A comprehensive evaluation and safety prediction method for dam reinforcement work behavior

Abstract

The application discloses a kind of comprehensive evaluation and safety prediction method of dam reinforcement work nature, comprising: collecting dam multi-node monitoring data, data preprocessing is carried out;Monitoring equation is constructed, and data is fused and verified by improving batch estimation algorithm;Determine the evaluation index measurement value;Rely on set pair analysis to obtain each evaluation index connection number and normalization;Entropy method assigns index weight, calculates D-S evidence theory weighted basic probability assignment, and multi-source information fusion evaluation result is fused;Synthetic connection number is calculated by combining normalized index to determine safety grade, and then the development trend of nature is analyzed by partial connection number calculation, to obtain safety trend prediction;Two kinds of results are integrated to form dam reinforcement effect evaluation conclusion.The application can comprehensively integrate spatiotemporal multi-source monitoring data, combine fusion technology and theory, objectively evaluate dam reinforcement work nature, and scientifically predict its development trend.

Description

Technical Field

[0001] This invention relates to the field of water conservancy project safety monitoring technology, specifically to a method, equipment, and storage medium for comprehensive evaluation and safety trend prediction of dam reinforcement work performance based on multi-source information fusion. Background Technology

[0002] As a core component of water conservancy infrastructure, dams play an irreplaceable strategic role in flood control, irrigation, water supply, and power generation. Among them, some dams that were built a long time ago and were affected by the design standards, construction technology and long-term service environment at that time will generally experience irreversible structural aging and performance degradation, which seriously threaten the safe service of the dam and the safety of people's lives and property downstream.

[0003] To address this, existing technologies have proposed targeted dam reinforcement operations. The core objective of dam reinforcement is to enhance the dam's safe service capability. Accurately evaluating the dam's operational behavior after reinforcement and scientifically predicting its long-term development trend are crucial steps in verifying the reinforcement effect, optimizing operation and maintenance decisions, and ensuring the long-term safety of the dam.

[0004] Existing methods for evaluating the performance and predicting trends of dam reinforcement and upgrading work often directly adopt methods used during the normal operation period of dams, which have significant limitations. On the one hand, the multi-source fusion system is insufficient. The application of multi-source information fusion is fragmented and has not formed a technical system for the hazard removal and reinforcement scenario; the data fusion adaptability is insufficient, the working condition of the dam body fluctuates greatly before and after hazard removal and reinforcement, existing methods are mostly based on a single data source, do not take into account the spatiotemporal distribution characteristics of monitoring data, lack fusion verification links, the fusion accuracy of similar data is low, the decision fusion of dissimilar information is uncertain, and the adaptability to the hazard removal and reinforcement engineering scenario is insufficient. On the one hand, the objectivity and resilience of evaluation decisions are weak. Existing methods rely heavily on expert experience in aspects such as indicator weighting and trust allocation, resulting in strong subjectivity; they often adopt a single evaluation theory, which cannot resolve disagreements in multi-indicator evaluations, lack cross-verification mechanisms, and thus have weak evaluation robustness. On the one hand, the comprehensive evaluation dimensions are incomplete. Most of them only output a single safety level result, are often static models, lack prediction of the dynamic trends before and after dam reinforcement, cannot reflect the development trend of dam's working state, and are difficult to support the systematic evaluation of reinforcement effect and operation and maintenance decisions.

[0005] Furthermore, existing methods do not fully integrate advanced algorithms and theories, making it difficult to achieve objective evaluation and scientific prediction of dam performance, and thus failing to meet the actual needs of current large-scale dam reinforcement and renovation projects.

[0006] Therefore, this application proposes a comprehensive evaluation and safety prediction method for the performance of dam reinforcement work, which can comprehensively analyze spatiotemporal multi-source monitoring data, combine fusion technology and theory, objectively evaluate the performance of dam reinforcement work, and scientifically predict its development trend, so as to solve the above-mentioned technical problems. Summary of the Invention

[0007] The main objective of this invention is to provide a comprehensive evaluation and safety prediction method for the performance of dam reinforcement and safety improvement, in order to solve the technical problems mentioned in the background art, such as inaccurate evaluation, unscientific prediction, and difficulty in objectively reflecting the effect of dam reinforcement and safety improvement.

[0008] The present invention solves the above-mentioned technical problems by adopting the following technical solutions: On the one hand, this invention discloses a comprehensive evaluation and safety prediction method for the performance of dam reinforcement and upgrading, including: Step S1. Collect monitoring data, including water level and uplift pressure, for the same monitoring object using multi-node sensors on the dam, and perform data preprocessing. Step S2. Establish monitoring data equations, use an improved batch estimation algorithm to fuse the preprocessed data, and verify the fused data; Step S3. Determine the corresponding measurement value of the specified indicator based on existing monitoring data; Step S4. Based on the corresponding metric values ​​of the indicators and the principle of set pair analysis, calculate the single-indicator correlation coefficient of each indicator to the evaluation level; Step S5. The entropy value method is used to calculate the connection coefficient of a single indicator by combining the principle of set pair analysis and DS evidence theory. The weighted basic probability values ​​of each indicator are obtained. Based on the preset level of each indicator, the evidence synthesis rule of DS evidence theory is used to fused layer by layer to finally obtain the multi-source information fusion evaluation result of the dam's working performance. The purpose of set pair analysis here is to eliminate subjectivity. By integrating the "sameness, difference, and opposite" idea in set pair analysis with DS evidence theory, the basic probability values ​​are objectively determined by the indicator connection coefficients, which can overcome the subjective defects of traditional expert experience-based assignment. Step S6. The single index correlation coefficient is calculated by combining the entropy value assignment principle with the set pair analysis principle to obtain the composite index correlation coefficient. The partial correlation coefficient in the set pair analysis is then used to calculate the positive bias correlation coefficient, the negative bias correlation coefficient, and the full bias correlation coefficient. Finally, the full bias correlation coefficient is used to obtain the final safety trend prediction result. Step S7. The final evaluation is obtained by combining the final safety trend prediction result with the multi-source information fusion evaluation result of the working state. At this time, the safety level determination result and the safety trend prediction result obtained by set pair analysis are used as the main evaluation conclusions, and the multi-source information fusion confidence degree distribution and global uncertainty obtained by DS evidence theory are used as supplementary explanations and verification basis for the reliability of the evaluation results. The two types of results corroborate each other and jointly characterize the working state level and development trend of the dam after reinforcement.

[0009] Preferably, in step S1, during the data preprocessing calculation, gross errors and outliers in the dam monitoring data are removed using the tolerance function threshold method in fuzzy set theory. The specific removal process includes: Define the universe of discourse for sensor measurement data Calculate any two of the data. and Data deviation distance for: in This is the arithmetic mean of the original data; and This is the sequence number of the sensor measurement data point, with a value of [value missing]. ;in Q This represents the total number of original measurement data. The allowance function is constructed as follows: ; In the formula The closer the value is to 1, the better the data consistency. Define the threshold determination function as follows: ; In the formula, ε is the threshold of the allowable function, when If the consistency between two data points is deemed insufficient, data points that deviate further from the original mean are identified as significant anomalies and removed, ultimately resulting in a consistent and valid data sequence. ,in, g This represents the number of valid data points after removing anomalies.

[0010] Preferably, the specific operation process of fusing the preprocessed data using the improved batch estimation algorithm in step S2 includes: Based on the characteristics of the dam monitoring area, the sensors are divided into categories according to their deployment locations. Grouping data to ensure spatial correlation in each group; For each set of valid data Classified by time series Batch, each batch contains The data point, the first Group 1 Batch data is ;in, sFor sensor grouping sequence number, l Batch number (values ​​are 1, 2, ...). t ), For the first s The number of data points included in a batch; The calculated mean of the data is: ; The variance is calculated as follows: ; In the formula This is an outlier factor used to reduce the weight of extreme value data within a batch, thereby weakening the interference of extreme values ​​on mean estimation. The sequence number of the data point within a single batch; For the first s Group 1 l Variance of batch data. Dynamic weights are assigned based on the variance of each batch of data, then the... Group 1 The weights of the batch data are: ,in ; Calculate the first The fusion result of the group data is as follows: ; The final global fusion result of the dam monitoring data is as follows: ,in , Let be the variance of the fusion results for the s-th group. The index number for grouped summation; After fusion is completed, step-by-step verification is performed. Based on the preset allowable thresholds of each monitoring indicator, the batch fusion results, group fusion results and global fusion results are verified for reasonableness in sequence. If the fusion result exceeds the corresponding safety allowable range or the fusion variance exceeds the preset limit, the fusion result at that level is determined to be invalid. The process returns to the data preprocessing stage to remove abnormal data and then performs fusion again. After the verification is passed, the valid fusion result is output.

[0011] Preferably, the process for determining the corresponding metric value of the specified indicator in step S3 includes: The dam evaluation indicators and evaluation levels are systematically divided into 5 levels, and the set of evaluation levels is denoted as follows: These correspond to the dam's normal, basically normal, slightly abnormal, severely abnormal, and malignant abnormal states, respectively. The corresponding index value ranges for each level are [0.8,1], [0.6,0.8), [0.4,0.6), [0.2,0.4), and [0,0.2], respectively.

[0012] Preferably, the process for obtaining the weighted basic probability assignment in step S5 includes: Determining the weights of underlying information using the entropy method and initial basic probability assignment Calculate the weighted basic probability assignment for:

[0013] Indicates the first The evaluation index is for the first The weighted basic probability is assigned to each security level, and the remaining trust after weighting is allocated to the entire set of the recognition framework to ensure the normalization of the basic probability assignment. Among them, the initial basic probability assignment of the underlying information is generated by transforming the single index set pair connection number obtained in step S4, and the degree of sameness, degree of difference at each level, and degree of opposition are sequentially assigned to the five security levels as the initial trust assignment of each underlying evaluation index to each security level; then, the weighted basic probability assignment is obtained by combining the weight of the underlying index determined by the entropy method. Based on the pre-defined levels of each indicator, the basic probability assignment of the information in the next higher level is determined by the evidence synthesis rules of the DS evidence theory, with the lowest level designated as the first level and its basic probability assignment being... The basic probability of the second layer of information is assigned as follows: , ..., then the first The basic probability assignment of layer information is Then we have:

[0014] in, The subscripts 1, 2, and 3 represent the sequence numbers of different evidence sources within the same level.

[0015] Preferably, in step S5, the information entropy of each evaluation indicator is first calculated using the entropy method. Objective weights are then determined based on the dispersion of the indicator data and its contribution to effective information (the smaller the information entropy, the richer the information carried by the indicator and the stronger its ability to distinguish the dam's operational behavior, thus assigning a higher weight coefficient). This objective weight is then mapped to the similarity-dissimilarity-inverse correlation coefficient of the single indicator obtained from set pair analysis. The DS evidence basic probability allocation function corresponding to each single indicator is then weighted and corrected to match the trust allocation of each evidence source with the actual importance of the indicator, ultimately forming a weighted basic probability assignment. The revised basic probability assignments are used in the synthesis of DS evidence for three main reasons: First, they replace manual experience-based weighting with objective weighting using the entropy method, eliminating the randomness of subjective assignments and improving the objectivity and credibility of the evaluation results. Second, they strengthen the leading role of highly correlated and high-information indicators in decision fusion, weaken the interference of redundant and low-correlation indicators, and alleviate the degree of conflict between multi-source evidence, reducing the risk of fusion distortion caused by conflicting evidence. Third, they achieve a deep connection between the set pair analysis correlation number and DS evidence theory, making the multi-source information fusion evaluation results of the dam's working performance more objective, stable, and more in line with the actual working state of the dam after reinforcement and renovation.

[0016] Preferably, the probability acquisition process in the process of acquiring the underlying information weight set in step S5 includes: Step 1: For One sample, The original data matrix for each evaluation indicator is as follows:

[0017] in, This represents the measured value of the nth evaluation index for the mth sample. Step 2: Normalize the data of each indicator to obtain a matrix. for:

[0018] in, , These are the optimal and worst state limit values ​​corresponding to the same evaluation index. The corresponding values ​​are determined based on the differences in the safety attributes of the evaluation index to ensure that the normalization result is positively correlated with the dam's safety level. Step 3: Calculate the entropy of the nth evaluation index:

[0019] in, , indicating the first n Under the evaluation index, the first m The weighting of indicators for each sample; r Let be the entropy normalization constant. Its function is to constrain the information entropy value to the interval [0,1]. And when When =0, It's meaningless, to The calculation was corrected to ; Step 4: Calculate the weight of each evaluation indicator:

[0020] in, For the first n The weight of each evaluation indicator.

[0021] 7. The comprehensive evaluation and safety prediction method for the performance of dam reinforcement work as described in claim 6, characterized in that the specific formula for obtaining the composite correlation coefficient of the indicators in step S6 includes:

[0022] in, a, b, c, d, e These are the sameness, first-level difference, second-level difference, third-level difference, and opposition corresponding to the evaluation indicators of dam performance; These are the difference coefficients for Level 1, Level 2, and Level 3, respectively. j The degree of contrast is the coefficient of opposition. The same degree corresponds to the normal safe state of the dam, the degree of opposition corresponds to the abnormal state of the dam risk, and the three levels of difference correspond to the transitional state between the two in turn.

[0023] Safety level mapping: Substitute the composite correlation coefficient into the coefficients of identity, difference, and opposition of the engineering setpoint to calculate the comprehensive correlation coefficient. After normalizing to the [0,1] interval, match it with the preset five-level evaluation interval to determine the safety level of the dam.

[0024] Preferably, the weights of each indicator in step S6 are obtained using the same method as the weight set of the underlying information in step S5.

[0025] Preferably, the specific calculation formulas for calculating the partially positive correlation coefficient, partially negative correlation coefficient, and fully partial correlation coefficient in step S6, combined with the partially related coefficient in set pair analysis, include: (1) The partial positive correlation coefficient is:

[0026] ; ; ;

[0027] in, , , , The first-order positive components, namely the same part, first-level different part, second-level different part, and third-level different part, respectively, represent the trend strength of each evaluation component evolving towards the same safe state. (2) The negative correlation coefficient is:

[0028] ; ; ;

[0029] in, , , , The first-order negative components, in order, are the same part, first-level different part, second-level different part, and third-level different part, representing the trend strength of each evaluation component evolving towards a state of risk opposition. (3) The partial correlation coefficient is:

[0030] in, It is a first-order fully partial correlation coefficient, and its positive or negative value corresponds to the overall evolution direction of the dam's working performance: a positive value indicates that the overall working performance is developing in a positive direction, and a negative value indicates that the overall working performance is developing in a negative direction. The magnitude of the absolute value corresponds to the significance of the trend evolution.

[0031] Preferably, for the fully partial correlation coefficient, when When this occurs, it indicates that the trend prediction object is in a critical trend that is neither negative nor positive, and it is necessary to find a higher-order partial derivative. when When this occurs, it indicates that the trend prediction object is developing negatively and there is a risk of deterioration, requiring timely and effective measures to reverse this trend; when This indicates that the trend forecast is developing in a positive direction, the dam's operational performance is continuously improving, safety redundancy is steadily increasing, and the effectiveness of risk mitigation and reinforcement is steadily showing. It is possible to maintain the existing safety operation and maintenance management strategy and continue to carry out routine tracking and monitoring.

[0032] Trend determination mapping: The evolution trend is determined by the positive or negative value of the first-order partial correlation coefficient. A positive value indicates that the dam's working performance is improving, while a negative value indicates that there is a risk of deterioration. When the value is 0, a higher-order partial correlation coefficient needs to be calculated to clarify the deeper trend.

[0033] On the other hand, the present invention also discloses a comprehensive evaluation and safety prediction system for performing the aforementioned comprehensive evaluation and safety prediction method for dam reinforcement work. The system includes: Sensor networks are composed of multiple sets of sensors that are interconnected at multiple nodes on a dam. To prevent node failure and external interference, two or more sensors are often used to monitor a point to enhance the mutual verification between nodes and the robustness of the entire network system. Generally, sensors of the same precision are used for monitoring points of the same type, and the monitoring parameters and noise can be considered to be normally distributed. The dam safety information system is used to perform fusion operations of multi-source safety information of the dam based on sensor network data.

[0034] Preferably, the dam safety information system includes an information acquisition unit, an information fusion unit, and an information storage unit. The information acquisition unit mainly includes dam prototype monitoring information, model information, and expert knowledge. The model information subunit refers to the extraction of dam information through different fusion criteria. The expert knowledge subunit mainly integrates expert knowledge, relevant theories, and experience.

[0035] Preferably, the dam safety information system further divides the information fusion unit into three levels according to the data abstraction level in the system: data layer fusion, feature layer fusion, and decision layer fusion, wherein: Data layer fusion is the direct analysis and processing of the original monitoring data of the dam. It mainly includes real-time judgment and processing of illegal, erroneous or unreasonable data that occurs during the upload, as well as preliminary identification of gross errors in the monitoring data. It aims to preserve as much original monitoring information as possible and provide subtle information. It processes a large amount of data. Feature layer fusion is used to perform fusion analysis on similar measuring points using methods such as eigenvalue analysis and correlation analysis. Generally, it first performs eigenvalue statistics and modeling analysis on the data of a single measuring point (such as statistical model, Kalman filter, support vector machine, artificial neural network, etc.), and then fuses information from similar measuring points to construct a spatial model, quantifying the working state of the dam from a spatiotemporal perspective. The decision-level fusion is used to comprehensively infer the degree of support of all monitoring information for the comprehensive performance decision of the dam based on the characteristic fusion results of different monitoring data, and by applying methods such as fuzzy evaluation, pattern recognition, and DS evidence theory. It combines qualitative and quantitative analysis to achieve the integration and complementarity of dam information.

[0036] Preferably, in a sensor network, among sensors monitoring the same object, each sensor obtains a set of consistent monitoring data at a certain moment, and performs improved batch estimation. This allows for the generation of fusion results that are closer to the actual values, thereby eliminating uncertainties caused by the diversity of the dam system and node failures during the monitoring process.

[0037] In another aspect, the present invention also discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the method described above.

[0038] In another aspect, the present invention also discloses a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the method described above.

[0039] As can be seen from the above technical solution, the present invention provides a comprehensive evaluation and safety prediction method for the operational performance of dam reinforcement and upgrading. Compared with the prior art, the present invention has the following advantages: 1. This invention overcomes several core technologies in the field of dam reinforcement performance evaluation, including the introduction of multi-source monitoring data hierarchical fusion, improved batch estimation adaptive weighting, objective value assignment method combining set pair analysis similarity, difference, and inverse with DS evidence theory, and dynamic trend prediction of partial correlation coefficients. These technologies address various technical bottlenecks in the field, such as the one-sidedness of single-point modeling evaluation, insufficient accuracy of multi-source monitoring information fusion, strong subjectivity in basic probability assignment of evidence, lack of quantitative evaluation basis for reinforcement and treatment effects, and insufficient scientificity of dam safety trend prediction methods.

[0040] 2. This invention constructs a complete technical system that includes intelligent removal of abnormal data, high-precision fusion of similar monitoring information, objective decision-making fusion of multiple indicators, quantitative comparison of performance before and after reinforcement, and dynamic trend inference of working performance. This system can effectively break the inherent limitations of the independent evaluation mode of single measurement points and avoid the subjective bias caused by the traditional reliance on expert experience for weighting.

[0041] 3. This invention can promote a fundamental leap in the evaluation of the operational performance of dams after reinforcement and renovation, upgrading it from traditional static comprehensive evaluation to dynamic safety prediction. This facilitates the objective quantification of the improvement in the safe service capability of dams after reinforcement and renovation, and allows for the early identification of potential safety risks in the project.

[0042] 4. This invention does not require the addition of complex monitoring hardware equipment and can be directly applied based on the existing prototype monitoring system of the dam, making it highly adaptable to the needs of major projects for the reinforcement and renovation of large-scale dilapidated reservoirs in China.

[0043] It should be understood that the descriptions in this section are not intended to identify key or essential features of embodiments of the invention, nor are they intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Of course, implementing any product of the invention does not necessarily require achieving all of the advantages described above simultaneously. Attached Figure Description

[0044] The accompanying drawings, which form part of this application, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a schematic diagram of the overall process of the present invention; Figure 2 This is a schematic diagram of the DS evidence theory information data fusion of the present invention; Figure 3 This is a schematic diagram of the security information system structure of the present invention; Figure 4 This is a schematic diagram of the multi-source information layered execution and fusion process of the security information system of the present invention. Detailed Implementation

[0045] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0046] For details in the embodiments, please refer to Figures 1 to 4 .

[0047] On the one hand, the comprehensive evaluation and safety prediction method for the working performance of dam reinforcement proposed in the embodiments of the present invention, such as... Figure 1 As shown, it includes: Step S1. Collect monitoring data, including water level and uplift pressure, for the same monitoring object using multi-node sensors on the dam, and perform data preprocessing.

[0048] During the data preprocessing calculation, gross errors and outliers in the dam monitoring data are removed using the permissible function threshold method in fuzzy set theory. The specific removal process includes: Define the universe of discourse for sensor measurement data Calculate any two of the data. and Data deviation distance for: in This is the arithmetic mean of the original data; and This is the sequence number of the sensor measurement data point, with a value of [value]. ;in Q This represents the total number of original measurement data.

[0049] The allowance function is constructed as follows: ; In the formula The closer the value is to 1, the better the data consistency. Define the threshold determination function as follows: ; In the formula, ε is the threshold of the allowable function, when If the consistency between two data points is deemed insufficient, data points that deviate further from the original mean are identified as significant anomalies and removed, ultimately resulting in a consistent and valid data sequence. ( ),in, g This represents the number of valid data points after removing anomalies.

[0050] Step S2. Establish monitoring data equations, use an improved batch estimation algorithm to fuse the preprocessed data, and verify the fused data.

[0051] The specific operational procedures for fusing the preprocessed data using the batch estimation algorithm include: Based on the characteristics of the dam monitoring area, the sensors are divided into categories according to their deployment locations. Groups (such as the left side group of the dam foundation, the middle group of the dam foundation, and the right side group of the dam foundation), each group contains One sensor, of which To ensure that each set of data has spatial correlation; Inter-sequences are divided into Batch, each batch contains The data point, the first Group 1 Batch data is ;in, s Sensor group number ，l This is the batch number (values ​​are 1, 2, ..., t). For the first s The number of data points included in a batch.

[0052] The calculated mean of the data is: ; The variance is calculated as follows: ; In the formula This is an outlier factor used to reduce the weight of extreme value data within a batch, thereby weakening the interference of extreme values ​​on mean estimation. The sequence number of the data point within a single batch; For the first s Group 1 l Variance of batch data.

[0053] Dynamic weights are assigned based on the variance of each batch of data, then the... Group 1 The weights of the batch data are: ,in ; Calculate the first The fusion result of the group data is as follows: ; The final global fusion result of the dam monitoring data is as follows: ,in , For the first s The variance of the group fusion results The index number for grouped summation.

[0054] After fusion is completed, step-by-step verification is performed. Based on the preset allowable thresholds of each monitoring indicator, the batch fusion results, group fusion results and global fusion results are verified for reasonableness in sequence. If the fusion result exceeds the corresponding safety allowable range or the fusion variance exceeds the preset limit, the fusion result at that level is determined to be invalid. The process returns to the data preprocessing stage to remove abnormal data and then performs fusion again. After the verification is passed, the valid fusion result is output.

[0055] Step S3. Determine the corresponding measurement value of the specified indicator based on the existing monitoring data.

[0056] The specific procedures include: The dam evaluation indicators and evaluation levels are systematically divided into 5 levels. Obtain each evaluation level and its corresponding range value: The set of evaluation levels is denoted as These correspond to the dam's normal, basically normal, slightly abnormal, severely abnormal, and malignant abnormal states, respectively. The corresponding index value ranges for each level are [0.8,1], [0.6,0.8), [0.4,0.6), [0.2,0.4), and [0,0.2], respectively.

[0057] Furthermore, the underlying evaluation indicators (i.e., monitoring points) in the dam performance information system are quantitative indicators with a large amount of measured data. Their measurement mainly considers both the numerical and trend performance of the monitoring effect. Based on the measured data, a mathematical model is established between the effect quantity and influencing factors (upstream and downstream water levels, temperature, rainfall, and timeliness). The measured values ​​are compared with the model fitting values, and the characteristic values ​​of the effect quantity are examined to determine whether they are within the normal range, thus obtaining the numerical score range, as shown in Table 1 below. The final numerical score is then obtained using linear interpolation. Trend performance refers to the trend change process of the timeliness component of the dam's measured effect quantity over time. The timeliness component of the effect quantity can be separated through the established mathematical model. The trend score is mainly obtained by analyzing the changing characteristics and patterns of the timeliness component.

[0058] Table 1: Measurement Table of Evaluation Indicators

[0059] in, This is the measured effect size. To monitor the predicted values ​​of the model, This is the maximum effect size. To monitor the mean squared error of the model.

[0060] It is also necessary to add that, in specific dam examples, for concrete gravity dams, taking seepage behavior analysis as an example, due to the existence of transverse joints, seepage behavior analysis can be performed on each dam section separately, constructing a concrete gravity dam seepage behavior information system. The top-level objective of this concrete gravity dam seepage behavior information system is the seepage behavior of the concrete gravity dam, and the whole is divided into multiple independent dam section evaluation units along the dam body. Taking any dam section as an example, the seepage behavior is further decomposed into three major seepage monitoring indicators: uplift pressure, seepage around the dam, and leakage. Each type of indicator has several observation points. By carrying out measurement attribute measurement on the observation data of all measurement points, the quantitative characterization of seepage information of multiple measurement points in a single dam section can be realized. This can provide a structured data carrier at the unit level, indicator level, and measurement point level for the hierarchical fusion of multi-source monitoring data.

[0061] Step S4. Based on the corresponding metric values ​​of the indicators and the principle of set pair analysis, calculate the single-indicator correlation coefficient of each indicator to the evaluation level.

[0062] Step S5. The entropy value method is used in combination with the set pair analysis principle and DS evidence theory to calculate the connection coefficient of a single indicator, thereby obtaining the weighted basic probability value of each indicator. Based on the preset level of each indicator, the evidence synthesis rule of DS evidence theory is used to recursively fuse the data layer by layer, and finally obtain the multi-source information fusion evaluation result from the dam to the working state. The purpose of set pair analysis here is to eliminate subjectivity. By integrating the "sameness, difference, and opposite" idea in set pair analysis with DS evidence theory, the basic probability value can be objectively determined using the indicator connection coefficient, which can overcome the subjective defects of traditional expert experience-based value assignment.

[0063] The process of obtaining the weighted basic probability assignment at this point is based on the DS evidence theory, such as... Figure 2 As shown, it includes: Determining the weights of underlying information using the entropy method and initial basic probability assignment Calculate the weighted basic probability assignment for:

[0064] Indicates the underlying evaluation indicators For the first kThe weighted basic probability is assigned to each security level, and the remaining trust after weighting is allocated to the entire set of the identification framework (i.e. uncertainty) to ensure the normalization of the basic probability assignment.

[0065] The initial basic probability assignment of the underlying information is generated by transforming the single index set connection coefficient obtained in step S4. The degree of sameness, the degree of difference at each level, and the degree of opposition are sequentially assigned to the five security levels as the initial trust assignment of each underlying evaluation index to each security level. Then, the weighted basic probability assignment is obtained by combining the weight of the underlying index determined by the entropy method.

[0066] Then, based on the preset levels of each indicator, the basic probability assignment of the information in the next higher level is determined by the evidence synthesis rules of the DS evidence theory, and the lowest level is designated as the first level, with its basic probability assignment being... The basic probability of the second layer of information is assigned as follows: , ..., then the first The basic probability assignment of layer information is Then we have:

[0067] in, The subscripts 1, 2, and 3 represent the sequence numbers of different evidence sources within the same level.

[0068] In the specific implementation process, the information entropy of each evaluation indicator is first calculated using the entropy method. Objective weights are then determined based on the dispersion of the indicator data and its contribution to effective information (the smaller the information entropy, the richer the information carried by the indicator and the stronger its ability to distinguish the dam's operational behavior, thus assigning a higher weight coefficient). This objective weight is then mapped to the similarity-dissimilarity-inverse correlation coefficient of the single indicator obtained from set pair analysis. The basic probability allocation function of the DS evidence corresponding to each single indicator is then weighted and corrected to match the trust allocation of each evidence source with the actual importance of the indicator. Finally, a weighted basic probability assignment is formed, and this weighted basic probability assignment is used in the DS evidence analysis. The synthesis process has three main aspects: First, it replaces manual experience-based weighting with objective weighting using the entropy method, eliminating the randomness of subjective weighting and improving the objectivity and credibility of the evaluation results. Second, it strengthens the leading role of highly correlated and information-rich indicators in decision fusion, weakens the interference of redundant and low-correlation indicators, and alleviates the degree of conflict between multi-source evidence, reducing the risk of fusion distortion caused by conflicting evidence. Third, it achieves a deep integration of set pair analysis correlation coefficients and DS evidence theory, making the DS basic probability assignment no longer solely dependent on expert experience and more objective. Ultimately, this makes the multi-source information fusion evaluation results of the dam's working performance more objective, stable, and more closely reflect the actual working state of the dam after reinforcement and renovation.

[0069] The probability acquisition process in the process of obtaining the underlying information weight set includes: Step 1: For One sample, The original data matrix for each evaluation indicator is as follows:

[0070] in, For the first m The first sample n The measured values ​​of each evaluation indicator; Step 2: Normalize the data of each indicator to obtain a matrix. for:

[0071] in, , These are the optimal and worst state limit values ​​corresponding to the same evaluation index. The corresponding values ​​are determined based on the differences in the safety attributes of the evaluation index to ensure that the normalization result is positively correlated with the dam's safety level. Step 3: Calculate the... n The entropy of each evaluation indicator is:

[0072] in, , indicating the first n Under the evaluation index, the first m The weighting of indicators for each sample; r Let be the entropy normalization constant. Its function is to constrain the information entropy value to the interval [0,1]. And when When =0, It's meaningless, to The calculation was corrected to ; Step 4: Calculate the weight of each evaluation indicator:

[0073] in, For the first n The weight of each evaluation indicator.

[0074] Step S6. The single-index correlation coefficient is calculated using the entropy value assignment principle combined with the set pair analysis principle to obtain the composite index correlation coefficient. The partial correlation coefficient in the set pair analysis is then used to calculate the first-order partial positive correlation coefficient, the first-order partial negative correlation coefficient, and the full partial correlation coefficient. Finally, the full partial correlation coefficient is used to obtain the final safety trend prediction result.

[0075] The specific formula for obtaining the composite correlation coefficient of the indicator includes:

[0076] in, a, b, c, d, e These are the sameness, first-level difference, second-level difference, third-level difference, and opposition corresponding to the evaluation indicators of dam performance; These are the difference coefficients for Level 1, Level 2, and Level 3, respectively. j This is the degree of contrast coefficient. The same degree corresponds to the normal safe state of the dam, the degree of contrast corresponds to the abnormal risk state of the dam, and the three-level difference degree corresponds to the transitional state between the two. At this time, the safety level of the dam can be determined based on the resulting composite index correlation coefficient. Among them, the safety level mapping is as follows: the comprehensive connection degree is calculated by substituting the composite connection number into the difference and opposition coefficient of the engineering set value, normalizing it to the [0,1] interval, and matching it with the preset five-level evaluation interval to determine the safety level of the dam.

[0077] Furthermore, in the DS evidence theory here, the identification frame is the basis for the entire judgment, the basic probability distribution is the foundation of fusion, the synthesis rule is the fusion process, and the likelihood function and the confidence function are used to express the upper and lower limits of the fusion theory's support for a certain hypothesis.

[0078] It can be further explained at this point that, in one example, the weights of each indicator in this step are obtained using the same method as the underlying information weight set in step S5.

[0079] The specific formulas for calculating the first-order partially positive, first-order partially negative, and fully partial correlation coefficients in set pair analysis include: (1) The partial positive correlation coefficient is:

[0080] ; ; ;

[0081] in, , , , The first-order positive components, namely the same part, first-level different part, second-level different part, and third-level different part, respectively, represent the trend strength of each evaluation component evolving towards the same safe state. (2) The negative correlation coefficient is:

[0082] ; ; ;

[0083] in, , , , The first-order negative components, in order, are the same part, first-level different part, second-level different part, and third-level different part, representing the trend strength of each evaluation component evolving towards a state of risk opposition. (3) The partial correlation coefficient is:

[0084] in, It is a first-order fully partial correlation coefficient, and its positive or negative value corresponds to the overall evolution direction of the dam's working performance: a positive value indicates that the overall working performance is developing in a positive direction, and a negative value indicates that the overall working performance is developing in a negative direction. The magnitude of the absolute value corresponds to the significance of the trend evolution.

[0085] Among them, for the fully partial correlation coefficient, when When this occurs, it indicates that the trend prediction object is in a critical trend that is neither negative nor positive, and it is necessary to find a higher-order partial derivative. when When this occurs, it indicates that the trend prediction object is developing negatively and there is a risk of deterioration, requiring timely and effective measures to reverse this trend; when This indicates that the trend forecast is developing in a positive direction, the dam's operational performance is continuously improving, safety redundancy is steadily increasing, and the effectiveness of risk mitigation and reinforcement is steadily showing. It is possible to maintain the existing safety operation and maintenance management strategy and continue to carry out routine tracking and monitoring.

[0086] Among them, the trend determination mapping: the evolution trend is determined by the positive or negative value of the first-order partial correlation coefficient. A positive value indicates that the dam's working performance is developing in a positive direction, a negative value indicates that there is a risk of deterioration, and when the value is 0, a higher-order partial correlation coefficient needs to be calculated to clarify the deeper trend.

[0087] Step S7. The final evaluation is obtained by combining the final safety trend prediction result with the multi-source information fusion evaluation result of the working performance.

[0088] On the other hand, the present invention also discloses a comprehensive evaluation and safety prediction system for performing the aforementioned comprehensive evaluation and safety prediction method for the performance of dam reinforcement work. The system includes: (1) Dam safety information system, such as Figure 3As shown, the structure is divided into multiple layers from top to bottom, further subdivided into three major information source modules: dam foundation, dam body, and reservoir area near the dam. Dam body information is further divided into three data collection methods: inspection, instrument monitoring, and verification. The instrument monitoring module, as the core data collection unit for multi-source monitoring, is connected to the sensor network and is further subdivided into three monitoring categories: seepage, stress-strain, and deformation. Seepage monitoring includes indicators such as uplift pressure, seepage around the dam, seepage pressure, seepage flow rate, pore pressure, and groundwater level. Seepage pressure data is collected by deploying multiple piezometer points using equipment such as piezometers and piezometers. Deformation monitoring includes indicators such as vertical displacement, horizontal displacement, cracks, and deflection. Horizontal displacement data is obtained by deploying several vertical measurement points using monitoring methods such as plumb lines, lines of sight, and tension lines. This hierarchical framework can aggregate prototype monitoring data from different locations, categories, and measurement points, providing a complete structured data source for the hierarchical fusion of multi-source dam safety information, thus enabling the fusion of multi-source dam safety information. (2) Sensor network is formed by connecting multiple sets of sensors with each other through multiple nodes on the dam. In order to prevent node failure and external interference, two or more sensors are often used to monitor a certain point to enhance the mutual verification between nodes and the robustness of the entire network system. Generally, sensors of the same type are used for monitoring points, and the monitoring parameters and noise can be considered to be normally distributed.

[0089] The dam safety information system includes an information acquisition unit, an information fusion unit, and an information storage unit. The information acquisition unit mainly includes dam prototype monitoring information, model information, and expert knowledge. The model information sub-unit refers to the extraction of dam information through different fusion criteria. The expert knowledge sub-unit mainly integrates expert knowledge, relevant theories, and experience.

[0090] In its implementation, the dam safety information system, based on the levels of data abstraction within the system, divides its information fusion units into three levels: data layer fusion, feature layer fusion, and decision layer fusion. Data layer fusion is the direct analysis and processing of the original monitoring data of the dam. It mainly includes real-time judgment and processing of illegal, erroneous or unreasonable data that occurs during the upload, as well as preliminary identification of gross errors in the monitoring data. It aims to preserve as much original monitoring information as possible and provide subtle information. It processes a large amount of data. Feature layer fusion is used to perform fusion analysis on similar measuring points using methods such as eigenvalue analysis and correlation analysis. Generally, it first performs eigenvalue statistics and modeling analysis on the data of a single measuring point (such as statistical model, Kalman filter, support vector machine, artificial neural network, etc.), and then fuses information from similar measuring points to construct a spatial model, quantifying the working state of the dam from a spatiotemporal perspective. The decision-level fusion is used to comprehensively infer the degree of support of all monitoring information for the comprehensive performance decision of the dam based on the characteristic fusion results of different monitoring data, and by applying methods such as fuzzy evaluation, pattern recognition, and DS evidence theory. It combines qualitative and quantitative analysis to achieve the integration and complementarity of dam information.

[0091] In the specific implementation process, please refer to the specific integration and execution process. Figure 4 First, relying on the dam safety information system, raw monitoring data from multiple monitoring points under various monitoring categories are collected. In the data-level fusion stage, the raw observation data from multiple monitoring points are preprocessed, spatiotemporally registered, and integrated with similar monitoring point data. Abnormal observation information is removed to obtain standardized integrated monitoring point data. Then, feature-level fusion is performed to extract safety features from the integrated data of various monitoring indicators and to perform feature fusion of similar monitoring indicators to generate intermediate fusion results corresponding to various monitoring dimensions. Finally, decision-level fusion is performed to integrate the intermediate evaluation results of multiple monitoring indicators. Combining set pair analysis partial correlation coefficients and DS evidence theory, objective values ​​are assigned, comprehensive safety evaluation and trend prediction are completed, and comprehensive evaluation of dam performance and safety monitoring decision conclusions are output. This realizes hierarchical information deduction from single-point observation information to conclusions on the overall working performance of the dam.

[0092] Furthermore, in the specific implementation process, among the sensors monitoring the same object, each sensor obtains a set of consistent monitoring data at a certain moment, and improves the batch estimation. This can yield a fusion result that is closer to the actual value, thereby eliminating the uncertainty caused by the diversity of the dam system and node failures during the monitoring process.

[0093] Based on the above system, in a specific embodiment, taking a set of concrete gravity dams (dam section 35) as an example, the dam body has low concrete strength, poor integrity, insufficient flood discharge capacity, serious dam leakage, and high uplift pressure. Therefore, multiple reinforcement and safety improvements have been carried out, including heightening, concrete cladding for seepage prevention, prestressed anchor cables, grouting, and drainage. To monitor its seepage, monitoring items were set up including longitudinal and transverse uplift pressure of the dam foundation, dam foundation leakage, and dam body leakage. Simultaneously, this dam section is located in the central area of ​​a major fault fracture zone, with low physical and mechanical parameters of the foundation surface and the lowest safety factor for the dam foundation. The longitudinal uplift pressure of the dam foundation of dam section 35 was selected from data collected by sensors at three locations: the left side, the middle, and the right side of the dam foundation.

[0094] The sensor monitoring data are divided into three groups according to the location: sensor group 1 (left side of the dam foundation, measuring points 1-10), sensor group 2 (middle of the dam foundation, measuring points 11-20), and sensor group 3 (right side of the dam foundation, measuring points 21-30).

[0095] The original monitoring data of longitudinal uplift pressure on the dam foundation are shown in the table below: Table 2: Original Monitoring Data of Longitudinal Uplift Pressure on Dam Foundation

[0096] Among them, the data 12.82 and 13.15 in the middle of the dam foundation in Group 2 are outliers, which are caused by the blockage of the pressure measuring tube. They are used to verify the effectiveness of this algorithm in outlier removal and group fusion.

[0097] Then, outlier removal, grouping and calculating the mean and variance within each group, adaptive weight values, and the improved batch estimation algorithm proposed in the above embodiments are used for grouping and fusion. Finally, the fused estimate of the improved batch estimation algorithm of this application is obtained. The operation results are compared with those of the traditional method in Table 3 below: Table 3: Comparison of Results between Improved Batch Estimation Algorithm and Traditional Method

[0098] The calculation results show that the traditional mean method and the traditional batch estimation algorithm, due to the lack of preprocessing to remove outliers, have variances of 2.0021 and 0.6342, respectively, and absolute errors of 0.31. These are all significantly larger than the 0.07 and 0.0053 of the algorithm used in this embodiment (improved batch estimation algorithm). The variances are significantly smaller, and the data stability is better. Considering the actual situation that piezometer blockage or other abnormal interference often occurs during dam monitoring, causing errors in monitoring data, the algorithm used in this embodiment is superior in all aspects under real engineering conditions and can better support real-time early warning and quantitative assessment of dam status.

[0099] Subsequently, through quantitative analysis (establishing a statistical model of the monitoring effect, analyzing time-dependent components, characteristic values, etc.) and qualitative analysis (process lines, distribution maps, engineering operation status, etc.) of the seepage monitoring data of this dam section, the measurement values ​​of each bottom layer index of seepage were determined as shown in Table 4 below. The grade correlation coefficients for each indicator are calculated and shown in Tables 5 and 6 below: Table 4: Seepage Index Measurement Values

[0100] Table 5: Table of Correlation between Indicator Levels

[0101] Table 6: Correlation Table of Normalized Indicator Levels

[0102] In the DS evidence theory, the basic probability assignment can be determined by the index correlation coefficients calculated from set pair analysis. Here, integrating existing multi-source monitoring information for this dam section, the normalized values ​​of the index correlation coefficients for longitudinal and lateral uplift pressures of the dam foundation, dam foundation leakage, and dam body leakage are calculated, resulting in the basic probability assignments shown in Table 7 below: Table 7: Basic Probability Assignment Table for DS Evidence Theory

[0103] Based on the measured data before and after the reinforcement and repair, the weights of the seepage evaluation indicators of the dam were calculated using the entropy weight method, as shown in Table 8 below: Table 8: Weighting Table of Seepage Evaluation Indicators

[0104] The data in Table 7 were integrated according to the DS evidence theory, and a comparative analysis was conducted using the average evidence theory. The final results are shown in Table 9 below: Table 9: Analysis of Fusion Results

[0105] Analysis shows that the dam was in a slightly abnormal state before reinforcement, and was in a basically normal state after reinforcement, indicating that the reinforcement measures had a good effect and the seepage characteristics had improved.

[0106] The calculation results of the two methods are consistent. Compared with the average evidence theory, the fusion results of the DS theory have higher distinguishability, which can reduce the uncertainty of decision-making and make the evaluation of the dam's working performance more intuitive and accurate.

[0107] The final compositional correlation coefficient of the seepage behavior before reinforcement is:

[0108] Similarly, the correlation coefficient of the composite index after reinforcement can be obtained as follows: .

[0109] The dam's seepage state is evaluated using the obtained composite correlation coefficient. The normalized composite index correlation coefficient is The corresponding seepage behavior level of this dam section is classified as slightly abnormal.

[0110] Further dynamic analysis of the seepage behavior of this dam section is conducted using multivariate correlation coefficients and partial correlation coefficients from set pair analysis. The first-order partial positive correlation coefficient of the seepage behavior is used. First-order negative correlation coefficient Then the fully negative correlation coefficient The value is less than 0. This indicates that the dam's seepage behavior is slightly abnormal, but there is a trend towards developing into a severe abnormality. It is a dangerous dam and urgently needs to be reinforced and comprehensively treated.

[0111] In addition, the composite index connection number after risk elimination and reinforcement is used. Normalized connectivity This indicates that the seepage behavior of this dam section is basically normal, and the seepage performance has improved after reinforcement.

[0112] Seepage trend prediction is performed based on multivariate correlation coefficients and partial correlation coefficients. The first-order partial positive correlation coefficient is: First-order negative correlation coefficient Then the fully negative correlation coefficient The value is close to but slightly less than 0, indicating that this state is a critical condition of basic normality. Although the condition has improved after reinforcement, there is still a tendency to develop towards a slightly abnormal state. If the factors affecting the seepage behavior change, it may lead to the abnormal development of the dam's seepage behavior, which is consistent with the actual situation. This dam section is located in a major fault fracture zone, with complex geological and working conditions, a low safety factor, high uplift pressure, and large leakage. There are many problems with the seepage working behavior. Although it has been reinforced, it is still necessary to further strengthen the monitoring and inspection of the dam, and if necessary, install additional specialized monitoring instruments to ensure the safe operation of the dam.

[0113] In summary, this method, by employing a specific system, introduces a hierarchical fusion of multi-source monitoring data, improved batch estimation with adaptive weighting, an objective value assignment method that integrates set pair analysis with similarities, differences, and inverses and DS evidence theory, and a dynamic trend prediction technique for partial correlation coefficients. Its purpose is to address the technical bottlenecks in existing dam reinforcement performance evaluation, such as one-sided single-point modeling, low accuracy of multi-source information fusion, strong subjectivity in basic probability assignment, lack of quantitative basis for reinforcement effects, and insufficient scientific rigor in safety trend prediction.

[0114] Therefore, by employing a full-chain technical system encompassing intelligent removal of abnormal data, high-precision fusion of similar monitoring information, objective decision-making integration of multiple indicators, quantitative comparison of performance before and after reinforcement, and dynamic trend deduction of operational performance, the inherent limitations of single-point evaluation are overcome, and the subjective bias of traditional expert experience-based assessments is eliminated. This achieves a fundamental leap from static comprehensive evaluation to dynamic safety prediction of the operational performance of dams after reinforcement. It can objectively quantify the improvement in the safe service capability of dams after reinforcement, identify potential safety risks in advance, and requires no additional complex monitoring equipment. It can be implemented entirely based on existing dam prototype monitoring systems, combining technological innovation with engineering universality.

[0115] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0116] Furthermore, it should be noted that if any directional indication (such as up, down, left, right, front, back, etc.) is involved in the embodiments of the present invention, the directional indication is only used to explain the relative positional relationship and movement of each component in a specific posture. If the specific posture changes, the directional indication will also change accordingly.

[0117] Furthermore, those skilled in the art should understand that in the actual use of the embodiments of this application, there may be preset thresholds used as the basis for judging the corresponding technical solutions. These thresholds are conventional technical means commonly used in the field to implement functions such as state judgment, condition recognition, and control logic switching. The specific values, setting basis, value selection methods, determination methods, and adjustment rules of the thresholds involved in this technical solution are all conventional technical choices that can be reasonably determined by those skilled in the art based on conventional technical factors such as actual application scenarios, system working states, characteristics of the detection object, hardware performance parameters, and functional requirements, through conventional experiments, calibrations, and debugging. The specific setting and adjustment of the aforementioned thresholds will not cause this technical solution to be unimplementable as a whole, nor will it affect the realization of the core concept and the achievement of the technical effects of this technical solution.

[0118] Furthermore, if the embodiments of this invention involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the meaning of "and / or" throughout the text includes three parallel solutions; for example, "A and / or B" includes solution A, solution B, or a solution where both A and B are satisfied simultaneously. Furthermore, in the embodiments of this invention, "multiple" refers to two or more. Moreover, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

Claims

1. A comprehensive evaluation and safety prediction method for the performance of dam reinforcement and safety improvement work, characterized in that, include: Step S1. Collect monitoring data, including water level and uplift pressure, for the same monitoring object using multi-node sensors on the dam, and perform data preprocessing. Step S2. Establish monitoring data equations, use an improved batch estimation algorithm to fuse the preprocessed data, and verify the fused data; Step S3. Determine the corresponding measurement value of the specified indicator based on existing monitoring data; Step S4. Based on the corresponding metric values ​​of the indicators and the principle of set pair analysis, calculate the single-indicator correlation coefficient of each indicator to the evaluation level; Step S5. The entropy value method is used to calculate the connection coefficient of a single indicator by combining the set pair analysis principle and DS evidence theory. The weighted basic probability values ​​of each indicator are obtained. Based on the preset level of each indicator, the evidence synthesis rule of DS evidence theory is used to recursively fuse the data layer by layer, and finally the multi-source information fusion evaluation result of the dam's working performance is obtained. Step S6. The single index correlation coefficient is calculated by combining the entropy value assignment principle with the set pair analysis principle to obtain the composite index correlation coefficient. The partial correlation coefficient in the set pair analysis is then combined to calculate the partial positive correlation coefficient, the partial negative correlation coefficient, and the full partial correlation coefficient. Finally, the full partial correlation coefficient is used to obtain the final safety trend prediction result. Step S7. The final evaluation is obtained by combining the final safety trend prediction results with the multi-source information fusion evaluation results of the working performance.

2. The comprehensive evaluation and safety prediction method for the performance of dam reinforcement and safety improvement work as described in claim 1, characterized in that, In step S1, during the data preprocessing calculation, gross errors and outliers in the dam monitoring data are removed using the tolerance function threshold method in fuzzy set theory. The specific removal process includes: Define the universe of discourse for sensor measurement data Calculate any two of the data. and Data deviation distance for: in This is the arithmetic mean of the original data; and This is the sequence number of the sensor measurement data point, with a value of [value missing]. ;in Q This represents the total number of original measurement data. The allowance function is constructed as follows: ; In the formula The closer the value is to 1, the better the data consistency. Define the threshold determination function as follows: ; In the formula, ε is the threshold of the allowable function, when If the consistency between two data points is deemed insufficient, data points that deviate further from the original mean are identified as significant anomalies and removed, ultimately resulting in a consistent and valid data sequence. ,in, g This represents the number of valid data points after removing anomalies.

3. The comprehensive evaluation and safety prediction method for the performance of dam reinforcement and safety improvement work as described in claim 2, characterized in that, The specific operational procedures for fusing the preprocessed data using the improved batch estimation algorithm in step S2 include: Based on the characteristics of the dam monitoring area, the sensors are divided into categories according to their deployment locations. Grouping data to ensure spatial correlation in each group; For each set of valid data Classified by time series Batch, each batch contains The data point, the first Group 1 Batch data is ;in, s For sensor grouping sequence number, l Batch number (values ​​are 1, 2, ...). t ), For the first s The number of data points included in a batch; The calculated mean of the data is: ; The variance is calculated as follows: ; In the formula This is an outlier factor used to reduce the weight of extreme value data within a batch, thereby weakening the interference of extreme values ​​on mean estimation. The sequence number of the data point within a single batch; For the first s Group 1 l Variance of batch data. Dynamic weights are assigned based on the variance of each batch of data, then the... Group 1 The weights of the batch data are: ,in ; Calculate the first The fusion result of the group data is as follows: ; The final global fusion result of the dam monitoring data is as follows: ,in , For the first s The variance of the group fusion results The index number for grouped summation; After fusion is completed, step-by-step verification is performed. Based on the preset allowable thresholds of each monitoring indicator, the batch fusion results, group fusion results and global fusion results are verified for reasonableness in sequence. If the fusion result exceeds the corresponding safety allowable range or the fusion variance exceeds the preset limit, the fusion result at that level is determined to be invalid. The process returns to the data preprocessing stage to remove abnormal data and then performs fusion again. After the verification is passed, the valid fusion result is output.

4. The comprehensive evaluation and safety prediction method for the performance of dam reinforcement and safety improvement work as described in claim 1, characterized in that, The process for determining the corresponding metric value of the specified indicator in step S3 includes: The dam evaluation indicators and evaluation levels are systematically divided into 5 levels, and the set of evaluation levels is denoted as follows: These correspond to the dam's normal, basically normal, slightly abnormal, severely abnormal, and malignant abnormal states, respectively. The corresponding index value ranges for each level are [0.8,1], [0.6,0.8), [0.4,0.6), [0.2,0.4), and [0,0.2], respectively.

5. The comprehensive evaluation and safety prediction method for the performance of dam reinforcement and safety improvement work as described in claim 1, characterized in that, The process for obtaining the weighted basic probability value in step S5 includes: Determining the weights of underlying information using the entropy method and initial basic probability assignment Calculate the weighted basic probability assignment for: Indicates the first The evaluation index is for the first The weighted basic probability is assigned to each security level, and the remaining trust after weighting is allocated to the entire set of the recognition framework to ensure the normalization of the basic probability assignment. Among them, the initial basic probability assignment of the underlying information is generated by transforming the single index set pair connection number obtained in step S4, and the degree of sameness, first-level difference, second-level difference, third-level difference, and opposition are sequentially assigned to the five security levels as the initial trust assignment of each underlying evaluation index to each security level; then, the weighted basic probability assignment is obtained by combining the weight of the underlying index determined by the entropy method. Based on the pre-defined levels of each indicator, the basic probability assignment of the information in the next higher level is determined by the evidence synthesis rules of the DS evidence theory, with the lowest level designated as the first level and its basic probability assignment being... The basic probability of the second layer of information is assigned as follows: , ..., then the first The basic probability assignment of layer information is Then we have: in, The subscripts 1, 2, and 3 represent the sequence numbers of different evidence sources within the same level.

6. The comprehensive evaluation and safety prediction method for the performance of dam reinforcement and safety improvement work as described in claim 5, characterized in that, The process for obtaining the underlying information weights includes: Step 1: For One sample, The original data matrix for each evaluation indicator is as follows: in, For the first m The first sample n The measured values ​​of each evaluation indicator; Step 2: Normalize the data of each indicator to obtain a matrix. for: in, , These are the optimal and worst state limit values ​​corresponding to the same evaluation index. The corresponding values ​​are determined based on the differences in the safety attributes of the evaluation index to ensure that the normalization result is positively correlated with the dam's safety level. Step 3: Calculate the... n The entropy of each evaluation indicator is: in, , indicating the first n Under the evaluation index, the first m The weighting of indicators for each sample; r Let be the entropy normalization constant. Its function is to constrain the information entropy value to the interval [0,1], and when When =0, It's meaningless, to The calculation was corrected to ; Step 4: Calculate the weight of each evaluation indicator: in, For the first n The weight of each evaluation indicator.

7. The comprehensive evaluation and safety prediction method for the performance of dam reinforcement and safety improvement work as described in claim 6, characterized in that, The specific formula for obtaining the composite correlation coefficient of the indicators in step S6 includes: in, a, b, c, d, e These are the sameness, first-level difference, second-level difference, third-level difference, and opposition corresponding to the evaluation indicators of dam performance; These are the difference coefficients for Level 1, Level 2, and Level 3, respectively. j The degree of contrast is the coefficient of opposition. The same degree corresponds to the normal safe state of the dam, the degree of opposition corresponds to the abnormal state of the dam risk, and the three levels of difference correspond to the transitional state between the two in turn.

8. The comprehensive evaluation and safety prediction method for the performance of dam reinforcement work as described in claim 7, characterized in that, The weights of each indicator in step S6 are obtained using the same method as the weight set of the underlying information in step S5.

9. The comprehensive evaluation and safety prediction method for the performance of dam reinforcement and safety improvement work as described in claim 6, characterized in that, The specific calculation formulas for calculating the partially positive correlation coefficient, partially negative correlation coefficient, and fully partial correlation coefficient in step S6, combined with the partial correlation coefficient in set pair analysis, include: (1) The partial positive correlation coefficient is: ； ； ； in, , , , The first-order positive components, namely the same part, first-level different part, second-level different part, and third-level different part, respectively, represent the trend strength of each evaluation component evolving towards the same safe state. (2) The negative correlation coefficient is: ； ； ； in, , , , The first-order negative components, in order, are the same part, first-level different part, second-level different part, and third-level different part, representing the trend strength of each evaluation component evolving towards a state of risk opposition. (3) The partial correlation coefficient is: in, It is a first-order fully partial correlation coefficient, and its positive or negative value corresponds to the overall evolution direction of the dam's working performance: a positive value indicates that the overall working performance is developing in a positive direction, and a negative value indicates that the overall working performance is developing in a negative direction. The magnitude of the absolute value corresponds to the significance of the trend evolution.

10. The comprehensive evaluation and safety prediction method for the performance of dam reinforcement and safety improvement work as described in claim 9, characterized in that, For the fully partial correlation coefficient, when When this occurs, it indicates that the trend prediction object is in a critical trend that is neither negative nor positive, and it is necessary to find a higher-order partial derivative. when When this occurs, it indicates that the trend prediction object is developing negatively and there is a risk of deterioration, requiring timely and effective measures to reverse this trend; when This indicates that the trend forecast is developing in a positive direction, the dam's operational performance is continuously improving, its safety redundancy is steadily increasing, and the effectiveness of reinforcement and safety improvement is steadily becoming apparent.

Images