High arch dam deformation safety analysis method and system driven by monitoring point group monitoring data

Abstract

The invention discloses a high arch dam deformation safety analysis method and system driven by monitoring point group monitoring data, and belongs to the technical field of hydraulic structure safety intelligent monitoring. In the principal component space, constructing an SPE statistic and a CSPE statistic, and providing a gross error diagnosis criterion in combination with hypothesis testing; and constructing a gross error and missing component estimation method by applying a co-integration theory, an improved whale algorithm and an expanded Dybase-Fowler test. Constructing a deformation feature similar measuring point group division method by means of Euclidean distance, average connection distance and hierarchical clustering; establishing a measuring point group deformation behavior monitoring model, and optimizing and determining model parameters in combination with an improved whale algorithm; and comprehensively considering the deviation elastic state degree of the deformation principal component and the combined control limit of the deformation principal component, and proposing a measurement point group deformation safety evaluation criterion. The method can effectively represent the structural characteristics and the operation state of the high arch dam, is beneficial to actual operation, provides technical support for engineering safety monitoring, energy improvement and life prolonging, and has wide application prospects.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of safety intelligent monitoring of hydraulic structures, and more particularly to a high arch dam deformation safety analysis method and system driven by monitoring data of a group of measuring points. BACKGROUND

[0002] Up to now, China has built, is building and plans to build many arch dams with a height of more than 200 m. These dams are located in mountainous and gorges areas, and the engineering safety is restricted by many factors such as high reservoir water thrust, high stress level, high and steep slope stability, deep fault, earthquake, temperature crack, high-speed water flow and the like. Due to the catastrophic consequences of accidents, the state and society have put forward very high requirements for the safe service of high arch dams, and it is necessary to ensure that the project fully plays its social and economic benefits on the premise of long-term safe operation. The deformation change comprehensively reflects the dynamic change process of the service behavior of the high arch dam body and dam foundation, and the scientific analysis of the spatio-temporal variation law of the high arch dam deformation by using in-situ monitoring data is an important means to ensure the safe service of the project.

[0003] A large amount of deformation monitoring data can be obtained from the monitoring instruments arranged in the high arch dam body and dam foundation, and in-depth analysis of the in-situ monitoring information of the high arch dam deformation is not only the requirement of relevant specifications, but also the most direct means for the safety analysis of the high arch dam deformation. In the process of monitoring data collection, due to the influence of instruments, monitoring methods, human factors and the like, there are often contaminated information in the in-situ monitoring data, mainly including gross errors (represented by abnormal jumps of measured values at a certain moment or a certain period, i.e. isolated or spot-type gross errors) and missing components. If the gross errors and missing components in the monitoring data are not effectively processed, it is likely to affect the objectivity of the dam safety analysis, causing false alarms and missing of dangerous conditions. With the increase of service life, the performance and stability of the monitoring instruments become worse, and this problem will be particularly prominent. Summarizing the existing research results, the conventional monitoring data processing methods mostly take single-point deformation as the analysis object, and cannot effectively capture the deformation relationship between the measuring points, which is easy to cause problems such as misjudgment of gross errors, omission of gross errors, poor estimation accuracy of missing components and the like. Taking a group of measuring points as the analysis object has advantages, compared with one-dimensional time series, multi-dimensional data series contains hidden, novel and potentially valuable information, has the advantages of rich information quantity, high degree of freedom and strong stability, and provides an effective way for the processing of high arch dam deformation monitoring data.

[0004] The research on the deformation behavior monitoring and safety assessment method of high arch dam has made certain progress, and three basic analysis modes have been formed, namely: typical small probability method, confidence interval method and structure calculation method. High arch dam is a kind of spatial shell structure, and the load is transmitted to the mountain and dam foundation rock mass on both sides through the arch-beam system to realize the joint action of the dam body and the dam foundation. Most of the existing researches are aimed at single point deformation, and for high arch dam, single point deformation only reflects the local characteristics of the structure, and cannot effectively represent the spatial distribution of deformation and the mutual influence between the deformations of each part. With the deepening of research, the diagnosis method taking the deformation of the measuring point group as the analysis object has been put forward, mainly aiming at improving the typical small probability method and the confidence interval method, such as information entropy method, confidence ellipsoid method and the like. However, the above methods still have certain deficiencies, the information entropy method is based on the principle of probability statistics, which is convenient to use, but cannot effectively represent the influence of structure type, load, dam material, terrain, geology and the like on the deformation of the dam; the confidence ellipsoid method is based on the monitoring model, which is related to the actual state of the project to a certain extent, and has strict probability and physical meaning, but the construction process of the confidence ellipsoid is complex, and it is difficult to promote to the super high dimension. Considering the structure characteristics and operation state of high arch dam, a convenient deformation behavior monitoring and safety assessment method of measuring point group is researched and put forward, which has important theoretical significance and application value.

[0005] In summary, by applying advanced theories such as mathematics, mechanics, dam engineering and artificial intelligence, a high arch dam deformation safety analysis method driven by measuring point group monitoring data is researched, and on this basis, an advanced visual analysis system is constructed to provide theoretical, methodological and technical support for the safe operation of high arch dams, which is a key scientific problem and technical bottleneck that needs to be solved by technical personnel in the field of intelligent monitoring of hydraulic structures, and is of great significance to ensure the long-term safe service of projects. SUMMARY

[0006] Therefore, the application provides a high arch dam deformation safety analysis method and system driven by measuring point group monitoring data to solve the bottleneck problem in the background art.

[0007] In order to achieve the above purpose, the application adopts the following technical solutions.

[0008] On the one hand, the application provides a high arch dam deformation safety analysis method driven by measuring point group monitoring data, which comprises the following steps:

[0009] S1, a rough error and missing component processing method for high arch dam measuring point group deformation monitoring data is constructed, comprising:

[0010] S11, a rough error diagnosis method for measuring point group deformation monitoring data is established, and the specific steps are as follows:

[0011] S111, based on the in-situ monitoring data, the deformation principal component of the measuring point group is extracted;

[0012] S112, establishing SPE statistics and CSPE statistics;

[0013] S113, establishing a gross error identification criterion combined with hypothesis testing;

[0014] S12, constructing a gross error and missing component estimation method for deformation monitoring data of a measuring point group, and the specific steps are:

[0015] S121, establishing a cointegration model for deformation monitoring sequences of multiple measuring points;

[0016] S122, searching for an optimal measuring point set for cointegration analysis based on an improved whale algorithm;

[0017] S123, establishing an optimal cointegration model using the searched optimal measuring point set, and testing the stationarity of the residual sequence of the optimal cointegration model combined with an extended Durbin-Watson test;

[0018] S124, estimating the true values of gross errors and missing components using the optimal cointegration model, and correcting the original in-situ monitoring sequence;

[0019] S2, constructing a high arch dam measuring point group deformation behavior monitoring and safety assessment method, including:

[0020] S21, constructing a deformation characteristic similar measuring point group division method, and the specific steps are:

[0021] S211, establishing a deformation similarity degree representation method between measuring points based on Euclidean distance;

[0022] S212, establishing a deformation similarity degree representation method between measuring point groups based on average connection distance;

[0023] S213, establishing a deformation characteristic similar measuring point group division criterion based on hierarchical clustering and mutation principle;

[0024] S22, constructing a measuring point group deformation behavior monitoring model for deformation characteristic similar measuring point groups, and the specific steps are:

[0025] S221, extracting deformation principal components of deformation characteristic similar measuring point groups, and establishing a measuring point group deformation behavior monitoring model;

[0026] S222, establishing a monitoring model parameter estimation method combined with an improved whale algorithm;

[0027] S223, evaluating the effectiveness of monitoring model parameter estimation;

[0028] S23, establishing a measuring point group deformation behavior safety assessment method, and the specific steps are:

[0029] S231, according to the analysis result of the deformation state monitoring model of the measuring point group, a judging method of the deviation degree of the deformation principal component from the elastic state is constructed based on the confidence interval method;

[0030] S232, in the principal component space, a Hotelling T 2 statistic is constructed, and by means of the multivariate statistical theory and the typical small probability principle, a joint control limit of the deformation principal component is established;

[0031] S233, the deviation degree of the deformation principal component from the elastic state and the joint control limit of the deformation principal component are comprehensively considered, and the evaluation criterion of the deformation state of the measuring point group is established, including normal, basically normal, mild abnormality and severe abnormality.

[0032] On the other hand, the application provides a measuring point group monitoring data driven high arch dam deformation safety analysis system, comprising a measuring point group deformation in-situ monitoring data processing module, a monitoring model-safety evaluation module and a three-dimensional visualization display module connected in sequence;

[0033] The measuring point group deformation in-situ monitoring data processing module is used to realize S1, and comprises a gross error diagnosis unit and a gross error and missing component estimation unit connected in sequence.

[0034] The gross error diagnosis unit is used to realize S11, and comprises a deformation principal component extraction subunit, a statistical quantity construction subunit and a gross error identification criterion subunit connected in sequence.

[0035] The deformation principal component extraction subunit is used to realize S111.

[0036] The statistical quantity construction subunit is used to realize S112.

[0037] The gross error identification criterion subunit is used to realize S113.

[0038] The gross error and missing component estimation unit is used to realize S12, and comprises a cointegration model subunit, an optimal measuring point set search subunit, a residual sequence stationarity test subunit and a monitoring sequence correction subunit connected in sequence.

[0039] The cointegration model subunit is used to realize S121.

[0040] The optimal measuring point set search subunit is used to realize S122.

[0041] The residual sequence stationarity test subunit is used to realize S123.

[0042] The monitoring sequence correction subunit is used to realize S124.

[0043] The monitoring model-safety evaluation module is used to realize S2, and comprises a deformation characteristic similar measuring point group division unit, a measuring point group deformation state monitoring model unit and a measuring point group deformation state safety evaluation criterion unit connected in sequence.

[0044] The deformation feature similar measuring point group division unit is used to implement S21, and includes a measuring point interval deformation similarity degree representation subunit, a measuring point group interval deformation similarity degree representation subunit and a clustering criterion subunit connected in sequence;

[0045] The measuring point interval deformation similarity degree representation subunit is used to implement S211;

[0046] The measuring point group interval deformation similarity degree representation subunit is used to implement S212;

[0047] The clustering criterion subunit is used to implement S213;

[0048] The measuring point group deformation state monitoring model unit is used to implement S22, and includes a model construction subunit, a model parameter estimation subunit and a parameter estimation effectiveness evaluation subunit connected in sequence;

[0049] The model construction subunit is used to implement S221;

[0050] The model parameter estimation subunit is used to implement S222;

[0051] The parameter estimation effectiveness evaluation subunit is used to implement S223;

[0052] The measuring point group deformation state safety assessment criterion unit is used to implement S23, and includes a deformation principal component deviation from elastic state degree assessment subunit, a deformation principal component joint control limit subunit and a comprehensive assessment criterion subunit, the comprehensive assessment criterion subunit being connected with the deformation principal component deviation from elastic state degree assessment subunit and the deformation principal component joint control limit subunit simultaneously;

[0053] The deformation principal component deviation from elastic state degree assessment subunit is used to implement S231;

[0054] The deformation principal component joint control limit subunit is used to implement S232;

[0055] The comprehensive assessment criterion subunit is used to implement S233;

[0056] The three-dimensional visual display module is used to present the analysis results to an operator in the form of a cloud chart, an isogram, a distribution chart, a process line and a table.

[0057] According to the technical solution, compared with the prior art, the application provides a measuring point group monitoring data driven high arch dam deformation safety analysis method and system, and has the following beneficial effects:

[0058] (1) Taking advantage of the rich information, high degree of freedom and strong stability of multidimensional data sequences, we can process the gross errors and missing components in the in-situ monitoring data of deformation of the measuring point group. In the principal component space, we can construct the SPE statistic and CSPE statistic, and combine hypothesis testing to propose gross error diagnosis criteria. We can apply cointegration theory, improved whale algorithm and extended Dickie-Fowler test to build a method for estimating gross errors and missing components. The aim is to improve the objectivity and scientificity of deformation behavior monitoring and safety assessment of the high arch dam measuring point group, and reduce the probability of false alarms and missed dangerous conditions.

[0059] (2) By using Euclidean distance, average connectivity distance and hierarchical clustering, a method for dividing measurement point groups with similar deformation characteristics is constructed; a monitoring model for the deformation behavior of measurement point groups is established, and the model parameters are optimized and determined by combining the improved whale algorithm; considering the degree of deviation of the deformation principal components from the elastic state and the joint control limit of the deformation principal components, a safety assessment method for the deformation behavior of measurement point groups is proposed; compared with the information entropy method, the safety assessment method of this invention has more rigorous identification conditions for abnormal deformation, and is easier to operate in practice than the confidence ellipsoid method, thereby improving the accuracy and efficiency of deformation behavior monitoring and safety assessment of in-service high arch dams;

[0060] (3) To address the shortcomings of the whale algorithm, an adaptive adjustment strategy is introduced to enable the algorithm to adapt from global fast search to local precise search; a differential evolution mutation strategy is used to enhance the diversity of the whale population; Gaussian mutation perturbation and Tent chaotic perturbation are introduced to reduce the probability of the algorithm getting trapped in local optima; a binary algorithm is introduced to convert the solution space from continuous values ​​to discrete values ​​to solve the discrete search problem of measurement points; based on the improved whale algorithm, an optimal measurement point set search method for cointegration analysis and a parameter optimization method for monitoring the deformation behavior of measurement point groups are constructed; the improved whale algorithm can significantly improve the search efficiency of the optimal measurement point set and the accuracy of monitoring model parameter optimization, and enhance the model's adaptability and reliability.

[0061] (4) This invention can not only effectively characterize the structural characteristics and operating status of high arch dams, but also facilitate practical operation, providing technical support for engineering safety monitoring and service life extension. Attached Figure Description

[0062] Figure 1 A flowchart of a high arch dam deformation safety analysis method driven by monitoring data from a group of measuring points provided by this invention;

[0063] Figure 2 This invention provides a flowchart of a method for processing gross errors and missing components in deformation monitoring data of a high arch dam measuring point group;

[0064] Figure 3 This invention provides a flowchart of a method for monitoring the deformation behavior and assessing the safety of a group of measuring points for a high arch dam.

[0065] Figure 4 This invention provides an architecture for a high arch dam deformation safety analysis system driven by monitoring data from a group of measuring points.

[0066] Figure 5 This invention provides a layout diagram of vertical monitoring points for a high arch dam.

[0067] Figure 6 This invention provides a radial deformation monitoring sequence for vertical measuring points;

[0068] Figure 7(a) shows the first deformation principal element t1 process line of the 32 vertical monitoring points provided by the present invention; Figure 7(b) shows the SPE of the 32 vertical monitoring points provided by the present invention. k With SPE α＝0.01 relation;

[0069] Figure 8(a) shows the iterative process of IWOA search for the optimal measurement point set provided by the present invention; Figure 8(b) shows the iterative process of WOA search for the optimal measurement point set provided by the present invention.

[0070] Figure 9(a) shows the monitoring value, the fitted value of the optimal cointegration model, and the residual sequence of PL9-1 provided by the present invention; Figure 9(b) shows the estimation results of the gross errors and missing components of PL9-1 by the autoregressive model, statistical model, and optimal cointegration model provided by the present invention.

[0071] Figure 10 Distribution diagram of similar measurement point group for deformation features provided by the present invention;

[0072] Figure 11 The process line of the first deformable principal element t1 of the measurement point group A to C provided by the present invention;

[0073] Figure 12 The fitting result of the IWOA optimization model for the first deformable principal element t1 of the measurement point group A to C provided by the present invention;

[0074] Figure 13 The prediction results of the IWOA optimization model, WOA optimization model and GLS optimization model for the first deformable principal element t1 of the measurement point group A to C provided by the present invention;

[0075] Figure 14 The relationship between the residuals of the IWOA optimization model for measurement point group A to C provided by this invention and the 2S and 3S control limits;

[0076] Figure 15 Hotelling T for measuring point group A to C provided by this invention 2 The relationship between statistics and joint control limits of transformed principal components. Detailed Implementation

[0077] 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 some embodiments of the present invention, and not all embodiments. 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.

[0078] This invention discloses a method for deformation safety analysis of high arch dams driven by monitoring data from a group of measuring points, such as... Figure 1 As shown.

[0079] S1. Methods for handling gross errors and missing components in deformation monitoring data of high arch dam measuring point group.

[0080] S11. Establish a method for diagnosing gross errors in deformation monitoring data of measuring point groups.

[0081] S111. Based on in-situ monitoring data, extract the deformation principal elements of the measuring point group.

[0082] The in-situ monitoring data matrix δ is constructed as follows:

[0083]

[0084] In formula (1): m is the total number of monitoring points; n is the monitoring duration; i and j (i≠j and i,j=1~m) are the monitoring point numbers; k is a certain monitoring date; δ i and δ j These are the deformation monitoring sequences for measuring points i and j, respectively; δ k,i and δ k,j These are the monitoring values ​​of monitoring points i and j on the k-th monitoring day, respectively.

[0085] If no abnormal conditions occur when analyzing environmental monitoring data, the in-situ monitoring data matrix δ is standardized to obtain a standardized matrix. The matrix form is consistent with equation (1), denoted as For standardized δ k,i The formula is:

[0086]

[0087] In equation (2): k, i and δ k,i The meaning is consistent with equation (1); δ i,max and δ i,min The deformation monitoring sequence δ at measuring point i is respectively i The maximum and minimum values.

[0088] Calculate the normalized matrix The correlation coefficient matrix R is given by the formula:

[0089]

[0090]

[0091] In equations (3) and (4): i, j, m, k, and n have the same meanings as in equation (1); r i,j for and Correlation coefficient; and For standardized δ k,i and δ k,j .

[0092] Construct the eigenvector matrix P and eigenvalue matrix Λ of the correlation coefficient matrix R:

[0093]

[0094] The calculation formula is:

[0095] (R-ΛG)P=0 (7)

[0096] In equations (5) to (7): m and j have the same meaning as in equation (1); φ is the number of the eigenvector and eigenvalue; λ φ Let φ be the eigenvalue of the correlation coefficient matrix R, satisfying λ1≥λ2≥…λ φ …≥λ m ;p φ p is the φ-th eigenvector of the correlation coefficient matrix R; φ,j φ is the element in the φ-th row and j-th column of the eigenvector matrix P; 0 is the m-th zero matrix; G is the m-th identity matrix.

[0097] Will Perform a rotation transformation to construct the rotation effect quantity matrix T:

[0098]

[0099] In equation (8): P, φ, m, k, and n have the same meanings as in equations (1) and (5); t φ t is the φth rotational effect, which is the φth column of the rotational effect matrix T; k,φ Let be the element in the k-th row and φ-th column of the rotation effect matrix Τ; This is a standardized matrix.

[0100] t φ The explanatory power of the original sequence δ is e φ The calculation formula is:

[0101]

[0102] The meaning of the parameters in equation (9) is the same as that in equation (6).

[0103] Based on cumulative explanatory power The number of morphing principal components, l, is determined, with a cumulative explanatory power exceeding 90% used as the criterion for identifying morphing principal components. This yields the morphing principal component matrix T′={t1 t2…t ψ …t l}, where ψ is the principal component number, and the principal component is the first l columns of the rotation effect matrix T.

[0104] S112. Establish SPE and CSPE statistics.

[0105] Since the eigenvector matrix P is an orthogonal matrix, it satisfies:

[0106] P T =P -1 (10)

[0107] Therefore, the standardized matrix Established as:

[0108]

[0109] Establish a standardized matrix Main regularity matrix The formula is:

[0110]

[0111] In equations (10) to (12): the meaning of T is consistent with that of equation (8); t ψ p is the ψth deformable principal element; ψ Let ψ be the ψth eigenvector of the correlation coefficient matrix R.

[0112] The SPE statistic for monitoring day k is denoted as SPE. k The calculation formula is:

[0113]

[0114] In equation (13): m, j, k, and The meaning is consistent with equations (1), (11), and (12); and Each is a matrix and The element in the k-th row and j-th column.

[0115] S113. Combine hypothesis testing to establish gross error identification criteria.

[0116] Based on hypothesis testing, at a significance level of α, a SPE is established. k SPE control limits α The formula is:

[0117]

[0118] In equations (14) to (16): C α This represents the value of the normal distribution function corresponding to the significance level α. h0 is used to calculate SPE α Intermediate parameters; for The numbers; m, φ and λ φ The meaning is consistent with equations (1), (5) and (6).

[0119] The criteria for identifying the types of gross errors and the timing of their occurrence are as follows:

[0120] If SPE k Not exceeding SPE α Then there are no gross errors;

[0121] If SPE k Exceeding SPE at a certain time period α If the monitoring value exceeds the time period, then the value is a speckled gross error;

[0122] If SPE k Exceeding SPE at some point α If the monitoring value exceeds the specified time, then the value is an isolated gross error.

[0123] If gross errors exist, the identification criteria for measurement points containing gross errors are as follows:

[0124] Measurement point j for SPE on monitoring day k k Exceeding SPE α Contribution level using CSPE k,j Statistical representation, namely:

[0125]

[0126] The meaning of the parameters in equation (17) is the same as that in equation (13).

[0127] The CSPE during the analysis period was calculated based on equation (17). k,j Value, CSPE k,j The larger the value, the better the SPE of monitoring point j on the k-th monitoring day. k The greater the contribution of exceeding the limit, the higher the CSPE. k,j The measuring point with the largest value is the measuring point containing gross errors.

[0128] S12. Construct a method for estimating gross errors and missing components in deformation monitoring data of measuring point groups.

[0129] S121. Establish a cointegration model for multi-point deformation monitoring sequence.

[0130] Assuming there are m measuring points, f of which contain gross errors, the deformation monitoring sequence is denoted as δ1′…δ h ′…δ′ f Where h is the number of the measurement points containing gross errors; the remaining w = mf measurement points do not contain gross errors, and the deformation monitoring sequence is denoted as δ1″…δ q "…δ′ w ′, where q is the number of the measurement point excluding gross errors.

[0131] Establish a cointegration model:

[0132]

[0133] In equation (18): c qg For δ q The coefficient of ″; g is δ q The number of operations for ""; a q For sequence δ q The highest frequency of ″ is determined based on the relationship of the deformed scatter points; ε is the residual sequence.

[0134] By applying the stepwise regression algorithm, the coefficient c of equation (18) is determined. qg The residual sequence ε is obtained.

[0135] S122. Based on the improved whale algorithm, search for the optimal set of measurement points for cointegration analysis.

[0136] The Whale Optimization Algorithm (WOA) has advantages such as simple parameter settings and fast convergence speed. This embodiment establishes a search strategy for the measurement point set based on WOA. The basic principle of WOA will not be elaborated here.

[0137] The standard WOA has certain drawbacks. On the one hand, the convergence factor a = 2 - 2·τ / τ max The linear decrease from 2 to 0 causes the coefficient parameter A = 2a·Rand(0,1)-a to continuously shrink. When using contraction and spiral ascent for position updates, the distance between individual whales and the optimal position continuously decreases, causing the convergence speed of WOA to gradually slow down in the later stages of iteration. On the other hand, the algorithm gradually transitions from global search to local search in the later stages of iteration, reducing the differences between individuals and making the population more concentrated and similar, thus reducing population diversity. Furthermore, as the coefficient parameter A decreases, the update range of the whale population also continuously shrinks. Once trapped in a local optimum, it is difficult to escape effectively, thus causing premature convergence.

[0138] To address the aforementioned issues, we present an improvement strategy for the algorithm, establishing an improved WOA (IWOA).

[0139] (1) To address the problem of slow convergence speed, a nonlinear convergence factor is introduced. and nonlinear adaptive weights χ=(τ / τ max ) 2 In the formula: τ is the number of iterations; τ max This represents the maximum number of iterations.

[0140] The shrink wrap update mode has been improved to:

[0141] U=|C·bI z (τ)| (19)

[0142] I z (τ+1)=b·χ-A·U(20)

[0143] In equations (19) and (20): ρ is I z Element number; d is I z The dimension of the whale; U is the distance between the individual whale and the optimal position; τ is the number of iterations; τ max The maximum number of iterations is denoted by z; the individual whale number is denoted by I. z (τ)={I z,1 I z,2 …I z,ρ …I z,d} represents the position of the z-th whale during the τ-th iteration; I z,ρ For I z The ρ-th dimension element; I z (τ+1) represents the position of the z-th whale in the (τ+1)-th iteration; b = {b1 b2 … b} ρ …b d} represents the optimal position; A is the coefficient parameter; C = 2·Rand(0,1); Rand(0,1) is a random number within [0,1].

[0144] The spiral ascent mechanism has been improved to:

[0145] U=|bI z (τ)| (21)

[0146]

[0147] In equations (21) and (22): The constants controlling the shape of the spiral; β∈[-1,1] are random numbers, where the whale is closest to its prey when β=-1 and furthest when β=1; τ, z, U, b, χ, I z (τ+1) has the same meaning as equations (19) and (20).

[0148] The random search position update method has been improved to:

[0149] U=|C·b Rand -I z (τ)| (23)

[0150] I z (τ+1)=b Rand ·χ-A·U(24)

[0151] In equations (23) and (24): b Rand For randomly selected individual whales; τ, z, U, b, χ, I z (τ+1) has the same meaning as equations (19) and (20).

[0152] (2) To increase the individual diversity of the whale population, a differential evolution mutation strategy is introduced. Two random whale individuals in the population are differentially scaled and combined with another random whale individual to form a new individual. The expression is as follows:

[0153]

[0154] In equation (25): and Three distinct individuals are randomly selected from the population; F is a scaling factor between [0,2]; τ and I z (τ+1) has the same meaning as equations (19) and (20).

[0155] For IWOA, based on the relationship between the random number σ in the range [0,1] and |A|, different position update mechanisms are implemented, as follows:

[0156] When σ≥0.5, the spiral ascent mechanism is executed;

[0157] When σ < 0.5 and |A| ≥ 1, a random search mechanism is executed;

[0158] When σ < 0.5 and |A| < 1 and Rand(0,1) ≤ Cr, the shrinking and wrapping mechanism is executed to update the individual position, where Rand(0,1) is a random number in [0,1] and Cr is the crossover probability;

[0159] When σ < 0.5 and |A| < 1 and Rand(0,1) > Cr, the differential evolution mutation strategy is executed.

[0160] (3) To address the premature convergence problem, Gaussian mutation perturbation and Tent chaotic perturbation are executed alternately for I. z (τ+1) is used for perturbation update.

[0161] The expression for the Gaussian mutation perturbation is:

[0162] Gaussian(I) = I z,ρ +Iz,ρ ·Norm(0,1)(26)

[0163] In equation (26): Gaussian(I) is the position after Gaussian mutation perturbation; Norm(0,1) is a random number that follows a standard normal distribution; z and ρ have the same meaning as in equations (19) and (20).

[0164] The expression for the Tent chaotic mapping is:

[0165]

[0166] In equation (27): v is the Tent perturbation mapping value; K is the number of individuals in the whale population; z, ρ, I z,ρ The meaning of Rand(0,1) is consistent with that of equations (19) and (20).

[0167] The expression for the Tent chaotic perturbation is:

[0168] v′=I min +v(I max -I min (28)

[0169]

[0170] In equations (28) to (30): v′ is the intermediate parameter of the Tent chaotic perturbation; γ is the attenuation coefficient; I max and I min These are the upper and lower bounds of the solution space, respectively; The direction coefficient has a value of 1 or -1; Tent(I) is the perturbation position after the Tent chaotic perturbation; z, τ, ρ and I z,ρ The meaning is consistent with equations (19) and (20).

[0171] Using the multiple correlation coefficient and residual standard deviation of the cointegration model as the individual fitness f z The formula for calculating the average fitness of a whale population is: The meaning of the parameters in the formula is consistent with that in formula (27). If f z <f avg If necessary, implement a Gaussian mutation perturbation; otherwise, implement a Tent chaotic perturbation.

[0172] (4) The solution space of IWOA is continuous. To solve the discrete search problem of measurement points, a binary algorithm is introduced to transform the solution space from a continuous domain to a discrete domain. The transformation function is:

[0173]

[0174] The meaning of the parameters in equation (31) is consistent with that in equations (19) and (20).

[0175] Introduce the label matrix ξ = {ξ1ξ2…ξ} j …ξ m},ξ j The value of ξ is either 0 or 1. j =1 indicates the selection of the j-th measurement point, ξ j =0 means abandoning the j-th measurement point, and determine the optimal measurement point set based on the label matrix.

[0176] The parameters that IWOA needs to initialize are K, F, and C. r τ max I max I min and The parameter values ​​were determined through trial and error.

[0177] S123. Using the optimal set of measurement points found, establish an optimal cointegration model, and combine it with the extended Dickie-Fowler test to test the stationarity of the residual sequence of the optimal cointegration model.

[0178] If ε is stationary, then the optimal cointegration model holds; if it is not stationary, then the optimal cointegration model does not hold.

[0179] S124. Using the optimal cointegration model, estimate the true values ​​of gross errors and missing components, and correct the in-situ monitoring sequence.

[0180] Execute S11 to identify whether there are still gross errors in the corrected matrix δ. If there are, execute S12 again and repeat the above steps until there are no gross errors.

[0181] This invention provides a process for processing gross errors and missing components in deformation monitoring data from a group of measuring points for high arch dams, as follows: Figure 2 As shown.

[0182] S2. Construct a method for monitoring the deformation behavior and assessing the safety of a group of measuring points for high arch dams.

[0183] S21. Construct a method for dividing deformation feature similar measurement point groups.

[0184] S211. Based on Euclidean distance, establish a method for characterizing the degree of deformation similarity between measuring points.

[0185] To characterize the degree of deformation similarity between measurement points i and j, a comprehensive distance d is introduced. i,j , is represented as:

[0186]

[0187] In equation (32): ω1 and ω2 are weights; Δδ k,i =δ k,i -δ k-1,i ;d i,j,1 and d i,j,2These represent the degree of similarity in absolute deformation features and the degree of similarity in trend features, respectively; δ k-1,i and δ k-1,j Let i, j, k, n, and δ be the monitoring values ​​of monitoring points i and j on the (k-1)th monitoring day, respectively; i, j, k, n, and δ are the values ​​of the monitoring points i and j, respectively. k,i and δ k,j The meaning is consistent with equation (1).

[0188] ω1 and ω2 are obtained using the entropy weight method.

[0189] S212. Based on the average connection distance, establish a method for characterizing the degree of deformation similarity among measurement point groups.

[0190] For the measurement point group G p and G q Using average connection distance D p,q metric G p and G q The degree of similarity in deformation between them, that is:

[0191]

[0192] In equation (33): n p and n q G p and G q The number of measurement points; i, j, and d i,j The meaning is consistent with equation (32).

[0193] When n p and n q When both are 1, equation (33) degenerates into equation (32), and d i,j and D p,q This is collectively referred to as the composite distance.

[0194] S213. Based on hierarchical clustering and mutation principles, establish a criterion for dividing groups of measurement points with similar deformation characteristics.

[0195] Initially, all measurement points form their own groups. The two groups of measurement points with the smallest distance are merged into a new group. This process is repeated until all measurement points become the same comprehensive group. A phylogenetic clustering tree diagram is drawn, and the optimal grouping scheme is obtained based on the mutation principle.

[0196] S22. For a group of measuring points with similar deformation characteristics, construct a monitoring model for the deformation behavior of the measuring point group.

[0197] S221. Extract the principal components of the deformation of the measurement point group with similar deformation characteristics, and establish a monitoring model of the deformation behavior of the measurement point group.

[0198] For details on the extraction steps of the principal elements of the deformation feature similar measurement point group, please refer to S111.

[0199] The principal component ψ is deformed. ψRepresented as:

[0200] t ψ =p ψ,1 δ1+p ψ,2 δ2+…+p ψ,j δ j +…+p ψ,m δ m (34)

[0201] In equation (34): j, m and δ j The meaning is consistent with equation (1); ψ is the principal component number of the transformation; p ψ,j is the element in the ψ-th row and j-th column of the correlation coefficient matrix R.

[0202] For δ j ,satisfy:

[0203] δ j =δ j,H +δ j,T +δ j,θ (35)

[0204] In equation (35): δ j,H δ j,T and δ j,θ δ j Water pressure, temperature, and aging time components.

[0205] Substituting equation (35) into equation (34) and combining like terms, we get:

[0206]

[0207] In equation (36): j, m, ψ, p ψ,j δ j,H δ j,T and δ j,θ The meaning is consistent with equations (1), (34), and (35); and These are the water pressure component, temperature component, and aging component of the deformation principal element, respectively.

[0208] δ j,H δ j,T and δ j,θ Represented as:

[0209]

[0210] δ j,θ =c j,1 θ+c j,2 lnθ(39)

[0211] In equations (37) to (39): f and s are integer variables; a j,fb j,1,s b j,2,s c j,1 and c j,2 The coefficients are undetermined; H is the reservoir water level sequence; n is the cumulative monitoring time series; δ j,H δ j,T and δ j,θ The meaning is consistent with equation (35).

[0212] Substituting equations (37) to (39) into equation (36), we get:

[0213]

[0214] Therefore, t ψ Build as:

[0215]

[0216] In equations (40) to (43): a0 is a constant term; j, m, ψ, H, n, f, s, θ and p ψ,j The meaning is consistent with equations (1), (34) and (36) to (39).

[0217] S222. Combine the improved whale algorithm to establish a parameter estimation method for the monitoring model.

[0218] The parameters that need to be estimated for the deformation behavior monitoring model of the measuring point group include: a0, a j,f b j,1,s b j,2,s c j,1 and c j,2 .

[0219] The WOA is improved by introducing nonlinear convergence factors, nonlinear adaptive weights, differential evolution mutation strategy, Gaussian mutation perturbation and Tent chaotic perturbation, as shown in equations (19) to (30).

[0220] The root mean square error of the deformation behavior monitoring model of the measurement point group is used as the individual fitness.

[0221] S223. Evaluate the effectiveness of the parameter estimation for the monitoring model.

[0222] The effectiveness of IWOA in estimating parameters for the deformation behavior monitoring model of a group of measuring points was evaluated using the multiple correlation coefficient R, residual standard deviation S, and mean absolute percentage error MAPE.

[0223] S23. Establish a method for assessing the safety of deformation behavior of a group of measuring points.

[0224] S231. Based on the analysis results of the deformation behavior monitoring model of the measuring point group, a method for determining the degree of deviation of the principal deformation element from the elastic state is constructed based on the confidence interval method.

[0225] By utilizing the relationship between the residuals of the deformation behavior monitoring model of the measuring point group and the 2S and 3S control limits, where S is the residual standard deviation of the deformation behavior monitoring model of the measuring point group, the degree of deviation is quantitatively characterized.

[0226] S232. In the main dimension space, construct Hotelling T 2 Using statistics, and drawing on multivariate statistical theory and the principle of typical small probability, we establish joint control limits for modified principal components.

[0227] For l mutually independent t1~t2 l Construct a deformable principal component matrix T′={t1 t2…t ψ …t l} and deformed principal component eigenvalue matrix

[0228] In the principal space, Hotelling T on the k-th monitoring day 2 The statistic is denoted as The expression is:

[0229]

[0230] In equation (44): υ is the k-th row of the deformed principal matrix Τ′.

[0231] Based on hypothesis testing, let the confidence level be α, and the joint control limits of the transformed principal components be denoted as... The expression is:

[0232]

[0233] In equation (45): F α (l,nl) represents the F-distribution function value with confidence level α, first degree of freedom l, and second degree of freedom nl; l is the number of deformed principal components; and n is the monitoring duration.

[0234] S233. Taking into account the degree of deviation of the principal deformation element from the elastic state and the joint control limit of the principal deformation element, establish the evaluation criteria for the deformation behavior of the measuring point group as normal, basically normal, slightly abnormal and severely abnormal.

[0235] normal: And all t ψ All satisfy |ε|≤2S;

[0236] Basically normal: And there exists any t ψ The condition 2S < |ε| ≤ 3S is satisfied;

[0237] Mild abnormality: And there exists any t ψ The condition |ε| > 3S is satisfied;

[0238] Severe abnormality:

[0239] Where: ε is the residual of the deformation behavior monitoring model of the measuring point group; S is the residual standard deviation of the deformation behavior monitoring model of the measuring point group; ψ, t ψ , and The meaning is consistent with equations (36), (44) and (45).

[0240] The present invention provides a method for monitoring the deformation behavior and safety assessment of a group of measuring points for a high arch dam, as follows: Figure 3 As shown.

[0241] Based on the above method, a deformation safety analysis system for high arch dams driven by monitoring data from a group of measuring points is provided below, such as... Figure 4 As shown, it includes a data processing module for in-situ deformation monitoring of a group of measuring points, a monitoring model-safety assessment module, and a three-dimensional visualization display module, which are connected in sequence.

[0242] The in-situ monitoring data processing module for deformation of the measuring point group is used to realize S1, including a gross error diagnosis unit and a gross error and missing component estimation unit connected in sequence.

[0243] The gross error diagnosis unit is used to implement S11, including the deformable principal component extraction subunit, the statistics construction subunit, and the gross error identification criterion subunit connected in sequence.

[0244] The deformable principal element extraction subunit is used to implement S111;

[0245] The statistical construct sub-unit is used to implement S112;

[0246] The gross error identification criterion subunit is used to implement S113;

[0247] The gross error and missing component estimation unit is used to implement S12, which includes a cointegration model subunit, an optimal measurement point set search subunit, a residual sequence stationarity test subunit, and a monitoring sequence correction subunit connected in sequence.

[0248] The cointegration model subunit is used to implement S121;

[0249] The optimal measurement point set search subunit is used to implement S122;

[0250] The residual sequence stationarity test subunit is used to implement S123;

[0251] The monitoring sequence correction subunit is used to implement S124;

[0252] The monitoring model-safety assessment module is used to implement S2, which includes a deformation feature similarity measurement point group division unit, a measurement point group deformation behavior monitoring model unit, and a measurement point group deformation behavior safety assessment criterion unit connected in sequence.

[0253] The deformation feature similarity measurement point group division unit is used to realize S21, including the deformation similarity characterization subunit between measurement points, the deformation similarity characterization subunit between measurement point groups, and the clustering criterion subunit connected in sequence.

[0254] The sub-unit representing the degree of deformation similarity between measurement points is used to implement S211;

[0255] The sub-unit representing the degree of deformation similarity among measurement point groups is used to implement S212;

[0256] The clustering criterion subunit is used to implement S213;

[0257] The deformation behavior monitoring model unit of the measuring point group is used to implement S22, including a model construction sub-unit, a model parameter estimation sub-unit, and a parameter estimation effectiveness evaluation sub-unit connected in sequence.

[0258] The model building subunit is used to implement S221;

[0259] The model parameter estimation subunit is used to implement S222;

[0260] The parameter estimation validity evaluation subunit is used to implement S223;

[0261] The deformation behavior safety assessment criterion unit of the measuring point group is used to realize S23, including the deformation principal element deviation from the elastic state assessment subunit, the deformation principal element joint control limit subunit, and the comprehensive assessment criterion subunit. The comprehensive assessment criterion subunit is connected to both the deformation principal element deviation from the elastic state assessment subunit and the deformation principal element joint control limit subunit.

[0262] The sub-evaluation unit for assessing the degree of deviation of the deformable principal element from the elastic state is used to implement S231;

[0263] The deformable principal element is used to jointly control the finite element to implement S232;

[0264] The comprehensive evaluation criterion subunit is used to implement S233;

[0265] The 3D visualization module is used to present the analysis results to the operator in the form of cloud maps, contour maps, distribution maps, process lines, and tables.

[0266] Taking the radial deformation of a high arch dam as an example, the effectiveness of the method and system is verified. The dam body is a parabolic double-curvature arch dam with a crest elevation of 1885m, a maximum height of 305m, a crest width of 16m, a maximum span of 480m, a normal water level of 1880m, and a dead water level of 1800m. The vertical measuring points are arranged as follows... Figure 5 As shown, Figure 6 This is the process line for in-situ monitoring of radial deformation at vertical measuring points.

[0267] a. Gross error diagnosis results

[0268] Executing S111 yields the first deformation principal element t1 for 32 vertical monitoring points, as shown in Figure 7(a). Executing S112 yields SPE. k The process curve is shown in Figure 7(b). With the confidence level α set to 0.01, the calculated SPE is... α＝0.01 =18.2032. As shown in Figure 7(b), SPE k <SPE α＝0.01 Execution of S113 showed that the in-situ deformation monitoring data of the 32 measuring points did not contain gross errors.

[0269] To verify the effectiveness of the gross error diagnosis method of this invention, the radial deformation monitoring value of PL5-4 on October 3, 2014, was 7mm, which was used to construct isolated gross errors of different degrees, denoted as cases 0# to 6#, namely: 7.5mm, 8.5mm, 9.5mm, 11mm, 13mm, 16mm, and 19mm. Under cases 0# to 6#, SPE k All exceeded SPE α＝0.01 Furthermore, the CSPE statistic of PL5-4 is significantly higher than that of the other measurement points. Therefore, under conditions 0# to 6#, the monitoring data of PL5-4 on October 3, 2014, are isolated gross errors. Table 1 compares the performance of the gross error identification method of this invention, the Laida method, the Dixon method, and the t method. The gross error identification method of this invention has the best performance.

[0270] Table 1 Performance comparison of the four methods

[0271]

[0272] b. Estimation results of gross errors and missing components

[0273] Assuming the monitoring data from measuring point PL9-1 from July 15, 2016 to September 15, 2016 are speckled gross errors, and using in-situ monitoring data from January 1, 2014 to July 14, 2016, S121 is executed to establish a cointegration model. The parameters of IWOA and WOA are initialized to K = 1000, F = 0.6, and C... r =0.8, t max =500, Imax =1、I min =0 and Executing S122 yields the iterative processes of IWOA and WOA in searching for the optimal measurement point set, as shown in Figures 8(a) and (b). IWOA and WOA converge on the 189th and 238th iterations, respectively, with cointegration model fitting multiple correlation coefficients of 0.99996 and 0.91442, respectively. IWOA's search performance is superior to WOA's. The optimal measurement point set searched by IWOA is PL9-2, PL9-5, PL11-1, PL11-4, PL13-3, PL16-1, PL16-2, and IP16-1. Based on this, an optimal cointegration model is established.

[0274] Based on the scatter plot relationship, the highest power of equation (18) is always taken as 1. Using the stepwise regression method, the parameters of the optimal cointegration model are estimated, and the following is obtained:

[0275]

[0276] In equation (46): δ′ PL9-1 ,δ″ PL9-2 ,δ″ PL9-5 ,δ″ PL11-1 ,δ″ PL11-4 ,δ″ PL13-3 ,δ″ PL16-1 ,δ″ PL16-2 ,δ″ IP16-1 and δ″ PL19-3 The radial deformation in-situ monitoring sequences for measuring points PL9-1, PL9-2, PL9-5, PL11-1, PL11-4, PL13-3, PL16-1, PL16-2, IP16-1, and PL19-3 are shown in Figure 9(a). After performing S123, the extended Dickey-Fowler test results show that the null hypothesis was rejected at both the 1% and 5% significance levels, indicating that the residual sequence ε is stationary. Figure 9(b) shows the estimation results of the autoregressive model, statistical model, and optimal cointegration model for gross errors. The multiple correlation coefficients of the three models are 0.7432, 0.8715, and 0.9997, respectively, with the optimal cointegration model showing the best estimation performance. S124 is then performed to correct the original deformation monitoring sequence.

[0277] c. Monitoring model and security assessment results

[0278] After passing through S211 to S213, the 32 measuring points were finally divided into three groups of measuring points with similar deformation characteristics. Figure 10It can be seen that measuring point group A is located in the upper part of the structure, measuring point group B is located in the middle part of the structure, and measuring point group C is adjacent to the dam abutment and dam foundation. The first deformation principal element t1 of measuring point groups A through C is extracted respectively. The explanatory power of the first deformation principal element t1 of measuring point groups A through C for the original monitoring sequence is 98.2%, 97.4%, and 93.3%, respectively, all greater than 90%. Therefore, the first deformation principal element t1 reflects the main characteristics of the deformation of measuring point groups A through C, such as... Figure 11 As shown.

[0279] The deformation behavior monitoring model of the measuring point group A to C is constructed using S221 to S223, targeting the first deformation principal element t1. The parameters of IWOA are initialized as follows: K = 1500, F = 0.6, C... r =0.8, t max =500, I max =1、I min =0 and The fitting results of the IWOA optimization model are as follows: Figure 12 As shown, for measuring point groups A to C, the R values ​​are 0.9980, 0.9986, and 0.9971, respectively; the S values ​​are 0.1885, 0.1569, and 0.2667, respectively; and the MAPE values ​​are 0.1643, 0.2730, and 0.2952, respectively. The model shows a good fit and generally reflects the deformation variation law of the dam. To test the extensionality of the IWOA optimization model, an optimization model was established by combining generalized least squares estimation (GLS) and WOA, and compared with the IWOA optimization model. Data from September 27, 2016 to December 31, 2016 were used as the prediction sample. Figure 13 As shown in Table 2, the IWOA optimization model has the best extensional performance, followed by the WOA optimization model, and the GLS optimization model has the worst performance.

[0280] Table 2 Predictive Performance Evaluation Indicators

[0281]

[0282] Executing S231 yields the relationship between the residuals of the IWOA optimized model for measurement point group A to C and the 2S and 3S control limits, as follows: Figure 14 Execute S232, with a confidence level of α = 0.01, and calculate... Hotelling T of measuring point group A~C 2 The relationship between the statistic and the joint control limits of the transformed principal components is as follows: Figure 15 As shown. Therefore, the radial deformation of the three measuring point groups all satisfy... And |ε|≤2S. Executing S233, the deformation behavior of the dam is judged to be normal, but due to the divergence of the time-dependent components of the principal deformation elements, deformation monitoring needs to be strengthened and the subsequent development trend needs to be continuously monitored.

[0283] Calculation results show that the time-dependent deformation components of measuring point group A exhibit a unique upstream trend, inconsistent with common understanding. Analysis of in-situ monitoring data reveals two main points: First, in the initial stage of impoundment, the riverbank rock mass creeps towards the riverbed, the valley width continuously shrinks, and the dam body is compressed by the mountain, resulting in an upstream deformation trend. Second, under the influence of upstream water pressure, the reservoir subsides, and the difference between upstream and downstream subsidence drives the dam body to exhibit upstream-oriented deformation. Therefore, for the locations of measuring point group A, such as the top of the high cantilever beam and the crown of the large-span arch ring, the dam body is relatively thin, less constrained by the dam foundation and riverbank, and has greater flexibility, making it more sensitive to load and environmental changes. The structure is prone to exhibiting unconventional deformation patterns. Furthermore, the narrowing of the valley width and the compression of the dam body by the slope may lead to buckling failure of the dam structure. Therefore, these areas should be given special attention during deformation monitoring.

[0284] Compared to the information entropy method, the safety assessment method of this invention connects structural features and operational status, making the identification of abnormal deformation more rigorous. When a condition changes from basically normal to slightly abnormal or from slightly abnormal to severely abnormal, the spatiotemporal similarity characteristics of the deformation of the surface measuring point group change to varying degrees. Furthermore, compared to the confidence ellipsoid method, the safety assessment method of this invention is more conducive to practical applications.

[0285] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0286] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for deformation safety analysis of high arch dams driven by monitoring data from a group of measuring points, characterized in that, Includes the following steps; S1. Constructing methods for handling gross errors and missing components in deformation monitoring data of high arch dam measuring point groups, including: S11. Establish a method for diagnosing gross errors in deformation monitoring data of measuring point groups. The specific steps are as follows: S111. Based on in-situ monitoring data, extract the deformation principal components of the measuring point group; S112. Establish the SPE statistic and CSPE statistic; S113. Establish gross error identification criteria by combining hypothesis testing; S12. Construct a method for estimating gross errors and missing components in deformation monitoring data of measuring point groups. The specific steps are as follows: S121. Establish a cointegration model for multi-point deformation monitoring sequences; S122. Based on the improved whale algorithm, search for the optimal set of measurement points for cointegration analysis; S123. Using the optimal set of measurement points found, establish an optimal cointegration model, and combine it with the extended Dickie-Fowler test to test the stationarity of the residual sequence of the optimal cointegration model. S124. Using the optimal cointegration model, estimate the true values ​​of gross errors and missing components, and correct the in-situ monitoring sequence. S2. Construct a method for monitoring the deformation behavior and assessing the safety of a group of measuring points for high arch dams, including: S21. Construct a method for dividing deformation feature similar measurement point groups. The specific steps are as follows: S211. Based on Euclidean distance, establish a method for characterizing the degree of deformation similarity between measuring points; S212. Based on the average connection distance, establish a method for characterizing the degree of deformation similarity among measurement point groups; S213. Based on hierarchical clustering and mutation principles, establish the criteria for dividing groups of measurement points with similar deformation characteristics; S22. For a group of measuring points with similar deformation characteristics, construct a deformation behavior monitoring model for the measuring point group. The specific steps are as follows: S221. Extract the principal components of the deformation of the measurement point group with similar deformation characteristics, and establish a monitoring model of the deformation behavior of the measurement point group. S222. Combine the improved whale algorithm to establish a parameter estimation method for the monitoring model; S223. Evaluate the effectiveness of the parameter estimation for the monitoring model; S23. Establish a method for assessing the deformation behavior of a group of measuring points. The specific steps are as follows: S231. Based on the analysis results of the deformation behavior monitoring model of the measuring point group, a method for determining the degree of deviation of the principal deformation element from the elastic state is constructed based on the confidence interval method. S232. In the main dimension space, construct Hotelling T 2 Statistics, using multivariate statistical theory and the principle of typical small probability, establish joint control limits for modified principal components; S233. Taking into account the degree of deviation of the principal deformation element from the elastic state and the joint control limit of the principal deformation element, establish the evaluation criteria for the deformation behavior of the measuring point group as normal, basically normal, slightly abnormal and severely abnormal.

2. The method for deformation safety analysis of high arch dams driven by monitoring data from a group of measuring points according to claim 1, characterized in that, In step S113, the gross error identification criteria are as follows; The SPE statistic for monitoring day k is denoted as SPE. k The calculation formula is: In the formula: j is the measurement point number; m is the total number of monitoring points; k is a certain monitoring date; δ is the in-situ monitoring data matrix; Let δ be the normalized matrix; The main regularity matrix of δ; and Each is a matrix and The element in the k-th row and j-th column; Based on hypothesis testing, at a significance level of α, a SPE is established. k SPE control limits α , is represented as: In the formula: m is the total number of monitoring points; C α This represents the value of the normal distribution function corresponding to the significance level α. h0 is used to calculate SPE α Intermediate parameters; for The eigenvalues ​​are numbered; φ is the number of the eigenvectors and eigenvalues; λ φ This represents the φ-th eigenvalue of the correlation coefficient matrix R. The criteria for identifying the types of gross errors and the timing of their occurrence are as follows: If SPE k Not exceeding SPE α Then there are no gross errors; If SPE k Exceeding SPE at a certain time period α If the monitoring value exceeds the time period, then the value is a speckled gross error; If SPE k Exceeding SPE at some point α If the monitoring value exceeds the specified time, then the value is an isolated gross error. If gross errors exist, the identification criteria for measurement points containing gross errors are as follows: Measurement point j for SPE on monitoring day k k Exceeding SPE α Contribution level using CSPE k,j Statistical representation, namely: The meanings of the parameters in the formula are the same as those of SPE. k Consistent; Calculate CSPE during the analysis period k,j Value, CSPE k,j The larger the value, the better the SPE of monitoring point j on the k-th monitoring day. k The greater the contribution of exceeding the limit, the higher the CSPE. k,j The measuring point with the largest value is the measuring point containing gross errors.

3. The method for deformation safety analysis of high arch dams driven by monitoring data from a group of measuring points according to claim 1, characterized in that, In step S122, the improvement strategy for the whale algorithm is as follows; By introducing nonlinear convergence factors and nonlinear adaptive weights, the position update modes of contraction-encirclement, spiral ascent, and random search are improved; By using a differential evolution mutation strategy, two random whale individuals in the population are scaled using a differential vector and then combined with another random whale individual to form a new individual, thereby enhancing the diversity of the population. Alternately apply Gaussian mutation perturbation and Tent chaotic perturbation to update the whale's position, reducing the probability of getting trapped in local optima; To solve the discrete search problem of measurement points, a binary algorithm is introduced to transform the solution space of the whale algorithm from a continuous domain to a discrete domain. The transformation function is as follows: In the formula: z is the individual whale number; I z ={I z,1 I z,2 …I z,ρ …I z,d } represents the position of the z-th whale; ρ represents I. z Element number; d is I z Dimensions; I z,ρ For I z The ρ-th element; Rand(0,1) is a random number in [0,1].

4. The method for deformation safety analysis of high arch dams driven by monitoring data from a group of measuring points according to claim 1, characterized in that, In step S221, the method for constructing the deformation behavior monitoring model of the measuring point group is as follows; Extract the principal components of the deformation feature group of similar measurement points; The principal component t of the ψth deformation ψ Represented as: t ψ =p ψ,1 δ1+p ψ,2 d2+…+p ψ,j d j +…+p ψ,m d m In the formula: j is the measuring point number; m is the total number of monitoring points; ψ is the deformation principal element number; δ j For the deformation monitoring sequence of measuring point j; p ψ,j Let be the element in the ψ-th row and j-th column of the correlation coefficient matrix R; δ j satisfy: d j =d j,H +d j,T +d j,θ Where: δ j,H δ j,T and δ j,θ δ j The water pressure component, temperature component, and aging component; Let δ1, δ2…δ j …δ m Substitute t ψ By combining like terms, we get: In the formula: j is the measuring point number; m is the total number of monitoring points; ψ is the deformation principal element number; p ψ,j Let be the element in the ψ-th row and j-th column of the correlation coefficient matrix R; and These are respectively the hydraulic pressure component, the temperature component, and the aging component of the deformation principal element; δ j,H δ j,T and δ j,θ Build as: d j,θ =c j,1 θ+c j,2 lnθ In the formula: f and s are integer variables; a j,f b j,1,s b j,2,s c j,1 and c j,2 The coefficients are undetermined; H is the reservoir water level sequence; u is the cumulative monitoring time series; δ j,H δ j,T and δ j,θ Substitute into t ψ,H t ψ,T and t ψ,θ In the middle, we get: Therefore, t ψ Build as: In the formula: a0 is a constant term; j is the measurement point number; m is the total number of monitoring points; ψ is the deformation principal element number; p ψ,j Let be the element in the ψ-th row and j-th column of the correlation coefficient matrix R; f, s, a j,f b j,1,s b j,2,s c j,1 c j,2 The meanings of H, u, and θ are the same as those of δ. j,H δ j,T and δ j,θ The expressions are consistent.

5. The method for deformation safety analysis of high arch dams driven by monitoring data from a group of measuring points according to claim 1, characterized in that, In step S222, the parameter estimation method for the deformation behavior monitoring model of the measuring point group is as follows; The parameters that need to be estimated for the deformation behavior monitoring model of the measuring point group include: the constant term a0 and the undetermined coefficients a. j,f b j,1,s b j,2,s c j,1 and c j,2 Parameter optimization is performed based on the whale algorithm; To improve the performance of parameter optimization, nonlinear convergence factor, nonlinear adaptive weight, differential evolution mutation strategy, Gaussian mutation perturbation and Tent chaotic perturbation are introduced to improve the whale algorithm. The root mean square error of the deformation behavior monitoring model of the measurement point group is used as the individual fitness.

6. The method for deformation safety analysis of high arch dams driven by monitoring data from a group of measuring points according to claim 1, characterized in that, In step S232, the method for constructing the joint control limit of the deformed principal components is as follows: For l mutually independent deformable principal elements t1~t l Construct a deformable principal component matrix T′={t1 t2…t ψ …t l } and deformed principal component eigenvalue matrix Based on multivariate probability and statistics theory, Hotelling T on the k-th monitoring day 2 The statistic is denoted as The expression is: In the formula: υ is the k-th row of the deformed principal component matrix T′; Based on hypothesis testing, let the confidence level be α, and the joint control limits of the transformed principal components be denoted as... The expression is: In the formula: F α (l,nl) represents the F-distribution function value with confidence level α, first degree of freedom l, and second degree of freedom nl; l is the number of deformed principal components; and n is the monitoring duration.

7. The method for deformation safety analysis of high arch dams driven by monitoring data from a group of measuring points according to claim 1, characterized in that, In step S233, the safety assessment criteria for the deformation behavior of the measuring point group are as follows; Taking into account both the degree of deviation of the principal deformation elements from the elastic state and the joint control limit of the principal deformation elements, evaluation criteria for normal, basically normal, slightly abnormal, and severely abnormal deformation behavior of the measuring point group are established. normal: And all t ψ All satisfy |ε|≤2S; Basically normal: And there exists any t ψ The condition 2S < |ε| ≤ 3S is satisfied; Mild abnormality: And there exists any t ψ The condition |ε| > 3S is satisfied; Severe abnormality: In the formula: ψ is the principal component number of the deformation; t ψ Let ψ be the ψth principal deformation element; ε be the residual of the deformation behavior monitoring model of the measurement point group; and S be the residual standard deviation of the deformation behavior monitoring model of the measurement point group. Hotelling T for monitoring day k 2 Statistic; The joint control limit for the deformed principal components.

8. A deformation safety analysis system for high arch dams driven by monitoring data from a group of measuring points, characterized in that, A method for analyzing the deformation safety of a high arch dam driven by monitoring data from a group of measuring points, as described in any one of claims 1 to 7, comprises a module for processing in-situ monitoring data of deformation from a group of measuring points, a monitoring model-safety assessment module, and a three-dimensional visualization display module connected in sequence. The in-situ monitoring data processing module for deformation of the measuring point group is used to realize S1, including a gross error diagnosis unit and a gross error and missing component estimation unit connected in sequence. The gross error diagnosis unit is used to implement S11, including the deformable principal component extraction subunit, the statistics construction subunit, and the gross error identification criterion subunit connected in sequence. The deformable principal element extraction subunit is used to implement S111; The statistical construct sub-unit is used to implement S112; The gross error identification criterion subunit is used to implement S113; The gross error and missing component estimation unit is used to implement S12, which includes a cointegration model subunit, an optimal measurement point set search subunit, a residual sequence stationarity test subunit, and a monitoring sequence correction subunit connected in sequence. The cointegration model subunit is used to implement S121; The optimal measurement point set search subunit is used to implement S122; The residual sequence stationarity test subunit is used to implement S123; The monitoring sequence correction subunit is used to implement S124; The monitoring model-safety assessment module is used to implement S2, which includes a deformation feature similarity measurement point group division unit, a measurement point group deformation behavior monitoring model unit, and a measurement point group deformation behavior safety assessment criterion unit connected in sequence. The deformation feature similarity measurement point group division unit is used to realize S21, including the deformation similarity characterization subunit between measurement points, the deformation similarity characterization subunit between measurement point groups, and the clustering criterion subunit connected in sequence. The sub-unit representing the degree of deformation similarity between measurement points is used to implement S211; The sub-unit representing the degree of deformation similarity among measurement point groups is used to implement S212; The clustering criterion subunit is used to implement S213; The deformation behavior monitoring model unit of the measuring point group is used to implement S22, including a model construction sub-unit, a model parameter estimation sub-unit, and a parameter estimation effectiveness evaluation sub-unit connected in sequence. The model building subunit is used to implement S221; The model parameter estimation subunit is used to implement S222; The parameter estimation validity evaluation subunit is used to implement S223; The deformation behavior safety assessment criterion unit of the measuring point group is used to realize S23, including the deformation principal element deviation from the elastic state assessment subunit, the deformation principal element joint control limit subunit, and the comprehensive assessment criterion subunit. The comprehensive assessment criterion subunit is connected to both the deformation principal element deviation from the elastic state assessment subunit and the deformation principal element joint control limit subunit. The sub-evaluation unit for assessing the degree of deviation of the deformable principal element from the elastic state is used to implement S231; The deformable principal element is used to jointly control the finite element to implement S232; The comprehensive evaluation criterion subunit is used to implement S233; The 3D visualization module is used to present the analysis results to the operator in the form of cloud maps, contour maps, distribution maps, process lines, and tables.

Images