Multi-target identification and defect detection method in intelligent inspection of hydraulic structure

Abstract

The invention relates to the technical field of hydraulic structure detection, in particular to a multi-target identification and defect detection method in hydraulic structure intelligent inspection, which comprises the following steps of: acquiring a current hydraulic structure detection range, dividing the current hydraulic structure detection range into a plurality of detection areas, endowing the detection areas with detection priorities according to the type characteristics of the detection areas, static and dynamic data of the detection area are respectively collected; extracting target and defect features in the static and dynamic data of the detection area, and dynamically adjusting the detection priority of the current detection area; acquiring detection priority data of the current detection area, executing defect grading detection operation, and recording a detection result; through the mode, the detection range of the hydraulic structure is divided into regions, and the detection priority of each region is dynamically adjusted, so that the detection efficiency and effect are improved.

Description

Technical Field

[0001] This invention relates to the field of hydraulic structure inspection technology, and in particular to a method for multi-target identification and defect detection in intelligent inspection of hydraulic structures. Background Technology

[0002] Hydraulic structures, as the core carriers of water conservancy projects, encompass key facilities such as dams, sluices, and dikes. Their structural integrity directly determines the safe and stable operation of public functions such as flood control, power generation, and water supply. As water conservancy projects develop towards larger scale and greater complexity, the service environment of hydraulic structures is becoming increasingly harsh (such as long-term water flow erosion, alternating wet and dry conditions, and corrosive media erosion). Defects such as cracks, leaks, and concrete spalling occur frequently. If these defects are not detected and addressed in a timely and accurate manner, they may lead to major safety accidents such as structural instability and dam failure.

[0003] Traditional inspection methods have many limitations. First, in determining the inspection scope, there is often a lack of systematic methods, relying solely on experience to roughly determine the inspection area, which makes it difficult to comprehensively and accurately cover all key parts of hydraulic structures, potentially leading to the omission of some areas with problems. For example, for some complex hydraulic structures, determining the inspection scope based solely on experience may overlook some hidden but important structural parts.

[0004] Secondly, after the testing areas are divided, the setting of testing priorities for each area lacks a scientific basis. A uniform or relatively fixed priority order is often adopted without fully considering the actual characteristics and risk levels of different areas. This results in some high-risk areas that urgently need testing not receiving timely attention, while some low-risk areas may be over-tested, leading to a waste of resources. For example, for areas with different functions and structural types within hydraulic structures, reasonable priorities are not assigned based on their characteristics, affecting testing efficiency and effectiveness.

[0005] Therefore, it is essential to propose a multi-target identification and defect detection method for intelligent inspection of hydraulic structures that divides the inspection scope of hydraulic structures into regions and dynamically adjusts the inspection priority of each region to improve inspection efficiency and effectiveness. Summary of the Invention

[0006] The purpose of this invention is to provide a multi-target identification and defect detection method for intelligent inspection of hydraulic structures, which aims to divide the inspection range of hydraulic structures into regions and dynamically adjust the inspection priority of each region, thereby improving inspection efficiency and effectiveness.

[0007] To achieve the above objectives, the present invention employs a multi-target identification and defect detection method for intelligent inspection of hydraulic structures, comprising the following steps:

[0008] The current inspection scope of the hydraulic structure is obtained, divided into multiple inspection areas, and inspection priorities are assigned to the inspection areas according to their type characteristics. Static and dynamic data of the inspection areas are collected separately.

[0009] Extract target and defect features from static and dynamic data of the detection area, and dynamically adjust the detection priority of the current detection area;

[0010] Obtain the detection priority data for the current detection area, perform defect classification detection operations, and record the detection results.

[0011] Among the steps are: obtaining the current inspection range of the hydraulic structure, dividing it into multiple inspection zones, assigning inspection priorities to the inspection zones based on their type characteristics, and collecting static and dynamic data for each inspection zone;

[0012] Obtain the engineering drawings of the current hydraulic structures and, in conjunction with the site survey, define the scope of this inspection;

[0013] Acquire boundary data for each detection area, mark the area boundaries in the BIM model, and identify overlapping areas.

[0014] For each segmented detection region, its core type features are extracted, and each region is assigned an initial priority based on the extracted type features;

[0015] Collect and record static and dynamic data of the detection area.

[0016] Before the steps of acquiring boundary data for each detection area, marking area boundaries in the BIM model, and identifying overlapping adjacent areas:

[0017] The detection range is divided into multiple detection zones, and a unique identifier is assigned to each zone.

[0018] Prior to the steps of collecting and recording static and dynamic data from the detection area:

[0019] Bind the initial priority and the region unique identifier to generate the initial priority allocation table.

[0020] Among the steps involved in collecting and recording static and dynamic data from the detection area:

[0021] By employing BIM model extraction and on-site measurement methods, focusing on the inherent attribute data of the area, static data of the inspection area is collected;

[0022] Dynamic data of the detection area is collected by using database retrieval and real-time monitoring.

[0023] Among them, in the step of extracting target and defect features from static and dynamic data of the detection area and dynamically adjusting the detection priority of the current detection area:

[0024] Static and dynamic data are preprocessed to remove noise and outliers, and a standardized dataset is generated after preprocessing.

[0025] Extracting regional risk target features from static data;

[0026] Extracting regional defect target features from dynamic data;

[0027] Set adjustment indicators and weights for the features to form a priority adjustment scoring system, and adjust the detection priority of the current detection area.

[0028] Before setting the adjustment indicators and weights corresponding to the features, forming a priority adjustment scoring system, and adjusting the detection priority of the current detection region:

[0029] Create a feature-region association table for each region and record the feature type, feature strength, and feature confidence level.

[0030] Among them, after setting the adjustment indicators and weights corresponding to the features, forming a priority adjustment scoring system, and adjusting the detection priority of the current detection region:

[0031] Based on the adjustment records, a dynamic priority adjustment table is generated, and the priority corresponding to the area unique identifier is updated.

[0032] Among the steps, the following steps are involved: obtaining the detection priority data for the current detection area, performing defect classification detection operations, and recording the detection results:

[0033] Obtain priority data for all current detection areas and generate a detection task sequence;

[0034] Perform defect classification detection operations according to the detection task sequence.

[0035] After performing the defect classification detection operation according to the detection task sequence:

[0036] For each testing area, the differentiated test results are recorded according to priority, and then linked and archived to generate an overview report of the test results.

[0037] This invention discloses a multi-target identification and defect detection method for intelligent inspection of hydraulic structures. The method involves acquiring the current inspection range of the hydraulic structure, dividing it into multiple inspection areas, assigning inspection priorities to each area based on its type characteristics, and collecting static and dynamic data for each area. It then extracts target and defect features from the static and dynamic data of each area, dynamically adjusts the inspection priority of the current area, acquires the inspection priority data for the current area, performs a defect classification inspection operation, and records the inspection results. Through this method, the inspection range of the hydraulic structure is divided into areas, and the inspection priority of each area is dynamically adjusted, thereby improving inspection efficiency and effectiveness. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0039] Figure 1 This is a flowchart of the steps in the intelligent inspection method for hydraulic structures of the present invention, which involves multi-target identification and defect detection.

[0040] Figure 2 This is a flowchart of steps S100 of the present invention.

[0041] Figure 3 This is a flowchart of steps S200 of the present invention.

[0042] Figure 4 This is a flowchart of steps S300 of the present invention.

[0043] Figure 5 This is a structural principle diagram of the multi-target identification and defect detection system in the intelligent inspection of hydraulic structures of the present invention.

[0044] Figure 6 This is a schematic diagram of the electronic device of the present invention.

[0045] 401 - Regional Feature Acquisition Module, 402 - Regional Detection Priority Dynamic Adjustment Module, 403 - Defect Detection Module. Detailed Implementation

[0046] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application.

[0047] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0048] It should be understood that although the terms first, second, third, etc., may be used in this application to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to determination."

[0049] Please see Figures 1-4 This invention provides a method for multi-target identification and defect detection in intelligent inspection of hydraulic structures, comprising the following steps:

[0050] S100: Obtain the current inspection range of the hydraulic structure, divide it into multiple inspection areas, assign inspection priorities to the inspection areas according to the characteristics of the inspection area type, and collect static and dynamic data of the inspection areas respectively.

[0051] In this embodiment, the current inspection range of the hydraulic structure is obtained and divided into multiple inspection zones. Inspection priorities are assigned to each inspection zone based on its type characteristics, and static and dynamic data are collected for each zone. The specific process is as follows:

[0052] S101: Obtain the engineering drawings of the current hydraulic structure and, in conjunction with the site survey, define the scope of this inspection;

[0053] S102: Divide the detection range into multiple detection areas and assign a unique identifier to each area;

[0054] S103: Acquire the boundary data of each detection area, mark the area boundaries in the BIM model, and identify the overlap of adjacent areas;

[0055] S104: For each segmented detection region, extract its core type features and assign an initial priority to each region based on the extracted type features;

[0056] S105: Bind the initial priority and the unique identifier of the region, and generate the initial priority allocation table;

[0057] S106: Employ BIM model extraction and on-site measurement methods to focus on the inherent attribute data of the area and collect static data of the detection area;

[0058] S107: Dynamic data of the detection area is collected by using database retrieval and real-time monitoring.

[0059] In the above process, the physical scope of the inspection is first determined by combining the engineering drawings of hydraulic structures (such as dam structure drawings and sluice gate layout drawings) with the results of on-site surveys. Taking a concrete dam as an example, the inspection scope needs to cover the upstream and downstream sides of the dam, the dam crest, the dam gallery, and underwater structures from 5 to 20 meters deep. Taking a sluice gate as an example, it needs to cover key structures such as the gate panel, the gate chamber floor, the stilling basin, and the foundations of the opening and closing equipment. At the same time, by binding GPS positioning with the BIM model, the inspection scope is mapped to the digital model, generating a boundary file containing latitude, longitude, elevation, and structure numbers.

[0060] Adopting the principle of "structural function + spatial continuity", the detection range is divided into several independent minimum detection areas, with the area of ​​each area controlled between 50 and 200 m². 2 (To ensure a balance between detection accuracy and efficiency). The classification rules are as follows:

[0061] The dam is divided according to "dam section - elevation - slope", such as "dam section #3 - elevation 95-100m - water-facing side" and "dam section #5 - gallery layer - inner wall".

[0062] The sluice gate is divided according to "structural unit - part", such as "Gate No. 1 - left side of the panel" and "gate chamber - upstream bottom plate area 1";

[0063] After division, each region is assigned a unique ID (e.g., “D3-100-Y”, where D represents the dam, 3 represents the dam section, 100 represents the elevation, and Y represents the water-facing side), and the region boundaries are marked in the BIM model to ensure that the regions do not overlap or omit any.

[0064] For each segmented detection region, its core type features are extracted, including:

[0065] Functional type: Determine whether the area is a core functional area (such as water retention, flood discharge, or water diversion area) or an auxiliary functional area (such as a dam crest walkway or gate chamber maintenance platform);

[0066] Structural type: Identify the structural material (concrete, steel structure, earth and stone) and structural form (flat walls, corners, holes, hinged joints) of the area;

[0067] Environment type: Determine the environment in which the area is located (water scouring zone, underwater zone, dry zone, corrosive environment).

[0068] Based on the extracted type features and referring to the preset priority grading standard, each region is assigned an initial priority (high / medium / low):

[0069] High priority: core functional areas + stress concentration structures (such as the water-retaining area on the upstream side of the dam, gate hinge nodes), underwater key structures (such as the dam body below 10m underwater), and areas with strong scouring / corrosive environments (such as the stilling basin bottom plate).

[0070] Medium priority: core functional area + ordinary structure (such as the straight area on the backwater surface of the dam, the side wall of the gate chamber), auxiliary functional area + high-risk environment (such as the walkway on the dam crest near the retaining wall);

[0071] Low priority: auxiliary functional areas + ordinary structures + low-risk environments (such as maintenance access on the dam crest away from the retaining wall, and slope protection in the non-scour zone downstream of the sluice gate);

[0072] After being assigned, the initial priority is bound to the unique identifier of the region, generating an initial priority allocation table, which serves as the basis for subsequent data collection and dynamic adjustment.

[0073] Static data collection focuses on the inherent attribute data of the area, employing a combination of "BIM extraction + on-site measurement":

[0074] Obtain from the BIM model: regional structural dimensions (length, width, thickness), design load parameters (design values ​​of water pressure and seepage pressure), and distances from the core structure (such as horizontal distances from the dam's cutoff wall);

[0075] Additional on-site measurements: regional surface flatness, concrete strength (rebound test), and initial state of steel structure corrosion (visual inspection + coating thickness measurement);

[0076] After collection, the data is entered into the regional basic database and associated with the regional unique identifier to form a static data archive (e.g., “D3-100-Y region: length 20m, width 5m, concrete design strength C30, distance from the seepage barrier wall 3m”).

[0077] Dynamic data acquisition targets dynamically changing regional data, obtained through a combination of database retrieval and real-time monitoring.

[0078] Historical dynamic data: Extract the defect records (defect type, occurrence time, severity, and repair status) and sensor historical data (annual monitoring maximum values ​​of piezometers and strain gauges) for the area from the water conservancy project operation and maintenance database for the past 3 to 5 years.

[0079] Real-time dynamic data: Current data (seepage pressure value and displacement change in the past 24 hours) are collected by sensors deployed on site (such as real-time seepage pressure monitors and displacement monitoring stations), and the current surface condition image of the area is obtained by drone aerial photography (resolution not less than 5472×3648).

[0080] Meanwhile, during the data collection process, the data collection time and the number of the collection device (such as sensor ID, drone number) must be recorded to ensure data traceability; after the data collection is completed, the regional basic database is updated to form a dynamic data time series archive.

[0081] S200: Extract target and defect features from static and dynamic data of the detection area and dynamically adjust the detection priority of the current detection area.

[0082] In this embodiment, target and defect features are extracted from the static and dynamic data of the detection area, and the detection priority of the current detection area is dynamically adjusted. The specific process is as follows:

[0083] S201: Preprocess static and dynamic data to eliminate noise and outliers, and generate a standardized dataset after preprocessing;

[0084] S202: Extracting regional risk target features from static data;

[0085] S203: Extracting regional defect target features from dynamic data;

[0086] S204: Establish a feature-region association table for each region and record the feature type, feature strength, and feature confidence.

[0087] S205: Set the adjustment indicators and weights corresponding to the features to form a priority adjustment scoring system, and adjust the detection priority of the current detection area;

[0088] S206: Based on the adjustment records, generate a dynamic priority adjustment table and update the priority corresponding to the area unique identifier.

[0089] In the above process, static and dynamic data are preprocessed to eliminate noise and outliers, preparing for feature extraction:

[0090] Static data: Outlier values ​​of concrete strength test results are removed (using the 3σ principle), and structural dimension data are calibrated (compared with BIM design values; if the error exceeds 5%, re-measurement is performed).

[0091] Dynamic data: Real-time sensor data is smoothed (using moving average) to remove instantaneous interference signals; UAV aerial images are preprocessed (Gaussian filtering for noise reduction, CLAHE contrast enhancement, and underwater image dehazing) to ensure image clarity;

[0092] After preprocessing, a standardized dataset is generated, with a uniform format of "regional unique identifier-data type-data value-time stamp" to facilitate subsequent feature extraction.

[0093] Feature extraction from static data: Extracting target features related to regional risk from preprocessed static data.

[0094] Structural risk characteristics: such as concrete strength measured value being more than 10% lower than design value (marked as "insufficient strength characteristic"), steel structure coating thickness being less than standard value (marked as "coating failure characteristic"), structural dimensions deviating from design value by more than 8% (marked as "structural deformation characteristic");

[0095] Spatial correlation characteristics: such as the distance between the area and the anti-seepage wall is less than 2m (marked as "near anti-seepage structure characteristics"), or the area is located within the influence range of geological faults (marked as "fault influence characteristics").

[0096] Defect feature extraction from dynamic data:

[0097] Historical defect characteristics: Extract "defect recurrence characteristics" (the same defect has occurred ≥2 times in the past 2 years) and "serious defect characteristics" (a serious defect has occurred, such as a crack with a width ≥0.5mm) from historical defect records;

[0098] Real-time defect features: For the pre-processed aerial images, a lightweight target detection model (such as YOLOv5-nano) is used for preliminary identification to extract "suspected crack features" (linear areas with abrupt changes in grayscale values ​​in the image), "suspected leakage features" (wet or water-stained areas in the image), and "suspected spalling features" (blocky areas where concrete surfaces have detached in the image).

[0099] Real-time monitoring features: Extract “abnormal fluctuation features” (pressure value changes by more than 0.5 MPa within 24 hours) and “approaching threshold features” (displacement value reaches 80% of the design warning threshold) from sensor data.

[0100] After extraction, a "feature-region" association table is created for each region, recording the feature type, feature intensity (e.g., "crack feature - length approximately 1.2m"), and feature confidence level (e.g., "leakage feature confidence level 0.85").

[0101] Set adjustment indicators and weights for each feature to form a priority adjustment scoring system, such as:

[0102] Adjustment index: Structural risk level; Corresponding characteristics: Insufficient strength, coating failure, structural deformation; Index weight: 0.3; Scoring rules (0-10 points): 6 points for 1 characteristic, 8 points for 2 characteristics, and 10 points for 3 characteristics.

[0103] Adjustment indicator: Historical defect risk level; Corresponding characteristics: defect recurrence, critical defect; Indicator weight: 0.25; Scoring rules (0-10 points): 5 points for 1 item, 9 points for 2 items;

[0104] Adjusted index: Real-time defect suspicion level; Corresponding features: suspected cracks, leaks, peeling; Index weight: 0.25; Scoring rules (0-10 points): 1 suspected feature gets 4 points, 2 features get 7 points, and 3 features get 10 points.

[0105] Adjustment indicator: Real-time monitoring of anomalies; Corresponding characteristics: Abnormal fluctuations, approaching the threshold; Indicator weight: 0.2; Scoring rules (0-10 points): 5 points for 1 anomaly, 10 points for 2 anomalies.

[0106] For each region, determine its score under each adjustment indicator based on the "feature-region" association table, and calculate the total adjustment score according to "total adjustment score = ∑(indicator score × indicator weight)".

[0107] Set adjustment rules:

[0108] If the total adjustment score is ≥8 points: if the initial priority is high, it remains; if it is medium, it is upgraded to high; if it is low, it is upgraded to medium.

[0109] If 4 points ≤ total adjusted score < 8 points: the initial priority remains unchanged;

[0110] If the total adjustment score is less than 4 points: if the initial priority is high, it will be reduced to medium; if medium, it will be reduced to low; if low, it will remain unchanged.

[0111] For example, if a region has an initial priority of medium, and "insufficient strength feature," "suspected leakage feature," and "near threshold feature" are extracted, then:

[0112] Structural risk score = 6 points (1 structural risk feature), historical defect risk score = 0 points (no historical defect feature), real-time defect suspicion score = 4 points (1 suspected feature), real-time monitoring anomaly score = 5 points (1 anomaly).

[0113] Total adjusted score = 6 × 0.3 + 0 × 0.25 + 4 × 0.25 + 5 × 0.2 = 1.8 + 0 + 1 + 1 = 3.8 points < 4 points, so the initial priority is reduced from medium to low;

[0114] After adjustment, a dynamic priority adjustment table is generated, updating the priority corresponding to the unique identifier of the area, and synchronizing it to the BIM model and inspection execution, serving as the basis for subsequent defect detection.

[0115] S300: Obtain the detection priority data of the current detection area, perform defect classification detection operation, and record the detection results.

[0116] In this embodiment, the detection priority data of the current detection area is obtained, a defect classification detection operation is performed, and the detection results are recorded. The specific process is as follows:

[0117] S301: Obtain priority data for all current detection areas and generate a detection task sequence;

[0118] S302: Perform defect classification detection operation according to the detection task sequence;

[0119] S303: For each testing area, record differentiated testing results according to priority, and archive them together to generate a summary report of testing results.

[0120] In the above process, priority data synchronization and parsing: The detection execution system reads the priority data of all current detection areas from the dynamic priority adjustment table, sorts them in the order of "high → medium → low", and generates a detection task sequence; at the same time, it parses the detection requirements corresponding to the priority of each area:

[0121] High-priority areas: Inspection accuracy requirement ≥95%, high-precision inspection equipment (such as drones equipped with 50-megapixel cameras, portable crack width gauges) must be used, and inspection data must be transmitted in real time;

[0122] Medium priority area: Detection accuracy requirement ≥90%, use conventional detection equipment (such as drones with 20-megapixel cameras, ordinary appearance inspection tools), and transmit detection data in batches;

[0123] Low priority areas: Detection accuracy requirement ≥85%, lightweight detection equipment (such as handheld devices with 12-megapixel cameras) can be used, and detection data is temporarily stored locally and then summarized.

[0124] Based on the priority of the testing requirements, allocate testing equipment and time resources:

[0125] Equipment allocation: High-precision drones, underwater robots, and other core equipment will be prioritized for high-priority area detection tasks;

[0126] Time allocation: High-priority areas will be assigned testing tasks on a "real-time response" basis, starting within 1 hour of receiving the task; medium-priority areas will be assigned testing tasks on a "batch execution" basis, with one batch each in the morning and afternoon; low-priority areas will be assigned testing tasks on a "centralized execution" basis, with centralized testing conducted once every 2 days.

[0127] High-priority area defect detection operations:

[0128] Image acquisition: Low-altitude aerial photography using drones (flight altitude 5-8m, overlap 80%) or close-range photography using underwater robots (1-2m from the structure surface) is used to acquire ultra-high resolution images (single pixel ≥5472×3648) to ensure that minute defects (cracks <0.2mm in width) are clearly visible;

[0129] Defect identification: Enable the high-priority branch of the priority-adaptive multi-target identification model (CSPDarknet-75 network + CBAM + ECA attention mechanism), input the preprocessed image, identify the defect type (crack, leakage, spalling, exposed rebar), and output the defect bounding box and confidence score (threshold ≥ 0.8).

[0130] Defect quantification: Using a defect quantification module, combined with image acquisition parameters (focal length, shooting distance), the actual size of the defect (crack length, width, and spalling area) is calculated, with the error controlled within ≤5%; at the same time, on-site measurement tools (crack width gauge, ultrasonic thickness gauge) are used to verify key defects to ensure quantification accuracy;

[0131] Severity assessment: The severity of the defect is determined based on the quantitative results (slight / moderate / severe), such as "crack length 3.2m, width 0.6mm → severe".

[0132] Medium priority area defect detection operation:

[0133] Image acquisition: High-resolution images (single pixel ≥2048×1536) are obtained by using drones for mid-to-high-altitude aerial photography (flight altitude 10-15m, overlap 60%) or handheld cameras (distance from the structure surface 3-5m).

[0134] Defect identification: Enable priority branches in the model (CSPDarknet-53 network + CBAM attention mechanism) to identify defect types and output bounding boxes and confidence scores (threshold ≥ 0.7);

[0135] Defect quantification: Only key parameters (crack length, spalling area) are quantified, with the error controlled within ≤8%, and no on-site measurement verification is required (unless the defect confidence level is <0.8);

[0136] Severity assessment: The severity is determined based on the quantitative results, such as "crack length 1.5m, width 0.3mm → moderate".

[0137] Low-priority area defect detection operations:

[0138] Image acquisition: Use a handheld camera or a regular drone to take pictures (flying altitude 20-25m, overlap 40%) to obtain images with standard resolution (single pixel ≥1280×960);

[0139] Defect identification: Enable the low-priority branch of the model (CSPDarknet-35 network + CBAM channel attention only) to identify obvious defects (cracks with a width ≥ 0.5 mm, peeling with an area ≥ 0.5 m²), and output bounding boxes and confidence scores (threshold ≥ 0.6);

[0140] Defect assessment: No need for precise dimensional quantification, the severity of the defect is determined solely based on image features (e.g., "crack visible to the naked eye, length approximately 1m → minor").

[0141] Rapid screening: Areas showing no signs of defects are marked as "defect-free" and require no further testing.

[0142] For each testing area, differentiated testing results are recorded according to priority. The testing results are associated with the area's unique identifier, priority data, and historical testing records to form a complete "area-test-result" file, which facilitates subsequent priority updates, defect tracking, and operation and maintenance decisions.

[0143] Once the data is archived, an overview report of the test results will be generated, including the completion rate of tests in each priority area, defect distribution statistics, and a list of high-risk defects, providing a basis for subsequent inspection plan adjustments and emergency response.

[0144] Corresponding to the aforementioned embodiments of the multi-target identification and defect detection method in intelligent inspection of hydraulic structures, this application also provides embodiments of the multi-target identification and defect detection system in intelligent inspection of hydraulic structures.

[0145] Figure 5 This is a block diagram illustrating a multi-target identification and defect detection system for intelligent inspection of hydraulic structures, according to an exemplary embodiment. (Refer to...) Figure 5 The system may include: a region feature acquisition module 401, a region detection priority dynamic adjustment module 402, and a defect detection module 403, wherein:

[0146] The regional feature acquisition module 401 is used to obtain the current detection range of the hydraulic structure, divide it into multiple detection areas, assign detection priorities to the detection areas according to the characteristics of the detection area type, and collect static and dynamic data of the detection areas respectively.

[0147] The regional detection priority dynamic adjustment module 402 is used to extract target and defect features from the static and dynamic data of the detection area and dynamically adjust the detection priority of the current detection area.

[0148] The defect detection module 403 is used to acquire the detection priority data of the current detection area, perform defect classification detection operation, and record the detection results.

[0149] In this embodiment, the area feature acquisition module 401 acquires the current detection range of the hydraulic structure, divides it into multiple detection areas, assigns detection priorities to the detection areas according to their type characteristics, and collects static and dynamic data for each detection area. The area detection priority dynamic adjustment module 402 extracts target and defect features from the static and dynamic data of the detection areas and dynamically adjusts the detection priority of the current detection area. The defect detection module 403 acquires the detection priority data of the current detection area, performs defect classification detection operations, and records the detection results. Through the above method, the detection range of the hydraulic structure is divided into areas, and the detection priority of each area is dynamically adjusted, thereby improving detection efficiency and effectiveness.

[0150] Regarding the system in the above embodiments, the specific ways in which each module performs operations have been described in detail in the embodiments related to the method, and will not be elaborated here.

[0151] For the system embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this application according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0152] Accordingly, this application also provides an electronic device, including: one or more processors; a memory for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the multi-target recognition and defect detection method in the intelligent inspection of hydraulic structures as described above. Figure 6 The diagram shown is a hardware structure diagram of any device with data processing capabilities in an intelligent inspection system for hydraulic structures, provided by an embodiment of the present invention. (Except for...) Figure 6 In addition to the processor, memory, and network interface shown, any data processing device in the embodiment may also include other hardware depending on the actual function of the data processing device, which will not be described in detail here.

[0153] Accordingly, this application also provides a computer-readable storage medium storing computer instructions, which, when executed by a processor, implement the multi-target identification and defect detection method in the intelligent inspection of hydraulic structures described above. The computer-readable storage medium can be an internal storage unit of any data-processing device as described in any of the foregoing embodiments, such as a hard disk or memory. The computer-readable storage medium can also be an external storage device, such as a plug-in hard disk, smart media card (SMC), SD card, flash card, etc., equipped on the device. Furthermore, the computer-readable storage medium can include both internal storage units of any data-processing device and external storage devices. The computer-readable storage medium is used to store the computer program and other programs and data required by the data-processing device, and can also be used to temporarily store data that has been output or will be output.

[0154] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the disclosure herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein.

[0155] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope.

Claims

1. A method for multi-target recognition and defect detection in intelligent inspection of hydraulic structures, characterized in that, The method comprises the following steps: Obtain the current detection range of the hydraulic structure, divide it into multiple detection areas, assign detection priorities to the detection areas according to the type characteristics of the detection areas, and collect static and dynamic data of the detection areas respectively; Extract target and defect characteristics from the static and dynamic data of the detection area, and dynamically adjust the detection priority of the current detection area; Obtain the detection priority data of the current detection area, perform defect classification detection operation, and record the detection results.

2. The method of claim 1, wherein the method comprises: In the step of obtaining the current detection range of the hydraulic structure, dividing it into multiple detection areas, assigning detection priorities to the detection areas according to the type characteristics of the detection areas, and collecting static and dynamic data of the detection areas: Obtain the engineering drawings of the current hydraulic structure, and define the detection range of this detection according to the site survey; Obtain the boundary data of each detection area, mark the area boundary in the BIM model, and identify the overlapping situation of adjacent areas; For each divided detection area, extract its core type characteristics, and assign an initial priority to each area according to the extracted type characteristics; Collect static and dynamic data of the detection area and record it.

3. The method of claim 2, wherein the method further comprises: Before the step of obtaining the boundary data of each detection area, marking the area boundary in the BIM model, and identifying the overlapping situation of adjacent areas: Divide the detection range into multiple detection areas, and assign a unique identifier to each area.

4. The method of claim 3, wherein the method further comprises: Before the step of collecting static and dynamic data of the detection area and recording it: Bind the initial priority and the unique identifier of the area to generate an initial priority allocation table.

5. The method of claim 2, wherein the method further comprises: In the step of collecting static and dynamic data of the detection area and recording it: Use BIM model extraction and on-site measurement to focus on the inherent attribute data of the area, and collect static data of the detection area; Use database retrieval and real-time monitoring to collect dynamic data of the detection area.

6. The method of claim 4, wherein the method further comprises: In the step of extracting target and defect characteristics from the static and dynamic data of the detection area, and dynamically adjusting the detection priority of the current detection area: Preprocess the static and dynamic data to eliminate noise and outliers, and generate a standardized data set after preprocessing; Extract the target characteristics of the area from the static data; Extract the target characteristics of the area from the dynamic data; Set the adjustment index and weight corresponding to the characteristics to form a priority adjustment scoring system, and adjust the detection priority of the current detection area.

7. The method of claim 6, wherein the method further comprises: Before the step of setting the adjustment index and weight corresponding to the characteristics to form a priority adjustment scoring system, and adjusting the detection priority of the current detection area: Establish a feature-area association table for each area, and record the feature type, feature intensity, and feature confidence.

8. The method of claim 6, wherein the method further comprises: After the step of setting the adjustment index and weight corresponding to the characteristics to form a priority adjustment scoring system, and adjusting the detection priority of the current detection area: According to the adjustment record, generate a dynamic priority adjustment table, and update the priority corresponding to the unique identifier of the area.

9. The method of claim 1, wherein the method further comprises: In the step of obtaining the detection priority data of the current detection area, performing defect classification detection operation, and recording the detection results: Obtain the priority data of all current detection areas, and generate a detection task sequence; According to the detection task sequence, perform defect classification detection operation.

10. The method of claim 8, wherein the method further comprises: After the step of performing the defect classification detection operation according to the detection task sequence: For each detection area, the differentiated detection results are recorded in priority and associated with archives to generate a detection result overview report.

Images