The steel frame structure safety assessment method based on the boundary adaptive cloud network model according to claim 1 is characterized in that: Variable Fuzzy Cloud Expectation Ex a,VFC As shown below: In the formula, x t,norm (i,j) represents the standardized training data of the i-th monitoring indicator at the j-th safety level, t represents the training data sequence number, μ t ′(i,j) is the standardized training data x of the ith monitoring indicator at the jth safety level t,norm For the relative membership of the jth security level threshold interval, n t is the number of standardized training data for the i-th monitoring indicator at the j-th safety level; Entropy a,VFC As shown below: Where η R (i,j) is the right limit value of the threshold interval of the i-th monitoring indicator at the j-th safety level obtained by appropriately expanding the softening factor, η L (i,j) is the left limit value of the indicator threshold interval obtained by appropriately expanding the i-th monitoring indicator at the j-th safety level through the softening factor. Super entropy He a,VFC As shown below: He a,VFC =0.1En a,VFC (i,j) (4) Antecedent - Membership degree μ of variable fuzzy cloud a,VFC As shown below: μ a,VFC (i,j)=exp[-(x norm (i,j)-Ex a,VFC ) 2 / (2N a ′ ,VFC (i,j) 2 )] (5) In the formula, x norm (i,j) is the standardized measured value of the i-th monitoring indicator at the j-th safety level; N a ' ,VFC (i,j) is En a,VFC (i,j) is the expectation, He a,VFC is a normal random distribution with a standard deviation of .

6. The steel frame structure safety assessment method based on the boundary adaptive cloud network model according to claim 1 is characterized in that: Boundary adaptive antecedent cloud certainty μ a,BACN As shown below: In the formula, m1(A i ) is the evidence source from the antecedent-"3En" rule cloud, which is the i-th piece of evidence data in the evidence source; m2(B j ) comes from the evidence source in the antecedent-variable fuzzy cloud, which is the jth piece of evidence data in the evidence source; Among them, the basic probability assignment m1(A i )’s kth monitoring indicator item m 1k (A i ) is as follows: In the formula, μ ki,3En Assign m1(A) to the basic probability i )’s cloud membership corresponding to the kth monitoring indicator item; n j Indicates the number of security levels; Basic probability assignment m2(B j )’s kth monitoring indicator item m 2k (B j ) is as follows: Where n j Indicates the number of security levels; μ kj,VFC Assign m2(B j ) corresponds to the cloud membership of the kth monitoring indicator item.

7. The steel frame structure safety assessment method based on the boundary adaptive cloud network model according to claim 1 is characterized in that: The characteristic parameters of the post-process cloud of the boundary adaptive cloud network model include the expected value of the post-process cloud of the boundary adaptive cloud network model Ex b,BACN , Boundary Adaptive Cloud Network Model Post-Cloud Entropy En b,BACN 、Super entropy He b,BACN .

8. A steel frame structure safety assessment method based on a boundary adaptive cloud network model according to claim 7, characterized in that: Boundary Adaptive Cloud Network Model Post-Cloud Expectation Ex b,BACN As shown below: Where, Ex b,BACN (j) is the expected cloud of the subsequent component of the j-th safety level score of the structure, s R (j) is the right boundary of the jth safety level scoring interval of the structure, s L (j) is the left boundary of the jth safety level scoring interval of the structure; Boundary Adaptive Cloud Network Model Post-processing Cloud Entropy En b,BACN As shown below: In the formula, En b,BACN (j) is the subsequent cloud entropy of the j-th safety level score of the structure; Super entropy He b,BACN As shown below:

9. The steel frame structure safety assessment method based on the boundary adaptive cloud network model according to claim 1 is characterized in that: In the boundary adaptive cloud network model based on weighted DS fusion, the output of the post-processing cloud is the safety evaluation quantitative value x b,BACN (i,j), that is: Where N b ' ,BACN (i,j) is En b,BACN For the expectation, He b,BACN is a normal random distribution with standard deviation; Ex b,BACN (j) is the expected cloud of the subsequent component of the j-th safety level score of the structure; μ a,BACN is the membership degree of the subsequent cloud; norm Ex is the standardized measured value; a,3En is the characteristic parameter of the antecedent-"3En" rule cloud.

Images