4.1. Establishing Analytic Hierarchy Process Model
As more and more attention has been paid to the structural and operational safety of bridges in recent years, the research and application of methods for evaluating the safety condition of bridges have received great attention, and extensive studies have been carried out. These assessment methods include but are not limited to, traditional comprehensive evaluation, analytical hierarchy process, expert system evaluation, and load test evaluation. In this article, the analytical hierarchy process is used to assess the safety of a cable-stayed bridge across the sea. The fundamental aspect of the analytical hierarchy process is to establish an analytical hierarchy process model that comprehensively reflects the safety status of the bridge, with each evaluation criterion reflecting its importance and independence [
31,
32]. Taking into account the safety of various evaluation criteria and the operational state of the bridge, the final established AHP model for the safety assessment of the cable-stayed bridge over the sea is shown in
Figure 6.
As shown in
Figure 6, the AHP model for the safety assessment of the cable-stayed bridge over the sea consists of three levels. The top level represents the overall structural condition of the cable-stayed bridge. The middle level includes the evaluation of static, dynamic, and load-related indicators of the bridge. The lowest level includes strains, cable forces, and displacements of the bridge, as well as acceleration and frequency data, temperature influences, and vehicle loads.
Relying solely on dynamic characteristics when assessing the safety of cable-stayed bridges over the sea can lead to false alarms and false detections. The reasons for these false alarms and missing detections may be due to the harsh environmental conditions in which the cable-stayed bridge is located. The dynamic characteristics of the bridge are subject to the effects of temperature, wind, and other natural factors, which can significantly interfere with on-site monitoring results and cause false alarms. Furthermore, due to the large span and complex structure of cable-stayed bridges over the sea, on-site monitoring systems may be unable to detect changes in dynamic characteristics caused by localized minor damage, resulting in false detections.
In accordance with the hierarchical structure and the subordination relationship, experts are invited to make pairwise comparisons of indicators at the same level, assigning the degree of importance of each level of indicators on a scale from 1 to 9, thus constructing a judgment matrix. The meanings represented by the judgment matrix scale are shown in
Table 4.
After calculating the weight vector of indicators and the maximum eigenvalue using the judgment matrix, it’s important to note that the scores in the judgment matrix are derived from the subjective expertise of experts. As a result, inevitable decision errors or one-sided outcomes may occur. Therefore, it is necessary to conduct a consistency check on the judgment matrix to demonstrate the correctness of the assessment results. The specific method for this verification test is as follows:
In this context, n represents the number of elements within the same hierarchical level,
denotes the maximum eigenvalue of the judgment matrix, and
RI stands for the Random Consistency Index (the value of
RI is determined based on the value of n, as indicated in
Table 5). When the
CR is less than 0.1, it is considered that the determined values for the importance of elements are reasonable. This signifies that the inconsistency level in constructing the judgment matrix is within an acceptable range, confirming the validity of the judgment matrix. If
CR exceeds 0.1, it implies that the judgment matrix is not valid, and adjustments are required for the determined values of the element’s importance.
Experts, based on their own experience, organized discussions and used Santy’s ‘1–9 Scale Method’ to assign relative importance values to primary elements in the safety assessment of the cross-sea cable-stayed bridge (
Table 6). An assessment matrix is then created using the AHP. The above formula is then applied for a one-time check. If the test is passed, it indicates that the judgment matrix has satisfactory consistency.
After calculation,
= 3.009, and referring to the table,
RI = 0.58. This results in
CR = 0.0078 < 0.1. Therefore, the judgment matrix for the first-level elements in the construction of the cable-stayed bridge is deemed consistent through the consistency test. After normalization, the calculated weights for the first-level elements in the safety state assessment of the cross-sea cable-stayed bridge are shown in
Table 7.
Based on the calculation results in this section, static indicators with a weight of 60% are considered as the primary basis for analysis. Although using dynamic features for analysis results in false positives and missing positives, dynamic feature analysis is also indispensable, so it is assigned a weight ratio of 20%. The external load reflects the current environment of the bridge and assigns the remaining 20% to the load index. According to the weights of 60%, 20%, and 20%, the overall safety status of the cable-stayed bridge over the sea is comprehensively evaluated.
When the number of elements at the same level is less than 3,
RI = 0, and a judgment matrix cannot be constructed. Based on expert experience and previous research [
33], in the mean static indicator assessment of cable-stayed bridges over the sea, the strain can reflect the local performance of the bridge, while the displacement can reflect the overall performance. Within this level, the paper assigns a weight of 40% to displacements that can reflect overall performance, a weight of 40% to strains that can reflect local performance, and a weight of 20% to cable forces.
When evaluating the dynamic index of the middle layer, the frequency that can better reflect the dynamic characteristics of the structure is assigned a weight ratio of 80% at this level. Accelerations that are less important relative to frequency receive a weight of 20%.
Since the temperature effect and the vehicle load are equally important when evaluating the load index of the middle layer, in this document a weight ratio of 50% is assigned to the temperature action and the vehicle load.
4.2. Determination of Weights for Bottom-Level Monitoring Indicators
In this paper, the determination of the weights of different indicators and their respective levels ultimately arises from the different importance of the structural information provided by different types of sensors. This results in different sensor types being assigned different weights. Therefore, this paper uses the strength reserve comparison method as a basis [
34], in conjunction with the marginal reserve quantity and marginal reserve ratio approaches, to determine the weights of different indicators.
4.2.1. Determination of Weights for Strain Monitoring
In this paper, the weight values of each strain monitoring point are determined using the limit reserve quantity method. First, a comprehensive synthesis of a significant amount of historical data from each strain monitoring point is conducted to determine the maximum strain value at each point. The maximum strain value is processed to obtain the maximum stress, and the maximum stress and the ultimate tensile strength are subtracted, and the difference between the two is the measure of the strain reserve strength at that point. By using the reserve strength measurement of each strain monitoring point, it is possible to determine the weight ratio of each monitoring section as well as the weight ratio of different positions within the same section.
4.2.2. Determination of Weights for Cable Force Measuring Points
This article uses the 10/10~18/2 scaling method to determine the weighting values of each cable force measurement point. In this approach, we integrate the practical considerations of a cable-stayed bridge over the sea, collect expert opinions, and use an extended AHP model with the 10/10~18/2 scaling method to create a hierarchical judgment matrix to calculate the weights of individual cable force measurement points. For example, consider as the weight of various indicators with respect to the overall goal . An analysis of the decision matrix is carried out, focusing on the maximum characteristic vector (), the consistency index (CI), and the consistency ratio (CR). If the CR value is less than 0.1, it indicates that the decision matrix has relatively good consistency in the context of bridge construction.
4.2.3. Weight Determination of Displacement Monitoring Points
The determination of the section weights for displacement monitoring points differs from that for strain monitoring points. In this paper, the weight values for displacement monitoring points are determined using the marginal reserve degree method. Typically, the monitoring values of displacement sensors are relatively small. In order to facilitate subsequent analytical calculations, in this article, the numerical results obtained under the combined effects of phase two loading, temperature, prestress, and moving loads are selected as the final calculated values.
According to the specifications, the maximum deflection resulting from the combined effects of internal loads, external loads, and environmental loads on a cable-stayed bridge should not exceed 1/600th of the bridge’s maximum span. Direct analysis of the difference between the specified displacement limits and the calculated displacement values would result in an equal weighting of all displacement monitoring points and thus not take into account the differences in cross-sectional variations between the monitoring points. Thus, a calculation is made by dividing the two, and the quotient is used to create a judgment matrix. This, in conjunction with the limit reserve method, yields the weight percentages for the sections of each displacement monitoring point.
4.2.4. Determination of Weights for Acceleration Measurement Points
The weights for acceleration monitoring points are also determined using the marginal reserve degree method. After calculating and analyzing the monitoring data during the operating period of the bridge, it was found that the acceleration monitoring data stabilizes weekly during normal operation. Therefore, the monitoring values of the acceleration points over a period of 7 days were selected for analysis. The maximum values of each acceleration monitoring point within a week were recorded, and the weighted proportion was calculated using the average of these maximum values over the week.
According to the structural specifications for cable-stayed bridges over the sea, the acceleration limits satisfy Equation (3):
Whereas, represents the acceleration limit; represents the first-order frequency of the bridge.
The situation is similar between acceleration monitoring values and displacement monitoring values. Since the values measured on site are significantly below the specified values, a different analysis of both methods would lead to the same weighting of the individual sensor monitoring points and would therefore not reflect the differentiation of different acceleration monitoring points. Therefore, the method of dividing the two values and establishing an importance matrix based on the calculation results is used. Then the weights of each acceleration monitoring point are calculated using the same method.
4.4. Procedure for the Safety Assessment of Cable-Stayed Bridges across the Sea
As part of the research on safety assessment methods for cable-stayed bridges over the sea, the specific steps required in the structural safety assessment of such bridges are as follows:
First, the collected monitoring data from various sensors is organized and analyzed to determine the weight proportions of the individual sensor measurement points. Secondly, the normalization values of various types of indicator parameters are calculated, which are used as the final evaluation values. Then, the displacement indicators are subjected to non-uniformity processing, and the resulting values represent the final displacement indicator evaluation values. Finally, the weighted synthesis method is used to calculate the weights of the intermediate layer in the AHP model, resulting in an overall assessment of the safety of the cable-stayed bridge leads over the sea.
According to the Urban Bridge Condition Index (BCI) regulations, the structural condition of bridge structures is classified into five levels from A to E, which represent the range from excellent condition to dangerous condition [
35]. Based on the above rating system, the bridge safety rating classification is determined as shown in
Table 8.
When assessing the process of a cable-stayed bridge over the sea as part of the AHP, the intermediate layer includes three independent assessment criteria: dynamic analysis, static analysis, and load assessment, each of which gives a corresponding score. By processing the data at the lowest level, scores for the intermediate layer are derived. Then, the overall safety assessment rating of the cable-stayed bridge can be calculated using the weighted synthesis method.