A precise control method for adjusting posture of a rotary bridge based on deep learning

Abstract

The application relates to the technical field of bridge construction, in particular to a precise control method for posture adjustment of a swivel bridge based on deep learning, which obtains posture evaluation information from multiple deployment posture evaluation points and outputs posture evaluation features; according to the rotation speed features of a reference bridge swivel main pier set and the attribute features of the reference bridge swivel main pier set, posture stability of the reference bridge swivel main pier set is determined, and posture stability features are output; according to the difference between the posture evaluation features and the posture stability features, whether each traction cable is faulty is detected. The application can realize real-time recording of the posture of a bridge body, automatic control of a hydraulic traction system during swiveling, and does not require a large amount of manual observation during the whole swiveling process, thereby saving manpower and material resources and guaranteeing the accuracy and timeliness of observation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of bridge construction, and in particular to a precise control method for adjusting the posture of a swivel bridge based on deep learning. BACKGROUND

[0002] In recent years, with the popularization and application of swivel construction technology, various bridges involving crossing railways and three-dimensional intersections often consider using swivel construction technology to solve spatial conflicts. In traditional swivel construction, the beam body often over-rotates or under-rotates. Since swivel operations are often performed at the main pier, the micro-operation at the main pier is often amplified by more than several dozen times through the amplification effect of the main beam span. The beam body often hovers between over-rotation and under-rotation during adjustment and positioning. Due to the very low dynamic and static friction coefficients of the rotating hinge and the problem of rotational inertia, the beam body often cannot be stopped at the ideal position.

[0003] At present, most of the bridge swivel construction process is not monitored in real time, but relies on manual observation and on-site shouting through a wireless intercom to control the transmission and reception of instructions. The posture of the bridge body during construction is not recorded in real time, and the hydraulic traction system during swivel is manually controlled. A large amount of manual observation is used during the entire swivel process, which not only consumes manpower and resources, but also cannot guarantee the accuracy and timeliness of observation.

[0004] Therefore, how to provide a technical solution for accurately and automatically adjusting the posture of a swivel bridge in real time has become a technical problem to be solved in the field. SUMMARY

[0005] To achieve the above-mentioned purpose, the present application provides the following technical solutions:

[0006] According to the first aspect of the present application, the present application claims a precise control method for adjusting the posture of a swivel bridge based on deep learning, which comprises:

[0007] Obtain posture evaluation information from a plurality of deployment posture evaluation points, and output posture evaluation features; the posture evaluation points are used to evaluate the posture depth information of the traction cable in the reference bridge swivel main pier, and a plurality of posture evaluation points are deployed for each traction cable; the posture evaluation features represent the posture stable state of each traction cable obtained in real time during rotation;

[0008] Determine the posture stability of the reference bridge swivel main pier set according to the rotation speed features of the reference bridge swivel main pier set and the attribute features of the reference bridge swivel main pier set, and output posture stability features; the posture stability features represent the expected posture stability state of the reference bridge swivel main pier set in the associated rotation speed;

[0009] According to the difference between the attitude evaluation feature and the attitude stability feature, whether each traction cable has a fault is detected.

[0010] Further, the detection of whether each traction cable has a fault according to the difference between the attitude evaluation feature and the attitude stability feature comprises:

[0011] For each traction cable, according to the attitude evaluation feature, real-time rigidity attributes and real-time texture attributes of the associated attitude evaluation points are determined, and first rigidity attributes and first texture attributes are output.

[0012] According to the attitude stability feature, target rigidity attributes and target texture attributes of the associated attitude evaluation points are determined, and second rigidity attributes and second texture attributes are output.

[0013] According to the comparison record of the first rigidity attributes and the second rigidity attributes, whether the real-time obtained attitude stability is in a fault persistent stability in the rigidity dimension is detected, and a first detection record is output.

[0014] According to the comparison record of the first texture attributes and the second texture attributes, whether the real-time obtained attitude stability is in a fault persistent stability in the texture roughness dimension is detected, and a second detection record is output.

[0015] In a state where the first detection record or the second detection record represents that the real-time obtained attitude stability has a fault persistent stability, it is detected that the associated traction cable has a fault.

[0016] Further, the detection of whether the real-time obtained attitude stability is in a fault persistent stability in the rigidity dimension according to the comparison record of the first rigidity attributes and the second rigidity attributes, and the output of the first detection record, comprises:

[0017] According to the second rigidity attributes, a total rigidity variance of the attitude depth information of the associated traction cable along the time sequence is determined, and a rigidity variance threshold value is determined according to the total rigidity variance.

[0018] According to the first rigidity attributes, an object rigidity variance of the attitude depth information in the real-time obtained target object set is determined.

[0019] According to the object rigidity variance and the rigidity variance threshold value, whether the real-time obtained attitude stability information in the target object set has a fault in the rigidity dimension is detected, and the first detection record is output.

[0020] Further, the detection of whether the real-time obtained attitude stability is in a fault persistent stability in the texture roughness dimension according to the comparison record of the first texture attributes and the second texture attributes, and the output of the second detection record, comprises:

[0021] determine a total texture roughness standard deviation of the pose depth information associated with the traction cable according to the second texture attribute, and determine a texture roughness standard deviation threshold according to the total texture roughness standard deviation;

[0022] determine an object texture roughness standard deviation of the pose depth information in the real-time obtained target object set according to the first texture attribute;

[0023] detect whether the real-time obtained pose stability information in the texture roughness dimension in the target object set has a fault according to the object texture roughness standard deviation and the texture roughness standard deviation threshold, and output a second detection record.

[0024] Further, the determining the pose stability of the reference bridge swivel main pier set according to the rotation speed feature of the reference bridge swivel main pier set and the attribute feature of the reference bridge swivel main pier set, and outputting the pose stability feature, comprises:

[0025] obtaining historical rotation information of the reference bridge swivel main pier set; the historical rotation information comprises indexes and obtained pose stability information of the reference bridge swivel main pier set when the reference bridge swivel main pier set rotates normally in multiple rotation poses; the rotation speed features of different rotation poses are different;

[0026] constructing a pose determination model of the reference bridge swivel main pier set according to the historical rotation information;

[0027] determining a target rotation pose of the reference bridge swivel main pier set from multiple deployed candidate poses of the determination model according to the rotation speed feature of the reference bridge swivel main pier set, and outputting a pose feature;

[0028] determining the pose stability of the reference bridge swivel main pier set in the current rotation pose by using the pose determination model according to the attribute feature of the reference bridge swivel main pier set and the pose feature, and outputting a pose stability feature.

[0029] Further, the rotation speed feature comprises multiple rotation speed indexes; the determining a target rotation pose of the reference bridge swivel main pier set from multiple deployed candidate poses of the determination model according to the rotation speed feature of the reference bridge swivel main pier set, and outputting a pose feature, comprises:

[0030] for each first rotation pose in the multiple deployed first rotation poses, determining an association degree between each rotation speed index in the rotation speed feature and a rotation inclination index associated with the first rotation pose;

[0031] The correlation degrees of each rotation speed index are weighted and summed, and a matching score of the rotation speed feature and the associated first rotation posture is output; wherein the weight of each rotation speed index is determined according to the influence degree of the associated rotation speed index on the stability of the reference bridge pier set;

[0032] According to the matching score of the rotation speed feature and each first rotation posture, a second rotation posture is determined from the plurality of first rotation postures, and the second rotation posture is taken as a target rotation posture of the reference bridge pier set, and a posture feature is output.

[0033] Further, the rotation speed feature includes at least one of the following rotation speed indexes:

[0034] Rotation speed;

[0035] Bridge deck type;

[0036] Ambient temperature;

[0037] Rotation time;

[0038] Rotation arc length;

[0039] Ball hinge center azimuth.

[0040] Further, the attribute feature of the reference bridge pier set includes mechanical indexes and rotation indexes; wherein,

[0041] The mechanical indexes include tetrafluoro sliding plate features and ring sliding track features;

[0042] The rotation indexes include at least one of closure section contact attributes, bridge mass distribution attributes, bridge segmented system attributes, ring sliding track features and bridge maintenance states.

[0043] Further, the method further includes:

[0044] According to the historical stability information of each traction cable, the correlation degree of each traction cable posture stability and geological features is determined;

[0045] According to the structure feature and the correlation degree of each traction cable, a deployment scheme of each traction cable posture evaluation point is determined; the deployment scheme includes a deployment number and an evaluation position; wherein the deployment number increases as the correlation degree increases.

[0046] The present application relates to the technical field of bridge construction, and particularly relates to a precise control method for posture adjustment of a swivel bridge based on deep learning, which obtains posture evaluation information from multiple deployment posture evaluation points and outputs posture evaluation features; determines posture stability of a reference bridge swivel main pier set according to rotation speed features of the reference bridge swivel main pier set and attribute features of the reference bridge swivel main pier set, and outputs posture stability features; and detects whether each traction cable is faulty according to differences between the posture evaluation features and the posture stability features. The present application can achieve real-time recording of the posture of a bridge body, automatic control of a hydraulic traction system during swiveling, and does not require a large amount of manual observation during the entire swiveling process, thereby saving manpower and resources and ensuring the accuracy and timeliness of observation. BRIEF DESCRIPTION OF DRAWINGS

[0047] Figure 1 A work flow chart of a precise control method for posture adjustment of a swivel bridge based on deep learning as claimed in the present application embodiment;

[0048] Figure 2 A structure module chart of a precise control device for posture adjustment of a swivel bridge based on deep learning as claimed in the present application embodiment. DETAILED DESCRIPTION

[0049] The technical solutions in the present application embodiments will be described clearly and completely below with reference to the drawings in the present application embodiments. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art without creative labor on the basis of the embodiments in the present application fall within the scope of protection of the present application.

[0050] The terms "first", "second", "third" in the present application are only for descriptive purpose, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the technical features indicated. Therefore, the features defined with "first", "second", "third" can explicitly or implicitly include a plurality of the features. In the description of the present application, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise explicitly and specifically limited. All directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present application are only used to explain the relative position relationship, movement state, etc. between components in a certain posture (as shown in the drawings), and if the certain posture changes, the directional indications also change accordingly. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or structure including a series of steps or units is not limited to the listed steps or units, but can optionally include steps or units not listed, or can optionally include other steps or units inherent to the process, method, product or structure.

[0051] Reference herein to "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment can be included in a variety of embodiments of the present application. The presence of the phrase in various places in the specification does not necessarily all refer to the same embodiment, nor is it necessarily independent or alternative to other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0052] The bridge swivel main pier set is an important bridge construction structure, and the existence of the bridge brings great convenience to people's daily travel. However, the more powerful the functions that a system can provide, the more complex its structure set will be, and the closer the connection between the parts will be, which leads to the difficulty of system maintenance. How to ensure the effective and safe rotation of the bridge swivel main pier set has become an urgent problem, and therefore, in improving the evaluation effect and fault diagnosis ability, it is particularly important and important to establish an effective and reliable evaluation method.

[0053] In the related art, the traction cable evaluation of the bridge swivel main pier set can be realized by visual inspection, that is, the traction cable is visually inspected regularly to detect whether the traction cable structure has a fault; however, since the traction cable fault often exists in the rotation process, the regular maintenance method cannot guarantee the real-time of maintenance, and there is a great safety hazard.

[0054] In view of the above problems, the rotation information of the traction cable can be obtained in real time during rotation, and then the performance of each structure is analyzed according to the obtained rotation index to detect whether the structure rotation is normal; however, the rotation index can only reflect whether the structure is currently rotating normally, and cannot reliably detect the fatigue degree and damage degree of the structure in a timely manner, so that potential safety hazards cannot be detected in a timely manner, for example, the failure of the external baffle of the bridge body caused by structure vibration cannot be detected, and for example, the over-stress and fatigue damage of the structure and the structure cannot be detected; therefore, the evaluation method using real-time rotation information of the structure has a certain lag, so that the comprehensiveness and reliability of the evaluation record are low.

[0055] Therefore, in various embodiments of the present application, the stability of the posture of the traction cable is obtained by setting a plurality of posture evaluation points, and the real-time obtained stability of the posture is compared with the expected stability of the posture, so as to detect whether the traction cable has a fault according to whether the stability of the posture has a fault, so that the fault characteristics of the traction cable can be obtained in a timely manner when the performance information of the traction cable does not have a fault but the structure has a safety hazard state, the early fault signal of the structure is detected, the comprehensive evaluation of the traction cable is ensured, the risk of failure is reduced, and the comprehensiveness and accuracy of the evaluation record of the traction cable are improved; further, since the real-time rotation speed feature is introduced in the process of determining the expected stability of the posture, the comprehensiveness of the model input feature is improved, so that the accuracy of the expected stability of the posture output by the model is improved, and the accuracy of the final detection record is further improved, so as to ensure the safe and stable rotation of the traction cable.

[0056] The embodiment of the present application provides a precise control method for posture adjustment of a rotating bridge based on deep learning, which is applied to electronic structures, and can be applied to computers, bridge rotating body vehicle-mounted computers, servers, cloud servers and the like. Figure 1 As shown in the figure, the method can include:

[0057] Step 101: Obtain posture evaluation information from a plurality of deployment posture evaluation points, and output posture evaluation features; the posture evaluation points are used to evaluate the posture depth information of the traction cable in the reference bridge rotating body, and a plurality of posture evaluation points are deployed for each traction cable; the posture evaluation features represent the stable state of the posture of each traction cable obtained in real time during rotation.

[0058] In real-time application, in order to ensure the accuracy of the posture evaluation information, a plurality of posture evaluation points can be deployed for each traction cable.

[0059] In real-time application, since the structure characteristics of the traction cable are different, the position points at which the traction cable generates stability can be different, and the degree of influence of the ground structure can also be different; therefore, in order to improve the accuracy of the information obtained by the posture evaluation points, how to set the posture evaluation points can be determined according to the correlation degree of the structure characteristics of the traction cable and the performance index of the geological characteristics.

[0060] According to this, in an embodiment, the method can further comprise:

[0061] According to the historical stability information of each traction cable, the correlation degree of the posture stability of each traction cable and the geological feature is determined;

[0062] According to the structural characteristics and the correlation degree of each traction cable, a deployment scheme of the posture evaluation point of each traction cable is determined; the deployment scheme includes the deployment quantity and the evaluation position; wherein, when the correlation degree increases, the deployment quantity increases accordingly.

[0063] In real-time application, determining the correlation degree of the posture stability of the traction cable and the geological feature can be understood as determining the correlation degree of the posture stability of the traction cable and the performance index of the geological feature; wherein, the performance index of the geological feature can include the ground structure and the azimuth angle of the spherical hinge center; for the posture evaluation point of the traction cable, the side of the traction cable related to the posture evaluation point can be called the evaluation surface, and the azimuth angle of the spherical hinge center can be understood as the included angle between the rotation direction and the detection surface.

[0064] In real-time application, when determining the deployment scheme, a plurality of stress surfaces, that is, the sides that will be affected by the ground structure, can be determined according to the structural characteristics of the traction cable; then, a plurality of evaluation surfaces are determined from the plurality of stress surfaces according to the correlation degree of each stress surface and the geological feature, and the posture evaluation point is deployed on the evaluation surface.

[0065] Specifically, for each stress surface of the traction cable, the posture stability of the stress surface under different ground structures in the same state of the same azimuth angle of the spherical hinge center can be determined according to the historical stability information, the ground structure correlation degree of the stress surface is calculated, specifically, the ground structure correlation degree of the stress surface can be calculated according to the ratio of the posture stability and the ground structure; then, the posture stability of the stress surface under different azimuth angles of the spherical hinge center under the same ground structure is determined according to the historical stability information, so as to calculate the azimuth angle correlation degree of the spherical hinge center of the stress surface, specifically, the azimuth angle correlation degree of the spherical hinge center of the stress surface can be calculated according to the ratio of the posture stability and the azimuth angle of the spherical hinge center; then, the ground structure correlation degree and the azimuth angle correlation degree of the spherical hinge center are weighted and summed to output the geological feature correlation degree of the stress surface and the geological feature; then, it is detected whether the geological feature correlation degree of each stress surface is greater than the deployment correlation degree threshold value, and in the state of being greater than the deployment correlation degree threshold value, the stress surface is used as the monitoring surface; in real-time application, the first N stress surfaces with the largest geological feature correlation degree can also be selected as the monitoring surface, and N is an integer greater than or equal to 1.

[0066] In real-time application, the number of deployment of the posture evaluation points can be determined according to the structural characteristics of the traction cable. Specifically, the traction cable can be classified according to the structural characteristics of the traction cable, and a first type of traction cable and a second type of traction cable are output. The first type of traction cable is a structure with simple structure and less affected by geological characteristics, and the number of deployment of the posture evaluation points thereof can be N1. The second type of traction cable is a structure with complex structure and more affected by geological characteristics, and the number of deployment of the posture evaluation points thereof can be N2. N1 and N2 are both integers not less than 1, and N1 < N2. The number of N1 and N2 can be deployed according to the real-time evaluation requirements. In this application,

[0067] Step 102: According to the rotation speed characteristics of the reference bridge turning main pier set and the attribute characteristics of the reference bridge turning main pier set, the posture stability of the reference bridge turning main pier set is determined, and the posture stability characteristics are output. The posture stability characteristics represent the expected posture stability state of the reference bridge turning main pier set in the associated rotation speed.

[0068] In real-time application, a posture determination model can be established according to simulation or historical information, and then the reference posture stability curve, i.e. the target posture stability curve, is determined by using the posture determination model. Thus, according to the difference between the associated values in the posture stability and the target posture stability curve under the same rotation speed, it is determined whether the real-time obtained posture stability has a fault.

[0069] Accordingly, in an embodiment, the determination of the posture stability of the reference bridge turning main pier set according to the rotation speed characteristics of the reference bridge turning main pier set and the attribute characteristics of the reference bridge turning main pier set, and the output of the posture stability characteristics, comprise:

[0070] Obtain the historical rotation information of the reference bridge turning main pier set. The historical rotation information includes indexes and obtained posture stability information of the reference bridge turning main pier set in normal rotation in multiple rotation postures. The rotation speed characteristics of different rotation postures are different.

[0071] According to the historical rotation information, a posture determination model of the reference bridge turning main pier set is constructed.

[0072] According to the rotation speed characteristics of the reference bridge turning main pier set, a target rotation posture of the reference bridge turning main pier set is determined from multiple deployed determination model candidate postures, and a posture feature is output.

[0073] According to the attribute characteristics of the reference bridge turning main pier set and the posture feature, the posture stability of the reference bridge turning main pier set in the current rotation posture is determined by using the posture determination model, and the posture stability feature is output.

[0074] In real-time application, the rotation speed of different time periods is difficult to be completely consistent, therefore, the closest rotation posture can be retrieved from the model information according to the similarity between the rotation speed characteristics, so as to guarantee the reliability of the posture determination model output record.

[0075] According to this, in an embodiment, the rotation speed characteristics include a plurality of rotation speed indicators; the target rotation posture of the reference bridge pier set is determined from a plurality of deployed determination model candidate postures according to the rotation speed characteristics of the reference bridge pier set, and the output posture characteristics include:

[0076] For each first rotation posture in the plurality of deployed first rotation postures, the correlation degree of each rotation speed indicator in the rotation speed characteristics and the associated rotation inclination indicator in the first rotation posture is determined;

[0077] The correlation degrees of each rotation speed indicator are weighted and summed to output the matching score of the rotation speed characteristics and the associated first rotation posture; wherein the weight of each rotation speed indicator is determined according to the influence degree of the associated rotation speed indicator on the stability of the reference bridge pier set;

[0078] According to the matching score of the rotation speed characteristics and each first rotation posture, a second rotation posture is determined from the plurality of first rotation postures, and the second rotation posture is taken as the target rotation posture of the reference bridge pier set, and the output posture characteristics.

[0079] In real-time application, when determining the correlation degree of the rotation speed indicator and the associated rotation inclination indicator, the ratio of the rotation speed indicator and the associated rotation inclination indicator can be calculated to output the environment correlation degree, and the calculated environment correlation degree is compared with the deployed correlation degree threshold value to detect whether the environment correlation degree is greater than the deployed correlation degree threshold value, and in the state that the environment correlation degree is greater than the deployed matching threshold value, the associated first rotation posture is taken as the candidate posture; then the candidate posture with the highest environment correlation degree is selected from the determined candidate postures as the target rotation posture, i.e., as the current rotation posture, and the posture characteristics of the target rotation posture are output.

[0080] In an embodiment, the rotation speed characteristics include at least one of the following rotation speed indicators:

[0081] Rotation speed;

[0082] Bridge deck type;

[0083] Ambient temperature;

[0084] Rotation time;

[0085] Rotation arc length;

[0086] Ball joint center azimuth.

[0087] In an embodiment, the attribute features of the reference bridge pier set include mechanical indexes and rotation indexes; wherein,

[0088] The mechanical indexes include tetrafluoro sliding plate features and ring sliding channel features.

[0089] The rotation indexes include at least one of closure section contact attributes, bridge mass distribution attributes, bridge segmented system attributes, ring sliding channel features, and bridge maintenance states.

[0090] In real-time application, the attribute features of the reference bridge pier set can be indexes pre-deployed in the production process.

[0091] In real-time application, the shape indexes, rotation indexes, and traction cable types of different bridge pier sets are different, so the rotation information and the attitude stability characteristics at the same rotation speed are also different. Using the same determination model to determine the attitude stability has low accuracy. Therefore, different determination models can be established for different types of bridge pier sets to achieve more accurate and reliable determination.

[0092] Accordingly, in an embodiment, the constructing the attitude determination model of the reference bridge pier set according to the historical rotation information can include:

[0093] According to the historical rotation information, constructing the attitude determination model of at least one type of reference bridge pier set, and outputting a plurality of first attitude determination models; the attribute features of different types of reference bridge pier sets are different.

[0094] On this basis, in an embodiment, the determining the attitude stability of the reference bridge pier set in the current rotation attitude according to the attribute features of the reference bridge pier set and the attitude features, using the attitude determination model, and outputting attitude stability features can include:

[0095] According to the attribute features of the reference bridge pier set, determining a target attitude determination model from the constructed plurality of first attitude determination models;

[0096] Using the target attitude determination model and the associated attitude features, determining the attitude stability of the reference bridge pier set in the current rotation attitude, and outputting attitude stability features.

[0097] Step 103: According to the difference between the attitude evaluation features and the attitude stability features, detecting whether each traction cable has a fault.

[0098] In real-time application, in order to ensure the comprehensiveness of the detection record, thereby improving the accuracy of the detection record, detection can be performed from two dimensions of rigidity and texture roughness.

[0099] According to this, in an embodiment, the detection of whether each traction cable is faulty according to the difference between the attitude evaluation feature and the attitude stability feature can include:

[0100] For each traction cable, according to the attitude evaluation feature, the real-time rigidity attribute and the real-time texture attribute of the associated attitude evaluation point are determined, and the first rigidity attribute and the first texture attribute are output.

[0101] According to the attitude stability feature, the target rigidity attribute and the target texture attribute of the associated attitude evaluation point are determined, and the second rigidity attribute and the second texture attribute are output.

[0102] According to the comparison record of the first rigidity attribute and the second rigidity attribute, it is detected whether the real-time obtained attitude stability is in a state of fault persistence and stability in the rigidity dimension, and a first detection record is output.

[0103] According to the comparison record of the first texture attribute and the second texture attribute, it is detected whether the real-time obtained attitude stability is in a state of fault persistence and stability in the texture roughness dimension, and a second detection record is output.

[0104] In a state where the first detection record or the second detection record represents that the real-time obtained attitude stability is in a state of fault persistence and stability, it is detected that the associated traction cable is faulty.

[0105] In an embodiment, the detection of whether the real-time obtained attitude stability is in a state of fault persistence and stability in the rigidity dimension according to the comparison record of the first rigidity attribute and the second rigidity attribute, and the output of the first detection record can include:

[0106] According to the second rigidity attribute, the overall rigidity variance of the attitude depth information of the associated traction cable obtained along the time sequence is determined, and a rigidity variance threshold value is determined according to the overall rigidity variance.

[0107] According to the first rigidity attribute, the object rigidity variance of the attitude depth information in the real-time obtained target object set is determined.

[0108] According to the object rigidity variance and the rigidity variance threshold value, it is detected whether the real-time obtained attitude stability information in the target object set is faulty in the rigidity dimension, and the first detection record is output.

[0109] In real-time application, the determination of the overall rigidity variance of the attitude depth information of the associated traction cable according to the second rigidity attribute and the determination of the rigidity variance threshold value according to the overall rigidity variance can be that the average value of the smooth objects at the same moment is obtained according to the traction cable associated attitude evaluation point, then the time sequence of the objects is constructed according to the average value of the information obtained at each moment, and then the overall reference variance, that is, the overall rigidity variance, is determined by using the time object sequence, and the overall reference variance and the deployed coefficient extreme rigidity variance threshold value.

[0110] In real-time application, the determination of the object rigidity variance of the attitude depth information in the real-time obtained target object set according to the first rigidity attribute can be understood as that the information set obtained by evaluation is periodically obtained, and each obtained information set includes a plurality of target objects in the current acquisition period, the target object information at the same moment in each attitude evaluation point is informationally fused, then the target object set is constructed according to the time extraction arrangement, and the object rigidity variance of the attitude depth information in the target object set is calculated.

[0111] In real-time application, when the real-time obtained attitude smooth information in the target object set is detected according to the object rigidity variance and the rigidity variance threshold value to determine whether there is a fault in the rigidity dimension, it can be that whether the rigidity variance of each object in the target object set is greater than the rigidity variance threshold value is detected, and when it is greater, it can be considered that the variance of the current object is much larger than the normal range, so that the detected attitude smooth information has a fault.

[0112] In real-time application, for the attitude smooth, whether some dynamic friction factor components in the friction texture are significantly higher than the normal range can be detected to identify the fault persistence and stability.

[0113] According to this, in an embodiment, the detection of whether the real-time obtained attitude smooth is in fault persistence and stability in the texture roughness dimension according to the comparison record of the first texture attribute and the second texture attribute, and the output of the second detection record can include:

[0114] According to the second texture attribute, the overall texture roughness standard deviation of the associated traction cable attitude depth information is determined, and the texture roughness standard deviation threshold value is determined according to the overall texture roughness standard deviation;

[0115] According to the first texture attribute, the object texture roughness standard deviation of the attitude depth information in the real-time obtained target object set is determined.

[0116] According to the object texture roughness standard deviation and the texture roughness standard deviation threshold value, whether the real-time obtained attitude smooth information in the target object set is in fault in the texture roughness dimension is detected, and the second detection record is output.

[0117] In real-time application, when determining the overall texture roughness standard deviation of the associated traction cable posture depth information, and determining the texture roughness standard deviation threshold value according to the overall texture roughness standard deviation, each dynamic friction factor component in the friction texture can be calculated, and statistical characteristics such as mean and standard deviation can be further calculated, and then a normal range can be established according to the statistical characteristics, that is, the texture roughness standard deviation threshold value is determined.

[0118] It should be noted that the threshold value deployed in the embodiment of the application can be set according to historical experience, or can be determined by using an associated simulation model, and the embodiment of the application does not limit this.

[0119] In summary, the precise control method for posture adjustment of a rotating bridge based on deep learning provided by the embodiment of the application acquires the posture stability of the traction cable by setting multiple posture evaluation points, and compares the real-time acquired posture stability with the expected posture stability, so as to detect whether the traction cable has a fault according to whether the posture stability has a fault, so that the fault characteristics of the traction cable can be obtained in time in the state that the performance information of the traction cable does not have a fault but the structure has a safety hidden danger, the early fault signal of the structure is detected, the all-around evaluation of the traction cable is ensured, the fault risk is reduced, and the comprehensiveness and accuracy of the evaluation record of the traction cable are improved. Further, since the real-time rotating speed feature is introduced in the process of determining the expected posture stability, the comprehensiveness of the model input feature is improved, so that the accuracy of the expected posture stability output by the model is improved, and the accuracy of the final detection record is further improved, so that the safe and stable rotation of the traction cable is ensured.

[0120] In order to realize the precise control method for posture adjustment of a rotating bridge based on deep learning, the embodiment of the application further provides a precise control device for posture adjustment of a rotating bridge based on deep learning, which is arranged on an electronic structure, as shown in the figure, and the device can include: Figure 2

[0121] The evaluation unit 201 is configured to acquire posture evaluation information from multiple deployed posture evaluation points, and output posture evaluation features; the posture evaluation points are used to evaluate the posture depth information of the traction cable in the reference bridge rotating main pier, and multiple posture evaluation points are deployed for each traction cable; and the posture evaluation features represent the real-time obtained posture stability state of each traction cable in the rotating process.

[0122] The calculation unit 202 is configured to determine the posture stability of the reference bridge rotating main pier set according to the rotating speed features of the reference bridge rotating main pier set and the attribute features of the reference bridge rotating main pier set, and output posture stability features; and the posture stability features represent the expected posture stability state of the reference bridge rotating main pier set in the associated rotating speed.

[0123] ​The processing unit 203 is configured to detect whether each traction cable is in failure according to a difference between the attitude evaluation feature and the attitude stability feature

[0124] In an embodiment, the processing unit 203 can be specifically configured to:

[0125] For each traction cable, determine real-time rigidity attribute and real-time texture attribute of an associated attitude evaluation point according to the attitude evaluation feature, and output first rigidity attribute and first texture attribute;

[0126] Determine target rigidity attribute and target texture attribute of the associated attitude evaluation point according to the attitude stability feature, and output second rigidity attribute and second texture attribute;

[0127] Detect whether real-time obtained attitude stability is in failure sustained stability in rigidity dimension according to a comparison record of the first rigidity attribute and the second rigidity attribute, and output first detection record;

[0128] Detect whether real-time obtained attitude stability is in failure sustained stability in texture roughness dimension according to a comparison record of the first texture attribute and the second texture attribute, and output second detection record;

[0129] In a case where the first detection record or the second detection record represents that real-time obtained attitude stability is in failure sustained stability, detect that the associated traction cable is in failure.

[0130] In an embodiment, the detecting whether real-time obtained attitude stability is in failure sustained stability in rigidity dimension according to the comparison record of the first rigidity attribute and the second rigidity attribute, and outputting the first detection record, comprises:

[0131] Determine overall rigidity variance of attitude depth information of the associated traction cable along a time sequence according to the second rigidity attribute, and determine a rigidity variance threshold value according to the overall rigidity variance;

[0132] Determine object rigidity variance of attitude depth information in a target object set obtained in real time according to the first rigidity attribute;

[0133] Detect whether real-time obtained attitude stability information in the target object set is in failure in rigidity dimension according to the object rigidity variance and the rigidity variance threshold value, and output the first detection record.

[0134] In an embodiment, the detecting whether real-time obtained attitude stability is in failure sustained stability in texture roughness dimension according to the comparison record of the first texture attribute and the second texture attribute, and outputting the second detection record, comprises:

[0135] determine a total texture roughness standard deviation of the pose depth information associated with the traction cable according to the second texture attribute, and determine a texture roughness standard deviation threshold according to the total texture roughness standard deviation;

[0136] determine an object texture roughness standard deviation of the pose depth information in the real-time obtained target object set according to the first texture attribute;

[0137] detect whether the real-time obtained pose stability information in the target object set has a fault in the texture roughness dimension according to the object texture roughness standard deviation and the texture roughness standard deviation threshold, and output a second detection record.

[0138] In an embodiment, the computing unit 202 can be specifically configured to:

[0139] obtain historical rotation information of the reference bridge turning main pier set; the historical rotation information includes indexes and obtained pose stability information of the reference bridge turning main pier set when the reference bridge turning main pier set is normally rotated in multiple rotation poses; rotation speed characteristics of different rotation poses are different;

[0140] construct a pose determination model of the reference bridge turning main pier set according to the historical rotation information;

[0141] determine a target rotation pose of the reference bridge turning main pier set from multiple deployed candidate poses of the determination model according to rotation speed characteristics of the reference bridge turning main pier set, and output a pose feature;

[0142] determine pose stability of the reference bridge turning main pier set in a current rotation pose by using the pose determination model according to attribute characteristics of the reference bridge turning main pier set and the pose feature, and output a pose stability feature.

[0143] In an embodiment, the rotation speed characteristics include multiple rotation speed indexes; the determination of the target rotation pose of the reference bridge turning main pier set from the multiple deployed candidate poses of the determination model according to the rotation speed characteristics of the reference bridge turning main pier set and the output of the pose feature include:

[0144] for each first rotation pose in the multiple deployed first rotation poses, determine an association degree between each rotation speed index in the rotation speed characteristics and a rotation inclination index associated with the first rotation pose;

[0145] weight and sum the association degrees of each rotation speed index to output a matching score of the rotation speed characteristics and the associated first rotation pose; wherein the weight of each rotation speed index is determined according to an influence degree of the associated rotation speed index on the stability of the reference bridge turning main pier set;

[0146] determine a second rotation posture from the plurality of first rotation postures according to the matching scores of the rotation speed features and each first rotation posture, and output the second rotation posture as a target rotation posture of the reference bridge pier set.

[0147] In an embodiment, the rotation speed feature includes at least one of the following rotation speed indicators:

[0148] rotation speed;

[0149] bridge type;

[0150] ambient temperature;

[0151] rotation time;

[0152] rotation arc length;

[0153] spherical hinge center azimuth.

[0154] In an embodiment, the attribute feature of the reference bridge pier set includes mechanical indicators and rotation indicators; wherein,

[0155] the mechanical indicators include fluorine sliding plate features and ring sliding channel features;

[0156] the rotation indicators include at least one of closure section contact attributes, bridge mass distribution attributes, bridge segmented system attributes, ring sliding channel features and bridge maintenance states.

[0157] In an embodiment, the processing unit 203 can be further configured to:

[0158] determine the correlation degree between each traction cable posture stability and geological features according to historical stability information of each traction cable;

[0159] determine a deployment scheme of each traction cable posture evaluation point according to the structure features and the correlation degree of each traction cable; the deployment scheme includes a deployment number and an evaluation position; wherein, the deployment number increases as the correlation degree increases.

[0160] It should be noted that the above embodiment provides a deep learning-based precise control device for adjusting the posture of a rotating bridge. When evaluating the traction cable, only the above-mentioned division of program modules is used as an example for illustration. In real-time applications, the above-mentioned processing can be completed by different program modules according to needs, that is, the internal structure of the device is divided into different program modules to complete all or part of the above-described processing. In addition, the deep learning-based precise control device for adjusting the posture of a rotating bridge and the deep learning-based precise control method for adjusting the posture of a rotating bridge provided by the above embodiment belong to the same concept, and the specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0161] In several embodiments provided by the present application, it should be understood that the disclosed system, device and method can be implemented in other manners. For example, the division of the above-described device embodiments is only a logical function division, and there can be another division manner for the actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between different units, or the among different units, can be indirect couplings or communication connections through some interfaces, devices or units, and can be in electrical, mechanical or other forms.

[0162] In addition, each functionally described unit in the various embodiments of the present application can be integrated into one processing unit, or each unit can exist alone physically, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware, or in the form of software functional units.

[0163] The specific implementation ways of the present application have been described in detail above, but they are only examples, and the present application is not limited to the above-described specific implementation ways. Any equivalent modifications or substitutions made by those skilled in the art to the present application are also within the scope of the present application, and thus, any equivalent transformations and modifications, improvements, etc. made without departing from the spirit and principle range of the present application should be covered within the scope of the present application.

Claims

1. A method for precise control of posture adjustment of a rotating bridge based on deep learning, characterized in that, The method comprises: Obtaining posture evaluation information from a plurality of deployment posture evaluation points, and outputting posture evaluation features; the posture evaluation points are used to evaluate the posture depth information of the traction cable in the reference bridge turning main pier, and a plurality of posture evaluation points are deployed for each traction cable; the posture evaluation features represent the posture stable state of each traction cable in real time during rotation; According to the rotation speed features of the reference bridge turning main pier set and the attribute features of the reference bridge turning main pier set, the posture stability of the reference bridge turning main pier set is determined, and posture stability features are output; the posture stability features represent the expected posture stability state of the reference bridge turning main pier set in the associated rotation speed; According to the difference between the posture evaluation features and the posture stability features, it is detected whether each traction cable has a fault; According to the difference between the posture evaluation features and the posture stability features, it is detected whether each traction cable has a fault, comprising: For each traction cable, according to the posture evaluation features, the real-time rigidity attribute and the real-time texture attribute of the associated posture evaluation point are determined, and the first rigidity attribute and the first texture attribute are output; According to the posture stability features, the target rigidity attribute and the target texture attribute of the associated posture evaluation point are determined, and the second rigidity attribute and the second texture attribute are output; According to the comparison record of the first rigidity attribute and the second rigidity attribute, it is detected whether the real-time obtained posture stability is in a fault persistent stable state in the rigidity dimension, and a first detection record is output; According to the comparison record of the first texture attribute and the second texture attribute, it is detected whether the real-time obtained posture stability is in a fault persistent stable state in the texture roughness dimension, and a second detection record is output; In the state that the first detection record or the second detection record represents that the real-time obtained posture stability has a fault persistent stable state, it is detected that the associated traction cable has a fault; According to the rotation speed features of the reference bridge turning main pier set and the attribute features of the reference bridge turning main pier set, the posture stability of the reference bridge turning main pier set is determined, and posture stability features are output, comprising: Obtaining historical rotation information of the reference bridge turning main pier set; the historical rotation information includes indexes and obtained posture stability information of the reference bridge turning main pier set when normally rotating in a plurality of rotation postures; the rotation speed features of different rotation postures are different; According to the historical rotation information, a posture determination model of the reference bridge turning main pier set is constructed; According to the rotation speed features of the reference bridge turning main pier set, a target rotation posture of the reference bridge turning main pier set is determined from a plurality of deployed determination model candidate postures, and posture features are output; According to the attribute features of the reference bridge turning main pier set and the posture features, the posture stability of the reference bridge turning main pier set in the current rotation posture is determined by using the posture determination model, and posture stability features are output.

2. The method of claim 1, wherein, According to the comparison record of the first rigidity attribute and the second rigidity attribute, it is detected whether the real-time obtained posture stability is in a fault persistent stable state in the rigidity dimension, and a first detection record is output, comprising: According to the second rigidity attribute, determine the overall rigidity variance of the attitude depth information of the associated traction cable along the time sequence, and determine a rigidity variance threshold value according to the overall rigidity variance; According to the first rigidity attribute, determine the object rigidity variance of the attitude depth information in the real-time obtained target object set; According to the object rigidity variance and the rigidity variance threshold value, detect whether the real-time obtained attitude stability information in the target object set has a fault in the rigidity dimension, and output a first detection record.

3. The method of claim 1, wherein, According to the comparison record of the first texture attribute and the second texture attribute, detect whether the real-time obtained attitude stability is in a fault persistent state in the texture roughness dimension, and output a second detection record, including: According to the second texture attribute, determine the overall texture roughness standard deviation of the attitude depth information of the associated traction cable, and determine a texture roughness standard deviation threshold value according to the overall texture roughness standard deviation; According to the first texture attribute, determine the object texture roughness standard deviation of the attitude depth information in the real-time obtained target object set; According to the object texture roughness standard deviation and the texture roughness standard deviation threshold value, detect whether the real-time obtained attitude stability information in the target object set has a fault in the texture roughness dimension, and output a second detection record.

4. The method of claim 1, wherein, The rotation speed feature includes a plurality of rotation speed indicators; and the determining, from the plurality of deployed determination model candidate attitudes, the target rotation attitude of the reference bridge deck rotation main pier set according to the rotation speed feature of the reference bridge deck rotation main pier set, and outputting the attitude feature, includes: For each first rotation attitude in the plurality of deployed first rotation attitudes, determining the association degree of each rotation speed indicator in the rotation speed feature and the associated rotation inclination indicator in the first rotation attitude; Weighted sum of the association degree of each rotation speed indicator, output the matching score of the rotation speed feature and the associated first rotation attitude; wherein the weight of each rotation speed indicator is determined according to the influence degree of the associated rotation speed indicator on the stability of the reference bridge deck rotation main pier set; According to the matching score of the rotation speed feature and each first rotation attitude, determine a second rotation attitude from the plurality of first rotation attitudes, and take the second rotation attitude as the target rotation attitude of the reference bridge deck rotation main pier set, and output the attitude feature.

5. The method of claim 4, wherein, The rotation speed feature includes at least one of the following rotation speed indicators: Rotation speed; Bridge deck type; Ambient temperature; Rotation time; Rotation arc length; Ball hinge center azimuth.

6. The method of claim 1, wherein, The attribute feature of the reference bridge deck rotation main pier set includes mechanical indicators and rotation indicators; wherein, The mechanical indicators include tetrafluoro sliding plate features and ring sliding channel features; The rotation indicators include at least one of the following: closure section contact attribute, bridge mass distribution attribute, bridge segmented system attribute, ring sliding channel feature, and bridge maintenance state.

7. The method of any one of claims 1 to 6, wherein, The method further includes: According to the historical stability information of each traction cable, determine the association degree of the attitude stability of each traction cable and the geological feature; According to the structural features and the correlation degree of each traction cable, a deployment scheme of each traction cable posture evaluation point is determined; the deployment scheme includes a deployment number and an evaluation position; wherein the deployment number increases with the increase of the correlation degree.

Images