Sluice multi-monitoring-point deformation monitoring model construction method and deformation prediction method

Abstract

The invention discloses a sluice multi-monitoring-point deformation monitoring model construction method, which comprises the steps of S1, constructing a group of bidirectional analysis frameworks, and forming a two-dimensional panel by fusing time sequence data and sluice cross section data; s2, expanding the two-dimensional panel containing the time dimension and the cross section dimension to a geographic space attribute level to construct a three-dimensional panel; s3, on the basis of three-dimensional panel data representation, performing clustering analysis according to the space-time similarity of the monitoring points, constructing an information criterion to evaluate candidate clustering numbers, performing parameter calibration by adopting a random coefficient model, constructing a Gaussian mixture model in each cluster as a multi-monitoring-point model, then estimating undetermined model parameters through the random coefficient model, and performing multi-monitoring-point monitoring on the multi-monitoring-point model. The method is used for constructing a deformation monitoring model of the sluice gate and carrying out fitting and prediction capability evaluation. According to the method, collaborative modeling can be carried out on data of a plurality of monitoring points, deformation space correlation characteristics between adjacent monitoring points are represented, and the accuracy and effectiveness of sluice deformation prediction are improved.

Description

Technical Field

[0001] The present invention relates to the field of water conservancy project safety technology, and in particular to a method for constructing a sluice multi-monitoring point deformation monitoring model based on IC-OGMM and RCM, a sluice multi-monitoring point deformation prediction method using the deformation monitoring model, a device and a storage medium. Background Art

[0002] As a low-head hydraulic structure built on rivers and canals, sluices regulate the flow rate through the sluices and the water levels upstream and downstream through gates. Their operation can bring comprehensive benefits such as flood control, irrigation, and drainage. They can serve as key control facilities in water conservancy projects and play a core role in regulating flood control, drainage, and ensuring navigation safety. Therefore, it is crucial to ensure the safe and reliable operation of sluices.

[0003] Analyses of the actual operation of multiple projects have shown that deformation is the most intuitive monitoring variable reflecting the operating status of hydraulic structures. Historical data show that most geological disasters and engineering accidents are related to deformation damage, which is mainly manifested as sudden deformation increases and crack expansion. Therefore, how to construct an accurate deformation monitoring model during operation has become a key task to ensure structural safety. Among them, the deformation monitoring value of the sluice is affected by complex environmental factors during operation, including water pressure, temperature, etc.

[0004] Based on the principles of structural mechanics, existing technologies conventionally adopt a water pressure-season-time (HST) model that comprehensively considers water level changes, ambient temperature changes, and time effects. The deformation of the sluice gate is predicted by fitting a polynomial equation of the upstream water level of the reservoir. Alternatively, deformation prediction methods based on SOA-LSTM and Markov models are used, which can achieve reliable results in short-term deformation prediction. However, from a mathematical theoretical perspective, existing statistical models and various artificial intelligence models all independently analyze and predict the explanatory variable coefficients of each monitoring point, ignoring the spatial correlation between monitoring points. In other words, existing sluice gate deformation prediction methods mostly focus on the deformation analysis of a single monitoring point, ignoring the correlation between deformation data from multiple monitoring points.

[0005] In addition, since the deformations of adjacent monitoring points are actually correlated, and the water pressure, temperature and other loads borne by the sluice are all gradually changing, and the existing models often introduce redundant explanatory variables to obtain higher fitting accuracy, this may weaken the generalization ability of the model.

[0006] To this end, this application specifically proposes a method for constructing a sluice multi-monitoring point deformation monitoring model and a deformation prediction method to solve the above technical problems. Summary of the Invention

[0007] The main purpose of the present invention is to provide a method for constructing a deformation monitoring model for multiple monitoring points of a sluice and a deformation prediction method, which uses IC-OGMM to classify deformation monitoring points based on the spatiotemporal correlation of monitoring data; then, a random coefficient prediction model for the deformation of multiple monitoring points is constructed within each category to realize the deformation prediction of multiple monitoring points of the sluice, thereby solving the technical problems raised in the background technology.

[0008] The present invention adopts the following technical solutions to solve the above technical problems: A method for constructing a deformation monitoring model for a sluice with multiple monitoring points, comprising: S1. Construct a two-way analysis framework to form a two-dimensional panel by fusing time series data with sluice cross-section data; S2. Expand the two-dimensional panel containing time and cross-sectional dimensions to the geospatial attribute level to construct a three-dimensional panel; S3. Based on the three-dimensional panel data representation, cluster analysis was performed according to the spatiotemporal similarity of the monitoring points, and an information criterion was constructed to evaluate the number of candidate clusters. A random coefficient model was used for parameter calibration, and a Gaussian mixture model was constructed within each cluster as a multi-monitoring point model. The random coefficient model was then used to estimate the unknown model parameters. This was used to construct a deformation monitoring model for the sluice gates, and the fitting and prediction capabilities were evaluated.

[0009] Preferably, the construction process of the Gaussian mixture model in step S3 includes: S31. The distance d from the central axis of the water gate and the maximum absolute value of the time series and the degree of convergence of the data series As the clustering criterion, to characterize the spatiotemporal characteristics of the monitoring points, where , and is the time-effect term coefficient; S32. Construct and iteratively converge a Gaussian mixture density model function to cluster the dataset into several categories; S33. The candidate number of clusters K is evaluated by minimizing the Bayesian information criterion or the Akaike information criterion, and is used to construct an optimized Gaussian mixture model based on the information criterion as a multi-monitoring point model. Preferably, the expression of the Gaussian mixture density model function in step S32 is:

[0010] Where k represents the number of Gaussian distribution; i represents the number of data dimension; n represents the number of data sequence; I represents the total number of dimensions of data; is the weight of the k-th Gaussian distribution; is the mean of the k-th Gaussian distribution in dimension i; is the variance of the k-th Gaussian distribution in dimension i; is the value of the nth data point in dimension i.

[0011] Preferably, the mean is respectively ,variance and weights Perform iteration, and the iteration formula is:

[0012]

[0013]

[0014] in Represents the Gaussian density function, data set is a set of samples independently sampled from this distribution.

[0015] Preferably, in the step S33, for each candidate K value, the candidate cluster number K is evaluated using the minimization Bayesian information criterion or Akaike information criterion, and the calculation expression is:

[0016]

[0017] Where L(K) is the log-likelihood of the Gaussian mixture model containing K clusters, and N is the number of samples; The optimal number of clusters K is determined by minimizing the AIC or BIC value.

[0018] Preferably, the specific operation process of estimating the undetermined model parameters by using the random coefficient model in step S3 includes: S31. Based on the multi-monitoring point model, the panel data regression coefficients without time variability are constructed as follows:

[0019] in, Represents the two-dimensional deformation data of the sluice gate; Two-dimensional data representing explanatory variables; Represents a time index; represents the cross-section index; represents the explanatory variable index; Indicates that it does not change over time and can be divided into and ; is the common mean coefficient vector, is the deviation of individual data from the common mean; u is a random interference term; S32. Construct a parameter estimator based on the random coefficient model to estimate the unknown parameters in the panel data regression coefficients.

[0020] Preferably, the specific operation process of step S32 includes: Will Assuming it is a random variable and integrating the NT observation data, the matrix expression can be obtained as follows:

[0021]

[0022]

[0023]

[0024] The optimal linear unbiased estimator is used to estimate the β value,

[0025] Where N is the number of panels and T is the number of data in each panel.

[0026] On the other hand, the present invention also discloses a method for predicting deformation of a sluice gate at multiple monitoring points, comprising: L1. Use any of the above-described methods for constructing a sluice multi-monitoring point deformation monitoring model to construct a deformation monitoring model; L2. Use the deformation monitoring model, input real-time sluice gate data, and execute and output deformation prediction results at multiple monitoring points.

[0027] In another aspect, the present invention further discloses a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the steps of the above method.

[0028] On the other hand, the present invention further discloses a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the above method.

[0029] As can be seen from the above technical solution, the present invention provides a method for constructing a deformation monitoring model for a sluice with multiple monitoring points and a deformation prediction method. Compared with the prior art, the present invention has the following advantages: 1. The present invention constructs a two-dimensional panel that integrates time series data and sluice cross-sectional data, and expands it to the geographic space attribute level to construct a three-dimensional panel. It can integrate time, cross-sectional and geographic space multi-dimensional data, thereby facilitating cluster analysis using spatiotemporal similarity to provide richer individual dynamic behavior information and more accurate estimation results, thereby providing more comprehensive data support for sluice deformation prediction and improving model prediction accuracy.

[0030] 2. The present invention uses IC-OGMM to perform cluster analysis based on the spatiotemporal similarity of monitoring points based on three-dimensional panel data, and constructs an information criterion to evaluate the number of candidate clusters. This allows for reasonable classification based on the spatiotemporal characteristics of monitoring points, avoiding the effects of the number of groups and overfitting that plague traditional clustering methods. The coefficients within each group can be modeled as sampling results that follow the same normal distribution, enhancing the generalization ability of the model, facilitating more objective and accurate classification of monitoring points, and improving model reliability.

[0031] 3. The present invention uses a random coefficient model for parameter calibration, constructs a multi-monitoring point model in each cluster, and estimates the parameters of the undetermined model. It can collaboratively model the data of multiple monitoring points and characterize the spatial correlation characteristics of deformation between adjacent monitoring points, thereby avoiding independent modeling of the data of each monitoring point and solving the problem that traditional models ignore the spatial correlation between monitoring points, so as to improve the accuracy and effectiveness of sluice deformation prediction.

[0032] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present invention, nor is it intended to limit the scope of the present invention. Other features of the present invention will become easy to understand through the following description. Of course, it is not necessary to achieve all of the above-mentioned advantages simultaneously in order to implement any product of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are intended to explain the present invention and do not constitute an undue limitation of the present invention. In the accompanying drawings: Figure 1 It is a schematic diagram of the overall process of the present invention; Figure 2 This is a schematic diagram of the multi-monitoring point deformation fitting and prediction calculation process of the present invention; Figure 3 is a schematic diagram of two-dimensional panel data of the present invention; Figure 4 A schematic diagram for constructing three-dimensional panel data of the present invention; Figure 5 This is a schematic diagram of the spatiotemporal representation of monitoring data of the present invention; Figure 6 This is a schematic diagram of the arrangement of gate vertical displacement monitoring points according to an embodiment of the present invention; Figure 7 A schematic diagram of a deformation sequence of monitoring points according to an embodiment of the present invention; Figure 8 Schematic diagram of monitoring point classification based on IC-OGMM clustering according to an embodiment of the present invention; Figure 9Schematic diagram of the distribution of monitoring point clustering results according to an embodiment of the present invention; Figure 10 Schematic diagram of fitting and prediction results according to an embodiment of the present invention; Figure 11 Schematic diagram comparing the correlation coefficients of the deformation monitoring model (multi-monitoring point model), statistical model, BP model and LSTM model of the present invention; Figure 12 Schematic diagram comparing the residual standard deviations of the deformation monitoring model (multi-monitoring point model), statistical model, BP model and LSTM model of the present invention. DETAILED DESCRIPTION

[0034] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. In the absence of conflict, the embodiments in this application and the features in the embodiments can be combined with each other. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0035] In the embodiment, see Figures 1 to 12 .

[0036] like Figure 1 and Figure 2 As shown, a method for constructing a multi-monitoring point deformation monitoring model for a sluice gate proposed in an embodiment of the present invention includes: S1. Construct a two-way analysis framework to form a two-dimensional panel by fusing time series data with sluice cross-section data; Specifically, the panel data here is Figure 3 As shown in Figure 2, this refers to data collected from longitudinal observations of the same group of individuals over a period of time. Unlike traditional time series, this data structure encompasses both temporal and cross-sectional dimensions, forming a bidirectional analytical framework that can simultaneously characterize deformation patterns at all monitoring points.

[0037] In a specific embodiment, there are n deformation monitoring points and t time points in the sluice gate. The structure of the panel data can be shown in the following table:

[0038] Obviously, panel data has two dimensions, n and t, and is composed of a time series of a set of monitoring points, while integrating time series data and cross-sectional data.

[0039] The advantage of the panel data proposed above is that its simultaneous time series and cross-sectional dimensions can provide richer individual dynamic behavior information and more accurate estimation results.

[0040] S2. Expand the two-dimensional panel containing time and cross-sectional dimensions to the geospatial attribute level to construct a three-dimensional panel.

[0041] It should be noted that, in sluice deformation monitoring, the spatial interactions between adjacent monitoring points may show mutual influence. Therefore, based on the traditional panel data framework, the spatial coordinates (or spatial relationships) of each monitoring point can be incorporated to effectively expand the two-dimensional panel data containing time and cross-sectional dimensions to the geographic spatial attribute level. In this case, the spatial panel data structure is shown in the following table:

[0042] As shown in the table, the evolution from time series data, cross-sectional data, panel data, and finally spatial panel data reflects the characteristics of the gradual increase in data dimensions and the exponential growth of sample capacity. Therefore, the expansion of this data modeling framework has geometrically magnified the information potential available for analysis.

[0043] The above steps S1 and S2 can refer to Figure 4 As shown, the constructed spatial panel data contains the geographic spatial coordinates of the monitoring points and has good spatial correlation characteristics. Compared with traditional representation methods, the information capacity of this data framework is significantly improved, making it an ideal carrier for spatiotemporal data analysis. Therefore, the spatiotemporal data analysis content of this application is all constructed based on spatial panel data.

[0044] In a specific embodiment, Figure 3 As shown in Figure 2, the gate monitoring data has two-dimensional spatiotemporal characteristics, which includes both the time series evolution and the distribution of cross-sectional measuring points. Therefore, a single panel data sequence reflects the deformation characteristics of the gate section at a specific moment, while the complete monitoring data constitutes a two-dimensional spatiotemporal panel at any monitoring moment.

[0045] At this time, by constructing a two-dimensional panel that integrates time series data and sluice cross-sectional data, and extending it to the geospatial attribute level to construct a three-dimensional panel, it is possible to integrate multi-dimensional data of time, cross-section and geospatial space, thereby facilitating cluster analysis using spatiotemporal similarity to provide richer individual dynamic behavior information and more accurate estimation results, thereby providing more comprehensive data support for sluice deformation prediction and improving model prediction accuracy.

[0046] S3. Based on the characterization of three-dimensional panel data characteristics, cluster analysis was performed according to the spatiotemporal similarity of monitoring points, and the information criterion was constructed to evaluate the number of candidate clusters. The random coefficient model was used for parameter calibration. A Gaussian mixture model was constructed within each cluster as a multi-monitoring point model. The parameters of the model were then estimated using the random coefficient model. This model was used to construct a deformation monitoring model for the sluice gate, and the fitting and prediction capabilities were evaluated. Figure 5 .

[0047] Among them, since the probability density distribution of each monitoring point is strongly correlated with its inherent characteristics (especially spatial location), cluster analysis is performed based on the spatiotemporal similarity of the monitoring points. At the same time, since the results of the traditional GMM clustering method are easily affected by the number of groups K, it is necessary to propose an automated framework based on the information criterion to objectively determine the K value. Through clustering, monitoring points with similar spatiotemporal evolution patterns are classified into the same group, so that the coefficients within the group can be modeled as sampling results that obey the same normal distribution.

[0048] The construction process of the Gaussian mixture model includes: S31. The distance d from the central axis of the water gate and the maximum absolute value of the time series and the degree of convergence of the data series As the clustering criterion, to characterize the spatiotemporal characteristics of the monitoring points, where , and is the time-effect term coefficient; It should be noted that according to traditional hydraulic engineering theory and mechanical principles, the deformation of the sluice gate is mainly caused by three components: water pressure component , temperature component and time-dependent components , according to the following analysis, we can construct clustering criteria that can characterize the spatiotemporal characteristics of monitoring points: Water pressure component and temperature components It mainly depends on the location of the monitoring points and the geometric dimensions of the sluice gate. For concrete sluice gates, the spatial relationship between the monitoring points can be characterized by their distance d to the center axis of the sluice gate flow direction; The time characteristics of the monitoring point mainly affect the timeliness component , the time-dependent component can be separated from the time series measurement data , the time characteristics are described by two factors: one is the maximum absolute value of the time series , the other is the degree of convergence of the data series ,use Indicates that and is the time-effect coefficient.

[0049] S32. Construct and iteratively converge a Gaussian mixture density model function to cluster the data set into several categories. The expression of the Gaussian mixture density model function is:

[0050] Where k represents the number of Gaussian distribution; i represents the number of data dimension; n represents the number of data sequence; I represents the total number of dimensions of data; is the weight of the k-th Gaussian distribution; is the mean of the k-th Gaussian distribution in dimension i; is the variance of the k-th Gaussian distribution in dimension i; is the value of the nth data point in dimension i; In the iterative convergence of Gaussian mixture density function, the mean ,variance and weights Perform iteration, and the iteration formula is:

[0051]

[0052]

[0053] in Represents the Gaussian density function, data set is a set of samples independently sampled from this distribution; It is important to explain in detail here that, according to the clustering principle of the Gaussian mixture model (GMM), the dataset is generated by several latent components, and the complete dataset represents a probabilistic mixture of these components. Therefore, the resulting model is a mixture model, and for the case of multivariate continuous data, the probability density function of each component takes the form of a multivariate normal distribution; Therefore, further, for a one-dimensional data set, assuming that the probability distribution of the random variable x follows a mixture model of two Gaussian distributions, then:

[0054] Among them, k=1 and k=2 represent two Gaussian distributions; the kth prior probability is ; and are the mean and standard deviation of the two Gaussian distributions respectively; You can use parameter collection To simplify the parameters, that is: Assume that the data set contains N data points is a set of samples independently sampled from this distribution, Represents the unknown category label of the nth data point. and When known, the category label of the nth data point The posterior probability can be expressed as:

[0055] In the parameters Unknown but dataset If it is known, it can be obtained from the data series Inferred from estimated value of; At this time, there is an iterative algorithm to solve the parameters by maximizing the likelihood estimate ,have:

[0056] At this time, the parameters The natural logarithm of the likelihood function can be expressed as:

[0057] in, represents the Gaussian density function; If you ignore If some items in , the second-order derivative with respect to the parameter {pk} can be approximately expressed as:

[0058] Then, the initial parameters , It is gradually updated through the approximate Newton-Raphson iterative steps to , , and its update formula is:

[0059] S33. For each candidate K value, the candidate number of clusters K is evaluated using the minimization of the Bayesian Information Criterion or the Akaike Information Criterion. These two criteria reward model goodness of fit while penalizing model complexity to avoid overfitting. They are used to construct an optimized Gaussian mixture model based on the information criterion as a multi-monitoring point model. The calculation expression is:

[0060]

[0061] Where L(K) is the log-likelihood of the Gaussian mixture model containing K clusters, and N is the number of samples; Since for each candidate K value, the corresponding log-likelihood value L(K) can be calculated according to the aforementioned method of performing Gaussian mixture model clustering, the optimal number of clusters K can be determined by minimizing the AIC or BIC value.

[0062] That is, the optimized Gaussian mixture model based on the information criterion can preliminarily cluster the deformation monitoring points, divide the monitoring points into several independent categories according to the deformation characteristics, and then apply the random coefficient model to each cluster for independent calibration.

[0063] At this time, based on three-dimensional panel data, IC-OGMM is used to perform cluster analysis according to the spatiotemporal similarity of monitoring points, and an information criterion is constructed to evaluate the number of candidate clusters. This can reasonably classify the monitoring points according to their spatiotemporal characteristics, avoid the influence of the number of groups and the overfitting phenomenon of traditional clustering methods, and make the coefficients within the group modeled as sampling results that obey the same normal distribution, thereby enhancing the generalization ability of the model, facilitating more objective and accurate classification of monitoring points, and improving the reliability of the model.

[0064] Furthermore, the specific operation process of estimating the parameters of the undetermined model through the random coefficient model includes: S31. Based on the multi-monitoring point model, the panel data regression coefficients without time variability are constructed as follows:

[0065] in, Represents the two-dimensional deformation data of the sluice gate; Two-dimensional data representing explanatory variables; Represents a time index; represents the cross-section index; represents the explanatory variable index; Indicates that it does not change over time and can be divided into and ; is the common mean coefficient vector, is the deviation of individual data from the common mean; u is a random interference term; S32. Construct a parameter estimator based on the random coefficient model to estimate the unknown parameters in the panel data regression coefficients. The specific operation process includes: Will Let be a random variable and derive the following assumptions:

[0066] By integrating the NT observation data, the matrix expression can be obtained as follows:

[0067]

[0068]

[0069]

[0070] The optimal linear unbiased estimator is used to estimate the β value,

[0071] Where N is the number of panels, T is the number of data in each panel; Composite error term is a diagonal matrix, the i-th diagonal block Converges to a non-zero constant. Since the estimate of β by ordinary least squares (OLS) is biased. Once Converging to a non-zero constant matrix gives a consistent but inefficient estimate.

[0072] At this time, after estimating the β value, the coefficient is constrained by the asymptotic normal distribution. The optimal linear unbiased estimator of β is the generalized least squares estimate, which is:

[0073] Therefore, the variance of the estimator through the β value is expressed as: ; in, Obeying the asymptotically normal distribution, the parameter An effective estimator of .

[0074] In summary, by using the random coefficient model for parameter calibration, constructing a multi-monitoring point model in each cluster and estimating the unknown model parameters, it is possible to collaboratively model the data of multiple monitoring points and characterize the spatial correlation characteristics of deformation between adjacent monitoring points, thereby avoiding independent modeling of the data of each monitoring point and solving the problem that the traditional model ignores the spatial correlation between monitoring points, so as to improve the accuracy and effectiveness of sluice deformation prediction.

[0075] On the other hand, the present invention also discloses a method for predicting deformation of a sluice gate at multiple monitoring points, comprising: L1. Using the method for constructing a sluice deformation monitoring model using multiple monitoring points provided in the above embodiment, a deformation monitoring model is constructed; L2. Use the deformation monitoring model, input real-time sluice gate data, and execute and output deformation prediction results at multiple monitoring points.

[0076] In a specific embodiment, for Project A, the main body of the project includes a 12-hole regulating gate, a hydropower station, a ship lock and a diversion channel, which has comprehensive functions such as flood control, water storage and irrigation, navigation, power generation and urban water supply. The 12-hole regulating gate structure used is composed of 6 foundation bottom plates, which are numbered from left to right as 1# to 6# bottom plates. A vertical displacement monitoring point is arranged at each of the four corners of each bottom plate, and is marked in sequence according to the bottom plate number and in a clockwise direction (such as P1-1, P1-2 to P6-1, P6-2, with the prefix being the bottom plate number and the suffix being the point position). In addition, two settlement observation points are set downstream of the piers on both sides of the river channel of Project A, and six vertical displacement monitoring points are symmetrically arranged on the upstream and downstream sections of each side wing wall (left / right upstream / downstream wing wall measurement points 1–6), and the monitoring points are distributed as follows: Figure 6 shown.

[0077] Pre-flood vertical displacement monitoring was conducted on the 12-slot regulating gates at designated time points. At 8:00 a.m. on the measurement day, the upstream water level was 18.38 meters, the downstream water level was 13.18 meters, and the flow rate was 350 cubic meters per second. For comparison, the first monitoring point's records showed an upstream water level of 18.23 meters, a downstream water level of 12.89 meters, and a flow rate of 224 cubic meters per second. A statistical analysis of the vertical displacement changes between the two monitoring periods is shown in the following table:

[0078] Among them, vertical displacement data of 50 monitoring points were collected during the observation and monitoring period. The data in the table show that: (1) the maximum average settlement of the downstream pier during the observation interval was 3.67 mm; (2) the maximum settlement at the upstream pier monitoring point was 4.88 mm (P2-3); and (3) the downstream pier had the largest differential settlement, reaching 3.89 mm. The changes at all monitoring points were small, with limited differential settlement. The deformation changes indicate stable foundation performance with no signs of uneven settlement.

[0079] Based on the above settings, the vertical displacement data (vertically downward is positive, vertically upward is negative) of 12 monitoring points (such as P1-1, P1-2 to P6-1, P6-2) located on the six gate piers are selected for analysis. Figure 7 The deformation of each monitoring point after processing is shown. The analysis shows that all monitoring points show a consistent temporal deformation pattern: gradually increasing deformation followed by asymptotic convergence without obvious periodic fluctuations.

[0080] And according to the standard machine learning process, the dataset is divided into a training subset (80%) and a test subset (20%).

[0081] The monitoring data obtained from 12 monitoring points are clustered based on spatiotemporal characteristics. The distance d from the monitoring point to the central axis of the sluice flow is used as an indicator of spatial characteristics. The temporal characteristics are represented by two indicators: one is the maximum absolute value of the time series; the other is the degree of convergence of the data series. The three indicators are calculated for each monitoring point, and the data are as follows:

[0082] According to the table above, the indicator dataset is represented in different scales. The spatial indicator d ranges from 0.01 to 0.50, and the time indicator The value range is 1.04 to 4.31, The value range of is 0.53 to 4.18. In order to eliminate the dimensional influence between different indicators, the indicators d, and The values ​​of are normalized to the range of 0 to 1. Then, the clustering parameters are determined by AIC and BIC, where the initial number of groups is set to 3; the initial weight parameter is 0.25; initial variance 1; minimum number of elements is 2; the maximum allowed variance 3; minimum allowed distance is 0.1. Using the information criterion-based optimized Gaussian mixture model (IC-OGMM) method, based on d, and The clustering results of the monitoring points are as follows: Figure 8 shown.

[0083] The 12 monitoring points are divided into three groups. Figure 9 The cluster diagram of the monitoring points is shown. Monitoring points P3-1, P3-2, and P4-1 are classified into group 1. P1-1, P1-2, P2-1, and P2-2 are classified into group 2. P4-2, P5-1, P5-2, P6-1, and P6-2 belong to group 3. It is obvious that this classification roughly corresponds to their spatial location. For example, all monitoring points in group 1 are mainly located in the center part of the board, while the monitoring points in group 2 are roughly located in the left and right parts of the board (see Figure 9 Of course, due to the influence of the time index, the results are not strictly dependent on their spatial locations. For example, one monitoring point in group 3 is located in the central part, while the other four are located in the right area.

[0084] After clustering the monitoring points into three groups based on the IC-OGMM method, we constructed a random coefficient model for each group. To establish the deformation prediction model, the explanatory variables selected include the water level component H, , , , temperature component , , the time-dependent component t and , where the water level component H is represented by the water level exponential function commonly used in statistical models; the temperature component is characterized by the trigonometric function of time t (assuming that the temperature varies periodically over an annual period); and the aging component directly describes the long-term creep effect of the material by time t and its logarithm.

[0085] IC-OGMM and random coefficient model were used to train the model with deformation data of a specified time period, and the prediction performance was verified with the remaining data. Figure 10The fitting and prediction results of the three monitoring points in the first group are shown: the white area is the modeling dataset, and the blue area is the test dataset; the black solid line is the measured deformation data; the red, purple, green, and blue curves represent the fitting and prediction results of the statistical model, BP neural network model, LSTM model, and the IC-OGMM-RCM model proposed in this paper, respectively.

[0086] Despite the presence of noise in the measured data, the predictions from each model generally agree well with the measured trends. The fitting and prediction results for the remaining nine monitoring points in Groups 2 and 3 are detailed in Appendix A.

[0087] After the model is built, its performance is evaluated by the correlation coefficient R and the residual standard deviation s, and compared with the statistical model, BP model and LSTM model. The results are:

[0088]

[0089] in, Calculate values ​​for the model, is the measured value; is the mean of the monitoring data; n is the sample size.

[0090] The correlation coefficient R and the residual standard deviation s indicate the strength of the relationship between the measured and predicted datasets. R and s for the test dataset at each monitoring point are shown in the table below. Generally speaking, when R is greater than 0.9, the model is validated.

[0091]

[0092] The R values ​​for all monitoring points here range from 0.9779 to 0.9934, indicating a good fit between the predicted and measured data. The maximum R value for monitoring point P5-2 is 0.9934, meaning this dataset has the best fit with the model. The results for s range from 0.9187 to 1.9393.

[0093] In addition, in a specific embodiment, the proposed model is compared with the statistical model, the BP model and the LSTM model.

[0094] To evaluate the prediction performance of each model, Figure 11 and Figure 12The correlation coefficient R and residual standard deviation s calculated based on the test dataset are presented. Clearly, the random coefficient model (multi-monitoring point model) based on IC-OGMM clustering outperforms the other models. This model achieved higher R values ​​than the other models at seven of the 12 modeled monitoring points. For the remaining monitoring points, the LSTM model exhibited the best fit at three monitoring points (P1-1, P3-2, and P6-2), while the BP model (P4-1) and the statistical model (P3-1) each achieved the best fit at one monitoring point. Overall, the correlation coefficient R for the statistical, BP, and LSTM models was higher than 0.95 at all monitoring points.

[0095] For a test dataset of 12 monitoring points, the proposed model achieved the smallest s value at nine monitoring points, demonstrating the best prediction performance among the four models. The results showed that the s values ​​for the statistical model, BP model, LSTM model, and proposed model ranged from 1.214–2.595, 1.322–2.471, 1.034–2.145, and 0.919–1.939, respectively.

[0096] In summary, based on the comparative analysis results of the specific examples, the prediction model fits the monitoring data well, with correlation coefficients above 0.9 for all monitoring points. The proposed model was then compared with a statistical model based on single monitoring data, a BP model, and an LSTM model, and the results showed that the proposed model outperformed the other models.

[0097] In another aspect, the present invention further discloses a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the steps of the above method.

[0098] On the other hand, the present invention further discloses a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the above method.

[0099] In another embodiment provided in the present application, a computer program product comprising instructions is also provided, which, when executed on a computer, enables the computer to execute any one of the methods for constructing a multi-monitoring point deformation monitoring model and deformation prediction method for a sluice gate in the above-mentioned embodiments.

[0100] It is understandable that the system provided by the embodiment of the present invention corresponds to the method provided by the embodiment of the present invention, and the explanation, examples and beneficial effects of the relevant contents can refer to the corresponding parts of the above method.

[0101] The embodiment of the present application further provides an electronic device, comprising a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus. Memory for storing computer programs; The processor is used to implement the above-mentioned method for constructing a sluice multi-monitoring point deformation monitoring model and deformation prediction method when executing the program stored in the memory.

[0102] The communication bus mentioned in the above electronic device can be a peripheral component interconnect standard bus or an extended industry standard architecture bus, etc. The communication bus can be divided into an address bus, a data bus, a control bus, etc.

[0103] The communication interface is used for communication between the above electronic device and other devices.

[0104] The memory may include a random access memory, or a non-volatile memory, such as at least one disk memory. Optionally, the memory may also be at least one storage device located away from the aforementioned processor.

[0105] The above-mentioned processor can be a general-purpose processor, including a central processing unit, a network processor, etc.; it can also be a digital signal processor, an application-specific integrated circuit, a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component.

[0106] It should also be noted that electronic devices also include terminal devices, which can also be called terminals, user equipment, mobile stations, mobile terminals, etc. Terminal devices can be mobile phones, smart TVs, wearable devices, tablet computers, computers with wireless transceiver functions, virtual reality terminal devices, augmented reality terminal devices, wireless terminals in industrial control, wireless terminals in unmanned driving, wireless terminals in remote surgery, wireless terminals in smart grids, wireless terminals in transportation safety, wireless terminals in smart cities, wireless terminals in smart homes, etc. The embodiments of this application do not limit the specific technology and specific device form used by the terminal devices.

[0107] In the above embodiments, all or part of the embodiments can be implemented using software, hardware, firmware, or any combination thereof. When implemented using software, all or part of the embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, optical fiber, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., a floppy disk, hard disk, tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive).

[0108] 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 in the scope of protection of the present invention.

[0109] In addition, it should be noted that if the embodiments of the present invention involve directional indications (such as up, down, left, right, front, back, etc.), the directional indications are only used to explain the relative position relationship, movement status, etc. between the components in a certain specific posture. If the specific posture changes, the directional indications will also change accordingly.

[0110] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present invention, the descriptions of "first", "second", etc. are only for descriptive purposes and cannot be understood as indicating or suggesting their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of such features. In addition, the meaning of "and / or" appearing throughout the text includes three parallel schemes. Taking "A and / or B" as an example, it includes scheme A, or scheme B, or schemes in which A and B are satisfied at the same time. In addition, in the embodiments of the present invention, "multiple" refers to more than two. In addition, the technical solutions between the various embodiments can be combined with each other, but it must be based on the ability of ordinary technicians in this field to implement. When the combination of technical solutions is mutually contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

Claims

1. A method for constructing a deformation monitoring model for a sluice with multiple monitoring points, characterized in that: include: S1. Construct a two-way analysis framework to form a two-dimensional panel by fusing time series data with sluice cross-section data; S2. Expand the two-dimensional panel containing time and cross-sectional dimensions to the geospatial attribute level to construct a three-dimensional panel; S3. Based on the three-dimensional panel data representation, cluster analysis was performed according to the spatiotemporal similarity of the monitoring points, and an information criterion was constructed to evaluate the number of candidate clusters. A random coefficient model was used for parameter calibration, and a Gaussian mixture model was constructed within each cluster as a multi-monitoring point model. The random coefficient model was then used to estimate the unknown model parameters. This was used to construct a deformation monitoring model for the sluice gates, and the fitting and prediction capabilities were evaluated.

2. A method for constructing a multi-monitoring point deformation monitoring model and a deformation prediction method for a sluice gate according to claim 1, characterized in that: The construction process of the Gaussian mixture model in the S3 step includes: S31. The distance d from the central axis of the water gate and the maximum absolute value of the time series and the degree of convergence of the data series As the clustering criterion, to characterize the spatiotemporal characteristics of the monitoring points, where , and is the time-effect term coefficient; S32. Construct and iteratively converge a Gaussian mixture density model function to cluster the dataset into several categories; S33. The candidate number of clusters K is evaluated by minimizing the Bayesian information criterion or the Akaike information criterion, and is used to construct an optimized Gaussian mixture model based on the information criterion as a multi-monitoring point model.

3. A method for constructing a multi-monitoring point deformation monitoring model and a deformation prediction method for a sluice gate according to claim 2, characterized in that: The expression of the Gaussian mixture density model function in step S32 is: Where k represents the number of Gaussian distribution; i represents the number of data dimension; n represents the number of data sequence; I represents the total number of dimensions of data; is the weight of the k-th Gaussian distribution; is the mean of the k-th Gaussian distribution in dimension i; is the variance of the k-th Gaussian distribution in dimension i; is the value of the nth data point in dimension i.

4. A method for constructing a sluice multi-monitoring point deformation monitoring model and a method for deformation prediction according to claim 3, characterized in that: The iterative convergence of the Gaussian mixture density function is respectively ,variance and weights Perform iteration, and the iteration formula is: in Represents the Gaussian density function, data set is a set of samples independently sampled from this distribution.

5. The method for constructing a multi-monitoring point deformation monitoring model and a deformation prediction method for a sluice gate according to claim 3, characterized in that: In the step S33, for each candidate K value, the candidate cluster number K is evaluated using the minimization of the Bayesian information criterion or the Akaike information criterion, and the calculation expression is: Where L(K) is the log-likelihood of the Gaussian mixture model containing K clusters, and N is the number of samples; The optimal number of clusters K is determined by minimizing the AIC or BIC value.

6. A method for constructing a sluice multi-monitoring point deformation monitoring model and a method for predicting deformation according to claim 1, characterized in that: The specific operation process of estimating the undetermined model parameters by using the random coefficient model in step S3 includes: S31. Based on the multi-monitoring point model, the panel data regression coefficients without time variability are constructed as follows: in, Represents the two-dimensional deformation data of the sluice gate; Two-dimensional data representing explanatory variables; Represents a time index; represents the cross-section index; represents the explanatory variable index; Indicates that it does not change over time and can be divided into and ; is the common mean coefficient vector, is the deviation of individual data from the common mean; u is a random interference term; S32. Construct a parameter estimator based on the random coefficient model to estimate the unknown parameters in the panel data regression coefficients.

7. A method for constructing a multi-monitoring point deformation monitoring model and a deformation prediction method for a sluice gate according to claim 6, characterized in that: The specific operation process of the step S32 includes: Will Assuming it is a random variable and integrating the NT observation data, the matrix expression can be obtained as follows: The optimal linear unbiased estimator is used to estimate the β value, Where N is the number of panels and T is the number of data in each panel.

8. A method for predicting deformation of multiple monitoring points of a sluice gate, characterized in that: include: L1. Use the method for constructing a sluice multi-monitoring point deformation monitoring model according to any one of claims 1 to 7 to construct a deformation monitoring model; L2. Use the deformation monitoring model, input real-time sluice gate data, and execute and output deformation prediction results at multiple monitoring points.

Images