CN-122016847-A - Bionic self-repairing test method and system for object flow path side equipment

CN122016847ACN 122016847 ACN122016847 ACN 122016847ACN-122016847-A

Abstract

The invention discloses a bionic self-repairing test method for logistics road side equipment, and belongs to the technical field of intelligent operation and maintenance of logistics infrastructure and material performance test. The method aims to solve the technical problems that the existing self-repairing test technology is low in efficiency, relies on manual experience and cannot realize active management of the test process. The technical scheme is characterized by comprising the steps of acquiring multi-source time sequence monitoring data comprising a visual image sequence, an acoustic emission signal and an environmental parameter data stream in real time, extracting characteristics of the data to obtain a time sequence characteristic vector, inputting the characteristic vector into a pre-trained time sequence deep learning model, dynamically outputting a normalized repair completion degree score and predicted residual repair time, and finally automatically judging and triggering a test termination instruction based on the score, and simultaneously generating a test report comprising repair process curves and time comparison information. The method is mainly used for realizing rapid, accurate and automatic test and performance evaluation of the bionic self-repairing process of the logistics flow side equipment.

Inventors

LI KAI
QI ZHAOCHEN
ZHOU YANGYANG
LIANG YUAN

Assignees

中交投资咨询(北京)有限公司

Dates

Publication Date: 20260512
Application Date: 20251219

Claims (10)

1. The bionic self-repairing test method for the logistics roadside equipment is characterized by comprising the following steps of: S1, acquiring multi-source time sequence monitoring data of flow path side equipment of an object to be detected in real time in a self-repairing process, wherein the multi-source time sequence monitoring data at least comprise a visual image sequence representing surface morphology change of the multi-source time sequence monitoring data, acoustic emission signals representing micro deformation of equipment shells and connecting pieces, and environmental parameter data streams containing high-temperature and ultraviolet characteristics of the flow path side; S2, extracting characteristics of the multi-source time sequence monitoring data to obtain time sequence characteristic vectors representing the self-repairing process; S3, inputting the time sequence feature vector into a pre-trained repair state prediction model, and outputting predicted residual repair time required for reaching a preset repair completion standard and including a state evaluation of a current self-repair process, wherein the pre-trained repair state prediction model is a time sequence deep learning model trained based on historical self-repair data, the state evaluation of the current self-repair process is output as a normalized repair completion degree score which is synthesized with multi-source time sequence monitoring data and is between 0 and 1, and the predicted residual repair time is a time span required for reaching a preset threshold value by dynamically extrapolating the predicted repair completion degree score based on a historical sequence and a current state of the time sequence feature vector by the pre-trained repair state prediction model; And S4, automatically judging a test process based on the output state evaluation and the predicted residual repair time, and automatically generating a test termination instruction and a test report when a predetermined repair completion standard is met, wherein the automatic judgment logic is used for monitoring the repair completion degree score in real time, triggering the test termination instruction immediately when the repair completion degree score reaches or exceeds a preset threshold value for the first time, and the test report at least comprises a change curve of the repair completion degree along with time, a final repair completion degree score and comparison information of the predicted residual repair time and the actual measured repair time.
2. The method for bionic self-repairing test of an object flow path side device according to claim 1, wherein in step S2, feature extraction is performed on multi-source time sequence monitoring data, and the method specifically comprises: s21, extracting primary characteristics from a visual image sequence, an acoustic emission signal and an environmental parameter data stream respectively, extracting the area and the perimeter of a damaged area from the visual image sequence through an image segmentation algorithm to serve as the visual primary characteristics, extracting event count, signal amplitude and energy from the acoustic emission signal to serve as the acoustic primary characteristics, and extracting instantaneous values and change trends of temperature and humidity from the environmental parameter data stream to serve as the environmental primary characteristics; S22, performing weighted fusion on the visual primary features, the acoustic primary features and the environment primary features to generate time sequence feature vectors; the weight coefficient of each primary characteristic is obtained through back propagation learning in the training process of the repair state prediction model, and can be dynamically adjusted according to different environmental primary characteristics.
3. The method of claim 1, wherein the multi-source time sequence monitoring data are performed in a programmable multi-factor environmental simulation test pod in step S1, wherein the test pod is configured to dynamically and synchronously apply at least two of the following environmental stresses according to a predetermined environmental stress spectrum during a test period of the self-repairing process: cyclic temperature stress, namely carrying out cyclic change between a high temperature limit value and a low temperature limit value according to a preset time program; Mechanical vibration stress, namely simulating the vibration frequency and amplitude of the environment near the road side of the logistics transportation vehicle or equipment; simulating ultraviolet components in outdoor sunlight, and irradiating with preset irradiance; the environmental parameter data stream is derived from real-time monitoring data of the above-mentioned environmental stresses imposed in the test chamber.
4. The method for bionic self-repairing test of an object flow path side device according to claim 1, further comprising, after step S4: S5, after a test report is generated, the multisource time sequence monitoring data which is acquired in the test and has complete self-repairing process and the final measured repairing time are used as a group of new labeling samples; and performing incremental learning on the pre-trained repair state prediction model by using a new labeling sample so as to update the parameters of the pre-trained repair state prediction model for the prediction of a subsequent test task.
5. The method for bionic self-repairing test of physical distribution road side equipment according to claim 4, wherein in step S5, before incremental learning, a data reliability verification step is added, specifically: Checking the multisource time sequence monitoring data forming a new labeling sample based on a preset data quality rule set; The data quality rule set at least comprises whether the definition of the visual image sequence is higher than a set definition threshold, whether the signal to noise ratio of the acoustic emission signal is higher than a set signal to noise ratio threshold and whether the numerical value of the environmental parameter data stream is in a preset effective range; The method comprises the step that a party uses the new labeling sample to learn the increment of the pre-trained repair state prediction model only when the new labeling sample passes the data credibility check.
6. The bionic self-repairing test method of the object flow path side equipment according to claim 1, wherein the method further comprises a standardized initial damage preparation step before the step S1, specifically: Preparing initial damage with the same geometric shape and physical size on the surface of the self-repairing material of the device at the flow path side of the object to be tested according to a pre-stored standardized damage pattern and depth parameters by using programmable laser etching equipment integrated in a test system; the initial damage is used as the initial target monitored by the visual image sequence and the acoustic emission signal in the step S1.
7. The bionic self-repairing test method of the object flow path side equipment according to claim 1, wherein in step S1, the acquisition of the visual image sequence is jointly ensured by a main camera and a standby camera; The step S1 is to execute real-time acquisition of multi-source time sequence monitoring data of the device at the flow path side of the object to be detected in the self-repairing process, and simultaneously comprises a fault diagnosis step of monitoring the running state of a main camera and the quality of the visual image sequence in real time; When the failure of the main camera or the continuous unsatisfied requirement of the image quality is judged, the data acquisition source is automatically switched to the standby camera so as to ensure the continuity of the multi-source time sequence monitoring data.
8. The method for bionic self-repairing test of a physical distribution flow-side device according to claim 4, wherein in step S5, performing incremental learning on the pre-trained repair state prediction model using a new labeling sample specifically comprises: s51, incremental updating is carried out on the copy of the pre-trained repair state prediction model by using a new labeling sample, only parameters of a full-connection layer of the pre-trained repair state prediction model are updated, and a feature extraction layer is not retrained, so that a candidate model is obtained; S52, utilizing the historical data corresponding to the new labeling sample to evaluate the prediction accuracy of the candidate model and the current formally pre-trained repair state prediction model in parallel; And S53, replacing the current formally pre-trained repair state prediction model by the candidate model only when the prediction accuracy of the candidate model is higher than that of the current formally pre-trained repair state prediction model, and completing incremental learning.
9. The bionic self-repairing test method of the object flow path side equipment according to claim 1, wherein in the cyclic execution process of the steps S3 and S4, the method further comprises a dynamic regulation and control step linked with the environmental simulation and prediction step, specifically: receiving a repair completion degree score output by the pre-trained repair state prediction model in real time; Calculating to obtain real-time repair rate based on historical data of repair completion degree scores; And when the real-time repair rate is lower than a preset activity threshold value, sending a control instruction to a programmable multi-factor environment simulation test cabin so as to reduce the intensity or frequency of the applied environmental stress.
10. The bionic self-repairing test method of the object flow path side equipment according to claim 6, further comprising a damage effectiveness verification step between the standardized initial damage preparation step and S1, specifically: embedding conductive fiber net in self-repairing material of the flow path side equipment of the object to be detected, wherein the fiber line diameter is 50-100 mu m, and the grid spacing is 1-2 mm; amplifying and shooting an initial damage area through a 200-time optical microscope with the resolution of 0.5 mu m integrated in a test system to obtain a microscopic morphology image; And (2) analyzing the microcosmic morphology image based on an image recognition algorithm, combining with the conductivity detection of the damaged area to confirm that the initial damage penetrates through the pre-buried conductive fiber net, and starting the self-repairing process monitoring of the step S1 only when the initial damage is confirmed to be effective penetrating damage.

Description

Bionic self-repairing test method and system for object flow path side equipment Technical Field The invention relates to the technical field of intelligent operation and maintenance of logistics infrastructure and material performance test. More particularly, the invention relates to a bionic self-repairing test method and system for logistics road side equipment. Background The equipment at the side of the object flow path, such as a monitoring upright rod, a traffic signal lampshade, a sensor housing and the like, is exposed to outdoor complex environments for a long time, and is inevitably damaged by external impact, material aging and environmental corrosion to generate microcracks, scratches and the like. In order to prolong the service life of equipment and reduce maintenance cost, bionic self-repairing materials are gradually applied to the equipment. However, how to test and evaluate the self-repairing performance of these materials effectively and reliably has become a key element for restricting the engineering application. At present, a method for testing self-repairing materials for logistics flow side equipment has a plurality of limitations. First, the efficiency of testing is generally low. Common test formats rely on manual timing observations or laboratory offline analysis after destructive sampling. For example, the damage area is subjected to shooting comparison for a plurality of times at different time points by a microscope, or the recovery rate of the mechanical property is measured after a specific time interval. The whole process is long in time consumption and varies from a few hours to a few days, and the requirement of rapidly screening and evaluating the repair performance of the material cannot be met. The reason for this problem is that the conventional method lacks continuous and automatic monitoring means for the repair process, and determines whether the repair is completed or not mainly depends on experience of operators, so that subjectivity is strong, and it is difficult to determine an optimal test termination point. Secondly, the accuracy and objectivity of the test result are to be improved. Existing methods tend to focus on a single type of signal acquisition, such as relying solely on visual images to analyze crack closure. However, visual assessment is susceptible to factors such as lighting conditions, viewing angles, etc., and is insensitive to repair activities on the internal or microscopic scale of the material, possibly leading to erroneous decisions on repair status. Meanwhile, factors such as environment temperature and humidity have significant influence on repair speed and effect, but conventional testing is usually carried out in a constant laboratory environment, dynamic changes of real road side environment parameters are not considered, so that testing conditions are disjointed from actual working conditions, and the guiding value of testing results is limited. Attempts to integrate multiple types of sensor data have been faced with difficulties, mainly in that data from different sources (e.g., images, acoustic signals, environmental parameters) lacks efficient automated processing schemes for time alignment and feature fusion, and the contribution weights of each data source to the repair status are difficult to objectively determine. Furthermore, the standardization and reproducibility of the testing process is inadequate. The initial state of the test, namely the initial damage prepared on the surface of the material by people, has larger human operation errors in shape, size and depth. The inconsistency of the initial conditions directly leads to large fluctuation of test results among different batches, even different samples in the same batch, and effective transverse comparison cannot be performed, so that difficulties are brought to objective evaluation and classification of material properties. In addition, new challenges are faced in pursuing the intellectualization of test systems. For example, if it is desired that the test system can be self-optimized by the accumulated data, the requirement of incremental learning is faced, but how to ensure the quality of the newly added data and how to avoid performance fluctuation or rollback of the model in the learning process is a problem that needs to be carefully handled in practical applications. Meanwhile, in order to ensure the stability of long-term automatic testing, the fault tolerance of the system, such as how to ensure that the testing is not interrupted when a key sensor fails, is also a link to be considered in design. In summary, in the prior art, when the bionic self-repairing material of the equipment at the logistics road side is tested, improvements are needed in the aspects of testing efficiency, result accuracy, environment simulation authenticity, testing standardization, stability and self-optimizing capability of long-term operation of the system and the like. Disclosure of