Cable crane test hoisting multi-source parameter adaptive state representation method and system

Abstract

The application discloses a cable crane test hanging multi-source parameter adaptive state representation method and system, which comprises the following steps: when the cable crane test hanging is graded loaded, real-time collection of cable tower stress, tower top deflection, anchorage displacement, main cable sag and cable force constitutes a multi-source parameter set, and combination of parameter time sequence evolution characteristics and cable crane structure topological constraint to build a multi-source parameter state space. The current loading stage theoretical response benchmark is obtained, the coupling response deviation is calculated according to the deviation degree of the benchmark and the parameter evolution track, and the dynamic representation weight of each parameter is adaptively determined. The multi-dimensional state representation vector is constructed in combination with the state space, the parameter safety threshold and the dynamic weight, the comprehensive state representation index is obtained according to the mapping position of the vector in the preset state flow space, the test hanging state is distinguished in combination with the index, the stage determination condition and the vector evolution trend, and the corresponding test hanging control instruction is generated. The application provides reliable technical support for the structure safety dynamic perception and risk intelligent management and control of the cable crane test hanging operation.

Description

Technical Field

[0001] This invention belongs to the field of safety monitoring technology for bridge construction equipment, and more specifically, relates to a method and system for adaptive state characterization of multi-source parameters in cable crane trial lifting. Background Technology

[0002] In the construction of long-span bridges, cable-stayed cranes, as core heavy lifting equipment, must undergo graded loading test lifts before being officially put into use to verify the overall stress safety performance of key components such as towers, main cables, and anchorages. The test lift process of ultra-long-span cable-stayed cranes involves multiple monitoring indicators, including tower stress, tower top deviation, anchorage displacement, main cable sag, and cable force. These indicators have different physical meanings and dimensions, and the monitoring data at each stage under graded loading are fragmented. If safety judgment is based solely on single indicator thresholds, it is difficult to identify mechanical coupling risks between components in advance, which can easily lead to missed or incorrect judgments.

[0003] While existing cable-stayed crane technologies have achieved monitoring of some parameters during the trial lifting phase—for example, existing technologies disclose trial lifting operations after cable-stayed crane installation, enabling monitoring of mid-span sag and tower top deviation—and some solutions can achieve online monitoring and safety assessment of tower deviation, tilt angle, and multiple cable forces. However, most of these existing technologies remain at the level of sensor data acquisition and simple limit-over alarms, failing to form a complete multi-source parameter coupling analysis and state characterization system. They lack a comprehensive judgment logic for multi-parameter coupling deviation, evolution trends, and measurement point consistency.

[0004] Existing technologies cannot adaptively adjust the parameter evaluation weights according to the test load conditions, nor have they established a closed-loop handling logic that links load observation, retesting, downgrading, and unloading. Relying solely on static threshold evaluation is insufficient to adapt to the dynamic stress characteristics of the graded loading process, and cannot provide reliable risk prediction and automated handling basis for cable crane test operations. Summary of the Invention

[0005] This invention aims to overcome the shortcomings of existing cable crane test lifting methods that rely on single-index thresholds for judgment and are difficult to identify component coupling risks. By constructing a state space through multi-source parameters, adaptively determining the representation weights, and combining the manifold space to obtain a comprehensive state index, the test lifting state is judged and corresponding control commands are generated, thereby realizing dynamic evaluation and closed-loop handling of cable crane test lifting safety.

[0006] To address the aforementioned deficiencies or improvement needs of existing technologies, as a first aspect of this invention, the present invention provides a multi-source parameter adaptive state characterization method for cable crane trial lifting, comprising: S1. During the graded loading process of the cable crane trial lifting, multi-source monitoring data is collected in real time to construct a multi-source parameter set. Based on the temporal evolution characteristics of the multi-source parameter set under the current loading stage and the structural topological constraints of the cable crane, a multi-source parameter state space is constructed. The multi-source parameter set includes tower stress, tower top offset, anchor displacement, main cable sag, and cable force. S2. Obtain the theoretical response benchmark corresponding to the current loading stage, and determine the coupling response deviation between the multi-source parameters based on the degree of deviation between the theoretical response benchmark and the evolution trajectory of each parameter in the multi-source parameter state space, and adaptively determine the dynamic characterization weight of each parameter based on the coupling response deviation. S3. Based on the multi-source parameter state space, the safety threshold boundary of each parameter in the multi-source parameter set, and the dynamic representation weight, construct a multi-dimensional state representation vector, and determine the comprehensive state representation index of the cable crane based on the mapping position of the multi-dimensional state representation vector in the preset state manifold space. S4. Based on the comprehensive state characterization index, the preset state determination conditions of the current loading stage, and the evolution trend of the multidimensional state characterization vector, determine the trial lifting state characterization result of the cable crane, and generate corresponding trial lifting control instructions based on the trial lifting state characterization result.

[0007] Further, in S1, based on the temporal evolution characteristics of the multi-source parameter set under the current loading stage and the structural topological constraints of the cable-stayed crane, a multi-source parameter state space is constructed, including: Wavelet packet decomposition is performed on the time-series monitoring data of each parameter in the multi-source parameter set to extract the time-series evolution feature components of each parameter in different frequency bands; based on the structural topological constraints of the cable crane, a physical correlation matrix is ​​constructed among the parameters in the multi-source parameter set; the physical correlation matrix represents the mechanical coupling relationship of each parameter on the structural force transmission path; the state space of the multi-source parameter is constructed using the time-series evolution feature components as state variables and the physical correlation matrix as state transition constraints.

[0008] Furthermore, the specific construction process of the multi-source parameter state space is as follows: Calculate the time-delay mutual information between the time-series evolution characteristic components of any two parameters in the multi-source parameter set, and determine the time-delay mutual information. The optimal time delay order to reach the maximum value ; Based on the optimal time delay order, the time-series evolution feature components of each parameter are time-shifted and concatenated to construct a time-delay augmented feature matrix. ; Using the physical association matrix as the graph adjacency matrix, calculate the corresponding graph Laplacian matrix. The projection matrix is ​​solved with the objective of minimizing the local structural loss and the graph Laplacian regularization term of the time-delay augmented feature matrix after projection. The objective function is: in, Represents the trace of a matrix. Describing the Frobenius norm, The preset regularization coefficient; Using the projection matrix For the time-delay augmented feature matrix Dimensionality reduction mapping is performed to obtain the multi-source parameter state space.

[0009] Furthermore, the formula for calculating the time-delay mutual information is as follows: in, For the first The parameter and the first Each parameter in time delay The time-delay mutual information, For the first Each parameter in Time and the Each parameter in The joint probability distribution at time t, and These are the corresponding marginal probability distributions; The optimal time delay order .

[0010] Further, in S2, the deviation degree of coupling response between the multi-source parameters is determined based on the degree of deviation between the theoretical response benchmark and the evolution trajectory of each parameter in the multi-source parameter state space, including: Calculate the trajectory deviation distance between the evolution trajectory of each parameter in the multi-source parameter state space and the theoretical response benchmark; Based on the trajectory deviation distance and the structural topological constraints, a cooperative deviation index is determined between each parameter pair; the cooperative deviation index characterizes the degree of consistency between parameter pairs with mechanical coupling relationships and their deviation from the theoretical response benchmark. Based on the cooperative deviation index between each parameter pair, a coupling response deviation matrix is ​​constructed, and the norm of the coupling response deviation matrix is ​​used as the coupling response deviation degree between the multi-source parameters.

[0011] Further, in S2, the dynamic representation weights of each parameter are adaptively determined based on the coupling response deviation, including: When the deviation of the coupling response is greater than a preset deviation threshold, identify the abnormal parameter pair with the largest contribution in the coupling response deviation matrix; Based on the trajectory deviation distance of each parameter in the abnormal parameter pair, the initial weight of the parameter with a larger deviation in the abnormal parameter pair is increased, and the initial weight of the parameter with a smaller deviation is decreased, to obtain the dynamic characterization weight of each parameter; When the deviation of the coupling response is not greater than the preset deviation threshold, the preset benchmark weight of each parameter is used as the dynamic characterization weight.

[0012] Furthermore, the construction process of the multidimensional state representation vector in S3 specifically includes: Based on the aforementioned safety threshold boundary, calculate the normalized safety margin of each parameter in the multi-source parameter set within the multi-source parameter state space. , For the first Each parameter in Normalized safety margin at any given moment; The normalized safety margin is adaptively weighted and corrected using the dynamic representation weights to obtain the weighted state components of each parameter. , For the first Each parameter in The weighted state components at each time step; The weighted state components of all parameters in the multi-source parameter set at the same time are concatenated in a preset order to construct the multi-dimensional state representation vector; the expression of the multi-dimensional state representation vector is: in, for The multidimensional state representation vector at time step. The total number of parameters in the multi-source parameter set. This represents the transpose of a vector.

[0013] Further, in step S3, the comprehensive state representation index of the cable crane is determined based on the mapping position of the multidimensional state representation vector in the preset state manifold space, including: The preset state manifold space is constructed based on historical trial lifting data; the preset state manifold space includes a normal state submanifold, a warning state submanifold, and a dangerous state submanifold; Calculate the geodesic distances from the multidimensional state representation vector to the normal state submanifold, the warning state submanifold, and the dangerous state submanifold; Based on the geodesic distance and the preset state benchmark values ​​corresponding to each submanifold, the comprehensive state characterization index is obtained through weighted fusion calculation.

[0014] Further, in step S4, generating corresponding test lifting control commands based on the test lifting status characterization results includes: When the test lifting status characterization result is normal, a continue loading command is generated and output to the hoist control system of the cable crane; When the test lifting status characterization result is an early warning state, a load holding observation command is generated, and the rate of change of the comprehensive status characterization index is continuously monitored within a preset load holding time; if the rate of change converges to within a preset convergence threshold, a continue loading command is generated; if the rate of change does not converge to within the preset convergence threshold, an unloading command is generated. When the test lifting status characterization result is a dangerous state, an emergency unloading command is generated and output to the hoist control system, and an audible and visual alarm is triggered at the same time.

[0015] As a second aspect of the present invention, a multi-source parameter adaptive state characterization system for cable crane trial lifting is also provided, comprising: The parameter state space construction unit is used to collect multi-source monitoring data in real time during the graded loading process of the cable crane trial lifting to construct a multi-source parameter set, and construct a multi-source parameter state space based on the temporal evolution characteristics of the multi-source parameter set under the current loading stage and the structural topological constraints of the cable crane; the multi-source parameter set includes tower stress, tower top offset, anchor displacement, main cable sag, and cable force; The dynamic characterization weight determination unit is used to obtain the theoretical response benchmark corresponding to the current loading stage, and determine the coupling response deviation between the multi-source parameters based on the degree of deviation between the theoretical response benchmark and the evolution trajectory of each parameter in the multi-source parameter state space, and adaptively determine the dynamic characterization weight of each parameter based on the coupling response deviation. The comprehensive state index calculation unit is used to construct a multi-dimensional state representation vector based on the multi-source parameter state space, the safety threshold boundary of each parameter in the multi-source parameter set, and the dynamic representation weight, and to determine the comprehensive state representation index of the cable crane based on the mapping position of the multi-dimensional state representation vector in the preset state manifold space. The test lifting control command generation unit is used to determine the test lifting state characterization result of the cable crane based on the comprehensive state characterization index, the preset state judgment conditions of the current loading stage, and the evolution trend of the multi-dimensional state characterization vector, and to generate corresponding test lifting control commands based on the test lifting state characterization result.

[0016] In summary, compared with the prior art, the above-described technical solutions conceived by this invention can achieve the following beneficial effects: 1. The multi-source parameter adaptive state characterization method for cable-stayed crane trial lifting of the present invention collects multi-source monitoring data on tower stress, tower top deviation, anchor displacement, main cable sag, and cable force, and constructs a multi-source parameter state space by combining the temporal evolution characteristics of the parameters with the topological constraints of the cable-stayed crane structure. This method performs feature extraction on the temporal monitoring data under graded loading, and incorporates the mechanical coupling relationship of the structural force transmission path, transforming the original monitoring data into a standardized state expression with mechanical constraints. This avoids the data fragmentation problem caused by the difference in the dimensions of different parameters, eliminates the time-delay coupling interference in the monitoring time series, and provides a unified and reliable data calculation basis for subsequent parameter deviation analysis and state assessment.

[0017] 2. The cable crane trial lifting multi-source parameter adaptive state characterization method of the present invention calculates the coupling response deviation between multi-source parameters by comparing the parameter evolution trajectory with the theoretical response benchmark and adaptively outputs the dynamic characterization weight of each parameter. This method no longer uses fixed evaluation weights, but can adjust the evaluation ratio of each parameter according to the actual situation of the coupling deviation between parameters, identify abnormal parameter pairs in the coupling response deviation matrix, automatically amplify the weight of parameters with larger deviations when the system has coupling anomalies, and use the preset benchmark weights under normal working conditions, objectively reflecting the actual contribution level of each monitored parameter to the overall structural safety during the trial lifting loading process.

[0018] 3. The multi-source parameter adaptive state characterization method for cable crane trial lifting of the present invention constructs a multi-dimensional state characterization vector and obtains a comprehensive state characterization index based on a preset state manifold space mapping. It then outputs trial lifting control commands by combining the index results with the vector evolution trend. This method considers both parameter normalization safety margin and state evolution trend, relies on geodesic distance to complete state attribution judgment, and outputs commands for continued loading, load observation, or emergency unloading according to different working conditions such as normal, warning, and dangerous conditions. For warning conditions, it also monitors the rate of change of the index to complete a secondary judgment, realizing quantitative characterization of the cable crane trial lifting operation state and closed-loop handling of the operation process. Attached Figure Description

[0019] Figure 1 This is a flowchart of a multi-source parameter adaptive state characterization method for cable crane trial lifting according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the cable crane test lifting facility according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the system units according to an embodiment of the present invention; In all the accompanying drawings, the same reference numerals denote the same technical features, specifically: 1-left anchorage; 2-slope stability line; 3-traction cable; 4-lifting cable; 5-main cable; 6-cable support; 7-cable tower; 8-back cable; 9-right anchorage; 10-carrying trolley; 11-counterweight block. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0021] Example 1 Please refer to Figure 1 This embodiment 1 provides a multi-source parameter adaptive state characterization method for cable crane trial lifting, including: S1. During the graded loading process of the cable crane trial lifting, multi-source monitoring data is collected in real time to construct a multi-source parameter set. Based on the temporal evolution characteristics of the multi-source parameter set under the current loading stage and the structural topological constraints of the cable crane, a multi-source parameter state space is constructed. The multi-source parameter set includes tower stress, tower top offset, anchor displacement, main cable sag, and cable force. S2. Obtain the theoretical response benchmark corresponding to the current loading stage, and determine the coupling response deviation between the multi-source parameters based on the degree of deviation between the theoretical response benchmark and the evolution trajectory of each parameter in the multi-source parameter state space, and adaptively determine the dynamic characterization weight of each parameter based on the coupling response deviation. S3. Based on the multi-source parameter state space, the safety threshold boundary of each parameter in the multi-source parameter set, and the dynamic representation weight, construct a multi-dimensional state representation vector, and determine the comprehensive state representation index of the cable crane based on the mapping position of the multi-dimensional state representation vector in the preset state manifold space. S4. Based on the comprehensive state characterization index, the preset state determination conditions of the current loading stage, and the evolution trend of the multidimensional state characterization vector, determine the trial lifting state characterization result of the cable crane, and generate corresponding trial lifting control instructions based on the trial lifting state characterization result.

[0022] Please refer to Figure 2 The cable-stayed crane trial lifting facility involved in this embodiment mainly includes a main cable 5 spanning above the construction area. The two ends of the main cable 5 are respectively anchored to the left anchor 1 and the right anchor 9 located inside the slope stability line 2. Cable towers 7 are vertically provided on both sides to support the main cable 5. The top of the cable tower 7 is connected to the corresponding anchor via backing cables 8 to maintain structural stability. A trolley 10 is slidably provided on the main cable 5. A counterweight block 11 for trial lifting loading is connected below the trolley 10. The trolley 10 is also connected to the traction cable 3 to drive it to move laterally along the main cable 5. In addition, a cable support 6 is provided on the main cable 5 to support and guide the lifting cable 4 and the traction cable 3.

[0023] Based on this, please refer to Figure 2 This embodiment 1 further elaborates on the above steps.

[0024] (1) Construction of the parametric state space During the graded loading process of cable crane trial lifting, it is necessary to collect five types of monitoring data in real time: tower stress, tower top displacement, anchor displacement, main cable sag, and cable force, to construct a multi-source parameter set. These five types of parameters correspond to different mechanical response types of key load-bearing components of the cable crane: tower stress reflects the stress state of the main tower, tower top displacement characterizes the spatial displacement of the top of the tower, anchor displacement reflects the displacement deformation of the anchorage end, main cable sag reflects the geometric shape change of the main cable, and cable force reflects the tension state of each cable.

[0025] The acquired time-series monitoring data of each parameter needs to undergo feature extraction processing. Wavelet packet decomposition is performed on the time-series monitoring data of each parameter in the multi-source parameter set to extract the time-series evolution feature components of each parameter in different frequency bands. The original monitoring signal often contains high-frequency noise and low-frequency drift components, and direct analysis of the original signal is easily subject to interference. Wavelet packet decomposition can finely divide the signal according to frequency bands and extract effective feature components that reflect the stress evolution law of the structure, providing high-quality feature input for subsequent state space construction.

[0026] Based on the structural topological constraints of the cable-stayed crane, a physical correlation matrix is ​​constructed among the parameters in the multi-source parameter set. This physical correlation matrix characterizes the mechanical coupling relationship between each parameter along the structural force transmission path. The structural force transmission path of the cable-stayed crane is a continuous path from the main cable to the tower and then to the anchorage. The values ​​of each element in the physical correlation matrix are determined according to the positional relationship of each parameter along the structural force transmission path, reflecting the strength of coupling between corresponding parameter pairs along the mechanical transmission path.

[0027] Using temporal evolution feature components as state variables and physical correlation matrices as state transition constraints, a multi-source parameter state space is constructed.

[0028] To capture the potential time-delay coupling relationships between various parameters, the time-delay mutual information between the temporal evolution characteristic components of any two parameters in the multi-source parameter set is further calculated to determine the time-delay mutual information. The optimal time delay order to reach the maximum value The purpose of introducing time-delay mutual information is that the mechanical response transmission in cable-stayed structures is not instantaneous. The change of one parameter is transmitted to another parameter after a certain time delay. Considering only the synchronous correlation will miss this time-delay coupling relationship. Time-delay mutual information can measure the statistical dependence between parameters at different time offsets, thereby identifying the optimal time delay order.

[0029] The formula for calculating time-delay mutual information is: In the formula, For the first The parameter and the first Each parameter in time delay The time-delay mutual information, For the first Each parameter in Time and the Each parameter in The joint probability distribution at time t, and These are the corresponding marginal probability distributions; the formula for calculating the optimal time delay order is: .

[0030] Based on the optimal time delay order, the time-series evolution feature components of each parameter are time-shifted and concatenated to construct a time-delay augmented feature matrix. Time-shift concatenation stacks the characteristic components of each parameter at different time delay orders along the time axis, making the matrix... It can simultaneously reflect the amplitude information and time delay coupling information of each parameter.

[0031] Using the physical affinity matrix as the graph adjacency matrix, calculate the corresponding graph Laplacian matrix. The objective is to minimize the local structural loss of the time-delay augmented eigenvalue matrix after projection and the graph Laplacian regularization term, and to solve for the projection matrix. The objective function is expressed as follows: ,in, Represents the trace of a matrix. Describing the Frobenius norm, These are the preset regularization coefficients. The first term in the objective function. This is used to preserve the local graph structure of the data after dimensionality reduction projection, that is, to ensure that adjacent sample points in the original high-dimensional space remain adjacent in the low-dimensional projected space; the second term This is used to constrain the reconstruction error after projection and prevent excessive information loss during dimensionality reduction. By jointly optimizing these two terms, the projection matrix... It can satisfy the constraints of physical force transmission relationships while preserving the local geometry of the data.

[0032] Using projection matrix augmented characteristic matrix with time delay Dimensionality reduction mapping is performed to obtain a multi-source parameter state space. Dimensionality reduction mapping projects the high-dimensional time-delay augmented feature matrix into a low-dimensional space, which reduces computational complexity while preserving the key mechanical coupling features between parameters. The resulting multi-source parameter state space serves as the input basis for subsequent determination of dynamic representation weights and calculation of the comprehensive state index.

[0033] (2) Determination of dynamic representation weights After constructing the multi-source parameter state space, it is necessary to obtain the theoretical response benchmark corresponding to the current loading stage. The theoretical response benchmark is calculated based on the structural mechanics model of the cable-stayed crane, reflecting the theoretical expected response values ​​of each parameter under the current loading condition, and providing a comparison benchmark for the subsequent quantification of the degree of deviation.

[0034] Based on the deviation between the theoretical response benchmark and the evolution trajectories of each parameter in the multi-source parameter state space, the coupling response deviation between the multi-source parameters is determined, and the dynamic representation weights of each parameter are adaptively determined accordingly. First, the trajectory deviation distance between the evolution trajectories of each parameter in the multi-source parameter state space and the theoretical response benchmark is calculated. The formula for calculating the trajectory deviation distance is: ,in, For the first Trajectory deviation distance of each parameter For the first Each parameter in The measured value at time [time]. For the first Each parameter in The theoretical response baseline value at time t. and The first The measured maximum and minimum values ​​of each parameter during the current loading phase. This represents the total number of sampling moments within the current loading phase. The trajectory deviation distance has been normalized to eliminate the influence of differences in the dimensions of different parameters, allowing the degree of deviation of each parameter to be compared on the same scale.

[0035] Based on trajectory deviation distance and structural topological constraints, a cooperative deviation index is determined for each parameter pair. This cooperative deviation index characterizes the degree of consistency between parameter pairs with mechanical coupling and their deviation from the theoretical response benchmark. The formula for calculating the cooperative deviation index is as follows: ,in, For the first The parameter and the first Cooperative deviation index between parameters For the first The parameter and the first The mechanical correlation coefficients of each parameter in the topological constraints of the structure. For the first The parameter and the first The absolute value of the difference in the trajectory deviation distance of each parameter. For the first The parameter and the first The sum of the trajectory deviations of each parameter, For the first The parameter and the first The temporal correlation coefficients of each parameter during the current loading phase. For the first The sum of the absolute values ​​of the time-series correlation coefficients of one parameter with all other parameters. The total number of parameters in the multi-source parameter set. This is the preset correlation strength adjustment coefficient.

[0036] The design of the cooperative deviation index takes into account three factors: the mechanical correlation coefficient reflects the inherent coupling strength of the parameter pair on the structural force transmission path, the index term measures the difference in the degree of deviation between the two parameters, the smaller the difference, the greater the cooperative deviation, and the time correlation coefficient term further considers the response correlation between the parameters in the time dimension.

[0037] Based on the cooperative deviation index between each parameter pair, a coupling response deviation matrix is ​​constructed. , where matrix elements The Frobenius norm of the coupling response deviation matrix is ​​used as the coupling response deviation degree among multiple source parameters, and the calculation formula is as follows: ,in, The deviation of the coupling response among the multi-source parameters. The first element in the coupling response deviation matrix Line 1 The elements of the column. The coupled response deviation integrates the cooperative deviation information of all parameter pairs into a scalar value, which is used to characterize the overall severity of the deviation of the entire multi-source parameter set from the theoretical response benchmark.

[0038] The dynamic representation weights of each parameter are adaptively determined based on the coupling response deviation. When the coupling response deviation exceeds a preset deviation threshold, the abnormal parameter pairs with the largest contribution in the coupling response deviation matrix are identified. The method for identifying abnormal parameter pairs is to calculate the contribution ratio of each element in the matrix to the total norm; the parameter pair corresponding to the element with the largest contribution ratio is the abnormal parameter pair.

[0039] Based on the trajectory deviation distance of each parameter in the abnormal parameter pair, the initial weight of the parameter with a larger deviation is increased, while the initial weight of the parameter with a smaller deviation is decreased, resulting in the dynamic characterization weight of each parameter. This weight adjustment mechanism allows parameters with larger deviations to gain greater influence in the subsequent comprehensive state evaluation when coupling anomalies occur, thus more sensitively reflecting the actual stress state of the structure.

[0040] When the deviation of the coupling response is not greater than the preset deviation threshold, it indicates that the response of each parameter is basically consistent with the theoretical expectation. At this time, the preset benchmark weight of each parameter is used as the dynamic characterization weight to maintain the conventional evaluation weight allocation. The preset benchmark weight is pre-set according to the importance of each parameter in the safety of the cable-stayed crane structure, and maintains the stability of the evaluation under normal working conditions.

[0041] (3) Calculation of the comprehensive state index After obtaining the state space of the multi-source parameters and the dynamic representation weights, it is necessary to construct a multi-dimensional state representation vector based on the safety threshold boundary of each parameter in the multi-source parameter set and the dynamic representation weights, and determine the comprehensive state representation index of the cable crane based on the mapping position of the multi-dimensional state representation vector in the preset state manifold space.

[0042] First, based on the safety threshold boundary, the normalized safety margin of each parameter in the multi-source parameter set within the multi-source parameter state space is calculated. The formula for calculating the normalized safety margin is: ,in, For the first Each parameter in Normalized safety margin at any moment For the first Each parameter in The state value in the multi-source parameter state space at any given time. and The first Each parameter in The upper and lower bounds of the safety threshold corresponding to each moment.

[0043] The normalized safety margin maps the actual values ​​of each parameter to a range of 0 to 1. The closer the value is to 1, the farther the parameter is from the upper limit and the larger the safety margin; the closer the value is to 0, the closer the parameter is to the lower limit of the safety threshold and the smaller the safety margin. The upper and lower limits of the safety threshold are preset based on the design allowable values ​​of each component of the cable-stayed crane and engineering experience, reflecting the allowable range of variation of different parameters in structural safety evaluation.

[0044] An adaptive weighted correction of the normalized safety margin is performed using dynamic representation weights to obtain the weighted state components of each parameter. The formula for calculating the weighted state components is as follows: ,in, For the first Each parameter in The weighted state components at time step, For the first Each parameter in Dynamic representation weights at time points, For the first The normalized safety margin of each parameter is in The first time derivative at time t, For symbolic functions, The preset trend sensitivity coefficient, This is a preset nonlinear mapping index.

[0045] The construction of weighted state components introduces a trend-sensitive term based on the normalized safety margin. This trend-sensitive term reflects the changing trend of the parameter safety margin through the first-order time derivative of the normalized safety margin: when the safety margin is decreasing (the derivative is negative), the sign function outputs a negative value, causing the weighted state components to decrease, thus reflecting the deterioration trend of the safety state in advance; when the safety margin is increasing (the derivative is positive), the sign function outputs a positive value, causing the weighted state components to increase, reflecting the improvement trend of the safety state.

[0046] Trend Sensitivity Coefficient Controlling the influence strength of the trend term, nonlinear mapping exponent Used to nonlinearly compress the magnitude of the derivative, preventing the trend term from dominating the overall evaluation when the rate of change is too large. Dynamically representing weights. Here, the importance of each parameter is adaptively determined according to the degree of coupling deviation and assigned to the construction of the state components, so that the parameters with larger deviations receive higher weight contributions in the weighted state components.

[0047] The weighted state components of all parameters in the multi-source parameter set at the same time are concatenated in a predetermined order to construct a multi-dimensional state representation vector. The expression for the multi-dimensional state representation vector is: ,in, for The multidimensional state representation vector at time step. The total number of parameters in the multi-source parameter set. This represents the transpose of a vector. A multidimensional state representation vector integrates the state information of each parameter after adaptive weighting correction into a unified vector representation.

[0048] A pre-defined state manifold space is constructed based on historical test lifting data. This space includes three sub-manifolds: a normal state sub-manifold, a warning state sub-manifold, and a hazardous state sub-manifold. The normal state sub-manifold consists of multi-dimensional state representation vector samples from historical test lifting operations where all parameters were within safe ranges. The warning state sub-manifold consists of samples from historical test lifting operations where parameter responses showed abnormal deviations but had not yet reached a hazardous level. The hazardous state sub-manifold consists of samples from historical test lifting operations where parameter responses exceeded permissible safety ranges. These three sub-manifolds characterize the state distribution features of the cable crane under different safety levels, providing a reference benchmark for classifying the current state vector.

[0049] The multidimensional state representation vector at the current moment The mapping is performed by projecting the vectors onto a predefined state manifold space, determining their mapping positions within the manifold space. The mapping process uses multidimensional state representations of the vectors. For the query point, the geodesic distance between it and the submanifolds of the normal state, the warning state, and the danger state is calculated respectively, so as to characterize the spatial positional relationship of the current state in the manifold space.

[0050] The formula for calculating geodesic distance is: ,in, for The multidimensional state representation vector at time step [time] to the [time step] Geodesic distance of a submanifold And they respectively correspond to the normal state submanifold, the warning state submanifold, and the danger state submanifold. for The multidimensional state representation vector at time step [time]. For the first The set of sample points of each submanifold For the first Sample points in a submanifold For The Riemann logarithmic mapping operator with base point, Let be the Euclidean norm in the tangent space.

[0051] The reason for using geodesic distance instead of Euclidean distance is that the distribution of sample points in the state manifold space has a nonlinear geometric structure. The geodesic distance, being the shortest distance between the manifold surface metric vector and the submanifold, can more realistically reflect the degree of difference between the state vector and each security level. For each submanifold, the point that minimizes the distance among all sample points is selected as the nearest neighbor reference point, and the corresponding geodesic distance is taken as the distance metric for that submanifold.

[0052] Based on the geodesic distance and the preset state reference values ​​corresponding to each submanifold, a comprehensive state characterization index is obtained through weighted fusion calculation. The formula for calculating the comprehensive state characterization index is: ,in, for The comprehensive state characterization index at time 1. For the first The fusion weight coefficients of each submanifold satisfy the following conditions: , For the first The preset state baseline value corresponding to each submanifold For the first Distance decay scale parameter of individual manifolds It is a natural exponential function.

[0053] Comprehensive State Characterization Index The geodesic distances of the three submanifolds are fused into a single scalar value using an exponentially decaying weighted average. A smaller geodesic distance indicates that the current state vector is closer to that submanifold, and the corresponding exponential decay term is larger, contributing more to the overall state representation exponent. Fusion weighting coefficients. To reflect the relative importance of each submanifold in the comprehensive evaluation, a preset state baseline value is established. The reference security level scale value for each submanifold, and the distance attenuation scale parameter. The decay rate of the influence of the geodesic distance on the comprehensive index is controlled, and different decay scale parameters can be set for different submanifolds to adapt to their respective state distribution characteristics. The comprehensive state characterization index serves as the core criterion for generating subsequent trial lifting control commands, compressing multi-dimensional parameter information into a single value, which facilitates hierarchical state determination and closed-loop control decisions.

[0054] (4) Generation of trial lifting control commands After obtaining the comprehensive state characterization index, the trial lifting state of the cable crane is characterized by combining the preset state judgment conditions of the current loading stage and the evolution trend of the multi-dimensional state characterization vector, and corresponding trial lifting control commands are generated based on the characterization results. The preset state judgment conditions are pre-set according to the safety control requirements of each loading stage in the cable crane trial lifting plan, including normal state judgment intervals, early warning state judgment intervals, and dangerous state judgment intervals, each interval corresponding to a different value range of the comprehensive state characterization index. The evolution trend of the multi-dimensional state characterization vector is obtained by judging the consistency of direction of the vector at multiple consecutive moments, and is used to identify whether the state of each parameter shows a continuously deteriorating change direction.

[0055] When the comprehensive state characterization index falls within the normal state determination range, and the evolution trend of the multidimensional state characterization vector does not show a continuous deterioration direction, the trial lifting state characterization result is determined to be in a normal state. At this time, a continued loading command is generated and output to the hoist control system of the cable crane. After receiving the command, the hoist control system applies the next level of load according to the preset loading rate.

[0056] When the comprehensive state characterization index falls into the warning state determination range, or when the evolution trend of the multidimensional state characterization vector shows a continuous deterioration direction, the trial lifting state characterization result is determined to be a warning state. At this time, a load-bearing observation command is generated. After receiving the load-bearing observation command, the hoist control system stops the loading action and maintains the current load unchanged. The rate of change of the comprehensive state characterization index is continuously monitored within the preset load-bearing duration. During the load-bearing observation period, the value of the comprehensive state characterization index is collected at fixed time intervals, the rate of change between adjacent collection times is calculated, and it is determined whether the rate of change converges to within the preset convergence threshold.

[0057] If the rate of change converges to within the preset convergence threshold, it indicates that the mechanical response of the structure under the current load has become stable. The trial lifting status characterization result is corrected from the warning state to the normal state, and a continue loading command is generated and output to the hoist control system to resume the loading process.

[0058] If the rate of change does not converge to the preset convergence threshold, it indicates that the mechanical response of the structure under the current load is still changing and there is a risk of further deterioration. An unloading command is generated and output to the hoist control system. After receiving the unloading command, the hoist control system gradually reduces the load according to the preset unloading rate until the comprehensive state characterization index falls back to the normal state judgment range.

[0059] When the comprehensive condition characterization index falls into the danger zone, the trial lifting condition characterization result is determined to be in a dangerous state. At this time, an emergency unloading command is generated and output to the hoist control system. After receiving the emergency unloading command, the hoist control system executes the unloading action at the maximum safe unloading rate to reduce the load to a safe range as quickly as possible. At the same time, an audible and visual alarm is triggered to remind on-site personnel to take appropriate emergency measures.

[0060] The generation and output of the trial lifting control command are executed cyclically at every moment of the graded loading, forming a closed-loop control process of "state representation - command generation - execution feedback - re-representation". This enables the cable crane trial lifting process to dynamically adjust the loading strategy according to the actual response state of the structure, and complete the graded loading test under the premise of ensuring structural safety.

[0061] Example 2 Please refer to Figure 3 This embodiment 2 provides a multi-source parameter adaptive state characterization system for cable crane trial lifting, including: The parameter state space construction unit is used to collect multi-source monitoring data in real time during the graded loading process of the cable crane trial lifting to construct a multi-source parameter set, and construct a multi-source parameter state space based on the temporal evolution characteristics of the multi-source parameter set under the current loading stage and the structural topological constraints of the cable crane; the multi-source parameter set includes tower stress, tower top offset, anchor displacement, main cable sag, and cable force; The dynamic characterization weight determination unit is used to obtain the theoretical response benchmark corresponding to the current loading stage, and determine the coupling response deviation between the multi-source parameters based on the degree of deviation between the theoretical response benchmark and the evolution trajectory of each parameter in the multi-source parameter state space, and adaptively determine the dynamic characterization weight of each parameter based on the coupling response deviation. The comprehensive state index calculation unit is used to construct a multi-dimensional state representation vector based on the multi-source parameter state space, the safety threshold boundary of each parameter in the multi-source parameter set, and the dynamic representation weight, and to determine the comprehensive state representation index of the cable crane based on the mapping position of the multi-dimensional state representation vector in the preset state manifold space. The test lifting control command generation unit is used to determine the test lifting state characterization result of the cable crane based on the comprehensive state characterization index, the preset state judgment conditions of the current loading stage, and the evolution trend of the multi-dimensional state characterization vector, and to generate corresponding test lifting control commands based on the test lifting state characterization result.

[0062] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for adaptive state characterization of multi-source parameters in cable crane trial lifting, characterized in that, include: S1. During the graded loading process of the cable crane trial lifting, multi-source monitoring data is collected in real time to construct a multi-source parameter set. Based on the temporal evolution characteristics of the multi-source parameter set under the current loading stage and the structural topological constraints of the cable crane, a multi-source parameter state space is constructed. The multi-source parameter set includes tower stress, tower top offset, anchor displacement, main cable sag, and cable force. S2. Obtain the theoretical response benchmark corresponding to the current loading stage, and determine the coupling response deviation between the multi-source parameters based on the degree of deviation between the theoretical response benchmark and the evolution trajectory of each parameter in the multi-source parameter state space, and adaptively determine the dynamic characterization weight of each parameter based on the coupling response deviation. S3. Based on the multi-source parameter state space, the safety threshold boundary of each parameter in the multi-source parameter set, and the dynamic representation weight, construct a multi-dimensional state representation vector, and determine the comprehensive state representation index of the cable crane based on the mapping position of the multi-dimensional state representation vector in the preset state manifold space. S4. Based on the comprehensive state characterization index, the preset state determination conditions of the current loading stage, and the evolution trend of the multidimensional state characterization vector, determine the trial lifting state characterization result of the cable crane, and generate corresponding trial lifting control instructions based on the trial lifting state characterization result.

2. The method for adaptive state characterization of multi-source parameters in cable crane trial lifting according to claim 1, characterized in that, In step S1, based on the temporal evolution characteristics of the multi-source parameter set under the current loading stage and the structural topological constraints of the cable-stayed crane, a multi-source parameter state space is constructed, including: Wavelet packet decomposition is performed on the time-series monitoring data of each parameter in the multi-source parameter set to extract the time-series evolution feature components of each parameter in different frequency bands; based on the structural topological constraints of the cable crane, a physical correlation matrix is ​​constructed among the parameters in the multi-source parameter set; the physical correlation matrix represents the mechanical coupling relationship of each parameter on the structural force transmission path; the state space of the multi-source parameter is constructed using the time-series evolution feature components as state variables and the physical correlation matrix as state transition constraints.

3. The method for adaptive state characterization of multi-source parameters in cable crane trial lifting according to claim 2, characterized in that, The specific construction process of the multi-source parameter state space is as follows: Calculate the time-delay mutual information between the time-series evolution characteristic components of any two parameters in the multi-source parameter set, and determine the time-delay mutual information. The optimal time delay order to reach the maximum value ; Based on the optimal time delay order, the time-series evolution feature components of each parameter are time-shifted and concatenated to construct a time-delay augmented feature matrix. ; Using the physical association matrix as the graph adjacency matrix, calculate the corresponding graph Laplacian matrix. The projection matrix is ​​solved with the objective of minimizing the local structural loss and the graph Laplacian regularization term of the time-delay augmented feature matrix after projection. The objective function is: ； in, Represents the trace of a matrix. Describing the Frobenius norm, The preset regularization coefficient; Using the projection matrix For the time-delay augmented feature matrix Dimensionality reduction mapping is performed to obtain the multi-source parameter state space.

4. The multi-source parameter adaptive state characterization method for cable crane trial lifting according to claim 3, characterized in that, The formula for calculating the time-delay mutual information is: ； in, For the first The parameter and the first Each parameter in time delay The time-delay mutual information, For the first Each parameter in Time and the Each parameter in The joint probability distribution at time t, and These are the corresponding marginal probability distributions; The optimal time delay order .

5. The method for adaptive state characterization of multi-source parameters in cable crane trial lifting according to claim 1, characterized in that, In step S2, the deviation degree of coupling response between the multi-source parameters is determined based on the degree of deviation between the theoretical response benchmark and the evolution trajectory of each parameter in the multi-source parameter state space, including: Calculate the trajectory deviation distance between the evolution trajectory of each parameter in the multi-source parameter state space and the theoretical response benchmark; Based on the trajectory deviation distance and the structural topological constraints, a cooperative deviation index is determined between each parameter pair; the cooperative deviation index characterizes the degree of consistency between parameter pairs with mechanical coupling relationships and their deviation from the theoretical response benchmark. Based on the cooperative deviation index between each parameter pair, a coupling response deviation matrix is ​​constructed, and the norm of the coupling response deviation matrix is ​​used as the coupling response deviation degree between the multi-source parameters.

6. The method for adaptive state characterization of multi-source parameters in cable crane trial lifting according to claim 5, characterized in that, In step S2, the dynamic representation weights of each parameter are adaptively determined based on the coupling response deviation, including: When the deviation of the coupling response is greater than a preset deviation threshold, identify the abnormal parameter pair with the largest contribution in the coupling response deviation matrix; Based on the trajectory deviation distance of each parameter in the abnormal parameter pair, the initial weight of the parameter with a larger deviation in the abnormal parameter pair is increased, and the initial weight of the parameter with a smaller deviation is decreased, to obtain the dynamic characterization weight of each parameter; When the deviation of the coupling response is not greater than the preset deviation threshold, the preset benchmark weight of each parameter is used as the dynamic characterization weight.

7. The method for adaptive state characterization of multi-source parameters in cable crane trial lifting according to claim 1, characterized in that, The construction process of the multidimensional state representation vector in S3 specifically includes: Based on the aforementioned safety threshold boundary, calculate the normalized safety margin of each parameter in the multi-source parameter set within the multi-source parameter state space. , For the first Each parameter in Normalized safety margin at any given moment; The normalized safety margin is adaptively weighted and corrected using the dynamic representation weights to obtain the weighted state components of each parameter. , For the first Each parameter in The weighted state components at each time step; The weighted state components of all parameters in the multi-source parameter set at the same time are concatenated in a preset order to construct the multi-dimensional state representation vector; the expression of the multi-dimensional state representation vector is: ； in, for The multidimensional state representation vector at time step. The total number of parameters in the multi-source parameter set. This represents the transpose of a vector.

8. The multi-source parameter adaptive state characterization method for cable crane trial lifting according to claim 1, characterized in that, In step S3, based on the mapping position of the multidimensional state representation vector in the preset state manifold space, the comprehensive state representation index of the cable crane is determined, including: The preset state manifold space is constructed based on historical trial lifting data; the preset state manifold space includes a normal state submanifold, a warning state submanifold, and a dangerous state submanifold; Calculate the geodesic distances from the multidimensional state representation vector to the normal state submanifold, the warning state submanifold, and the dangerous state submanifold; Based on the geodesic distance and the preset state benchmark values ​​corresponding to each submanifold, the comprehensive state characterization index is obtained through weighted fusion calculation.

9. The method for adaptive state characterization of multi-source parameters in cable crane trial lifting according to claim 1, characterized in that, In step S4, a corresponding test lifting control command is generated based on the test lifting status characterization result, including: When the test lifting status characterization result is normal, a continue loading command is generated and output to the hoist control system of the cable crane; When the test lifting status characterization result is an early warning state, a load holding observation command is generated, and the rate of change of the comprehensive status characterization index is continuously monitored within a preset load holding time; if the rate of change converges to within a preset convergence threshold, a continue loading command is generated; if the rate of change does not converge to within the preset convergence threshold, an unloading command is generated. When the test lifting status characterization result is a dangerous state, an emergency unloading command is generated and output to the hoist control system, and an audible and visual alarm is triggered at the same time.

10. A multi-source parameter adaptive state characterization system for cable crane trial lifting, characterized in that, include: The parameter state space construction unit is used to collect multi-source monitoring data in real time during the graded loading process of the cable crane trial lifting to construct a multi-source parameter set, and construct a multi-source parameter state space based on the temporal evolution characteristics of the multi-source parameter set under the current loading stage and the structural topological constraints of the cable crane; the multi-source parameter set includes tower stress, tower top offset, anchor displacement, main cable sag, and cable force; The dynamic characterization weight determination unit is used to obtain the theoretical response benchmark corresponding to the current loading stage, and determine the coupling response deviation between the multi-source parameters based on the degree of deviation between the theoretical response benchmark and the evolution trajectory of each parameter in the multi-source parameter state space, and adaptively determine the dynamic characterization weight of each parameter based on the coupling response deviation. The comprehensive state index calculation unit is used to construct a multi-dimensional state representation vector based on the multi-source parameter state space, the safety threshold boundary of each parameter in the multi-source parameter set, and the dynamic representation weight, and to determine the comprehensive state representation index of the cable crane based on the mapping position of the multi-dimensional state representation vector in the preset state manifold space. The test lifting control command generation unit is used to determine the test lifting state characterization result of the cable crane based on the comprehensive state characterization index, the preset state judgment conditions of the current loading stage, and the evolution trend of the multi-dimensional state characterization vector, and to generate corresponding test lifting control commands based on the test lifting state characterization result.

Images