A stereo garage structure safety performance dynamic prediction method and system

Abstract

The application discloses a kind of stereo garage structure safety performance dynamic prediction method and system, it is related to safety performance dynamic prediction technical field.The method includes: determining target prediction time, and input target prediction time into fusion prediction model, obtain the structure safety degree prediction value of target stereo garage;Wherein, fusion prediction model includes: initial safety degree prediction model and safety degree error prediction model;Initial safety degree prediction model is used to output the initial safety degree prediction value of target stereo garage;Safety degree error prediction model is used to output the safety degree error prediction value of target stereo garage;Structure safety degree prediction value is the sum of initial safety degree prediction value and safety degree error prediction value.The application carries out error compensation to structure safety degree prediction by fusing lion group optimization heterogeneous core related vector machine model and difference autoregressive moving average model, improves the accuracy of prediction result, is conducive to avoiding the structure failure risk of sudden in the service process of stereo garage.

Description

Technical Field

[0001] This invention relates to the field of dynamic prediction technology for safety performance, and in particular to a method and system for dynamic prediction of the safety performance of a three-dimensional parking garage structure. Background Technology

[0002] Mechanical automated parking systems are a new type of parking structure, different from conventional parking lots. They have distinct advantages such as high space utilization and small footprint per unit, effectively addressing the conflict between urban planning and parking difficulties. However, with their rapid development, safety issues inevitably arise due to factors such as service environment, insulation aging, operational errors during storage and retrieval, collisions and falls of vehicle platforms, and structural performance degradation. Therefore, it is crucial to promptly and accurately detect failures in automated parking systems and conduct comprehensive assessments to determine their safety levels. This helps maintenance personnel identify early anomalies, quickly pinpoint the causes of failures, and address them effectively and promptly, reducing maintenance costs and significantly improving the reliability and safety of parking system operation.

[0003] In the field of safety assessment of mechanical parking equipment, the main approach is "scheduled inspection" assessment, based on GB17907-2010 "General Safety Requirements for Mechanical Parking Equipment." This involves periodic on-site inspections or measurements conducted by personnel during downtime, according to the provisions of this standard, to characterize obvious defects and provide qualitative or quantitative assessment results for the equipment's load-bearing metal structure. However, this method suffers from low utilization of historical equipment information, poor data integration, and an inability to comprehensively predict potential risks. Consequently, safety indicators are singular, assessment results are heavily influenced by human subjectivity, and confidence levels are low, making it difficult to meet the urgent need for dynamic, real-time safety assessments during the service life of special equipment. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for dynamic prediction of the structural safety performance of automated parking garages, so as to accurately predict the structural safety of automated parking garages and thus avoid the risk of sudden structural failure during the service of automated parking garages.

[0005] To achieve the above objectives, the present invention provides the following solution:

[0006] A method for dynamically predicting the structural safety performance of a multi-level parking garage, the prediction method comprising:

[0007] A target prediction time is determined and input into a fusion prediction model to obtain the structural safety prediction value of the target automated parking garage. The fusion prediction model includes an initial safety prediction model and a safety error prediction model. The initial safety prediction model outputs the initial safety prediction value of the target automated parking garage. The safety error prediction model outputs the safety error prediction value of the target automated parking garage. The structural safety prediction value is the sum of the initial safety prediction value and the safety error prediction value.

[0008] The method for determining the fusion prediction model includes:

[0009] Acquire multi-source information of the target automated parking garage at different historical moments, and determine the true value of the structural safety of the target automated parking garage at different historical moments based on the multi-source information;

[0010] The actual structural safety values ​​of the target automated parking garage at different historical moments are divided into training and testing sets.

[0011] The lion flock optimization algorithm is used to train the heterogeneous kernel correlation vector machine model, taking different historical moments in the training set as input and the true structural security values ​​corresponding to different historical moments in the training set as output, to obtain the initial security prediction model.

[0012] By inputting different historical moments from the test set into the initial security prediction model, the initial security prediction values ​​corresponding to different historical moments from the test set are obtained.

[0013] Based on the predicted initial safety value and the actual structural safety value at different historical moments in the test set, determine the actual safety error value at different historical moments in the test set.

[0014] Using different historical moments in the test set as inputs and the true values ​​of safety error corresponding to different historical moments in the test set as outputs, the differential autoregressive moving average model is trained to obtain the safety error prediction model.

[0015] The initial security prediction model and the security error prediction model are linearly superimposed to obtain a fused prediction model.

[0016] Optionally, the prediction method further includes:

[0017] Obtain the true value of the structural safety at the predicted target time.

[0018] Calculate the root mean square error between the actual structural safety value at the target prediction time and the predicted structural safety value at the target prediction time;

[0019] If the root mean square error is greater than a set threshold, the historical time is updated, multi-source information of the updated historical time is obtained, and the true value of structural safety at the updated historical time is determined based on the multi-source information of the updated historical time. The true value of structural safety at the updated historical time is divided into an updated training set and an updated test set. The fusion prediction model is re-determined based on the updated training set and the updated test set. The re-determined fusion prediction model is used to predict the structural safety value at times after the target prediction time. The updated historical time includes at least the target prediction time.

[0020] Optionally, the step of acquiring multi-source information of the target automated parking garage at different historical moments, and determining the true structural safety value of the target automated parking garage at different historical moments based on the multi-source information, specifically includes:

[0021] Acquire multi-source information about the target automated parking garage at different historical moments;

[0022] Based on the multi-source information, a structural safety performance index system for the target three-dimensional parking garage is constructed; the structural safety performance index system includes four levels, and each level includes at least one index.

[0023] The comprehensive weights of each indicator in the structural safety performance evaluation index system are determined based on the three-scale analytic hierarchy process and the CRITIC weighting method; the comprehensive weights include: a first weight and a second weight.

[0024] Based on the comprehensive weights, the true structural safety values ​​of the target multi-level parking garage at different historical moments are determined using the fuzzy comprehensive evaluation method.

[0025] Optionally, the determination of the comprehensive weights of each indicator in the structural safety performance evaluation index system based on the three-scale analytic hierarchy process and the CRITIC weighting method specifically includes:

[0026] The weights of each indicator in the structural safety performance evaluation index system are determined based on the three-scale analytic hierarchy process, and the first weight is obtained.

[0027] The weights of each indicator in the structural safety performance evaluation index system are determined based on the CRITIC weighting method, thus obtaining the second weight;

[0028] The first weight and the second weight are combined to obtain the comprehensive weight of each indicator in the structural safety performance evaluation index system.

[0029] Optionally, the determination of the weights of each indicator in the structural safety performance evaluation index system based on the three-scale analytic hierarchy process (AHP) to obtain the first weight specifically includes:

[0030] The importance of indicators of the same level in the structural safety performance evaluation index system is compared in a pairwise comparison manner to determine multiple comparison matrices;

[0031] The sum of the elements in each column of the multiple comparison matrices is compared in a pairwise manner to determine the judgment matrix;

[0032] Determine the optimization matrix based on the judgment matrix;

[0033] The weights of each indicator in the structural safety performance evaluation index system are determined based on the optimization matrix, thus obtaining the first weight.

[0034] Optionally, the determination of the weights of each indicator in the structural safety performance evaluation index system based on the CRITIC weighting method to obtain the second weights specifically includes:

[0035] The various indicators in the structural safety performance evaluation index system are quantified and subjected to stiffness-free processing to obtain the dimensionless processing results of each indicator.

[0036] The variance and conflict values ​​of each indicator are determined based on the dimensionless processing results of each indicator; the variance is determined by the standard deviation of the dimensionless processing results of each indicator, and the conflict value is determined by the correlation coefficient between the dimensionless processing results of each indicator.

[0037] The information content of each indicator is determined based on the variation value and conflict value of each indicator.

[0038] The weights of each indicator in the structural safety performance evaluation index system are determined based on the amount of information in each indicator, thus obtaining the second weight.

[0039] Optionally, the structural safety performance index system includes: primary index, secondary index, tertiary index and quaternary index;

[0040] The primary indicator includes: the structural safety of the target automated parking garage;

[0041] The secondary indicators include: the load characteristics, failure characteristics, bearing capacity, seismic characteristics, and characteristic life of the target automated parking system;

[0042] The three-level indicators include: dynamic load effect, load magnitude, load distribution, verticality, parallelism, frame diagonal condition, bolt connection condition, weld connection condition, static strength, fatigue strength, static stiffness, dynamic stiffness, stability, stress response, displacement response, design life, service life, and remaining life of the target multi-level parking garage.

[0043] The four-level indicators include: lifting impact, lifting dynamic load, inertial impact, vehicle load, vehicle storage distribution, frame, columns, floor beams, and floor tie rods of the target multi-level parking garage.

[0044] A dynamic prediction system for the structural safety performance of a multi-level parking garage, wherein the prediction system is applied to the above-mentioned prediction method, and the prediction system includes:

[0045] A structural safety prediction module is used to determine the target prediction time and input the target prediction time into a fusion prediction model to obtain the structural safety prediction value of the target multi-level parking garage. The fusion prediction model includes an initial safety prediction model and a safety error prediction model. The initial safety prediction model is used to output the initial safety prediction value of the target multi-level parking garage. The safety error prediction model is used to output the safety error prediction value of the target multi-level parking garage. The structural safety prediction value is the sum of the initial safety prediction value and the safety error prediction value.

[0046] Fusion prediction model determination module; the fusion prediction model determination module includes:

[0047] The historical structural safety value determination submodule is used to acquire multi-source information of the target multi-level parking garage at different historical moments, and determine the historical structural safety value of the target multi-level parking garage at different historical moments based on the multi-source information.

[0048] The sample set partitioning submodule is used to divide the true structural safety values ​​of the target automated parking garage at different historical moments into a training set and a test set.

[0049] The initial security prediction model determination submodule is used to train the heterogeneous kernel related vector machine model by using the lion pack optimization algorithm, taking different historical moments in the training set as input and the true structural security values ​​corresponding to different historical moments in the training set as output, to obtain the initial security prediction model.

[0050] The initial security prediction value determination submodule is used to input different historical moments in the test set into the initial security prediction model to obtain the initial security prediction values ​​corresponding to different historical moments in the test set.

[0051] The submodule for determining the true value of safety error is used to determine the true value of safety error corresponding to different historical moments in the test set based on the initial safety prediction value and the true value of structural safety at different historical moments in the test set.

[0052] The safety error prediction model determination submodule is used to train the differential autoregressive moving average model by taking different historical moments in the test set as input and the true safety error value corresponding to different historical moments in the test set as output, so as to obtain the safety error prediction model.

[0053] The fusion prediction model determination submodule is used to linearly superimpose the initial security prediction model and the security error prediction model to obtain the fusion prediction model.

[0054] Optionally, the prediction system further includes:

[0055] Fusion prediction model update module; the fusion prediction model update module includes:

[0056] The target structure safety value acquisition submodule is used to acquire the target structure safety value at the prediction time.

[0057] The root mean square error determination submodule is used to calculate the root mean square error between the actual value of structural safety at the target prediction time and the predicted value of structural safety at the target prediction time.

[0058] The fusion prediction model update submodule is used to update the historical time if the root mean square error is greater than a set threshold, obtain multi-source information of the updated historical time, determine the true value of structural safety at the updated historical time based on the multi-source information, divide the true value of structural safety at the updated historical time into an updated training set and an updated test set, and re-determine the fusion prediction model based on the updated training set and the updated test set; the re-determined fusion prediction model is used to predict the structural safety value at times after the target prediction time; the updated historical time includes at least the target prediction time.

[0059] Optionally, the historical structural security degree true value determination submodule specifically includes:

[0060] The historical multi-source information acquisition submodule is used to acquire multi-source information of the target automated parking garage at different historical moments;

[0061] A structural safety performance index system construction submodule is used to construct the structural safety performance index system of the target three-dimensional parking garage based on the multi-source information; the structural safety performance index system includes four levels, and each level includes at least one index;

[0062] The submodule for determining the comprehensive weight of indicators is used to determine the comprehensive weight of each indicator in the structural safety performance evaluation indicator system based on the three-scale analytic hierarchy process and the CRITIC weighting method; the comprehensive weight includes: a first weight and a second weight;

[0063] The historical structural safety value determination submodule is used to determine the true structural safety value of the target multi-level parking garage at different historical moments based on the comprehensive weight and the fuzzy comprehensive evaluation method.

[0064] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0065] This invention, based on multi-source information of the target automated parking garage, employs a method that fuses a lion pack optimized heterogeneous kernel correlation vector machine model with a differential autoregressive moving average model. The trained heterogeneous kernel correlation vector machine model (i.e., the initial safety prediction model) predicts the initial safety value of the target automated parking garage, while the trained differential autoregressive moving average model (i.e., the safety error prediction model) predicts the safety error value of the target automated parking garage for the target time period. The final structural safety prediction value is obtained by linearly superimposing the initial safety prediction value and the safety error prediction value. This achieves error compensation for structural safety prediction, improves the accuracy of the prediction results, and helps avoid the risk of sudden structural failure during the service of the automated parking garage. Attached Figure Description

[0066] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0067] Figure 1 A flowchart of the dynamic prediction method for the structural safety performance of a three-dimensional parking garage provided in an embodiment of the present invention;

[0068] Figure 2 This is a flowchart illustrating the specific process of the dynamic prediction method for the structural safety performance of a three-dimensional parking garage provided in an embodiment of the present invention.

[0069] Figure 3 This is a schematic diagram of the structural safety performance index system for a multi-level parking garage provided in an embodiment of the present invention;

[0070] Figure 4 This is a schematic diagram of the fusion mechanism of the fusion prediction model provided in the embodiments of the present invention;

[0071] Figure 5 This is a structural diagram of the dynamic prediction system for the structural safety performance of a three-dimensional parking garage provided in an embodiment of the present invention. Detailed Implementation

[0072] 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.

[0073] The purpose of this invention is to provide a method and system for dynamic prediction of the structural safety performance of automated parking garages, so as to accurately predict the structural safety of automated parking garages and thus avoid the risk of sudden structural failure during the service of automated parking garages.

[0074] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0075] Figure 1 This is a flowchart of the dynamic prediction method for the structural safety performance of a three-dimensional parking garage provided in an embodiment of the present invention. Figure 2 This is a flowchart illustrating the dynamic prediction method for the structural safety performance of a multi-level parking garage provided in an embodiment of the present invention. Figure 1 and Figure 2 As shown, the present invention provides a method for dynamic prediction of the structural safety performance of a multi-level parking garage, the prediction method comprising:

[0076] Step 100: Determine the fusion prediction model.

[0077] Step 200: Determine the target prediction time and input the target prediction time into the fusion prediction model to obtain the structural safety prediction value of the target multi-level parking garage; the fusion prediction model includes: an initial safety prediction model and a safety error prediction model; the initial safety prediction model is used to output the initial safety prediction value of the target multi-level parking garage; the safety error prediction model is used to output the safety error prediction value of the target multi-level parking garage; the structural safety prediction value is the sum of the initial safety prediction value and the safety error prediction value.

[0078] Specifically, the method for determining the fusion prediction model includes:

[0079] Step 101: Obtain multi-source information of the target automated parking garage at different historical times, and determine the true value of the structural safety of the target automated parking garage at different historical times based on the multi-source information.

[0080] In this embodiment, the step of acquiring multi-source information of the target automated parking garage at different historical moments, and determining the true structural safety value of the target automated parking garage at different historical moments based on the multi-source information, specifically includes:

[0081] Step 1: Obtain multi-source information of the target automated parking garage at different historical moments; the multi-source information includes the inherent information, service information, and virtual information of the target automated parking garage itself; the specific methods for obtaining the above information are as follows:

[0082] Step 1.1: For inherent information, it can be obtained through the drawings provided by the manufacturer, including the drawing numbers of the general drawing and the individual drawings, and the serial number, code, name, material, specifications, quantity, unit weight and dimensional relationship of each component in the corresponding drawing number.

[0083] Step 1.2: For service information, the embedded management system of the automated parking garage is used to collect vehicle information and access status for any time period. Based on the system log, the vehicle brand and license plate number are statistically analyzed to determine the vehicle load. Combined with the identification of each parking space, such as vacant, occupied, or locked, the distribution of vehicles is determined. The strain history of different measuring points is obtained using a data acquisition and transmission system composed of strain gauges, a wireless dynamic strain testing and analysis system, and a workstation. Through regular on-site inspections / measurements by personnel during downtime, obvious defects are detected qualitatively or quantitatively. This allows for the statistical analysis, summarization, and organization of periodic inspection information for each inspection item, including items such as column verticality, beam parallelism, frame diagonal tolerances, and connection status.

[0084] Step 1.3: For virtual information, from the perspective of information integration and communication, a combined strategy of theory + data acquisition + simulation is introduced. Based on the ASP.Net Core architecture and Visual Studio 2019 platform, the C# computer programming language, the finite element parametric design language APDL, and the SQL Server database are used to build a real-time analysis module for the structural performance of the three-dimensional parking garage. This module can acquire virtual information such as the overall stress cloud map, displacement cloud map, modal information, and characteristic life of the structure in real time during the working cycle (one working cycle corresponds to one historical moment).

[0085] Step 2: Construct a structural safety performance index system for the target multi-level parking garage based on the multi-source information; the structural safety performance index system includes four levels, with each level including at least one index.

[0086] Specifically, the structural safety performance index system is established based on the intended use of the automated parking garage structure and the workload it handles. It considers the structural composition, the mechanical properties of the structure under various load combinations, and the criteria for structural resistance to strength failure, elastic instability, and fatigue failure. From the perspective of the mutual influence and constraints of multiple factors, a structural safety performance evaluation index system (i.e., the structural safety performance index system) is established. For specific index content, please refer to [link / reference]. Figure 3 .

[0087] The structural safety performance index system is generally divided into four levels: primary, secondary, tertiary, and quaternary indicators. The primary indicators include the structural safety of the target automated parking system. The secondary indicators include the load characteristics, failure characteristics, bearing capacity, seismic characteristics, and characteristic life of the target automated parking system. The tertiary indicators include the dynamic load effect, load magnitude, load distribution, verticality, parallelism, frame diagonal conditions, bolted connections, welded connections, static strength, fatigue strength, static stiffness, dynamic stiffness, stability, stress response, displacement response, design life, service life, and remaining life of the target automated parking system. The quaternary indicators include the lifting impact, lifting dynamic load, inertial impact, vehicle load, vehicle storage distribution, frame, columns, floor beams, and floor tie rods of the target automated parking system.

[0088] Step 3: Determine the comprehensive weight of each indicator in the structural safety performance evaluation index system based on the three-scale analytic hierarchy process (AHP) and the CRITIC weighting method. The comprehensive weight includes a first weight and a second weight; the first weight corresponds to the subjective weight, and the second weight corresponds to the objective weight. The linear weighted combination weight calculation strategy based on the three-scale AHP and the CRITIC weighting method adopted in this invention specifically includes the following steps:

[0089] Step 3.1: Determine the weights of each indicator in the structural safety performance evaluation index system based on the three-scale analytic hierarchy process (AHP) to obtain the first weight. The specific steps are as follows:

[0090] First, the importance of the same level indicators in the structural safety performance evaluation index system is compared in a pairwise comparison manner to determine multiple comparison matrices.

[0091] Specifically, the importance of indicators at the same level is compared in a pairwise manner, and multiple n-order comparison matrices E are constructed accordingly:

[0092]

[0093] In the formula: e ij Let be the average of the comparison results of z experts on indicator i and indicator j, where i,j = 1, 2, ..., n. If the importance of indicator i is higher than that of indicator j, then e ij =2; if the importance of indicator i is lower than that of indicator j, e ij =0; when the two are equal, e ij =1, where n is the number of indicators in the same level and group.

[0094] by Figure 3Taking the structural safety performance evaluation index system in China as an example, each level of index consists of multiple indicators enclosed in boxes. In the four-level indexes, if the importance of indicators D1-D3 is compared, a third-order comparison matrix can be obtained.

[0095] Secondly, the sums of the columns of the multiple comparison matrices are compared in a pairwise manner to determine the judgment matrix.

[0096] Specifically, by comparing the sum of the elements in each column of the comparison matrix, the values ​​of each element in the judgment matrix are calculated by pairwise comparison, as shown in equation (2).

[0097]

[0098] Where: g ij To determine the corresponding element in the i-th row and j-th column of matrix G, e iw To compare corresponding elements in the i-th row and w-th column of matrix E, e jw To compare the corresponding elements in the j-th row and w-th column of matrix E.

[0099] Next, the optimization matrix is ​​determined based on the judgment matrix.

[0100] Finally, the weights of each indicator in the structural safety performance evaluation index system are determined based on the optimization matrix, and the first weight is obtained.

[0101] Specifically, based on the judgment matrix, the optimization matrix O is determined using equation (3), and the weights of each indicator are calculated using equation (4).

[0102]

[0103]

[0104] In the formula: o ij To optimize the corresponding element in the i-th row and j-th column of matrix O, g iw To determine the corresponding element in the i-th row and w-th column of matrix G, g jw To determine the corresponding element in the j-th row and w-th column of matrix G, wi is the single weight (i.e., the first weight, or subjective weight) corresponding to the i-th indicator. The weight is the normalized weight of the i-th indicator.

[0105] Step 3.2: Determine the weights of each indicator in the structural safety performance evaluation index system based on the CRITIC weighting method to obtain the second weights. The specific steps are as follows:

[0106] First, the various indicators in the structural safety performance evaluation index system are quantified and subjected to stiffness-free processing to obtain the dimensionless processing results of each indicator. Indicator quantification refers to transforming multi-source information into effective information that can be used for subsequent analysis through data quality analysis and preprocessing; that is, the quantified results of each indicator.

[0107] Specifically, the quantitative results of each indicator are processed without rigidity using equation (5), and the data are then divided into positive indicator ↑ and negative indicator ↓:

[0108]

[0109] In the formula: y ki Let k be the value of the k-th quantized result of index i, where k = 1, 2, ..., n k n k y represents the number of quantification results for indicator i; ki ′ represents the dimensionless processing result of index i; and These represent the maximum and minimum values ​​of the quantification result corresponding to index i, respectively.

[0110] Secondly, the variance values ​​and conflict values ​​of each indicator are determined based on the dimensionless processing results of each indicator; the variance values ​​are determined by the standard deviation of the dimensionless processing results of each indicator, and the conflict values ​​are determined by the correlation coefficient between the dimensionless processing results of each indicator.

[0111] Specifically, the standard deviation of the indicators and the correlation coefficients between the indicators are determined to characterize the variability and conflict of the indicators:

[0112]

[0113] Where: δ i R represents the standard deviation of index i; the larger the value, the more information it reflects, and the greater the weight it is assigned. i r is the conflict calculation value for the i-th indicator; ij Let be the correlation coefficient between index i and index j.

[0114] Next, the information content of each indicator is determined based on the variation value and conflict value of each indicator.

[0115] Finally, the weights of each indicator in the structural safety performance evaluation index system are determined based on the amount of information in each indicator, thus obtaining the second weight.

[0116] Specifically, the information content of an indicator is calculated using the results of its variation and conflict. The larger the value, the greater the role and weight of the indicator in the system.

[0117]

[0118] In the formula: X i ω represents the information content of the i-th indicator; i is the objective weight (i.e., the second weight) of the i-th indicator; p is the total number of indicators.

[0119] Step 3.3: Combine the first weight and the second weight to obtain the comprehensive weight of each indicator in the structural safety performance evaluation index system.

[0120] Specifically, the combination strategy of subjective and objective weights can be used to calculate the comprehensive weight of the indicators according to formula (8) and combine it with fuzzy comprehensive evaluation to give the structural safety of the three-dimensional parking garage.

[0121]

[0122] In the formula: The overall weight of indicator i; μ is the expert preference coefficient; n k This represents the number of samples (i.e., the number of quantification results for index i).

[0123] Step 4: Based on the comprehensive weight, and combined with the fuzzy comprehensive evaluation method, determine the true structural safety value of the target multi-level parking garage at different historical moments.

[0124] Specifically, the calculated true value F of the structural safety at a certain historical moment is as follows:

[0125] F=[90 80 70 60]·[0.6598 0.1930 0.0968 0.0772] T =86.23.

[0126] Step 102: Divide the true values ​​of the structural safety of the target automated parking garage at different historical moments into a training set and a test set.

[0127] Step 103: Using the lion pack optimization algorithm, with different historical moments in the training set as input and the true structural security values ​​corresponding to different historical moments in the training set as output, train the heterogeneous kernel correlation vector machine model to obtain the initial security prediction model.

[0128] Step 104: Input different historical moments in the test set into the initial security prediction model to obtain the initial security prediction values ​​corresponding to different historical moments in the test set.

[0129] Step 105: Based on the predicted initial safety value and the actual structural safety value at different historical moments in the test set, determine the actual safety error value at different historical moments in the test set.

[0130] Step 106: Using different historical moments in the test set as input and the true values ​​of the safety error corresponding to different historical moments in the test set as output, train the differential autoregressive moving average model to obtain the safety error prediction model.

[0131] Step 107: Linearly superimpose the initial security prediction model and the security error prediction model to obtain a fused prediction model.

[0132] Furthermore, the prediction method also includes:

[0133] Step 300: Obtain the true value of the structural safety at the target prediction time.

[0134] Step 400: Calculate the root mean square error between the actual value of structural safety at the target prediction time and the predicted value of structural safety at the target prediction time.

[0135] Step 500: If the root mean square error is greater than a set threshold, update the historical time, obtain multi-source information of the updated historical time, determine the true value of the structural safety at the updated historical time based on the multi-source information of the updated historical time, divide the true value of the structural safety at the updated historical time into an updated training set and an updated test set, and re-determine the fusion prediction model based on the updated training set and the updated test set; the re-determined fusion prediction model is used to predict the structural safety value at times after the target prediction time; the updated historical time includes at least the target prediction time.

[0136] Figure 4 This is a schematic diagram of the fusion mechanism of the fusion prediction model provided in an embodiment of the present invention. Figure 4 As shown, for the above-mentioned dynamic prediction scheme for the safety performance of the lion pride by fusing the heterogeneous kernel correlation vector machine and the differential autoregressive moving average model, the specific implementation plan is given below from four aspects: sample acquisition, model training, model fusion, and model updating. The specific steps are as follows:

[0137] Step 4.1: Regarding sample acquisition, based on multi-source information of the automated parking garage, and according to the dynamic evaluation requirements for the safety performance of the metal structure of the automated parking garage, sample data for the fusion model is constructed using the structural safety performance evaluation results of the automated parking garage at different times (i.e., based on the comprehensive weight determined by subjective and objective weights, combined with the true values ​​of structural safety at different historical times obtained by fuzzy comprehensive evaluation), forming a sample set. Wherein, the input vector X i′ Let i' represent the work cycle sequence number at different times t (i.e., corresponding to different moments), and i' = 1, 2, ..., N. The output target Y is... AThis corresponds to the garage safety performance assessment results (i.e., the actual structural safety values ​​at different times).

[0138] Step 4.2: In terms of model training, select the first n1 samples from the sample set. The training set was used to optimize the heterogeneous kernel correlation vector machine (LION-IRVM model) using the lion flock algorithm. At time T0, the remaining n2 samples were used to complete the LION-IRVM model test, and the output values ​​of the test samples were... and By subtracting the values, we obtain the error sequence. (i.e., the true value of the safety error corresponding to different historical moments), and train the (differential autoregressive moving average model) ARIMA model (see equation (9)) as follows:

[0139]

[0140] In the formula: d is the difference order, usually taken as 0 or 1; l is the lag step size; p′ is the autoregressive order of the model, q′ is the moving average order; η k θ is the first coefficient. k ε is the second coefficient, and neither of them is 0; k is a positive integer, and its value ranges from 1 to p; i-k For the error, ε i E is an independent error term, representing the current white noise. i Let E be the true value of the safety error at time i. i ′ represents the predicted safety error at time i, E i ′ -k X1 is the predicted safety error value at time ik; X2′ is the number of samples in the test set. Wherein, E... i The sequence formed by ′ It is a stationary, zero-mean, and normally distributed sequence.

[0141] Step 4.3: Regarding model fusion and updating, at time T1, the LION-IRVM model and the ARIMA model are fused using a linear superposition method to obtain the prediction results for the next time period t. in This is the predicted value of LION-IRVM (i.e., the initial safety prediction value). Let δ be the predicted value of the ARIMA model (i.e., the predicted value of the safety error). Then, the root mean square error is used to check whether the ARIMA prediction result (i.e., the prediction error of the LION-IRVM model) is within the constraints. If δ < δ * (where δ is the calculated root mean square error, δ *If the threshold is set to T1+n, it indicates that the model has good fusion and can continue predicting. Similarly, for T1+n... c time t (n c (where δ is a positive integer), δ≥δ * If prediction continues, it will lead to a large error. In this case, it is necessary to obtain T1+(n) c -1) From time t to T1+n c Using new sample data at time t, the concept of a sliding window is introduced. The LION-IRVM and ARIMA models are rebuilt using this new sample data, thereby updating the LION-IRVM-ARIMA fusion model and achieving dynamic prediction of the structural safety performance of the automated parking garage. The dynamic prediction method for the structural safety performance of the automated parking garage provided by this invention was used to predict the structural safety of a tower-type automated parking garage. The results are shown in Table 1. When the predicted structural safety value at a certain time is lower than the safety threshold (e.g., 90%), the equipment needs to be repaired to avoid the risk of sudden structural failure during the service of the automated parking garage.

[0142] Table 1. Dynamic Prediction Results of Safety Performance of Automated Parking Garage Structures

[0143] time security time security time security time security time security time security 1 year 95.2568 6 years 95.1461 11 years 94.8771 16 years 93.5859 21 years 89.1332 26 years 73.9809 2 years 95.2473 7 years 95.1049 12 years 94.8043 17 years 93.4372 22 years 83.4843 27 years 70.7529 3 years 95.2315 8 years 95.0575 13 years 94.7252 18 years 89.9402 23 years 83.0951 28 years 70.1928 4 years 95.2094 9 years 95.0037 14 years 93.8644 2019 89.6776 24 years 82.6995 29 years 62.9774 5 years 95.1809 10 years 94.9435 15 years 93.7283 20 years 89.4086 25 years 74.4967 30 years 56.9677

[0144] Figure 5 This is a structural diagram of the dynamic prediction system for the structural safety performance of a multi-level parking garage provided in an embodiment of the present invention. Figure 5 As shown, the present invention also provides a dynamic prediction system for the structural safety performance of a three-dimensional parking garage. The system is applied to the above method, and the prediction system includes: a fusion prediction model determination module 1 and a structural safety prediction module 2.

[0145] The structural safety prediction module 2 is used to determine the target prediction time and input the target prediction time into the fusion prediction model to obtain the structural safety prediction value of the target multi-level parking garage. The fusion prediction model includes an initial safety prediction model and a safety error prediction model. The initial safety prediction model is used to output the initial safety prediction value of the target multi-level parking garage. The safety error prediction model is used to output the safety error prediction value of the target multi-level parking garage. The structural safety prediction value is the sum of the initial safety prediction value and the safety error prediction value.

[0146] The fusion prediction model determination module 1 includes:

[0147] The historical structural safety value determination submodule 11 is used to obtain multi-source information of the target multi-level parking garage at different historical times, and determine the historical structural safety value of the target multi-level parking garage at different historical times based on the multi-source information.

[0148] The sample set partitioning submodule 12 is used to divide the true values ​​of the structural safety of the target three-dimensional parking garage at different historical moments into a training set and a test set.

[0149] The initial security prediction model determination submodule 13 is used to train the heterogeneous kernel correlation vector machine model by using the lion pack optimization algorithm, taking different historical moments in the training set as inputs and the true structural security values ​​corresponding to different historical moments in the training set as outputs, to obtain the initial security prediction model.

[0150] The initial security prediction value determination submodule 14 is used to input different historical moments in the test set into the initial security prediction model to obtain the initial security prediction values ​​corresponding to different historical moments in the test set.

[0151] The submodule 15 for determining the true value of safety error is used to determine the true value of safety error corresponding to different historical moments in the test set based on the initial safety prediction value and the true value of structural safety at different historical moments in the test set.

[0152] The safety error prediction model determination submodule 16 is used to train the differential autoregressive moving average model by taking different historical moments in the test set as inputs and the true values ​​of safety error corresponding to different historical moments in the test set as outputs, so as to obtain the safety error prediction model.

[0153] The fusion prediction model determination submodule 17 is used to linearly superimpose the initial security prediction model and the security error prediction model to obtain the fusion prediction model.

[0154] Furthermore, the prediction system further includes: a fusion prediction model update module; the fusion prediction model update module includes:

[0155] The target structure safety degree true value acquisition submodule is used to acquire the target structure safety degree true value at the prediction time.

[0156] The root mean square error determination submodule is used to calculate the root mean square error between the actual value of structural safety at the target prediction time and the predicted value of structural safety at the target prediction time.

[0157] The fusion prediction model update submodule is used to update the historical time if the root mean square error is greater than a set threshold, obtain multi-source information of the updated historical time, determine the true value of structural safety at the updated historical time based on the multi-source information, divide the true value of structural safety at the updated historical time into an updated training set and an updated test set, and re-determine the fusion prediction model based on the updated training set and the updated test set; the re-determined fusion prediction model is used to predict the structural safety value at times after the target prediction time; the updated historical time includes at least the target prediction time.

[0158] Furthermore, the historical structural security degree true value determination submodule 11 specifically includes:

[0159] The historical multi-source information acquisition submodule is used to acquire multi-source information of the target automated parking garage at different historical moments.

[0160] The structural safety performance index system construction submodule is used to construct the structural safety performance index system of the target three-dimensional parking garage based on the multi-source information; the structural safety performance index system includes four levels, and each level includes at least one index.

[0161] The submodule for determining the comprehensive weight of indicators is used to determine the comprehensive weight of each indicator in the structural safety performance evaluation indicator system based on the three-scale analytic hierarchy process and the CRITIC weighting method; the comprehensive weight includes: a first weight and a second weight.

[0162] The historical structural safety value determination submodule is used to determine the true structural safety value of the target multi-level parking garage at different historical moments based on the comprehensive weight and the fuzzy comprehensive evaluation method.

[0163] The present invention provides a method and system for dynamic prediction of the structural safety performance of automated parking garages, driven by multi-source information, comprising the following steps: 1) acquisition of multi-source information on automated parking garages; 2) construction of a structural safety performance index system; 3) a linear weighted combination weight calculation strategy based on the three-scale analytic hierarchy process and the CRITIC method, combined with fuzzy comprehensive evaluation to give the structural safety degree; and 4) a dynamic prediction scheme for safety performance fusion of a lion pack optimized heterogeneous kernel correlation vector machine and a differential autoregressive moving average model. Based on inherent information, service information, and virtual information acquired from multiple sources, this invention provides a dynamic prediction scheme for the structural safety performance of automated parking garages from aspects such as the construction of a multi-dimensional, multi-layered index system, index quantification methods, determination of subjective and objective comprehensive weights, fuzzy comprehensive evaluation, and dynamic prediction. This avoids the shortcomings of traditional inspection and testing methods for automated parking equipment, such as low information data utilization, poor integration, and weak connection with equipment, which lead to significant subjective influence on structural safety performance judgments and the inability to comprehensively predict potential risks. This is beneficial for proactively mitigating the risk of sudden structural failure during the service life of automated parking garages.

[0164] 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 systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.

[0165] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for dynamically predicting the structural safety performance of a three-dimensional parking garage, characterized in that, The prediction method includes: A target prediction time is determined and input into a fusion prediction model to obtain the structural safety prediction value of the target automated parking garage. The fusion prediction model includes an initial safety prediction model and a safety error prediction model. The initial safety prediction model outputs the initial safety prediction value of the target automated parking garage. The safety error prediction model outputs the safety error prediction value of the target automated parking garage. The structural safety prediction value is the sum of the initial safety prediction value and the safety error prediction value. The method for determining the fusion prediction model includes: Acquire multi-source information of the target automated parking garage at different historical moments, and determine the true value of the structural safety of the target automated parking garage at different historical moments based on the multi-source information; The actual structural safety values ​​of the target automated parking garage at different historical moments are divided into training and testing sets. The lion flock optimization algorithm is used to train the heterogeneous kernel correlation vector machine model, taking different historical moments in the training set as input and the true structural security values ​​corresponding to different historical moments in the training set as output, to obtain the initial security prediction model. By inputting different historical moments from the test set into the initial security prediction model, the initial security prediction values ​​corresponding to different historical moments from the test set are obtained. Based on the predicted initial safety value and the actual structural safety value at different historical moments in the test set, determine the actual safety error value at different historical moments in the test set. Using different historical moments in the test set as inputs and the true values ​​of safety error corresponding to different historical moments in the test set as outputs, the differential autoregressive moving average model is trained to obtain the safety error prediction model. The initial security prediction model and the security error prediction model are linearly superimposed to obtain a fused prediction model.

2. The method for dynamic prediction of the structural safety performance of a three-dimensional parking garage according to claim 1, characterized in that, The prediction method further includes: Obtain the true value of the structural safety at the predicted target time. Calculate the root mean square error between the actual structural safety value at the target prediction time and the predicted structural safety value at the target prediction time; If the root mean square error is greater than a set threshold, the historical time is updated, multi-source information of the updated historical time is obtained, and the true value of structural safety at the updated historical time is determined based on the multi-source information of the updated historical time. The true value of structural safety at the updated historical time is divided into an updated training set and an updated test set. The fusion prediction model is re-determined based on the updated training set and the updated test set. The re-determined fusion prediction model is used to predict the structural safety value at times after the target prediction time. The updated historical time includes at least the target prediction time.

3. The method for dynamic prediction of the structural safety performance of a three-dimensional parking garage according to claim 1, characterized in that, The process of acquiring multi-source information about the target automated parking garage at different historical moments, and determining the true structural safety value of the target automated parking garage at different historical moments based on the multi-source information, specifically includes: Acquire multi-source information about the target automated parking garage at different historical moments; Based on the multi-source information, a structural safety performance index system for the target three-dimensional parking garage is constructed; the structural safety performance index system includes four levels, and each level includes at least one index. The comprehensive weights of each indicator in the structural safety performance evaluation index system are determined based on the three-scale analytic hierarchy process and the CRITIC weighting method; the comprehensive weights include: a first weight and a second weight. Based on the comprehensive weights, the true structural safety values ​​of the target multi-level parking garage at different historical moments are determined using the fuzzy comprehensive evaluation method.

4. The method for dynamic prediction of the structural safety performance of a three-dimensional parking garage according to claim 3, characterized in that, The determination of the comprehensive weights of each indicator in the structural safety performance evaluation index system based on the three-scale analytic hierarchy process and the CRITIC weighting method specifically includes: The weights of each indicator in the structural safety performance evaluation index system are determined based on the three-scale analytic hierarchy process, and the first weight is obtained. The weights of each indicator in the structural safety performance evaluation index system are determined based on the CRITIC weighting method, thus obtaining the second weight; The first weight and the second weight are combined to obtain the comprehensive weight of each indicator in the structural safety performance evaluation index system.

5. The method for dynamic prediction of the structural safety performance of a three-dimensional parking garage according to claim 4, characterized in that, The weights of each indicator in the structural safety performance evaluation index system are determined using the three-scale analytic hierarchy process (AHP) to obtain the first weight, which specifically includes: The importance of indicators of the same level in the structural safety performance evaluation index system is compared in a pairwise comparison manner to determine multiple comparison matrices; The sum of the elements in each column of the multiple comparison matrices is compared in a pairwise manner to determine the judgment matrix; Determine the optimization matrix based on the judgment matrix; The weights of each indicator in the structural safety performance evaluation index system are determined based on the optimization matrix, thus obtaining the first weight.

6. The method for dynamic prediction of the structural safety performance of a three-dimensional parking garage according to claim 4, characterized in that, The second weight is obtained by determining the weights of each indicator in the structural safety performance evaluation index system based on the CRITIC weighting method, specifically including: The various indicators in the structural safety performance evaluation index system are quantified and subjected to stiffness-free processing to obtain the dimensionless processing results of each indicator. The variance and conflict values ​​of each indicator are determined based on the dimensionless processing results of each indicator; the variance is determined by the standard deviation of the dimensionless processing results of each indicator, and the conflict value is determined by the correlation coefficient between the dimensionless processing results of each indicator. The information content of each indicator is determined based on the variation value and conflict value of each indicator. The weights of each indicator in the structural safety performance evaluation index system are determined based on the amount of information in each indicator, thus obtaining the second weight.

7. The method for dynamic prediction of the structural safety performance of a three-dimensional parking garage according to claim 3, characterized in that, The structural safety performance index system includes: primary index, secondary index, tertiary index and quaternary index; The primary indicator includes: the structural safety of the target automated parking garage; The secondary indicators include: the load characteristics, failure characteristics, bearing capacity, seismic characteristics, and characteristic life of the target automated parking system; The three-level indicators include: dynamic load effect, load magnitude, load distribution, verticality, parallelism, frame diagonal condition, bolt connection condition, weld connection condition, static strength, fatigue strength, static stiffness, dynamic stiffness, stability, stress response, displacement response, design life, service life, and remaining life of the target multi-level parking garage. The four-level indicators include: lifting impact, lifting dynamic load, inertial impact, vehicle load, vehicle storage distribution, frame, columns, floor beams, and floor tie rods of the target multi-level parking garage.

8. A dynamic prediction system for the structural safety performance of a three-dimensional parking garage, characterized in that, The prediction system includes: A structural safety prediction module is used to determine the target prediction time and input the target prediction time into a fusion prediction model to obtain the structural safety prediction value of the target multi-level parking garage. The fusion prediction model includes an initial safety prediction model and a safety error prediction model. The initial safety prediction model is used to output the initial safety prediction value of the target multi-level parking garage. The safety error prediction model is used to output the safety error prediction value of the target multi-level parking garage. The structural safety prediction value is the sum of the initial safety prediction value and the safety error prediction value. Fusion prediction model determination module; the fusion prediction model determination module includes: The historical structural safety value determination submodule is used to acquire multi-source information of the target multi-level parking garage at different historical moments, and determine the historical structural safety value of the target multi-level parking garage at different historical moments based on the multi-source information. The sample set partitioning submodule is used to divide the true structural safety values ​​of the target automated parking garage at different historical moments into a training set and a test set. The initial security prediction model determination submodule is used to train the heterogeneous kernel related vector machine model by using the lion pack optimization algorithm, taking different historical moments in the training set as input and the true structural security values ​​corresponding to different historical moments in the training set as output, to obtain the initial security prediction model. The initial security prediction value determination submodule is used to input different historical moments in the test set into the initial security prediction model to obtain the initial security prediction values ​​corresponding to different historical moments in the test set. The submodule for determining the true value of safety error is used to determine the true value of safety error corresponding to different historical moments in the test set based on the initial safety prediction value and the true value of structural safety at different historical moments in the test set. The safety error prediction model determination submodule is used to train the differential autoregressive moving average model by taking different historical moments in the test set as input and the true safety error value corresponding to different historical moments in the test set as output, so as to obtain the safety error prediction model. The fusion prediction model determination submodule is used to linearly superimpose the initial security prediction model and the security error prediction model to obtain the fusion prediction model.

9. The dynamic prediction system for the structural safety performance of a three-dimensional parking garage according to claim 8, characterized in that, The prediction system also includes: Fusion prediction model update module; the fusion prediction model update module includes: The target structure safety value acquisition submodule is used to acquire the target structure safety value at the prediction time. The root mean square error determination submodule is used to calculate the root mean square error between the actual value of structural safety at the target prediction time and the predicted value of structural safety at the target prediction time. The fusion prediction model update submodule is used to update the historical time if the root mean square error is greater than a set threshold, obtain multi-source information of the updated historical time, determine the true value of structural safety at the updated historical time based on the multi-source information, divide the true value of structural safety at the updated historical time into an updated training set and an updated test set, and re-determine the fusion prediction model based on the updated training set and the updated test set; the re-determined fusion prediction model is used to predict the structural safety value at times after the target prediction time; the updated historical time includes at least the target prediction time.

10. The dynamic prediction system for the structural safety performance of a three-dimensional parking garage according to claim 8, characterized in that, The submodule for determining the true value of historical structural security specifically includes: The historical multi-source information acquisition submodule is used to acquire multi-source information of the target automated parking garage at different historical moments; A structural safety performance index system construction submodule is used to construct the structural safety performance index system of the target three-dimensional parking garage based on the multi-source information; the structural safety performance index system includes four levels, and each level includes at least one index; The submodule for determining the comprehensive weight of indicators is used to determine the comprehensive weight of each indicator in the structural safety performance evaluation indicator system based on the three-scale analytic hierarchy process and the CRITIC weighting method; the comprehensive weight includes: a first weight and a second weight; The historical structural safety value determination submodule is used to determine the true structural safety value of the target multi-level parking garage at different historical moments based on the comprehensive weight and the fuzzy comprehensive evaluation method.

Images