Dynamic weighting method, system and storage medium for multi-level indicators of bridge system status

Abstract

The present application discloses a method, system and storage medium for dynamically weighting multi-level indicators of bridge system status, including the following contents: constructing or calling an architecture model of a multi-level indicator system of bridge system status; at the initial moment, determining the initial weight value of the multi-level indicator of bridge system status; at other moments, based on the weight value of each level indicator at the previous moment and the bottom level indicator values ​​obtained multiple times from the previous moment to the current moment, calculating all other indicator values ​​layer by layer upward, establishing a Bayesian multivariate linear regression model of multiple parallel indicator values ​​with subordinate relationship of each level indicator value to the next level, normalizing the regression coefficient of non-constant terms in the regression model, calculating the average value of the normalized regression coefficient and the initial weight, and normalizing the average value again as the current moment weight of each level indicator; repeating the weight update process until the required moment is reached. The present application can realize the automatic dynamic weighting of multi-level indicators of bridge system status.

Description

Technical Field

[0001] The present application belongs to the field of bridge structure monitoring, detection and evaluation, and specifically relates to a dynamic weighting method, system and storage medium for multi-level indicators of bridge system status. Background Art

[0002] Large infrastructure such as bridges is a complex physical system with high redundancy. It is difficult for a single or single-level indicator to fully reflect certain comprehensive conditions of bridges. At present, a large number of studies have proposed to apply the analytic hierarchy process to the evaluation of the comprehensive condition of bridges; the document "Research on Safety Condition Assessment Method of High-speed Railway Steel Truss Arch Bridge, Zhao Hanwei, Southeast University, 2019, pp. 103-112" is based on a hierarchical analysis structure model, and uses the analytic hierarchy process to determine the weight value of each single indicator in the model, so as to grade and evaluate the overall service condition of the bridge structure.

[0003] The analytic hierarchy process can treat the bridge as a system and make decisions through decomposition, comparative judgment, and comprehensive thinking, making the evaluation process more comprehensive and scientific; at the same time, the method combines qualitative and quantitative methods to decompose complex problems into multiple levels, which helps to systematically consider and evaluate factors at all levels, and by determining the specific weights of each evaluation indicator, the evaluation process is more transparent and explainable. ‌However, the process of determining the weights of the multi-level indicators of the state of structures or physical systems to date has heavily relied on expert experience, and most of them are determined once, that is, once the weights are determined, they will generally not be changed again, which actually violates the objective fact that the state of the bridge system is in a long-term and slow change during the operation period.

[0004] The advent of the era of big data and artificial intelligence has laid a data foundation for the dynamic evaluation of bridge systems, making it possible to dynamically adjust the weights of multi-level indicators of bridge system status based on long-term monitoring and regular inspection data. Summary of the invention

[0005] Purpose of the invention: In view of the deficiency of the prior art that it is impossible to scientifically and dynamically assign weights to various multi-level indicators of the bridge system status, the present application proposes a method, system and storage medium for dynamic weighting of multi-level indicators of the bridge system status, which realizes the dynamic weighting of multi-level indicators of the bridge system status and facilitates the dynamic characterization of the real-time status of the bridge throughout its life cycle.

[0006] In a first aspect, the present application provides a method for dynamically weighting multi-level indicators of a bridge system state, comprising the following steps:

[0007] S1, construct or call an architecture model including a multi-level indicator system of a bridge system state, wherein the architecture model includes multiple levels with subordinate relationships; each level except the bottom level includes one or more indicators, each indicator corresponds to one or more parallel subordinate indicators downward, and multiple subordinate indicators corresponding to the same indicator constitute a parallel indicator group;

[0008] S2, define t0, t1, …, t m is each moment during the operation period of the bridge, where t0 is the initial moment, and m≥1; at the moment t0, the initial weight value of each level indicator in the architecture model is determined;

[0009] S3, at t m time, based on t m-1 to m The lowest level indicator values ​​and t in the architecture model collected or calculated multiple times during the time period m-1 The weight values ​​of indicators at each level at the time are used to calculate the indicator value data sets corresponding to all other indicators layer by layer. The indicator value data sets include N groups of indicator values, where N represents t m-1 to m The number of times the lowest level indicator values ​​are collected or calculated during a period of time;

[0010] S4, establish t m The Bayesian multiple linear regression dynamic model of the index values ​​of each level at the moment is related to the multiple parallel index values ​​of the next level with subordinate relations. The constant term regression coefficient in the Bayesian multiple linear regression dynamic model is determined, and the normalized non-constant term regression coefficient is summed and normalized to obtain the normalized non-constant term regression coefficient. The average value of the normalized non-constant term regression coefficient and the initial weight is calculated, and then the average value result is summed and normalized as the index of each level at t m The weight value of the moment;

[0011] S5, in the process of increasing the time forward, repeat steps S3 to S4 to calculate the dynamic weights of the indicators at each level at the current or required time.

[0012] In a possible implementation of the first aspect, step S2 uses a hierarchical analysis method and a geometric mean method to determine the initial weight value of each level indicator in the architecture model; specifically includes the following contents:

[0013] S21, at time t0, for the n parallel indicators in the parallel indicator group, the relative importance judgment matrix between the two indicators is defined as A=(a ij ) n×n :

[0014] ;

[0015] The element a of the relative importance judgment matrix A ij Satisfy the following properties:

[0016] ;

[0017] ;

[0018] ;

[0019] Among them, a ij Indicates the importance value of the i-th indicator relative to the j-th indicator, a ji Indicates the importance of the jth indicator relative to the ith indicator, a ij with a ji are reciprocals of each other, i, j = 1, 2, ..., n; the importance value is determined according to the scaling principle;

[0020] S22, calculate the product M of each row element in the relative importance judgment matrix A i :

[0021] ;

[0022] Calculate M i The nth root b i :

[0023] ;

[0024] b1, b2, ..., b n Perform sum normalization to obtain the initial weights of n parallel indicators :

[0025] .

[0026] In a possible implementation manner of the first aspect, step S21 determines the importance value using the following scaling principle:

[0027] When two indicators are equally important, the importance value is 1;

[0028] When the i-th indicator is slightly more important than the j-th indicator, the importance value is 3;

[0029] When the i-th indicator is significantly more important than the j-th indicator, the importance value is 5;

[0030] When the i-th indicator is strongly more important than the j-th indicator, the importance value is 7;

[0031] When the i-th indicator is extremely more important than the j-th indicator, the importance value is 9.

[0032] Furthermore, when the relative importance of the i-th indicator and the j-th indicator is between “equally important, slightly important, obviously important, strongly important, and extremely important”, the importance values ​​are 2, 4, 6, and 8 respectively.

[0033] Furthermore, when the relative importance of the ith indicator and the jth indicator is between "equally important, slightly important, and obviously important", the importance values ​​also include 7 / 5, 3 / 2, 5 / 3, 5 / 2, and 7 / 2.

[0034] In a possible implementation of the first aspect, the number of times step S3 collects or measures the values ​​of the various indicators at the bottom layer is greater than the number of indicators in any parallel indicator group. Preferably, the value range of the various indicator values ​​at the bottom layer is [0, 1].

[0035] In a possible implementation manner of the first aspect, step S4 specifically includes the following content:

[0036] S41, based on the bottom-level indicator values ​​collected or calculated multiple times in step S3 and the indicator value data set corresponding to all other indicators calculated layer by layer upward, establish t m The Bayesian multivariate linear regression dynamic model of multiple parallel index values ​​with subordinate relationships at the next level except the bottom level at each level:

[0037] ;

[0038] in, Indicates t m Any set of n-dimensional parallel index values ​​of any level collected or measured at any time, which includes the 1st, 2nd, ..., nth dimensional index value vectors, are respectively: , , …, ; Indicates the index value to which the n-dimensional parallel index value belongs; , , …, represents the non-constant regression coefficient, represents the regression coefficient of the constant term;

[0039] The non-constant regression coefficients are summed and normalized as follows:

[0040] ;

[0041] in, It represents the normalized non-constant regression coefficient;

[0042] S42, calculate the average of the normalized non-constant regression coefficient and the initial weight:

[0043] ;

[0044] S43, sum and normalize the average values ​​as the indicators of each level at t m The weight of the moment:

[0045] ;

[0046] in, Represents the initial weight of the kth indicator in the parallel indicator group at the initial moment; Indicates t m The unnormalized weight of the kth indicator in the parallel indicator group after being corrected by data regression at any moment; Indicates t m The weight of the kth indicator in the parallel indicator group after being corrected by data regression at any moment.

[0047] In a second aspect, the present application provides a system for executing the dynamic weighting method, comprising:

[0048] Multi-level indicator architecture model for bridge system status;

[0049] An initial weight value calculation unit, used to determine the initial weight value of each level indicator in the architecture model at time t0;

[0050] Index value calculation unit, at t m time, based on t m-1 to m The lowest level indicator values ​​and t in the architecture model collected or calculated multiple times during the time period m-1 The weight values ​​of indicators at each level at the time are used to calculate the indicator value data sets corresponding to all other indicators layer by layer. The indicator value data sets include N groups of indicator values, where N represents t m-1 to m The number of times the lowest level indicator values ​​are collected or calculated during a period of time;

[0051] Weight value update unit, used to establish t m The Bayesian multiple linear regression dynamic model of the index values ​​of each level at the moment is related to the multiple parallel index values ​​of the next level with subordinate relations. The constant term regression coefficient in the Bayesian multiple linear regression dynamic model is determined, and the normalized non-constant term regression coefficient is summed and normalized to obtain the normalized non-constant term regression coefficient. The average value of the normalized non-constant term regression coefficient and the initial weight is calculated, and then the average value result is summed and normalized as the index of each level at t m The weight value of the moment;

[0052] The dynamic weighting unit is electrically / communicatively connected to the index value calculation unit and the weight value update unit, respectively, and controls the index value calculation unit and the weight value update unit to calculate the dynamic weights of the indicators at each level at the current or required moment as time increases.

[0053] In a third aspect, the present application provides a computer-readable storage medium having computer program instructions stored thereon, wherein the computer program instructions implement the dynamic weighting method when executed by a processor.

[0054] Compared with the prior art, this application has the following beneficial effects:

[0055] Based on the classic analytic hierarchy process, this application introduces a weight update mechanism of Bayesian multivariate linear regression driven by online monitoring and detection data. It can effectively, scientifically and theoretically realize the dynamic weighting of multi-level indicators of bridge system status, realize the dynamic characterization and evaluation of the time-varying state of bridges during operation, and help improve the digital and intelligent operation and maintenance level of bridge infrastructure. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 A flowchart of a method for dynamically weighting multi-level indicators of a bridge system state provided in an embodiment of the present application;

[0057] Figure 2 An architectural model including a multi-level indicator system for bridge system status and a parallel indicator group (within a dotted box) involved in the embodiment of the present application are provided. DETAILED DESCRIPTION

[0058] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings.

[0059] It should be clear that the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0060] The terms used in the embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit the present application. The singular forms "a", "said" and "the" used in the embodiments of the present application are also intended to include plural forms, unless the context clearly indicates other meanings.

[0061] like Figure 1 As shown in FIG, a dynamic weighting method for multi-level indicators of bridge system status mainly includes the following steps:

[0062] S1, determine the architecture model of the multi-level indicator system of the bridge system status, the architecture model is composed of multiple layers of indicators, each level (except the bottom level) contains one or more indicators, each indicator (except the bottom level) corresponds to one or more subordinate indicators, when there are multiple subordinate indicators, these subordinate indicators constitute a parallel indicator group; the architecture model defines t0, t1, ..., t m , ... etc. belong to various moments in the bridge operation period, and t0 is the initial moment;

[0063] Optionally, the architecture model is constructed based on professional knowledge or directly acquired and called based on existing materials.

[0064] S2, at time t0, determining the initial weight values ​​of indicators at each level in the architecture model;

[0065] In a possible implementation, for each level of parallel indicator group, the relative importance of each pair of indicators in the parallel indicator group is judged, and these subjective relative importances are numerically quantified using appropriate scaling principles, and further assembled into a judgment matrix; specifically, for a parallel indicator group containing n parallel indicators, the relative importance judgment matrix between each pair of indicators is defined as A=(a ij ) n×n :

[0066] ;

[0067] Among them, a ij Indicates the importance value of the i-th indicator relative to the j-th indicator, a ji Indicates the importance of the jth indicator relative to the ith indicator, a ij with a ji They are reciprocals of each other, i, j = 1, 2, ..., n; importance values ​​can be scaled according to specific rules;

[0068] The relative importance judgment matrix A satisfies the following properties:

[0069] ;

[0070] ;

[0071] ;

[0072] It is preferred to use the scaling principle shown in Table 1 to assign values ​​to the relative importance, as follows:

[0073] Table 1a ij Scaling principle

[0074]

[0075] Columns 2 to 6 in the above table are only general recommended value strategies for relative importance judgment. In practical applications, those skilled in the art can further subdivide the importance values ​​according to actual conditions and interpolate them into decimals, such as shown in column 7;

[0076] The geometric mean method (i.e., square root method) is used to calculate the initial weight of each parallel index. The process is as follows:

[0077] Calculate the product M of each row index in the relative importance judgment matrix A i :

[0078] ;

[0079] Calculate M i The nth root b i :

[0080] ;

[0081] b i Perform sum normalization:

[0082] ;

[0083] The obtained It refers to the initial weights of the required parallel indicator groups, that is, the initial weight values ​​of indicators at each level.

[0084] S3, at t m time, based on t m-1 to m The lowest level index values ​​and t obtained by multiple collection or measurement during the time period m-1 The weight values ​​of indicators at each level determined at the moment are used to calculate all other indicator values ​​layer by layer to form an indicator value data set. The indicator value data set includes N groups of indicator values, where N represents t m-1 to m The number of times the values ​​of each indicator at the lowest level are collected or calculated during a certain period of time; preferably, the value range of each indicator value at the lowest level is recommended to be [0,1] to ensure the accuracy of subsequent regression calculations; and the number of multiple collections or calculations is recommended to be greater than the number of indicators in any parallel indicator group in the multi-level indicator system architecture model.

[0085] S4, based on the indicator value data sets at each level, establish t m The Bayesian multiple linear regression dynamic model of the index values ​​of each level (except the bottom level) at the moment of time with respect to the multiple parallel index values ​​of the next level with subordinate relationship is used to determine the index values ​​of each level at t m The weight value at the moment; the specific process is as follows:

[0086] The expression of the Bayesian multivariate linear regression dynamic model for defining any level and any group of parallel indicators is as follows:

[0087] ;

[0088] in, Indicates t m Any group of n-dimensional parallel index values ​​at any level collected or measured at any time, , , …, for 1st, 2nd, …, nth dimension index value vector; Indicates the index value to which the n-dimensional parallel index value belongs; , , …, is the non-constant regression coefficient of the model, is the constant term regression coefficient of the model (i.e., intercept);

[0089] Normalize the sum of the non-constant regression coefficients as follows:

[0090] ;

[0091] in, It represents the normalized non-constant regression coefficient;

[0092] Calculate the average of the normalized non-constant regression coefficient and the initial weight, then sum and normalize the average results again, and finally use this as the index of each level at t m The dynamic weights at each moment are as follows:

[0093] ;

[0094] ;

[0095] in, is the weight of the kth indicator in the group of parallel indicators at the initial moment; Indicates t m The unnormalized weight of the kth indicator in the group of parallel indicators after being corrected by data regression at any moment; Indicates t m The weight of the kth indicator in the group of parallel indicators after being corrected by data regression at any moment.

[0096] S5, repeat S3~S4 (increasing in time forward) until the dynamic weights of indicators at each level at the current or required time are calculated.

[0097] In a second aspect, the present application provides a system for executing the dynamic weighting method, comprising:

[0098] Multi-level indicator architecture model for bridge system status;

[0099] An initial weight value calculation unit, used to determine the initial weight value of each level indicator in the architecture model at time t0;

[0100] Index value calculation unit, at t m time, based on t m-1 to m The lowest level indicator values ​​and t in the architecture model collected or calculated multiple times during the time period m-1 The weight values ​​of indicators at each level at the time are used to calculate the indicator value data sets corresponding to all other indicators layer by layer. The indicator value data sets include N groups of indicator values, where N represents t m-1 to m The number of times the lowest level indicator values ​​are collected or calculated during a period of time;

[0101] Weight value update unit, used to establish t m The Bayesian multiple linear regression dynamic model of the index values ​​of each level at the moment is related to the multiple parallel index values ​​of the next level with subordinate relations. The constant term regression coefficient in the Bayesian multiple linear regression dynamic model is determined, and the normalized non-constant term regression coefficient is summed and normalized to obtain the normalized non-constant term regression coefficient. The average value of the normalized non-constant term regression coefficient and the initial weight is calculated, and then the average value result is summed and normalized as the index of each level at t m The weight value of the moment;

[0102] The dynamic weighting unit is electrically / communicatively connected to the index value calculation unit and the weight value update unit, respectively, and controls the index value calculation unit and the weight value update unit to calculate the dynamic weights of the indicators at each level at the current or required moment as time increases.

[0103] It should be understood that the division of the processing units in the above system is only a division of logical functions, and in actual implementation, they can be fully or partially integrated into one physical entity, or they can be physically separated. In addition, the processing units in the system can be implemented in the form of a processor calling software; for example, the system includes a processor, the processor is connected to a memory, and instructions are stored in the memory. The processor calls the instructions stored in the memory to implement any of the above methods or implement the functions of the processing units of the system, wherein the processor is a general-purpose processor, such as a central processing unit or a microprocessor, and the memory is a memory within the system or a memory outside the system.

[0104] In a third aspect, the present application further provides a computer-readable storage medium on which computer program instructions are stored, and the computer program instructions implement the method in the above embodiment when executed by a processor.

[0105] For example, the above can be achieved Figure 1 The steps of the dynamic weighting method shown in .

[0106] Computer readable storage media can be tangible devices that can hold and store instructions used by instruction execution devices. Computer readable storage media can be, for example, but not limited to, electronic storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination of the above.

[0107] The computer program instructions used to perform the operations of the present application may be source code or object code written in any combination of one or more programming languages, and the computer-readable program instructions may be executed entirely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server.

[0108] The following takes the service status of a large-span steel truss arch bridge in Jiangsu Province as an example to illustrate the specific implementation process of this application.

[0109] (1) Based on the existing data, the architecture model of the multi-level indicator system for the service status of the bridge system is obtained (e.g. Figure 2 As shown in Figure 2), the architecture model is divided into three levels.

[0110] (2) At time t0, Figure 2 Taking the second-level indicator {safety} in the dashed box and the eight parallel three-level indicators {main beam bearing capacity, main arch truss bearing capacity, support bearing capacity, expansion joint device bearing capacity, main beam fatigue, main arch truss fatigue, hanger fatigue, expansion joint device fatigue} with subordinate relationship as an example, the relative importance between the two indicators is judged. In this embodiment, the relative importance is assigned using the scaling principle shown in Table 1, and the importance values ​​of the eight parallel three-level indicators are obtained as follows:

[0111] Table 2 Importance values ​​of eight parallel three-level indicators in this embodiment

[0112]

[0113] The initial weight vector of the eight parallel three-level indicators calculated by the average geometry method is {0.2842, 0.2842, 0.1421, 0.0947, 0.0568, 0.0406, 0.0406, 0.0568}.

[0114] The process of determining the initial weights of other indicators is similar, and the initial weight values ​​of the multi-level indicators of the service status of the bridge system are obtained as shown in Table 3.

[0115] Table 3 Initial weights of multi-level indicators of bridge system status in this example

[0116]

[0117] (3) At time t1, based on the bottom-level indicator values ​​obtained by 10 calculations from time t1 to time t0 and the initial weight values ​​of the indicators at each level determined at time t0 (initial time), 10 groups of indicator values ​​of all other indicators are calculated layer by layer; where the number of calculations 10 is greater than the maximum number of elements of all indicator groups with parallel relationships in the multi-level indicator system architecture model (such as Figure 2 The number of elements in each parallel index group is 2, 8, and 3 respectively, and the maximum value is 8).

[0118] (4) Based on the 10 indicator values ​​of each level obtained in step (3), a Bayesian multiple linear regression dynamic model is established for the indicator values ​​of each level (except the bottom level) at time t1 with respect to multiple parallel indicator values ​​with subordinate relationships at the next level;

[0119] Taking the second-level indicator {safety} and its subordinate eight parallel three-level indicators in step (2) as an example, firstly, the initial prior model of its Bayesian multivariate linear regression dynamic model at the current moment is constructed based on the second-level parallel indicator vector and the weight value of the parallel indicator value at the previous moment (t0):

[0120] y0=0.2842 +0.2842 +0.1421 +0.0947 +0.0568 +0.0406 +0.0406 +0.0568 ;

[0121] Since 10 measurements were made from t1 to t0, a total of 10 groups of index values ​​were obtained. The 8 parallel index vectors were assembled into a 10-by-8 matrix, and the posterior model of the Bayesian multivariate linear regression dynamic model at the current moment was estimated:

[0122] y = 0.0352 + 0.2835 + 0.2534 + 0.1491 + 0.0796 + 0.0604 +0.0031 + 0.0444 + 0.0898 ;

[0123] The non-constant regression coefficients are summed and normalized to {0.2839, 0.2538, 0.1493, 0.0797, 0.0605, 0.0031, 0.0445, 0.0899};

[0124] The normalized non-constant regression coefficients {0.2839, 0.2538, 0.1493, 0.0797, 0.0605, 0.0031, 0.0445, 0.0899} and the corresponding initial weights {0.2842, 0.2842, 0.1421, 0.0947, 0.0568, 0.0406, 0.0406, 0.0568} are averaged, and then the average results are summed and normalized to obtain {0.2892, 0.2738, 0.1483, 0.0888, 0.0597, 0.0222, 0.0433, 0.0747}, that is, the dynamic weights of the three-level indicators at time t1 are obtained; the above values ​​are calculated by Matlab and taken to four decimal places;

[0125] The process of determining the dynamic weights of other indicators at time t1 is similar.

[0126] (5) Repeat steps (3) to (4) in ascending order until the dynamic weights of indicators at each level at the desired time are calculated.

[0127] The above embodiments prove that the present application realizes real-time dynamic update of weights through analytic hierarchy process and data-driven Bayesian multivariate linear regression, and the weight update mechanism is feasible and effective.

[0128] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A dynamic weighting method for multi-level indicators of bridge system status, characterized in that: The steps include: S1, construct or call an architecture model including a multi-level indicator system of a bridge system state, wherein the architecture model includes multiple levels with subordinate relationships; each level except the bottom level includes one or more indicators, each indicator corresponds to one or more parallel subordinate indicators downward, and multiple subordinate indicators corresponding to the same indicator constitute a parallel indicator group; S2, define t0, t1, …, t m is each moment during the operation period of the bridge, where t0 is the initial moment, and m≥1; at the moment t0, the initial weight value of each level indicator in the architecture model is determined; S3, at t m time, based on t m-1 to m The lowest level indicator values ​​and t in the architecture model collected or calculated multiple times during the time period m-1 The weight values ​​of indicators at each level at the time are used to calculate the indicator value data sets corresponding to all other indicators layer by layer. The indicator value data sets include N groups of indicator values, where N represents t m-1 to m The number of times the lowest level indicator values ​​are collected or calculated during a period of time; S4, establish t m The Bayesian multiple linear regression dynamic model of the index values ​​of each level at the moment is related to the multiple parallel index values ​​of the next level with subordinate relations. The constant term regression coefficient in the Bayesian multiple linear regression dynamic model is determined, and the normalized non-constant term regression coefficient is summed and normalized to obtain the normalized non-constant term regression coefficient. The average value of the normalized non-constant term regression coefficient and the initial weight is calculated, and then the average value result is summed and normalized as the index of each level at t m The weight value of the moment; S5, in the process of increasing the time forward, repeat steps S3 to S4 to calculate the dynamic weights of the indicators at each level at the current or required time.

2. The dynamic weighting method for multi-level indicators of bridge system status according to claim 1 is characterized in that: In step S2, the initial weight values ​​of the indicators at each level in the architecture model are determined by using the hierarchical analysis method and the geometric mean method; specifically, the following contents are included: S21, at time t0, for the n parallel indicators in the parallel indicator group, the relative importance judgment matrix between the two indicators is defined as A=(a ij ) n×n : ； The element a of the relative importance judgment matrix A ij Satisfy the following properties: ； ； ； Among them, a ij Indicates the importance value of the i-th indicator relative to the j-th indicator, a ji Indicates the importance of the jth indicator relative to the ith indicator, a ij with a ji are reciprocals of each other, i, j = 1, 2, ..., n; the importance value is determined according to the scaling principle; S22, calculate the product M of each row element in the relative importance judgment matrix A i : ； Calculate M i The nth root b i : ； b1, b2, …, b n Perform sum normalization to obtain the initial weights of n parallel indicators : 。 3. The dynamic weighting method for multi-level indicators of bridge system status according to claim 2 is characterized in that: In step S21, the scaling principle of the importance value is as follows: When two indicators are equally important, the importance value is 1; When the i-th indicator is slightly more important than the j-th indicator, the importance value is 3; When the i-th indicator is significantly more important than the j-th indicator, the importance value is 5; When the i-th indicator is strongly more important than the j-th indicator, the importance value is 7; When the i-th indicator is extremely more important than the j-th indicator, the importance value is 9.

4. The dynamic weighting method for multi-level indicators of bridge system status according to claim 3 is characterized in that: When the relative importance of the ith indicator and the jth indicator is between "equally important, slightly important, obviously important, strongly important, and extremely important", the importance values ​​are 2, 4, 6, and 8 respectively.

5. The dynamic weighting method for multi-level indicators of bridge system status according to claim 4 is characterized in that: When the relative importance of the ith indicator and the jth indicator is between "equally important, slightly important, and obviously important", the importance values ​​also include 7 / 5, 3 / 2, 5 / 3, 5 / 2, and 7 / 2.

6. The dynamic weighting method for multi-level indicators of bridge system status according to claim 1 is characterized in that: In step S3, the number of times of collecting or calculating the values ​​of each indicator at the bottom level is greater than the number of indicators in any parallel indicator group.

7. The dynamic weighting method for multi-level indicators of bridge system status according to claim 1 is characterized in that: In step S3, the value range of each indicator value at the bottom layer is [0, 1].

8. The dynamic weighting method for multi-level indicators of bridge system status according to claim 1 is characterized in that: Step S4 specifically includes the following contents: S41, based on the bottom-level indicator values ​​collected or calculated multiple times in step S3 and the indicator value data set corresponding to all other indicators calculated layer by layer upward, establish t m The Bayesian multivariate linear regression dynamic model of multiple parallel index values ​​with subordinate relationships at the next level except the bottom level at each level: ； in, Indicates t m Any set of n-dimensional parallel index values ​​of any level collected or measured at any time, which includes the 1st, 2nd, ..., nth dimensional index value vectors, are respectively: , , …, ; Indicates the index value to which the n-dimensional parallel index value belongs; , , …, represents the non-constant regression coefficient, represents the regression coefficient of the constant term; The non-constant regression coefficients are summed and normalized as follows: ； in, It represents the normalized non-constant regression coefficient; S42, calculate the average of the normalized non-constant regression coefficient and the initial weight: ； S43, sum and normalize the average values ​​as the indicators of each level at t m The weight of the moment: ； in, Represents the initial weight of the kth indicator in the parallel indicator group at the initial moment; Indicates t m The unnormalized weight of the kth indicator in the parallel indicator group after being corrected by data regression at any moment; Indicates t m The weight of the kth indicator in the parallel indicator group after being corrected by data regression at any moment.

9. A system for executing the dynamic weighting method according to any one of claims 1 to 8, characterized in that: include: Multi-level indicator architecture model for bridge system status; An initial weight value calculation unit, used to determine the initial weight value of each level indicator in the architecture model at time t0; Index value calculation unit, at t m time, based on t m-1 to m The lowest level indicator values ​​and t in the architecture model collected or calculated multiple times during the time period m-1 The weight values ​​of indicators at each level at the time are used to calculate the indicator value data sets corresponding to all other indicators layer by layer. The indicator value data sets include N groups of indicator values, where N represents t m-1 to m The number of times the lowest level indicator values ​​are collected or calculated during a period of time; Weight value update unit, used to establish t m The Bayesian multiple linear regression dynamic model of the index values ​​of each level at the moment is related to the multiple parallel index values ​​of the next level with subordinate relations. The constant term regression coefficient in the Bayesian multiple linear regression dynamic model is determined, and the normalized non-constant term regression coefficient is summed and normalized to obtain the normalized non-constant term regression coefficient. The average value of the normalized non-constant term regression coefficient and the initial weight is calculated, and then the average value result is summed and normalized as the index of each level at t m The weight value of the moment; The dynamic weighting unit is electrically / communicatively connected to the index value calculation unit and the weight value update unit, respectively, and controls the index value calculation unit and the weight value update unit to calculate the dynamic weights of the indicators at each level at the current or required moment as time increases.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that: When the computer program instructions are executed by a processor, the dynamic weighting method described in any one of claims 1 to 8 is implemented.

Images