Elevator Shaft Data Modeling Method Based on UAV Detection

Abstract

The present invention discloses an elevator shaft data modeling method based on drone detection, which relates to the technical field of data modeling. The method includes: mounting a detection sensor group on the drone, acquiring the design drawing information of the target elevator shaft for flight path planning, and determining the drone detection path information; detecting and recording the target elevator shaft according to the drone detection path information by using the detection sensor group to obtain a multi-directional shaft detection point cloud data set, filtering and denoising and point cloud reconstruction are performed thereon to build a three-dimensional point cloud model of the elevator shaft; at the same time, a geological structure model is established to perform fusion analysis and correction and optimization on the three-dimensional point cloud model of the elevator shaft to obtain an optimized point cloud model of the elevator shaft. The technical effect of achieving all-round and efficient acquisition of elevator shaft data through drone detection, improving the comprehensiveness of detection data processing and the accuracy of data modeling, and thus ensuring the data modeling efficiency is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of data modeling, and in particular to a method for modeling elevator shaft data based on drone detection. Background Art

[0002] Elevator shaft data modeling plays an important role in elevator engineering and elevator maintenance management. It can not only improve the design level of elevators, ensure the safety and operation efficiency of elevators, but also optimize the layout and configuration of elevator systems, reducing errors and risks during installation and maintenance. In addition, through accurate data modeling, real-time monitoring and prediction of elevator performance can be achieved, providing strong support for preventive maintenance of elevators.

[0003] However, the existing elevator shaft data modeling methods rely on manual measurement, with low intelligence, and there are technical problems such as long measurement cycles and limited measurement accuracy. Summary of the Invention

[0004] The present application provides a method for modeling elevator shaft data based on drone detection, which solves the technical problems of low intelligence, long measurement cycles and limited measurement accuracy in the existing elevator shaft data modeling methods, and achieves the technical effect of comprehensively and efficiently collecting elevator shaft data through drone detection, improving the comprehensiveness of detected data processing and the accuracy of data modeling, and thus ensuring the data modeling efficiency.

[0005] In view of the above problems, the present invention provides a method for modeling elevator shaft data based on drone detection.

[0006] In a first aspect, the present application provides a method for modeling elevator shaft data based on drone detection, the method comprising: mounting a detection sensor group on the drone, the detection sensor group including a three-dimensional laser scanning device, an infrared imaging sensor, a laser ranging sensor and a vision sensor; obtaining the design drawing information of the target elevator shaft, planning a flight path based on the design drawing information, and determining the drone detection path information; detecting and recording the target elevator shaft according to the drone detection path information by using the detection sensor group to obtain a multi-directional shaft detection point cloud data set; performing filtering and denoising and point cloud reconstruction on the multi-directional shaft detection point cloud data set to build a three-dimensional point cloud model of the elevator shaft; collecting and obtaining geological structure belt information, performing feature analysis and three-dimensional modeling on the geological structure belt information to establish a geological structure model; and performing fusion analysis and correction optimization on the three-dimensional point cloud model of the elevator shaft based on the geological structure model to obtain an optimized point cloud model of the elevator shaft.

[0007] One or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0008] By using a detection sensor group carried on a drone and planning a flight path based on the design drawing information of the target elevator shaft to determine the drone detection path information, detecting and recording the target elevator shaft according to the drone detection path information by using the detection sensor group to obtain a multi-directional shaft detection point cloud data set, filtering and denoising and point cloud reconstruction are carried out based on this, a three-dimensional point cloud model of the elevator shaft is built, geological structure belt information is collected and analyzed for features and three-dimensional modeling is carried out to establish a geological structure model, and thus based on the geological structure model, fusion analysis and correction and optimization are carried out on the three-dimensional point cloud model of the elevator shaft to obtain an optimized point cloud model of the elevator shaft. Furthermore, the technical effect of achieving all-round and efficient acquisition of elevator shaft data through drone detection, improving the comprehensiveness of detection data processing and the accuracy of data modeling, and thus ensuring the data modeling efficiency is achieved.

[0009] The above description is only an overview of the technical solution of this application. In order to be able to understand the technical means of this application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features and advantages of this application more obvious and understandable, the specific embodiments of this application are specifically given below. Brief Description of the Drawings

[0010] Figure 1 It is a flow chart of the method for modeling elevator shaft data based on drone detection in this application;

[0011] Figure 2 It is a flow chart of obtaining a multi-directional shaft detection point cloud data set in the method for modeling elevator shaft data based on drone detection in this application. Detailed Description of the Embodiment

[0012] This application provides a method for modeling elevator shaft data based on drone detection, which solves the technical problems of low intelligence in the existing method for modeling elevator shaft data, long measurement period and limited measurement accuracy, and achieves the technical effect of all-round and efficient acquisition of elevator shaft data through drone detection, improving the comprehensiveness of detection data processing and the accuracy of data modeling, and thus ensuring the data modeling efficiency.

[0013] In order to make the purpose, technical solution and advantages of this application clearer, the following further details this application in combination with the drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain this application and are not used to limit this application.

[0014] The following describes this application in combination with the drawings in this application.

[0015] Embodiment 1, as Figure 1 shown, this application provides a method for modeling elevator shaft data based on drone detection, and the method includes:

[0016] Step S1: Mount a detection sensor group on the drone. The detection sensor group includes a 3D laser scanning device, an infrared imaging sensor, a laser ranging sensor, and a vision sensor.

[0017] Step S2: Obtain the design drawing information of the target elevator shaft, and perform flight path planning based on the design drawing information to determine the drone detection path information.

[0018] Specifically, with the rapid development of drone technology, using drones equipped with sensors for data detection has become an efficient and accurate detection method. To achieve the full-range and high-efficiency collection of elevator shaft data, first mount a detection sensor group on the drone. The detection sensor group includes a 3D laser scanning device for obtaining high-precision spatial data of the shaft; an infrared imaging sensor that can detect heat sources in the shaft, such as the working state of elevator equipment, as well as potential obstacles or safety hazards; a laser ranging sensor that provides accurate distance measurements, such as shaft dimensions, distance data, etc.; and a vision sensor for identifying structural features, obstacles, and the entrances and exits of the elevator shaft in the shaft, etc., so as to collect elevator shaft detection data in all aspects. Then obtain the design drawing information of the target elevator shaft through the elevator design database, including floor plans, sectional views, three-dimensional views, etc. Use CAD software or other relevant tools to analyze the design drawings, extract key information such as the dimensions, shape, and floor position of the shaft, and then use path planning algorithms, such as the A* algorithm, Dijkstra algorithm, etc., to perform flight path planning based on the key information of the design drawings, and at the same time set the maximum flight height and speed of the drone, etc., to ensure the safety of the drone during flight, and plan and determine the drone detection path information with the fastest detection efficiency to ensure that the detection sensors can cover the elevator shaft range and improve the comprehensiveness of data collection.

[0019] Step S3: Detect and record the target elevator shaft according to the drone detection path information using the detection sensor group to obtain a multi-directional shaft detection point cloud data set.

[0020] As Figure 2 shown, furthermore, for the obtaining of the multi-directional shaft detection point cloud data set, the steps of this application further include:

[0021] Detect and record the target elevator shaft according to the drone detection path information using the detection sensor group, and obtain a multi-dimensional data stream of shaft detection. The multi-dimensional data stream of shaft detection includes three-dimensional laser scanning data, infrared imaging data, laser ranging data, and visual image data. Construct a detection data integration processing channel, and perform mapping processing on the three-dimensional laser scanning data, infrared imaging data, laser ranging data, and visual image data based on the detection data integration processing channel to obtain integrated detection features of the elevator shaft. According to the detection requirements of the elevator shaft, determine the entity detection standard of the elevator shaft. Based on the entity detection standard of the elevator shaft, mark the area of interest of the integrated detection features of the elevator shaft to determine the target detection area of interest entity. Based on the target detection area of interest entity, use the detection sensor group to perform omnidirectional detection and recording of the target elevator shaft to obtain the multi-directional shaft detection point cloud data set.

[0022] Furthermore, for constructing the detection data integration processing channel, the steps of this application further include:

[0023] Collect and obtain a multi-dimensional detection database of the shaft, classify and label the multi-dimensional detection database of the shaft according to the type of detection sensor to obtain a shaft sensor detection data set. According to the type of detection sensor, determine the sensor data feature processing program. Based on the sensor data feature processing program, perform feature extraction on the shaft sensor detection data set respectively to obtain a key feature set of sensor data. Perform network integration training and evaluation optimization on the key feature set of sensor data to construct the detection data integration processing channel.

[0024] Furthermore, for constructing the detection data integration processing channel, the steps of this application further include:

[0025] Align the key feature set of sensor data according to the time series to obtain a feature sample set of sensor data, and perform data characteristic analysis on the feature sample set of sensor data to obtain sensor data characteristic information. Construct a list of deep learning networks, match the sensor data characteristic information with the list of deep learning networks to determine a set of matching deep learning networks. Based on the set of matching deep learning networks, perform distributed training on the feature sample set of sensor data to generate a set of multi-sensor feature analysis networks. Use the model integration strategy to perform model integration training and evaluation optimization on the set of multi-sensor feature analysis networks to construct the detection data integration processing channel.

[0026] Specifically, according to the pre-planned drone detection path information, control the detection sensor group carried by the drone to detect and record the target elevator shaft. During the detection process, the multi-dimensional data stream of shaft detection is collected in real time through the detection sensor group. The multi-dimensional data stream of shaft detection includes: shaft cross-section width, shaft cross-section depth, top floor height, pit depth, reserved width of door opening, reserved height of door opening, width of left and right door piers, size of door header beam, size of sill beam, size of ring beam, pitch of ring beam, protrusion and indentation of structural columns in the shaft, overall perpendicularity of each wall surface of the shaft, holes in the shaft roof, hook identification, and other clear height data of the elevator shaft. The sensor source types of the collected elevator shaft data include: three-dimensional laser scanning data, infrared imaging data, laser ranging data, and visual image data, etc.

[0027] To improve the efficiency of detection data processing, it is necessary to construct a detection data integration processing channel, which is used to process the multi-dimensional data stream of shaft detection from different sensor source types in parallel and efficiently. The specific construction process is as follows: First, collect and obtain a multi-dimensional shaft detection database through big data technology. The multi-dimensional shaft detection database is a historical shaft sensor detection data set, including sensor detection source data and corresponding detection feature data. Classify and label the multi-dimensional shaft detection database according to the detection sensor type to obtain a shaft sensor detection data set, which includes three-dimensional laser scanning data, infrared imaging data, laser ranging data, and visual image data, etc. For each detection sensor type, determine a sensor data feature processing program. Exemplarily, the three-dimensional laser scanning data feature processing program includes denoising, normalization, and principal component analysis, etc., to extract the key features of the shaft data. Based on the sensor data feature processing program, perform feature extraction on the shaft sensor detection data set respectively to obtain the corresponding key feature sets of sensor data, such as shaft cross-section ranging features, wall defect infrared features, etc.

[0028] Then, perform network integration training and evaluation optimization on the key feature set of the sensor data. Among the multi-sensor data features, each sensor may collect data at different time points. To perform effective data analysis and pattern recognition, it is necessary to ensure that the data of all sensors are consistent in time. Therefore, align the key feature set of the sensor data according to the time series to obtain an integrated sensor data feature sample set, which includes sensor key feature data and corresponding detection feature data. Analyze the data characteristics of the sensor data feature sample set to obtain sensor data characteristic information, including the statistical properties, distribution characteristics, correlation, etc. of the data, as the basis for subsequent selection of appropriate deep learning models and parameters. Construct a list of deep learning networks, which includes various different types of network architectures (such as convolutional neural network CNN, long short-term memory network LSTM, support vector machine, etc.), and each architecture is suitable for processing different types of sensor data characteristics. Match the sensor data characteristic information with the deep learning network list to determine the matching deep learning network set that is most suitable for the current data characteristics. Based on the matching deep learning network set, perform distributed simultaneous training on the sensor data feature sample set to generate a multi-sensor feature analysis network set, which is used to extract key features and analyze feature results of corresponding sensor types according to the wellbore multi-dimensional detection data, such as wellbore cross-sectional dimension features, wall cavity features, structural column protrusions, wellbore distance features, etc. Adopt a model integration strategy to perform model integration training on the multi-sensor feature analysis network set. For example, increase the weight of the analysis network with higher accuracy in the integrated model to fuse the multi-sensor feature analysis network set, and evaluate and optimize the fused integrated model to obtain a sensor feature integrated analysis network with qualified model performance for integrated analysis of multi-sensor data features. Furthermore, embed the sensor feature integrated analysis network into the integrated processing channel of the detection data. This channel can receive data inputs from multiple sensors and perform analysis and processing through the network set to generate useful output feature information, such as wellbore target detection, wellbore dimension features, and obstacle recognition, etc.

[0029] Based on the integrated processing channel of the detection data, map the three-dimensional laser scanning data, infrared imaging data, laser ranging data, and visual image data to obtain the integrated detection features of the elevator shaft. The integrated detection features of the elevator shaft are the key feature information of the elevator shaft obtained according to the current detection data, including shaft dimensions, physical targets, etc. According to the detection requirements of the elevator shaft, determine the physical detection standards for the elevator shaft. The physical detection standards for the elevator shaft are the physical information of the elevator shaft to be detected that meets the modeling requirements, such as shaft cross-section, door opening, door pier, roof, hook, ring beam (span), etc. Based on the physical detection standards for the elevator shaft, mark the areas of concern for the entities in the integrated detection features of the elevator shaft that meet the detection standards, and determine the target detection area of concern for entities, such as the four walls, top, bottom, door opening, and track installation area of the shaft. Based on the target detection area of concern for entities, use the detection sensor group to conduct a full-range detection record of the target elevator shaft, adjust the position and angle of the sensors to ensure that these areas are fully covered and recorded, and thus collect a multi-directional shaft detection point cloud data set. These data sets include three-dimensional point clouds, infrared images, distance measurement values, and visual images, etc. Achieve the full-range and efficient collection of elevator shaft data through drone detection, improve the comprehensiveness of the detection data processing, and further improve the subsequent data modeling accuracy.

[0030] Furthermore, for obtaining the multi-directional shaft detection point cloud data set, the steps of this application further include:

[0031] Based on the target detection area of concern for entities, perform additional correction on the drone detection path information to obtain the corrected drone detection path information; perform obstacle recognition and obstacle avoidance adjustment and optimization on the corrected drone detection path information to determine the optimized drone detection path information; based on the optimized drone detection path information and the detection sensor group, conduct a full-range detection record of the target elevator shaft to obtain the multi-directional shaft detection point cloud data set.

[0032] Specifically, based on the previously determined target detection and attention entity areas (such as key areas like the four walls, top, bottom, and door openings of the shaft), the original drone detection path information is additionally corrected. The correction may include adjusting the flight altitude, changing the flight direction, or increasing the density of detection points to ensure that these key areas are fully covered and detected in detail. On the corrected drone detection path, an obstacle recognition sensor group is used for real-time obstacle recognition. Once an obstacle (such as equipment, pipelines, maintenance platforms, etc. in the shaft) is recognized, an obstacle avoidance adjustment is immediately made, which may include changing the flight trajectory, raising or lowering the flight altitude, or even pausing the flight to avoid collisions. Through algorithm optimization, such as the A* algorithm, Dijkstra algorithm, or reinforcement learning, etc., the best obstacle avoidance path is searched to ensure that the drone can continue the detection task safely and efficiently. Combining the results of obstacle recognition and obstacle avoidance adjustment, the optimized path information for drone detection is determined. This path should ensure full coverage of the target detection and attention entity areas while avoiding collisions with any obstacles in the shaft. Based on the optimized drone detection path information and the detection sensor group, a comprehensive detection record of the target elevator shaft is carried out. The drone flies and detects according to the optimized drone detection path information, and at the same time, uses the sensor group to collect three-dimensional laser scanning data, infrared imaging data, laser ranging data, and visual image data in the shaft in real time, including key elevator shaft data such as the width and depth of the shaft, the span of the ring beams, the width and height of the door opening, the width of the left and right door piers, and the condition of the header beam. These data will be integrated into a multi-directional shaft detection point cloud data set to provide a basis for subsequent analysis and processing. Ensure that the drone can safely and efficiently cover the target detection and attention entity areas during the detection process and collect comprehensive and accurate shaft detection data.

[0033] Step S4: Filter and denoise the multi-directional shaft detection point cloud data set and reconstruct the point cloud to build a three-dimensional point cloud model of the elevator shaft.

[0034] Furthermore, for building the three-dimensional point cloud model of the elevator shaft, the steps of this application also include:

[0035] Filter and denoise the multi-directional shaft detection point cloud data set to obtain an available multi-directional shaft detection point cloud data set; use a triangulation reconstruction algorithm to register, align, and triangulate the available multi-directional shaft detection point cloud data set to obtain a shaft triangular mesh model; perform edge detection and model fitting training on the shaft triangular mesh model to determine panoramic modeling fitting parameters; map the panoramic modeling fitting parameters onto the shaft triangular mesh model for point cloud model reconstruction to build the three-dimensional point cloud model of the elevator shaft.

[0036] Furthermore, for obtaining the available multi-directional shaft detection point cloud data set, the steps of this application also include:

[0037] Extract the noise characteristics of the multi-directional wellbore detection point cloud dataset to obtain point cloud noise characteristic information, and determine the wavelet selection threshold according to the point cloud noise characteristic information; perform multi-scale decomposition on the multi-directional wellbore detection point cloud dataset to obtain a point cloud multi-scale wavelet coefficient set; screen, filter and reconstruct the point cloud multi-scale wavelet coefficient set according to the wavelet selection threshold to obtain the available multi-directional wellbore detection point cloud dataset.

[0038] Specifically, perform filtering and denoising and point cloud reconstruction on the multi-directional wellbore detection point cloud dataset. To ensure the quality of the point cloud data, first perform filtering and denoising on the multi-directional wellbore detection point cloud dataset, including extracting the noise characteristics of the multi-directional wellbore detection point cloud dataset, which usually involves analyzing outliers, off-points and high-frequency noise in the point cloud data, etc., to obtain the corresponding point cloud noise characteristic information. And according to the point cloud noise characteristic information, use the soft threshold or hard threshold method to determine the retention threshold of the wavelet coefficients, that is, determine the wavelet selection threshold. Selecting an appropriate threshold can retain important geometric features while removing noise. Perform multi-scale decomposition on the multi-directional wellbore detection point cloud dataset to obtain a point cloud multi-scale wavelet coefficient set at different scales. Screen, filter and reconstruct the point cloud multi-scale wavelet coefficient set according to the wavelet selection threshold, traverse the wavelet coefficient sets at each scale, set the coefficients below the threshold to zero (hard threshold) or perform shrinkage (soft threshold), and perform inverse wavelet transform on the filtered wavelet coefficients to reconstruct the point cloud data to obtain the available multi-directional wellbore detection point cloud dataset after filtering and denoising.

[0039] The triangulation reconstruction algorithm is used to register and align and triangulate and reconstruct the available multi - azimuth wellbore detection point cloud dataset. First, according to the characteristics and requirements of the point cloud data, a suitable registration algorithm is selected, such as the ICP (Iterative Closest Point) algorithm, the NDT (Normal Distribution Transform) algorithm, etc. The selected registration algorithm is applied to spatially align the multi - azimuth wellbore detection point cloud dataset. Check the registered point cloud data to ensure that they are in the same coordinate system and there is no obvious misalignment. According to the characteristics of the point cloud data, such as density, noise level, etc., a suitable triangulation algorithm is selected, such as Delaunay triangulation, Poisson surface reconstruction, etc. The selected triangulation algorithm is applied to convert the registered and aligned point cloud data into a wellbore triangular mesh model to represent the geometric shape of the wellbore. An edge detection algorithm is used to extract the edge features of the wellbore from the wellbore triangular mesh model. The detected edge features are processed to obtain key geometric features such as corner points, straight line segments, etc. Then, the extracted features are used for model fitting training, such as using the least - squares method to fit geometric shapes such as planes, cylinders, etc., to determine the panoramic modeling fitting parameters, such as plane parameters, cylinder parameters, etc. The panoramic modeling fitting parameters are mapped onto the wellbore triangular mesh model for point cloud model reconstruction. According to the mapped parameters, new point cloud data is generated on the mesh model to represent the structural dimensions of the wellbore. The generated point cloud model is optimized, such as removing redundant points, smoothing the surface, etc. The optimized point cloud model is integrated into a complete three - dimensional point cloud model of the elevator wellbore, that is, the three - dimensional point cloud model of the elevator wellbore is built. The efficient and accurate processing of the multi - azimuth wellbore detection point cloud dataset is realized, thereby improving the data modeling accuracy and modeling efficiency.

[0040] Step S5: Collect and obtain geological structure zone information, conduct feature analysis and three - dimensional modeling on the geological structure zone information, and establish a geological structure model.

[0041] Specifically, the acquisition of geological structure belt information is the basis for establishing a geological structure model, which usually involves geological exploration, geological surveying, and geological investigation processes, aiming to obtain detailed data of the geological structure belt. Such data may include the location, shape, properties of geological bodies, and the intersection relationships of geological interfaces. After obtaining the geological structure belt information, it is necessary to conduct feature analysis to extract key geological structure features, which usually involves the drawing of geological cross-sections, the analysis of geological structures, and the analysis of geological properties. Based on the feature analysis, three-dimensional geological modeling technology is used to perform three-dimensional modeling of the geological structure model and establish a three-dimensional model of the geological interface. Identification and fitting are carried out according to the three-dimensional model of the geological interface, and the internal structure of the geological body is described to generate a closed geological body. The generated geological structure model is optimized, such as smoothing the surface and removing redundant points, to obtain an optimized geological structure model, so as to improve the accuracy and visualization effect of the model. The geological structure model can intuitively display the spatial distribution and geometric shape of the geological structure belt, so as to improve the accuracy and practicality of subsequent elevator shaft modeling.

[0042] Step S6: Based on the geological structure model, perform fusion analysis and correction optimization on the three-dimensional point cloud model of the elevator shaft to obtain an optimized point cloud model of the elevator shaft.

[0043] Furthermore, for obtaining the optimized point cloud model of the elevator shaft, the steps of this application further include:

[0044] Align and match the geological structure model with the three-dimensional point cloud model of the elevator shaft to obtain a fused point cloud model of the elevator shaft; verify and evaluate the fused point cloud model of the elevator shaft to obtain model performance evaluation parameters, and perform iterative correction and optimization on the fused point cloud model of the elevator shaft based on the model performance evaluation parameters to obtain the optimized point cloud model of the elevator shaft.

[0045] Specifically, for performing fusion analysis and correction optimization on the three-dimensional point cloud model of the elevator shaft based on the geological structure model, first align and match the geological structure model with the three-dimensional point cloud model of the elevator shaft to ensure that the data formats, coordinate systems, and scales of the geological structure model and the three-dimensional point cloud model of the elevator shaft are consistent. The surface of the geological structure model is preliminarily aligned with the corresponding part of the three-dimensional point cloud model of the elevator shaft through a point cloud registration algorithm. On the basis of the preliminary alignment, according to the actual situation of the geological structure and the design requirements of the elevator shaft, the model is finely adjusted to ensure precise alignment at key positions. Then, the aligned geological structure model and the three-dimensional point cloud model of the elevator shaft are fused, including operations such as point cloud merging, duplicate removal, and smoothing, to obtain the corresponding fused point cloud model of the elevator shaft. Verify and evaluate the fused point cloud model of the elevator shaft to obtain model performance evaluation parameters, such as accuracy, integrity, and consistency, for evaluating the performance of the fused point cloud model.

[0046] Based on the model performance evaluation parameters, iterative correction and optimization are carried out on the elevator shaft integrated point cloud model. First, according to the model performance evaluation parameters, problems existing in the integrated point cloud model are identified, such as data missing, insufficient accuracy, etc. Then, aiming at the identified problems, correction and optimization are carried out on the integrated point cloud model, including adding data points, adjusting the model shape, smoothing the surface, etc. The corrected and optimized model is verified and evaluated again to ensure that the model performance is improved. If there are still deficiencies, the iterative correction and optimization process continues. After multiple iterations of correction and optimization, the optimized point cloud model of the elevator shaft is finally obtained. This model is not only accurately aligned with the geological structure model but also has high accuracy and integrity. It realizes the intelligent modeling and model iterative optimization of elevator shaft data, improving the accuracy and efficiency of data modeling.

[0047] In summary, the method for modeling elevator shaft data based on drone detection provided by this application has the following technical effects:

[0048] Since a detection sensor group is carried on the drone, and the flight path is planned based on the design drawing information of the target elevator shaft to determine the drone detection path information, the target elevator shaft is detected and recorded by using the detection sensor group according to the drone detection path information to obtain a multi-directional shaft detection point cloud data set. Then, filtering and denoising and point cloud reconstruction are carried out based on this to build a three-dimensional point cloud model of the elevator shaft. The geological structure belt information is collected and analyzed for features and three-dimensional modeling to establish a geological structure model. Thus, based on the geological structure model, fusion analysis and correction optimization are carried out on the three-dimensional point cloud model of the elevator shaft to obtain the optimized point cloud model of the elevator shaft. Furthermore, the technical effect of achieving all-round and efficient collection of elevator shaft data through drone detection, improving the comprehensiveness of detection data processing and the accuracy of data modeling, and thus ensuring the data modeling efficiency is achieved.

[0049] This specification and the drawings are only exemplary descriptions of this application, but the protection scope of this application is not limited thereto. It should be noted that any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in this application, and all should be covered by the protection scope of this application. In some cases, the actions or steps recorded in this application can be executed in a different order from that in the embodiments and still achieve the desired results. Additionally, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In certain embodiments, multi-tasking and parallel processing are also possible or may be advantageous.

Claims

1. The elevator shaft data modeling method based on drone detection is characterized by: The method comprises: A detection sensor group is mounted on the drone, wherein the detection sensor group includes a three-dimensional laser scanning device, an infrared imaging sensor, a laser ranging sensor, and a visual sensor; Obtain design drawing information of the target elevator shaft, perform flight path planning based on the design drawing information, and determine the detection path information of the drone; According to the detection path information of the drone, the detection sensor group is used to detect and record the target elevator shaft, so as to obtain a multi-directional shaft detection point cloud data set; Perform filtering, denoising and point cloud reconstruction on the multi-directional shaft detection point cloud data set to build a three-dimensional point cloud model of the elevator shaft; Collecting and acquiring geological structure belt information, performing feature analysis and three-dimensional modeling on the geological structure belt information, and establishing a geological structure model; Based on the geological structure model, the three-dimensional point cloud model of the elevator shaft is subjected to fusion analysis and correction optimization to obtain an optimized point cloud model of the elevator shaft; The construction of the three-dimensional point cloud model of the elevator shaft includes: Filtering and denoising the multi-directional shaft detection point cloud data set to obtain a usable multi-directional shaft detection point cloud data set; Using a triangulated reconstruction algorithm to perform registration, alignment and triangulation reconstruction on the available multi-directional shaft detection point cloud data set to obtain a shaft triangular mesh model; Perform edge detection and model fitting training based on the hoistway triangle mesh model to determine panoramic modeling fitting parameters; Mapping the panoramic modeling fitting parameters to the hoistway triangular mesh model to reconstruct the point cloud model, and building the elevator hoistway three-dimensional point cloud model; The step of obtaining the optimized point cloud model of the elevator shaft comprises: Aligning and matching the geological structure model with the elevator shaft three-dimensional point cloud model to obtain an elevator shaft fused point cloud model; The elevator shaft fusion point cloud model is verified and evaluated to obtain model performance evaluation parameters, and the elevator shaft fusion point cloud model is iteratively corrected and optimized based on the model performance evaluation parameters to obtain the elevator shaft optimized point cloud model.

2. The elevator shaft data modeling method based on drone detection according to claim 1 is characterized in that: The multi-directional well detection point cloud data set is obtained, including: According to the detection path information of the drone, the detection sensor group is used to detect and record the target elevator shaft, and a shaft detection multi-dimensional data stream is obtained, wherein the shaft detection multi-dimensional data stream includes three-dimensional laser scanning data, infrared imaging data, laser ranging data, and visual image data; Constructing a detection data integrated processing channel, and mapping and processing the three-dimensional laser scanning data, infrared imaging data, laser ranging data, and visual image data based on the detection data integrated processing channel to obtain the elevator shaft integrated detection features; Determine the physical detection standard of the elevator shaft according to the detection requirements of the elevator shaft; Marking the area of ​​interest for the elevator shaft integrated detection feature based on the elevator shaft entity detection standard to determine the target detection entity area of ​​interest; Based on the target detection focus entity area, the detection sensor group is used to perform all-round detection and recording of the target elevator shaft to obtain the multi-directional shaft detection point cloud data set.

3. The elevator shaft data modeling method based on drone detection according to claim 2 is characterized in that: The construction of the detection data integration processing channel includes: Acquire a multi-dimensional well detection database, classify and identify the multi-dimensional well detection database according to the detection sensor type, and obtain a well sensor detection data set; Determining a sensor data feature processing procedure according to the detection sensor type; Based on the sensor data feature processing program, feature extraction is performed on the shaft sensor detection data set to obtain a key feature set of sensor data; Network integration training and evaluation optimization are performed on the key feature set of the sensor data to build the detection data integration processing channel.

4. The elevator shaft data modeling method based on drone detection according to claim 3 is characterized in that: The step of constructing the detection data integration processing channel comprises: Aligning the key feature set of the sensor data according to a time series to obtain a sensor data feature sample set, and performing data characteristic analysis on the sensor data feature sample set to obtain sensor data characteristic information; Constructing a deep learning network list, matching the sensor data characteristic information with the deep learning network list, and determining a matching deep learning network set; Performing distributed training on the sensor data feature sample set based on the matching deep learning network set to generate a multi-sensor feature analysis network set; A model integration strategy is adopted to perform model integration training and evaluation optimization on the multi-sensor feature analysis network set, and to construct the detection data integration processing channel.

5. The elevator shaft data modeling method based on drone detection according to claim 3 is characterized in that: The step of obtaining the multi-directional wellbore detection point cloud data set includes: Based on the target detection entity area of ​​interest, the drone detection path information is additionally corrected to obtain the drone detection corrected path information; Performing obstacle identification and obstacle avoidance adjustment optimization on the drone detection correction path information to determine the drone detection optimization path information; Based on the drone detection optimization path information and the detection sensor group, the target elevator shaft is detected and recorded in all directions to obtain the multi-directional shaft detection point cloud data set.

6. The elevator shaft data modeling method based on drone detection according to claim 1 is characterized in that: The method of obtaining a usable multi-directional well detection point cloud data set includes: Extracting noise characteristics from the multi-directional wellbore detection point cloud data set to obtain point cloud noise characteristic information, and determining a wavelet selection threshold according to the point cloud noise characteristic information; Performing multi-scale decomposition on the multi-directional wellbore detection point cloud data set to obtain a point cloud multi-scale wavelet coefficient set; The point cloud multi-scale wavelet coefficient set is screened, filtered and reconstructed according to the wavelet selection threshold to obtain the available multi-directional wellbore detection point cloud data set.

Images