Foundation base quality safety intelligent detection method and system

Abstract

The application discloses a kind of foundation quality safety wisdom detection method and system, comprising: according to the similarity matching result of engineering information, generate the foundation engineering template of new engineering;Based on the video stream data received in real time, obtain the detection video file of detection foundation pile;Identify abnormal events in video stream, mark abnormal image frame;Foundation pile review operation is carried out, and the foundation pile review record is obtained;According to the inspection result, identify the inspection items needing rectification, and send rectification notice;Obtain the detection result of ground rock-soil character, and identify the defect position, defect type and defect degree of foundation pile;Establish rendering model and generate detection report, configure unit authority and evaluate task allocation rationality.The application optimizes the foundation quality detection process by intelligent and automated foundation detection method, significantly improves the precision, efficiency and safety of foundation quality detection, and provides reliable protection for the smooth implementation of engineering project.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of engineering detection technology, and in particular to a foundation quality safety intelligent detection method and system. BACKGROUND

[0002] Currently, foundation engineering is a crucial link in construction engineering, and its quality is directly related to the safety and service life of the building. However, the traditional foundation quality detection method still faces a series of problems, which restricts the efficiency and accuracy of foundation quality detection. First, the existing detection methods mainly rely on manual inspection, sampling detection and periodic inspection, etc. These methods have large manual errors and subjective factors, especially in complex geological environments, which can easily lead to missed or incorrect detection, making it difficult to comprehensively and accurately assess the safety of the foundation. Secondly, the traditional foundation detection lacks real-time and continuity. Since the detection is usually carried out at fixed time points, it is difficult to obtain real-time data from the construction site in real time, especially in response to changing weather conditions and complex geological environments. Different weather and geological environments have a significant impact on the foundation. In good weather conditions, the operation behavior of construction personnel may comply with the specifications, but in adverse weather conditions, the quality of the foundation at the construction site will be severely affected, for example, in heavy rain or low temperature weather, the bearing capacity of the soil and the stability of the pile foundation will change, and the construction operation behavior may deviate from the specifications. This phenomenon shows that the same construction behavior has differences in different environmental conditions, and the traditional detection method often ignores this, making it difficult to achieve targeted and real-time evaluation. In addition, with the expansion of engineering scale and the complexity of construction environment, the traditional detection method cannot meet the requirements of modern engineering projects for high efficiency, high accuracy and intelligentization. The position and detection data of the pile are usually collected by different workers respectively, and there are problems of data dispersion and information asymmetry. The lag of data transmission and processing makes it difficult for engineers to understand the detection results of each pile in real time, leading to delays and information loss in engineering management. Further, the traditional detection method often relies on manual recording and uploading, making it difficult to ensure the integrity and accuracy of the data. Key data files such as images and videos in the detection process are often provided by different workers, lacking a unified standard and format, leading to data fragmentation and making it difficult to effectively integrate and analyze. Therefore, how to efficiently and accurately match historical engineering data with new engineering data, how to identify and mark abnormal events in real time through real-time video monitoring and intelligent analysis technology, and how to ensure the accurate matching of detection data and pile position information for each pile, are still technical problems to be solved. SUMMARY

[0003] The present application provides a foundation quality safety intelligent detection method and system to solve the above problems in the prior art.

[0004] The first aspect of the present application provides a foundation quality safety intelligent detection method, mainly comprising:

[0005] According to the similarity of historical engineering information and current new engineering information, the most similar historical engineering is matched, and a new engineering template is generated based on the matching result;

[0006] According to the user-selected pile number, the camera is determined and accessed, the working state of the camera in the monitoring process is identified based on the real-time received video stream data, and the detection video file of the detection pile is obtained;

[0007] According to the detection video data, weather data and geological environment data, the abnormal events in the video stream are identified, the identified abnormal event image frames are saved, and are marked as abnormal image frames;

[0008] By comparing the pile number coordinate information selected by the user with the preset pile position coordinate, it is judged whether the pile position is correct, the pile rechecking operation is performed, and the pile rechecking record is obtained;

[0009] Based on the selected inspection items and the pile number, the inspection photos or inspection videos with watermarks are shot, the inspection items that need to be rectified are identified according to the inspection results, and rectification notices are sent to the project members of the items that need to be rectified;

[0010] According to the pre-defined data format, the on-site detection data is collected, the foundation rock-soil property detection result is determined, and the defect position, defect type and defect degree of the pile are identified;

[0011] According to the matching of the detection data and the pile position information, the rendering model is established and the detection report is generated, the unit authority is configured and the task allocation rationality is evaluated.

[0012] Further, according to the similarity of historical engineering information and current new engineering information, the most similar historical engineering is matched, and a new engineering template is generated based on the matching result, comprising:

[0013] The historical engineering information obtained by the engineering management system is used to calculate the similarity of the historical engineering information and the engineering information of the current new engineering by using a cosine similarity calculation method, match the historical engineering with the largest similarity, and generate a basic engineering template for the new engineering based on the matching result, and the engineering information includes engineering basic information, project members, engineering documents, and detection object lists; through the engineering management system, the engineering information of the new engineering is modified and edited, and saved to the engineering database after modification is completed; according to the engineering address information, the related engineering is displayed on the map through the coordinate system, and different colors are used to mark different engineering states, including shelved, in progress and completed; according to the detection scheme and the detection method, the corresponding engineering detection project is generated, the detection project inherits the common information of the engineering information, and for the specific detection method, the individual project information of the specific detection method is edited, and the common information includes engineering number, engineering name and project members.

[0014] Further, the camera is determined and accessed according to the stake number selected by the user, the working state of the camera in the monitoring process is identified based on the real-time received video stream data, and the detection video file of the detection stake is obtained, including:

[0015] Through the mobile terminal APP of the foundation detection system, the detection project, stake number and detection method are selected and the detection is initiated; according to the stake number selected by the user in the APP, the monitoring scene is associated with the corresponding camera, the camera number is determined and the camera is accessed; the video stream is received and processed through the accessed camera, the video data is stored in real time, the start time and the completion time of the detection are recorded, the video data in the detection process is saved from the start time to the end time, and the video is marked as the detection video data of the detection task; according to the real-time received video stream data, the working state of the camera in the monitoring process is identified, including the camera being closed, signal loss or camera interruption; if the camera is closed or signal loss is detected, the current video segment is automatically recorded and saved as a video file; if camera interruption occurs during the detection process, the video segments at the breakpoint or interruption are spliced in time sequence to generate a continuous video file as the detection video file of the detection stake; the use state of the camera is obtained, if the detection task is in progress, the camera is marked as occupied state, and other detection tasks are prohibited from occupying the camera until the current detection task is completed; according to the video storage time and the preset storage time threshold, the video data is classified, including hot data and cold data; according to the classification of the video data, a hot and cold storage strategy is used to optimize the storage of the video data, and by setting a timing task, the cold data is automatically transferred to a low-cost storage area, and the hot and cold storage strategy includes storing hot data in a high-performance storage medium and storing cold data in a low-cost storage medium.

[0016] Further, the method comprises: detecting video data, weather data and geological environment data; identifying an abnormal event in the video stream according to the video data, the weather data and the geological environment data; saving an image frame of the identified abnormal event and marking the image frame as an abnormal image frame.

[0017] According to the detection video data, each frame image in the video stream is extracted frame by frame through a video decoding technology, the weather data and the geological environment data are obtained through an environment detection module, the convolutional neural network and the long short-term memory network are jointly trained to construct an abnormal event identification model, and the abnormal event in the video stream is identified, wherein the convolutional neural network processes the images extracted from the video stream, the long short-term memory network processes the weather data and the geological environment data, the weather data includes temperature, humidity, air pressure and wind speed, the geological environment data includes soil type and soil humidity, and the abnormal event includes a dangerous source, an abnormal behavior and an irregular operation; if the abnormal event in the detection video is detected, the identified abnormal event is marked, the timestamp, the location and the event description of the abnormal event are recorded, the identified abnormal event image frame is saved according to the detected abnormal event in the detection video, and the image frame is marked as an abnormal image frame; and an alarm mechanism is triggered according to the identification result of the abnormal event, and a project member or a supervisor is notified.

[0018] Further, the method comprises: comparing the pile number coordinate information of the selected pile with the preset pile coordinate to determine whether the position of the pile is correct, performing pile review operation, and obtaining pile review record.

[0019] The pile review request is initiated in the mobile terminal APP of the foundation detection system, and the pile number of the selected pile is determined; the coordinate information of the pile is obtained through RTK or a total station according to the pile number of the selected pile, and is compared with the preset pile coordinate to determine whether the position of the pile is correct; if the position is correct, the review operation is continued, and if the position is incorrect, the user is prompted to select the correct detection object again through a pop-up window; according to the detection project selected by the project member in the APP, the video stream of the detection link to be recorded is uploaded, the corresponding key frame is automatically intercepted through an image processing algorithm, the video stream is labeled using a preset identifier, and the preset identifier includes opening, perforating, final hole, hole depth review, recording and sampling; the key frame and the detection link video file are bound to form the pile review record.

[0020] Further, the method comprises: selecting an inspection item and a pile number, shooting an inspection photo or an inspection video with a watermark, identifying an inspection item that needs to be rectified according to an inspection result, and sending a rectification notice to a project member of the inspection item that needs to be rectified.

[0021] Through the mobile terminal APP of the foundation detection system, the inspection items and pile numbers are selected, the inspection photos or videos with watermarks are shot, the inspection records and note information are filled in, and the inspection photos, inspection videos, inspection records and note information are uploaded to the cloud server of the foundation detection system. The inspection items include site inspection, witnessing and safety inspection. The watermarks include time, inspector ID and project ID. If the inspection result judges that the inspection items need to be rectified, the inspection items are marked as needing rectification, and rectification notices are sent to project members through the foundation detection system. The rectification notice content includes rectification problem description, inspection photos and inspection videos. According to the effect description, pictures and videos uploaded by the project members after rectification, the inspection items are marked as having been rectified. By comparing the effect description, pictures and videos before and after rectification, it is judged whether the rectification effect meets the standard, and the rectification result is fed back to the inspector.

[0022] Further, the on-site detection data is collected according to the predefined data format, the foundation soil property detection result is determined, and the defect position, defect type and defect degree of the pile are identified, including:

[0023] According to the predefined data format, the on-site detection data is collected, the detection data collected on site is uploaded to the cloud server through the preset uploading mode, the data interface is used for formatting and stored in the detection database. The detection data includes foundation soil property data and pile foundation pile body and bearing stratum parameter data. The preset uploading mode includes mobile network, Bluetooth, wifi and mobile phone APP. The foundation soil property data includes load test curve, sounding test curve, vane shear test curve, standard penetration test curve, static sounding test curve, simple soil test curve and ground penetrating radar profile gray scale diagram. The pile foundation pile body and bearing stratum parameter data includes low strain waveform, high strain test curve, static load test curve, core image, ultrasonic waveform, hole wall profile curve, hole wall panoramic image, pressure angle-settling thickness conversion curve, wave velocity profile, three-dimensional CT imaging diagram, tube wave reflection wave train diagram, electromagnetic attenuation curve and thermal method pile integrity three-dimensional image. According to the statistical result of the foundation soil property data, the test value is calculated, and the test value is compared with the characteristic value to judge whether the foundation soil property detection result is qualified, wherein the characteristic value takes the high value of the design value and the required value in the expert knowledge base. According to the pile foundation pile body and bearing stratum parameter data of the historical engineering project, the deep neural network is used for model training to build a pile defect identification model. According to the pile foundation pile body and bearing stratum parameter data of the current engineering project, the pile defect identification model is used to identify the defect position, defect type and defect degree of the pile. Combined with the foundation soil property detection result, the foundation detection result is determined.

[0024] Furthermore, the process of matching detection data with pile location information, establishing a rendering model and generating a report, configuring unit permissions, and evaluating the rationality of task allocation includes:

[0025] Based on the test data and pile location information uploaded by project members, pile location coordinates are associated with the test objects using identifiers, and test results are matched with each pile location. Based on the pile location information, test items, associated test data, and complete documentation from the test object, a 2D or 3D foundation test model is established using a coordinate system. Color rendering is used to distinguish test result categories. The foundation test model includes test result categories such as qualified, unqualified, and requiring rectification. Complete documentation includes original records, test data, inspection records, test results, photos and videos of key steps, and process videos. Based on the test items, test plan, test data, and foundation information for each test object... Based on the basic test results, a test report is generated using a preset test report template. The foundation testing system allows for setting permissions for project construction units to view test progress, visualize results, and access electronic reports. According to the actual project progress, the system obtains the completion status of each test task and displays the real-time status of the test progress using different colored progress bars. It identifies projects with slow progress and sends warning messages to remind project leaders and testing personnel. The system also obtains task completion and workload data from testing personnel to assess the rationality of test task allocation. If the rationality of the test task allocation is lower than a preset threshold, it sends test task allocation adjustment suggestions to the scheduler for dynamic adjustment of testing personnel.

[0026] A second aspect of the present invention provides a smart detection system for foundation quality and safety, mainly comprising:

[0027] The basic engineering template matching module is used to match the historical project with the highest similarity to the project information of the current new project based on the similarity of historical project information and the project information of the current new project, and generate the basic engineering template of the new project based on the matching results;

[0028] The detection video acquisition module is used to determine and connect to the camera based on the pile number selected by the user, identify the working status of the camera during the monitoring process based on the real-time received video stream data, and acquire the detection video file of the detected pile.

[0029] The abnormal event recognition module is used to identify abnormal events in the video stream based on the detected video data, weather data, and geological environment data, and to save the identified abnormal event image frames as abnormal image frames.

[0030] The pile location verification module is used to compare the pile number coordinates selected by the user with the preset pile location coordinates to determine whether the pile location is correct, perform pile verification operations, and obtain pile verification records.

[0031] The pile position inspection and rectification notification module is used for shooting inspection photos or inspection videos with watermarks based on selected inspection items and pile numbers, identifying inspection items that need to be rectified according to inspection results, and sending rectification notifications to project members of the items that need to be rectified.

[0032] The detection data analysis module is used for collecting on-site detection data according to a predefined data format, determining the foundation rock-soil property detection result, and identifying the defect position, defect type and defect degree of the foundation pile.

[0033] The detection report generation module is used for completing matching according to the detection data and the pile position information, establishing a rendering model and generating a detection report, configuring unit permissions and evaluating the rationality of task allocation.

[0034] The technical scheme provided by the embodiment of the present application can include the following beneficial effects:

[0035] The present application provides a kind of foundation quality safety wisdom detection method and system.The present application can efficiently generate foundation engineering template by matching historical engineering information and new engineering similarity, improve the efficiency and accuracy of new engineering detection.The present application can accurately identify and mark abnormal events in complex environment by using real-time video stream monitoring and the combination of weather and geological environment data, to ensure the real-time and comprehensiveness of detection process.For the accurate review of foundation pile position, avoid position error and data inconsistency problem, ensure the accuracy of detection.The present application generates watermark photo and video record automatically, and automatically analyzes and rectifies the detection result, greatly improves work efficiency and data credibility, and generates detection report based on preset template, real-time detection progress tracking and task rationality evaluation, effectively improves the transparency and cooperation efficiency of project management.As a whole, the present application optimizes the foundation quality detection process by intelligent and automated foundation detection method, significantly improves the accuracy, efficiency and safety of foundation quality detection, and provides reliable guarantee for the smooth implementation of engineering project. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 It is a flow chart of a kind of foundation quality safety wisdom detection method of the present application;

[0037] Figure 2 It is a schematic diagram of a kind of foundation quality safety wisdom detection method of the present application;

[0038] Figure 3 It is a schematic diagram of a kind of foundation quality safety wisdom detection system of the present application. DETAILED DESCRIPTION

[0039] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application will be described in detail below in combination with drawings and specific embodiments.

[0040] As Figures 1-2 The foundation quality safety intelligent detection method can specifically include the following steps:

[0041] In step S101, the historical project with the largest similarity is matched according to the similarity of the historical engineering information and the engineering information of the current new project, and the foundation engineering template of the new project is generated based on the matching result.

[0042] The historical engineering information obtained through the engineering management system is used to calculate the similarity of the historical engineering information and the engineering information of the current new project by using the cosine similarity calculation method, the historical engineering with the largest similarity is matched, and the foundation engineering template of the new project is generated based on the matching result. Engineering information includes engineering basic information, project members, engineering documents, and detection object list. Through the engineering management system, the engineering information of the new project is modified and edited, and saved to the engineering database after modification. According to the engineering address information, the related engineering is displayed on the map through the coordinate system, and different colors are used to mark different engineering states, including shelved, in progress and completed. According to the detection scheme and detection method, the corresponding engineering detection project is generated, the detection project inherits the common information of the engineering information, and for the specific detection method, the individual project information of the specific detection method is edited, and the common information includes engineering number, engineering name and project members.

[0043] Exemplarily, a new foundation engineering project is XYZ Mansion Foundation Detection Project, the historical project engineering data of the project management system has recorded the historical project engineering, and the historical project engineering with the most similar engineering information to the current new project is extracted from the historical project information, and the cosine similarity calculation method is used for matching, wherein, in the historical project library, there is an ABC Industrial Park Foundation Detection Project with highly similar engineering information to the new project XYZ Mansion Foundation Detection Project, the engineering information includes engineering basic information, project members, engineering documents, detection object list, and the cosine similarity calculation obtains the similarity of the historical project and the current new project is 0.95. According to the similarity result, the foundation engineering template of the ABC Industrial Park Foundation Detection Project is extracted, and the engineering template of the new project XYZ Mansion Foundation Detection Project is automatically generated by taking it as a reference, wherein the engineering basic information in the template content includes the engineering number XYZ-202001, the engineering name XYZ Mansion Foundation Detection Project, the person in charge Li Ming, the engineering address U City V District W Road 123, the five parties including the construction unit A Company, the supervision unit B Company, the design unit C Company, the construction unit D Company and the detection unit E Company. The project member information in the template content includes project manager O, technical person in charge P, detection engineer Q, supervision engineer R, etc. The engineering document in the template content includes all the file records related to the project, such as technical file, meeting minutes, construction report, detection report, etc. The detection object list in the template content includes natural foundation and pile foundation, wherein the pile foundation of XYZ Mansion is 240, the pile length is 20-30 meters, the pile diameter is 1.2 meters, the construction date is May 15, 2025, the detection process information is that the detection method of this project follows the detection method of ABC Industrial Park Foundation Detection Project, the detection method includes core drilling method, high strain method, low strain method, acoustic wave transmission method, static load, flat plate load, anchor locking force test, supporting anchor and soil nail acceptance test, hole and slot quality, hole camera method, cross-hole acoustic CT method, tube wave method, side hole method, TIP thermal anomaly method, reinforcement cage length detection, light dynamic penetration test, heavy dynamic penetration test, super heavy dynamic penetration test, vane shear test, standard penetration test, static cone penetration test, simple soil test, ground penetrating radar test, detection instruments include static load machine, high strain meter, etc. The relevant detection personnel, process pictures and videos, and the pre-warning information recorded in the monitoring process, the detection result information based on the detection result of the historical project is generated. The preliminary framework of the detection report includes the analysis of the detection result, the integrity of each pile, the bearing capacity and other information.In the foundation detection project of ABC industrial park, the integrity of pile body and the identification of pile end bearing layer are detected by combining high strain method and core drilling method, and the bearing capacity of large diameter bored pile is comprehensively evaluated. For the foundation detection project of XYZ mansion, the high strain method used in the new project needs to be personalized edited, including the test depth of high strain method is 25 meters of pile body, at least three tests are used for each pile body, and laser settlement observation instrument is configured. The modified foundation detection project information of XYZ mansion will be saved to the project database in the system. Through the coordinate system, the position of the foundation detection project of XYZ mansion is automatically displayed, and it is marked as in progress state and displayed on the map. At the same time, other projects that have been shelved or completed will be displayed on the map, and the project status will be marked with different colors.

[0044] In step S102, the camera is determined and accessed according to the pile number selected by the user, the working state of the camera in the monitoring process is identified based on the real-time received video stream data, and the detection video file of the detection pile is obtained.

[0045] Through the mobile terminal APP of the foundation detection system, the detection project, pile number and detection method are selected and the detection is initiated. According to the pile number selected by the user in the APP, the monitoring scene is associated with the corresponding camera, the camera number is determined and the camera is accessed. The video stream is received and processed through the accessed camera, the video data is stored in real time, the start time and completion time of the detection are recorded, the video data from the start time to the end time in the detection process is saved, and the video is marked as the detection video data of the detection task. According to the real-time received video stream data, the working state of the camera in the monitoring process is identified, and the working state includes camera off, signal loss or camera interruption. If the camera is off or the signal is lost, the current video segment is automatically recorded and saved as a video file. If camera interruption occurs during the detection process, the video segments at the breakpoint or interruption are spliced in time sequence to generate a continuous video file as the detection video file of the detection pile. The use state of the camera is obtained, if the detection task is being performed, the camera is marked as occupied state, and other detection tasks are prohibited from occupying the camera until the current detection task is completed. According to the video storage time and the preset storage time threshold, the video data is classified, and the data classification includes hot data and cold data. According to the classification of video data, the cold and hot storage strategy is adopted to optimize the storage of video data, and the cold data is automatically transferred to the low-cost storage area by setting the timing task. The cold and hot storage strategy includes storing hot data in high-performance storage medium and storing cold data in low-cost storage medium.

[0046] Exemplary, there is a ongoing foundation detection task for ABD Mansion Pile Detection Project, which is located in M District of L City, involving the detection of 130 pile foundations. The project leader initiates a pile detection task through the mobile APP of the foundation detection system, and the user selects No. 5 pile for detection and selects high strain method as the detection method. According to the user's selection, the monitoring scene is automatically associated with the corresponding camera, the camera number of the pile is determined as CAM-05, and the camera is successfully accessed. Through the access of CAM-05 camera, video stream is received and processed, and real-time video data recording starts. The detection start time of the video is 09:00:00 on January 15, 2020, and the completion time is 10:00:00 on January 15, 2020. The video data from the start time to the end time is saved completely, and the video is marked as the detection video data of the ABD Mansion Pile Detection Project. The entire video data will include the complete detection process from the start to the end of the detection, including key images and detection states during the detection process. During the detection process, the working state of the real-time monitoring camera is monitored. If the signal of the CAM-05 camera is lost due to network fluctuations or device problems at some time, the current video segment is automatically saved as an independent video file to prevent loss of important data. If the signal loss lasts for about 3 minutes, the camera will resume normal operation, and the subsequent video data will be received and stored. If the camera has a short interruption lasting 5 seconds, the breakpoint time will be recorded after the interruption of the camera is detected, and the video segment at the interruption and the normal video segment after it will be spliced after the subsequent detection is completed to ensure that the final generated detection video file is continuous. The spliced video file is taken as the complete pile detection video file, and the specific time of the breakpoint is recorded. By monitoring the usage state of the camera, it is found that the detection task is in progress, so the CAM-05 camera is marked as occupied and other detection tasks are prohibited from occupying the camera until the current detection task is completed. In this way, multiple detection tasks can be prevented from occupying the same camera resource at the same time, ensuring the integrity of the video data. After the detection is completed, all video data is classified, and the video data is automatically classified into hot data and cold data according to the video storage time and the preset storage time threshold. Hot data refers to detection video data that is currently in use or needs to be frequently accessed in the near future, and cold data refers to video data of detection tasks that are no longer frequently accessed or have been completed. Hot data is stored in high-performance storage media such as SSD for fast access, while cold data is automatically transferred to low-cost storage areas such as tapes or low-cost cloud storage. The timing task is set to 30 days, and the data is automatically transferred to the low-cost storage area after being stored in the high-performance storage medium for 30 days, ensuring efficient use of storage space and reducing storage costs.

[0047] In step S103, according to the detection video data, weather data and geological environment data, the abnormal event in the video stream is identified, the identified abnormal event image frame is saved, and is marked as an abnormal image frame.

[0048] According to the detection video data, by using the video decoding technology, each frame image in the video stream is extracted frame by frame, the weather data and the geological environment data are obtained through the environment detection module, the convolutional neural network and the long short-term memory network are jointly trained to construct an abnormal event identification model, and the abnormal event in the video stream is identified, wherein the convolutional neural network processes the image extracted from the video stream, the long short-term memory network processes the weather data and the geological environment data, the weather data includes temperature, humidity, air pressure and wind speed, the geological environment data includes soil type and soil humidity, and the abnormal event includes a dangerous source, an abnormal behavior and an irregular operation. If the abnormal event in the detection video is detected, the identified abnormal event is marked, the timestamp, position and event description of the abnormal event are recorded, and according to the detected abnormal event in the detection video, the identified abnormal event image frame is saved and marked as an abnormal image frame. According to the identification result of the abnormal event, an alarm mechanism is triggered, and project members or supervisors are notified.

[0049] Exemplary, in the pile foundation detection process of the foundation detection project of XYZ Building, one frame of image in the received video data shows that the operator is setting up the core drilling operation platform, including lifting, placing and fixing the core drilling operation platform components such as horizontal rods and vertical rods. In the detection on that day, the weather data records a temperature of 28°C, a humidity of 60%, an air pressure of 1013 hPa, and a wind speed of 10 km / h. In the geological environment data, the soil type is clay, and the soil humidity is 15%. However, in another detection video, one frame of image shows that the operator is also setting up the core drilling operation platform, but the weather data is a temperature of 15°C, a humidity of 85%, an air pressure of 1005 hPa, and a wind speed of 30 km / h. The geological environment data shows that the soil type is clay, but the soil humidity is 30%, which is much higher than the normal value. The convolutional neural network and the long short-term memory network are jointly trained to build an abnormal event recognition model to recognize abnormal events in the video stream. The convolutional neural network processes the images extracted from the video stream, and the long short-term memory network processes the weather data and the geological environment data. According to the processing of the convolutional neural network, the image obtained on that day is first determined to be core drilling operation platform setting work, which itself does not belong to abnormal behavior. Under the influence of environmental data, the weather is mild, the wind speed is moderate, the ground is hard and stable, and the construction is not disturbed by any external environment. The convolutional neural network and the long short-term memory network are jointly trained to recognize that the operator's behavior of setting up the core drilling operation platform under normal weather and soil conditions is recognized as normal operation by the convolutional neural network, and no abnormal alarm is triggered. This operation is reasonable in the current environment and there is no risk of dangerous or non-standard operation. However, in another day, the operator is detected to be setting up the core drilling operation platform in harsh weather conditions, the convolutional neural network still recognizes the operator's behavior image as core drilling operation platform setting work, but the long short-term memory network adjusts the model according to the geological environment data. Due to strong wind and slippery ground, the safety of construction is seriously affected. The strong wind may cause the scaffold assembly to be unstable or the worker to lose balance, and the slippery soil may cause the worker to slip or stumble when carrying and fixing the scaffold, increasing the risk of falling injury. Therefore, an abnormal behavior alarm is triggered, considering that the operator did not take appropriate safety measures in this environment, the behavior is judged as non-standard operation, and the behavior is marked as non-standard behavior, and an alarm is issued to remind the project personnel to suspend work and improve the operation method, and a timestamp, location information and event description are generated. The specific event record is timestamp 2020-01-15 09:34:25, location pile No. 5, event description operator setting up core drilling operation platform in low temperature, strong wind and slippery conditions. The video frames related to the abnormal event are extracted and marked as abnormal image frames, saved to the system for later viewing, and specially marked.At the same time, according to the identification result of the abnormal event, the alarm mechanism is triggered, and instant notification is automatically sent to the project members and supervisors, and the alarm information shows that the non-standard operation is found in the detection site of pile No. 5, the operator carries out the drilling core method operation platform building work in the low temperature, strong wind and slippery condition, please rectify immediately. After receiving the notice, the project manager and the safety officer immediately carried out on-site inspection to ensure that the rectification measures are implemented.

[0050] In step S104, the pile number coordinate information of the selected pile is compared with the preset pile coordinate to determine whether the pile position is correct, the pile is reviewed, and the pile review record is obtained.

[0051] The pile review request is initiated through the mobile APP of the foundation detection system, and the pile number of the selected pile is determined. According to the pile number of the selected pile, the coordinate information of the pile is obtained through RTK or total station, and compared with the preset pile coordinate to determine whether the pile position is correct. If the position is correct, continue to review the operation, if the position is not correct, the user is prompted to select the correct detection object for review through the pop-up window. According to the detection project selected by the project member in the APP, the video stream of the detection link to be recorded is uploaded, the key frame corresponding to the detection link is automatically intercepted through image processing algorithm, and the video stream is labeled using preset identifier, the preset identifier includes opening, perforating, final hole, hole depth review, recording and sampling. The key frame and the detection link video file are bound to form the pile review record.

[0052] Exemplary, in the foundation detection project of XYZ Mansion, the project manager O initiates a pile review request through the mobile APP of the foundation detection system, and selects the pile ZB-003 for review. According to the project requirements, the real-time coordinates of the pile are obtained through the RTK positioning system, and compared with the preset pile coordinates. Assuming that the preset pile coordinates are (128.752, 34.190, 10.245), and the pile coordinates measured by the RTK system are (128.752, 34.190, 10.240), the error of the Z-axis coordinates is 0.005 meters, which is within the allowable error range ± 0.01 meters, so the pile position is correct. After confirming the correct position, the project manager O selects the final hole detection as the detection link that needs to be reviewed in the APP, and starts uploading the related detection video stream. The uploaded video is the video taken during the drilling detection process of the pile. After uploading the video, the key frames are extracted through image processing algorithm, and the last stage of drilling is recognized and marked as the final hole mark. This key frame in the video is cut off. The key frame is bound with the complete detection video file, such as detection video file number V12345, to generate a review record containing pile number, detection link, key frame and video file. In this process, there are four key links in the detection video, including hole opening, hole drilling, final hole, hole depth review, coding and sampling. Each link will automatically extract and mark the corresponding key frame. For example, in the drilling stage, the drilling key frame is automatically recognized and extracted, and marked as drilling. Similarly, in the hole opening, hole depth review, coding and sampling stages, the corresponding key frames are marked. Finally, the project manager O checks the review record of the pile ZB-003 through the APP, confirms that the video key frame of all detection links has been successfully uploaded, completes the review process, and saves the review record as a complete detection file for subsequent inspection and approval.

[0053] Step S105, based on the selected inspection items and pile number, take the inspection photos or inspection videos with watermarks, identify the inspection items that need to be rectified according to the inspection results, and send rectification notices to the project members of the project that need to be rectified.

[0054] Through the mobile terminal APP of the foundation detection system, the inspection items and pile numbers are selected, the inspection photos or inspection videos with watermarks are shot, the inspection records and note information are filled in, and the inspection photos, inspection videos, inspection records and note information are uploaded to the cloud server of the foundation detection system. The inspection items include on-site inspection, witnessing and safety patrol, and the watermarks include time, inspector ID and project ID. If the inspection result judges that the inspection items need to be rectified, the inspection items are marked as needing rectification, and rectification notices are sent to project members through the foundation detection system. The rectification notice content includes rectification problem description, inspection photos and inspection videos. According to the effect description, pictures and videos uploaded by the project members after rectification, the rectification feedback is determined, and the inspection items are marked as having been rectified. By comparing the effect description, pictures and videos before and after rectification, it is judged whether the rectification effect meets the standard, and the rectification result is fed back to the inspector.

[0055] Exemplary, in the ABE basement pile foundation engineering project, the project site engineer S uses the mobile APP of the foundation detection system for daily safety inspection, and the engineer S selects the safety inspection in the APP, and selects the pile number P-021 as the inspection object. Then, the engineer S uses the APP to shoot a 45-second video and 3 on-site photos, which clearly record the problem of construction personnel without wearing safety helmets around the pile, and automatically add watermarks to the photos and videos during the shooting process, including the shooting time 2020-01-1410:23:45, the inspector ID O0523, and the project ID JD2020. Subsequently, the engineer S fills in the inspection record, the record content is that two construction personnel on the northwest side of P-021 pile foundation are found without wearing safety helmets, and it is stated in the remarks that this area is a frequent mechanical operation area, which has a large safety hazard. All photos, videos and text records are uploaded to the cloud server of the foundation detection system. After the system automatically analyzes the inspection results, it is judged that there is obvious safety problem in the inspection matter, which needs to be rectified, and the inspection matter is marked as needing rectification, and a rectification notice is automatically generated, and the notification message is pushed to all members of the project safety group. The rectification notice contains the problem description personnel without wearing safety helmets, and the on-site inspection photos and videos uploaded by the engineer S. At 15:12 on the same day, the construction site responsible person T uploaded the rectification effect materials in the APP, and the responsible person T submitted a newly shot video with a length of 40 seconds and 4 rectification photos, and filled in the rectification effect description. The on-site workers have been educated on safety, and all construction personnel have worn safety helmets according to the regulations, and warning signs have been posted. The system automatically compares this batch of materials with the original inspection record, and updates the status of the inspection matter to rectified, and then analyzes the videos and photos before and after the rectification through image comparison and text content recognition, and confirms that the rectification effect meets the project safety management standard, and the on-site personnel have all worn safety helmets. Finally, the system feeds back the rectification result to the original inspector S, and the prompt content is rectification qualified, problem closed loop, and the safety inspection process ends.

[0056] Step S106, according to the pre-defined data format, the on-site detection data is collected, the foundation rock-soil property detection result is determined, and the defect position, defect type and defect degree of the pile are identified.

[0057] The field detection data is collected according to a predefined data format, the collected detection data is uploaded to a cloud server through a preset uploading mode, the data interface is used for formatting and storing the detection data into a detection database, the detection data includes foundation soil property data and pile foundation pile body and bearing stratum parameter data, the preset uploading mode includes a mobile network, Bluetooth, wifi and a mobile phone APP, the foundation soil property data includes a load test curve, a sounding test curve, a vane shear test curve, a standard penetration test curve, a static sounding test curve, a simple soil test curve and a ground penetrating radar profile grayscale graph, the pile foundation pile body and bearing stratum parameter data includes low strain waveform, high strain test curve, static load test curve, drill core image, ultrasonic waveform, hole wall profile curve, hole wall panoramic image, pressure angle-sediment thickness conversion curve, wave velocity profile graph, three-dimensional CT imaging graph, tube wave reflection wave train graph, side hole reflection wave train graph electromagnetic attenuation curve and thermal method pile body integrity three-dimensional image, according to the statistical result of the foundation soil property data, the test value is calculated, and the test value is compared with a characteristic value, whether the foundation soil property detection result is qualified is judged, wherein the characteristic value takes the high value of the design value and the required value in the expert knowledge base, according to the pile foundation pile body and bearing stratum parameter data of the historical engineering project, the model is trained using a deep neural network, and a pile foundation defect recognition model is constructed, according to the pile foundation pile body and bearing stratum parameter data of the current engineering project, the pile foundation defect recognition model is used to identify the defect position, defect type and defect degree of the pile foundation, and the foundation soil property detection result is combined to determine the foundation detection result.

[0058] For example, during the foundation pile detection process of the ABF construction project, the field technician G connects the static sounding device through the mobile phone APP and completes the data collection of the foundation rock-soil property test of the pile site numbered JH-105. This test collects static sounding test curve data, in which the maximum cone resistance is 12.5 MPa, the friction ratio is 0.015, and the hole depth is 18.2 meters. The data is uploaded to the cloud server through the preset uploading mode of the 4G mobile network. After the system receives the data, it automatically formats the original data through the API data interface and stores it in the static sounding data module of the foundation detection database. The preset uploading mode includes mobile network, Bluetooth, wifi and mobile phone APP. At the same time, technician G also uses Bluetooth to connect the low-strain detector to detect the pile integrity of the pile. The uploaded low-strain waveform data shows that there is an obvious abnormal reflection waveform at a distance of 8.6 meters from the top of the pile, with a reflection time of 3.2 milliseconds, indicating a suspected obvious integrity defect. In addition, technician G also uploads the hole wall profile curve and core image of the pile through the APP. The hole wall curve shows that the pile diameter shrinks at a depth of 9 meters, and the core image verifies that the honeycomb surface at this position is obvious. Then the test values of the static sounding, vane shear test and standard penetration test curve data are automatically counted. Taking the vane shear test as an example, the shear strength test value is 18.5 kPa. By querying the shear strength characteristic value of the corresponding stratum in the expert knowledge base, the design value is 16.0 kPa, and the specification requirement value is 17.5 kPa. The higher value, i.e. 17.5 kPa, is taken as the judgment criterion. Since the test value is higher than the characteristic value, the system determines that the foundation rock-soil property is qualified. The system calls the pile defect recognition deep neural network model trained in the past 200 similar engineering projects to automatically identify and analyze the uploaded pile detection data. The recognition model result shows that there is a medium intensity defect in the range of 8.5 to 8.8 meters from the top of the pile, the defect type is pile necking and honeycomb surface, and the defect degree rating is medium, with a confidence of 92.4%. Combined with the qualified foundation rock-soil property test result and the medium defect recognition result of the pile body, the conclusion of the foundation detection of the pile is given as "the pile body has a medium defect", and it is recommended to recheck and evaluate the influence on the bearing capacity through the load test. Finally, the detection record is archived by the project detection system and is synchronized to the project manager, regulatory unit, owner unit and supervision unit.

[0059] Step S107, according to the detection data and pile site information, the matching is completed, the rendering model is established and the detection report is generated, the unit authority is configured and the rationality of task allocation is evaluated.

[0060] According to the detection data and pile position information uploaded by the project members, the pile position coordinates are associated with the detection objects through identifiers, and the detection results are matched with each pile position. According to the pile position information, detection project, associated detection data and complete data of the detection object in the detection process, a 2D or 3D foundation detection model is established through the coordinate system, and a color rendering method is used to distinguish the detection result categories. The content of the foundation detection model includes the detection result categories, including qualified, unqualified and needing rectification, and the complete data includes the original record, detection data, inspection record, detection result, key link photo video and process video. According to the detection project, detection scheme, detection data and foundation detection result of each detection object, a detection report is generated based on a preset detection report template. Through the foundation detection system, the rights of the project construction unit to view the detection progress, result visualization and electronic report are set. According to the actual progress of the project, the completion status of each detection task of the project is obtained, and the real-time state of the detection progress is displayed through a progress bar of different colors, the project with slow detection progress is identified, and warning information is sent to remind the project leader and the detection personnel. The task completion data and workload data of the detection personnel are obtained, the rationality of the detection task allocation is evaluated, and if the rationality of the detection task allocation is lower than a preset threshold, a detection task allocation adjustment suggestion is sent to the dispatcher for dynamic adjustment of the detection personnel.

[0061] For example, during the construction phase of the ABG Science and Technology Park Underground Garage Project, project inspector H uploaded pile site inspection data numbered QM-P047 using the foundation inspection system. The inspection project was ultrasonic pile integrity testing and standard penetration testing. The system automatically identified the pile's identifier and matched it with the coordinates measured on site (119.2564, 32.1437, -8.30), confirmed its spatial position in the coordinate system, and bound the inspection data to the pile site. The system integrated the inspection object of the pile site with its inspection project, inspection plan, data records, and complete materials to establish a 2D foundation inspection model. When generating the model, the system used color rendering to visually mark the results based on the categories: green for qualified, red for unqualified, and yellow for needing rectification. The QM-P047 inspection data showed that the ultrasonic waveform continuity was interrupted, and the abnormal section of the wave velocity appeared between 13.8 meters and 14.2 meters of the pile length, with the ultrasonic wave velocity dropping to 3500 m / s, which was lower than the standard threshold of 3800 m / s. The system initially determined that the pile integrity was abnormal, and the inspection result was marked as needing rectification, with yellow rendering in the 2D model. The complete materials for the pile inspection included the inspection date 2020-01-13, the original waveform file, the detection process video, the supervision witness photo, and the inspection record filled out by inspector H, totaling 7 key material items. According to the project's inspection template, the system automatically retrieved the QM-P047 pile site's corresponding inspection project configuration and detection plan number, JCE-2020-0401, and generated a complete electronic inspection report based on the current data, report number QM-TB-P047-20200113, in PDF format uploaded to the system for subsequent queries. According to the project settings, the system automatically opened the detection progress, result visualization, and electronic report download permissions for the project construction unit, ABG Science and Technology Park Development Company. In the project progress monitoring module, the system displayed the completion rate of the overall project inspection task in the overview interface and used a progress bar to distinguish the status of each sub-task, such as the current project requiring 260 foundation piles to be inspected, with 173 completed as of January 14, a completion rate of 66.5%. The detection status was color-coded: green for completed, blue for in progress, and gray for not started. The system found that the completion rate of the B zone sub-task was only 42%, far lower than the current overall average progress, triggering the early warning mechanism and sending a reminder to project manager I and inspection manager J that the B zone detection progress is lagging, please allocate detection resources in a timely manner. Meanwhile, the system also analyzed the task completion and load data of the inspectors, showing that inspector H had completed 32 pile inspection tasks in the past 7 days, with an average daily workload of 4.6 piles, while inspector K had only completed 12 piles, with an average of less than 2 piles per day, indicating a significant difference in task allocation. The current system's reasonable threshold is 75%, and the personnel task rationality score for this time is only 58%.Therefore, the system generates a detection task allocation adjustment suggestion that the stake detection task No. 12-15 is allocated from H to K, and automatically pushes the suggestion to the work interface of the dispatcher M for adjusting personnel arrangement.

[0062] As Figure 3 The foundation quality safety intelligent detection system can specifically include:

[0063] The foundation engineering template matching module is configured to match the historical engineering with the largest similarity according to the similarity of the historical engineering information and the engineering information of the current new engineering, and generate a foundation engineering template of the new engineering based on the matching result.

[0064] The detection video acquisition module is configured to determine and access a camera according to the stake number selected by the user, identify the working state of the camera in the monitoring process based on the real-time received video stream data, and acquire a detection video file of the detection stake.

[0065] The abnormal event identification module is configured to identify abnormal events in the video stream according to the detection video data, weather data, and geological environment data, save the identified abnormal event image frames, and mark them as abnormal image frames.

[0066] The stake position rechecking module is configured to compare the stake number coordinate information of the selected stake with the preset stake position coordinate, determine whether the stake position is correct, perform a stake rechecking operation, and acquire a stake rechecking record.

[0067] The stake position checking and rectification notification module is configured to shoot a checking photo or a checking video with a watermark based on the selected checking items and the stake number, identify the checking items that need to be rectified according to the checking result, and send a rectification notification to the project members of the items that need to be rectified.

[0068] The detection data analysis module is configured to collect field detection data according to a predefined data format, determine a foundation rock-soil property detection result, and identify the defect position, defect type, and defect degree of the stake.

[0069] The detection report generation module is configured to complete matching according to the detection data and the stake position information, establish a rendering model, generate a detection report, configure unit permissions, and evaluate the rationality of task allocation.

[0070] The above description is only the preferred embodiment of the present application and the explanation of the technical principles. It should be understood by those skilled in the art that the scope of the protection of the present application is not limited to the technical solutions formed by the specific combinations of the above technical features. It should also cover other technical solutions formed by the combinations of the above technical features or their equivalents without departing from the concept of the present application. For example, the technical solutions formed by replacing the above features with the technical features with similar functions disclosed in the present application (but not limited to) and the like.

Claims

1. A method for detecting the quality of a foundation base, characterized in that, The method comprises: According to the similarity of historical engineering information and the engineering information of the current new project, the most similar historical project is matched, and the basic project template of the new project is generated based on the matching result; According to the user-selected pile number, the camera is determined and accessed, the working state of the camera in the monitoring process is recognized based on the real-time received video stream data, and the detection video file of the detection pile is obtained; According to the detection video data, weather data and geological environment data, the abnormal events in the video stream are identified, the identified abnormal event image frames are saved, and are marked as abnormal image frames; By comparing the pile number coordinate information selected by the user with the preset pile position coordinate, it is judged whether the pile position is correct, the pile rechecking operation is performed, and the pile rechecking record is obtained; Based on the selected inspection items and the pile number, the inspection photos or inspection videos with watermarks are shot, the inspection items that need to be rectified are identified according to the inspection results, and rectification notices are sent to the project members of the items that need to be rectified; According to the pre-defined data format, the on-site detection data is collected, the detection results of the foundation rock and soil properties are determined, and the defect position, defect type and defect degree of the pile are identified; According to the matching of the detection data and the pile position information, the rendering model is established and the detection report is generated, the unit authority is configured and the rationality of task allocation is evaluated; According to the detection video data, through video decoding technology, each frame of image in the video stream is extracted frame by frame, through the environment detection module, the weather data and the geological environment data are obtained, the convolutional neural network and the long short-term memory network are jointly trained, the abnormal event recognition model is constructed, and the abnormal events in the video stream are identified, wherein the convolutional neural network processes the images extracted from the video stream, the long short-term memory network processes the weather data and the geological environment data, the weather data includes temperature, humidity, air pressure and wind speed, the geological environment data includes soil type and soil humidity, and the abnormal events include dangerous sources, abnormal behaviors and non-standard operations; If an abnormal event is detected in the detection video, the detected abnormal event is marked, the timestamp, position and event description of the abnormal event are recorded, and the identified abnormal event image frames are saved and marked as abnormal image frames according to the detected abnormal event in the detection video; According to the identification result of the abnormal event, the alarm mechanism is triggered, and the project members or supervisors are notified; According to the pre-defined data format, the on-site detection data is collected, the detection results of the foundation rock and soil properties are determined, and the defect position, defect type and defect degree of the pile are identified; ​ The field detection data is collected according to a predefined data format, the collected detection data is uploaded to a cloud server through a preset uploading mode, the data interface is used for formatting and storing the detection data into a detection database, the detection data includes foundation soil property data and pile body and bearing stratum parameter data, the preset uploading mode includes a mobile network, Bluetooth, WiFi and a mobile phone APP, the foundation soil property data includes a load test curve, a sounding test curve, a vane shear test curve, a standard penetration test curve, a static sounding test curve, a simple soil test curve and a ground penetrating radar profile grayscale graph, and the pile body and bearing stratum parameter data includes low strain waveform, high strain test curve, static load test curve, drill core image, ultrasonic waveform, hole wall profile curve, hole wall panoramic image, pressure angle-sediment thickness conversion curve, wave velocity profile graph, three-dimensional CT imaging graph, tube wave reflection wave train graph, side hole reflection wave train graph electromagnetic attenuation curve and thermal method pile integrity three-dimensional image; according to a statistical result of the foundation soil property data, a test value is calculated, and the test value is compared with a characteristic value to determine whether the foundation soil property detection result is qualified, wherein the characteristic value takes a high value of a design value and a required value in an expert knowledge base; according to pile body and bearing stratum parameter data of historical engineering projects, a deep neural network is used for model training to construct a pile defect identification model; according to pile body and bearing stratum parameter data of a current engineering project, the pile defect identification model is used to identify a defect position, a defect type and a defect degree of the pile, and the foundation soil property detection result is combined to determine a foundation detection result of the current engineering project.

2. The method of claim 1, wherein, The historical engineering with the largest similarity is matched according to the similarity of the historical engineering information and the engineering information of the new engineering, and a foundation engineering template of the new engineering is generated based on the matching result, including: The historical engineering information obtained through the engineering management system is used to calculate the similarity of the historical engineering information and the engineering information of the new engineering by using a cosine similarity calculation method, the historical engineering with the largest similarity is matched, and a foundation engineering template of the new engineering is generated based on the matching result, the engineering information includes engineering basic information, project members, engineering documents and a detection object list; the engineering information of the new engineering is modified and edited through the engineering management system, and is saved to an engineering database after the modification is completed; according to the engineering address information, the related engineering is displayed on a map through a coordinate system, and different engineering states are marked with different colors, the engineering states include shelved, in progress and completed; according to the detection scheme and the detection method, corresponding engineering detection projects are generated, the detection projects inherit the common information of the engineering information, and the individual project information of the specific detection method is edited for the specific detection method, the common information includes an engineering number, an engineering name and project members.

3. The method of claim 1, wherein, The camera is determined and accessed according to the pile number selected by the user, the working state of the camera in the monitoring process is identified based on the real-time received video stream data, and the detection video file of the detected pile is obtained, including: The detection project, the pile number and the detection method are selected and the detection is initiated through the mobile terminal APP of the foundation detection system. According to the pile number selected by the user in the APP, the monitoring scene is associated with the corresponding camera, the camera number is determined, and the camera is accessed; Through the accessed camera, video stream is received and processed, video data is stored in real time, the start time and the end time of detection are recorded, video data from the start time to the end time in the detection process is saved, and the video is marked as the detection video data of the detection task; According to the real-time received video stream data, the working state of the camera in the monitoring process is identified, and the working state includes camera off, signal loss or camera interruption; If the camera is off or the signal is lost, the current video segment is automatically recorded and saved as a video file; If camera interruption occurs during the detection process, the video segments at the breakpoint or interruption are spliced in time sequence to generate a continuous video file as the detection video file of the detection pile; The use state of the camera is obtained, if the detection task is being performed, the camera is marked as an occupied state, and other detection tasks are prohibited from occupying the camera until the current detection task is completed; According to the video storage time and the preset storage time threshold, the video data is classified, the data classification includes hot data and cold data; According to the classification of video data, the cold and hot storage strategy is adopted to optimize the storage of video data, and the cold data is automatically transferred to the low-cost storage area by setting the timing task, and the cold and hot storage strategy includes storing hot data in high-performance storage medium and storing cold data in low-cost storage medium.

4. The method of claim 1, wherein, The pile number coordinate information selected by the user is compared with the preset pile position coordinate, whether the pile position is correct is judged, the pile rechecking operation is performed, and the pile rechecking record is obtained, including: The pile position rechecking request is initiated through the mobile terminal APP of the foundation detection system, the pile number selected by the user is determined; According to the pile number selected by the user, the coordinate information of the pile is obtained through RTK or total station, and is compared with the preset pile position coordinate, whether the position of the pile is correct is judged; If the position is correct, continue the rechecking operation, if the position is not correct, prompt the user to select the correct detection object for rechecking through the pop-up window; According to the detection project selected by the project member in the APP, the video stream of the detection link to be recorded is uploaded, the key frame corresponding to the detection link is automatically intercepted through image processing algorithm, the video stream is labeled using the preset identifier, and the preset identifier includes opening, perforating, final hole, hole depth rechecking, recording and sampling; The key frame and the detection link video file are bound to form the pile rechecking record.

5. The method of claim 1, wherein, The inspection photo or video with watermark is shot based on the selected inspection item and pile number, the inspection item needing rectification is identified according to the inspection result, and the rectification notice is sent to the project member of the project needing rectification, including: Through the mobile APP of the foundation detection system, the inspection items and pile numbers are selected, the inspection photos or videos with watermarks are shot, the inspection records and note information are filled in, and the inspection photos, videos, records and note information are uploaded to the cloud server of the foundation detection system. The inspection items include on-site inspection, witnessing and safety inspection. The watermarks include time, inspector ID and project ID. If the inspection result judges that the inspection items need to be rectified, the inspection items are marked as needing rectification, and rectification notices are sent to project members through the foundation detection system. The rectification notice content includes rectification problem description, inspection photos and videos. According to the effect description, pictures and videos uploaded by the project members after rectification, the rectification feedback is determined, and the inspection items are marked as having been rectified. By comparing the effect description, pictures and videos before and after rectification, it is judged whether the rectification effect meets the standard, and the rectification result is fed back to the inspector.

6. The method of claim 1, wherein, The matching is completed according to the detection data and pile position information, the rendering model is established and the report is generated, the unit authority is configured and the rationality of task allocation is evaluated, including: According to the detection data and pile position information uploaded by the project members, the pile position coordinates are associated with the detection objects through the identifier, and the detection results are matched with each pile position. According to the pile position information, detection project, associated detection data and complete data of the detection object in the detection process, a 2D or 3D foundation detection model is established through the coordinate system, and a color rendering method is used to distinguish the detection result categories. The content of the foundation detection model includes the detection result categories, including qualified, unqualified and needing rectification, and the complete data, including original records, detection data, inspection records, detection results, key link photos and videos and process videos. According to the detection project, detection scheme, detection data and foundation detection result of each detection object, a detection report is generated based on a preset detection report template; Through the foundation detection system, the authority of the project construction unit to view the detection progress, result visualization and electronic report is set. According to the actual progress of the project, the completion status of each detection task of the project is obtained, and the real-time state of the detection progress is displayed through a progress bar of different colors. The project with slow detection progress is identified, and warning information is sent to remind the project leader and the detection personnel. The task completion data and workload data of the detection personnel are obtained, the rationality of the detection task allocation is evaluated, and if the rationality of the detection task allocation is lower than the preset threshold, the detection task allocation adjustment suggestion is sent to the dispatcher for dynamic adjustment of the detection personnel.

7. A foundation quality safety intelligent detection system based on the foundation quality safety intelligent detection method according to any one of claims 1-6, characterized in that, The system includes the following modules: A foundation engineering template matching module is configured to match a historical engineering with the largest similarity according to the similarity of historical engineering information and engineering information of a current new engineering, and generate a foundation engineering template of the new engineering based on the matching result. A detection video acquisition module is configured to determine and access a camera according to a pile number selected by a user, identify a working state of the camera in a monitoring process based on real-time received video stream data, and acquire a detection video file of a detection pile. An abnormal event identification module is configured to identify an abnormal event in the video stream according to the detected video data, weather data and geological environment data, save the identified abnormal event image frame and mark the abnormal event image frame as an abnormal image frame; A pile position review module is configured to compare the pile number coordinate information of the base pile selected by the user with the preset pile position coordinate, determine whether the base pile position is correct, perform a base pile review operation and obtain a base pile review record; A pile position inspection and rectification notification module is configured to take a check photo or a check video with a watermark based on the selected inspection items and the pile number, identify the inspection items that need to be rectified according to the check result and send a rectification notification to the project member of the item that needs to be rectified; A detection data analysis module is configured to collect the field detection data according to a predefined data format, determine a foundation rock-soil property detection result and identify the defect position, defect type and defect degree of the base pile; A detection report generation module is configured to complete matching according to the detection data and the pile position information, establish a rendering model and generate a detection report, configure unit permissions and evaluate the rationality of task distribution.

Images