CN-122023672-A - Reservoir side slope cushion layer intelligent detection system and method based on three-dimensional laser scanning

CN122023672ACN 122023672 ACN122023672 ACN 122023672ACN-122023672-A

Abstract

The application provides an intelligent detection system and method for a reservoir slope cushion layer based on three-dimensional laser scanning, and relates to the technical field of hydraulic engineering detection, wherein the method comprises the steps of acquiring high-precision point cloud data of the slope cushion layer through three-dimensional laser scanning equipment; the method comprises the steps of preprocessing high-precision point cloud data to construct a three-dimensional point cloud model, extracting a target point cloud data set corresponding to a slope cushion distribution area from the high-precision point cloud data, amplifying the target point cloud data set to generate an enhanced point cloud sequence, carrying out plane fitting analysis on the enhanced point cloud sequence, calculating fitting planes of all local areas on the surface of the slope cushion, finally calculating thickness deviation data, flatness deviation data and gradient deviation data of the slope cushion based on spatial position relation between the fitting planes and all data points in the enhanced point cloud sequence, and generating a comprehensive quality detection result based on a calculation result. The application improves the precision and the automation level of the construction quality detection of the reservoir slope cushion layer.

Inventors

QU LIXIN
LIU SHUAI
WANG YUN
Yi Tianqi
WANG JIEHUI
QIAN YACHAO

Assignees

山东省水利勘测设计院有限公司
元宇智数(深圳)科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260409

Claims (8)

1. The intelligent detection method for the reservoir side slope cushion layer based on three-dimensional laser scanning is characterized by comprising the following steps of: non-contact scanning is carried out on the surface of the slope cushion layer of the reservoir through three-dimensional laser scanning equipment, and high-precision point cloud data of the slope cushion layer are obtained; preprocessing the high-precision point cloud data to construct a three-dimensional point cloud model, and automatically identifying and extracting a target point cloud data set corresponding to a slope cushion distribution area from the three-dimensional point cloud model by utilizing a characteristic identification algorithm; amplifying the target point cloud data set by utilizing a condition generation countermeasure network to generate an enhanced point cloud sequence; carrying out plane fitting analysis on the enhanced point cloud sequence by utilizing a least square plane fitting algorithm so as to calculate fitting planes of all local areas on the surface of the slope cushion layer; and calculating thickness deviation data, flatness deviation data and gradient deviation data of the slope cushion layer based on the spatial position relation between the fitting plane and each data point in the enhanced point cloud sequence, and generating a comprehensive quality detection result based on a calculation result.
2. The method of claim 1, wherein the calculating thickness deviation data, flatness deviation data, and gradient deviation data of the slope mat based on the spatial positional relationship between the fitting plane and each data point in the sequence of enhanced point clouds, and generating a composite quality detection result based on the calculation result, comprises: For each group of data in the enhanced point cloud sequence, calculating the vertical distance from each data point in each group of data to the corresponding fitting plane, screening the vertical distance of each group of data by using an outlier detection algorithm, and taking the standard deviation of all the screened vertical distances as flatness deviation data; for each grid unit, calculating an elevation average value of corresponding data in the enhanced point cloud sequence, and comparing the elevation average value with a design elevation to obtain thickness deviation data; calculating an included angle between the main slope direction represented by the enhanced point cloud sequence and the design direction as gradient deviation data; Comparing the flatness deviation data, the thickness deviation data and the gradient deviation data with a preset allowable deviation range respectively, and identifying and recording grid cells and deviation values exceeding the allowable deviation range; And according to the identification result, combining the fuzzy comprehensive evaluation model to obtain the quality grade score of each grid unit, and generating a comprehensive quality detection report according to the quality grade score.
3. The method of claim 2, wherein the obtaining a quality class score for each grid cell based on the recognition result in combination with the fuzzy comprehensive evaluation model, and generating a comprehensive quality detection report based on the quality class scores, comprises: taking the flatness deviation data, the thickness deviation data and the gradient deviation data as evaluation indexes, respectively calculating the membership degree of each evaluation index of each grid unit in the identification result to a plurality of preset quality grades according to a preset membership degree function, and distributing preset weight coefficients for each evaluation index; inputting the membership degree of each evaluation index and the corresponding weight coefficient into a fuzzy comprehensive evaluation model, calculating the membership degree and the weight coefficient according to a fuzzy synthesis rule by the fuzzy comprehensive evaluation model, and outputting the comprehensive evaluation result of each grid unit; according to the comprehensive evaluation result, determining the quality grade with highest possibility as the quality grade score of the grid unit; and generating a comprehensive quality detection report based on the quality grade scores and the spatial position information of all grid cells.
4. The method of claim 1, wherein performing a plane fitting analysis on the sequence of enhanced point clouds using a least squares plane fitting algorithm to calculate a fitting plane for each local area of the surface of the slope mat comprises: dividing the horizontal projection area of the enhanced point cloud sequence into a plurality of grid units based on the construction quality specification of the slope cushion layer; for each grid cell, extracting all data points located in the vertical projection space of the grid cell to form a cell data point set; and calculating each unit data point set by applying a least square plane fitting algorithm to obtain a reference plane corresponding to the grid unit, and taking the reference plane as a fitting plane corresponding to the grid unit.
5. The method of claim 1, wherein the generating an countermeasure network using conditions, augmenting the target point cloud dataset, generating an enhanced point cloud sequence, comprises: acquiring material parameters and environmental parameters of a slope cushion layer; combining the target point cloud data set, the material parameter and the environmental parameter into condition data; inputting the condition data into a generator of a condition generation countermeasure network to obtain simulated point cloud data; Introducing a cosine similarity algorithm, and calculating the matching degree between the distribution feature vector of the simulated point cloud data and the distribution feature vector of the target point cloud data set; Generating a discriminator of an countermeasure network through the condition, discriminating whether the simulated point cloud data accords with the distribution characteristics of the target point cloud data set or not in an auxiliary mode according to the matching degree, and iteratively optimizing the generation parameters of the generator through a back propagation mechanism based on a judging result; according to preset various working conditions, circularly generating a plurality of groups of simulated point cloud data by utilizing the optimized generator; and merging the target point cloud data set with the plurality of groups of simulated point cloud data, and arranging according to the working condition sequence to form an enhanced point cloud sequence.
6. The method according to claim 1, wherein the preprocessing the high-precision point cloud data to construct a three-dimensional point cloud model, and automatically identifying and extracting a target point cloud data set corresponding to a slope pad distribution area from the three-dimensional point cloud model by using a feature identification algorithm, includes: based on the design axis of the reservoir slope, carrying out coordinate registration on the high-precision point cloud data acquired by different scanning sites to form an initial point cloud model; processing the initial point cloud model by adopting a spatial clustering algorithm, identifying noise data points, and removing the noise data points from the initial point cloud model to obtain a three-dimensional point cloud model; Mapping a design boundary line of the slope cushion layer to a space position corresponding to the three-dimensional point cloud model, and defining a design boundary area; Calculating the normal direction of each point in the three-dimensional point cloud model in the design boundary area by utilizing a feature recognition algorithm, and screening out points with deviation of the normal direction from the normal direction of the surface of the design cushion layer within a preset allowable range to form a candidate point cloud set; and carrying out morphological closing operation processing on the candidate point cloud set to generate a target point cloud data set.
7. The method of claim 1, further comprising, after generating the composite quality detection result based on the calculation result: Performing global anomaly detection on each item of deviation data in the comprehensive quality detection result by adopting an isolated forest algorithm, and identifying an anomaly detection unit deviating from a normal construction mode on the combination characteristics of flatness, thickness and gradient deviation; inputting the space coordinates of the anomaly detection unit and corresponding multidimensional deviation data into a long-term and short-term memory network prediction model, and predicting the quality degradation trend of the anomaly detection unit in a future designated time window; And generating a quality early warning report based on the prediction result of the quality degradation trend, wherein the quality early warning report comprises abnormal unit positioning, current deviation condition and future risk level.
8. Reservoir side slope bed course intelligent detection system based on three-dimensional laser scanning, its characterized in that includes: The acquisition module is used for carrying out non-contact scanning on the surface of the slope cushion layer of the reservoir through the three-dimensional laser scanning equipment to acquire high-precision point cloud data of the slope cushion layer; The construction module is used for preprocessing the high-precision point cloud data to construct a three-dimensional point cloud model, and automatically identifying and extracting a target point cloud data set corresponding to a slope cushion distribution area from the three-dimensional point cloud model by utilizing a characteristic identification algorithm; the generation module is used for generating an countermeasure network by utilizing conditions, amplifying the target point cloud data set and generating an enhanced point cloud sequence; the analysis module is used for carrying out plane fitting analysis on the enhanced point cloud sequence by utilizing a least square plane fitting algorithm so as to calculate the fitting plane of each local area on the surface of the slope cushion layer; And the calculation module is used for calculating thickness deviation data, flatness deviation data and gradient deviation data of the slope cushion layer based on the spatial position relation between the fitting plane and each data point in the enhanced point cloud sequence, and generating a comprehensive quality detection result based on a calculation result.

Description

Reservoir side slope cushion layer intelligent detection system and method based on three-dimensional laser scanning Technical Field The application relates to the technical field of hydraulic engineering detection, in particular to an intelligent detection system and method for reservoir side slope cushion layers based on three-dimensional laser scanning. Background The reservoir side slope cushion layer is used as a key component of a dam body seepage-proofing system, the construction quality of the reservoir side slope cushion layer directly influences the operation safety and the service life of the reservoir, and the traditional manual detection means is difficult to meet the requirements of large-scale hydraulic engineering on detection efficiency and precision. With the development of the three-dimensional laser scanning technology, the technology has wide application prospect in the field of hydraulic engineering quality detection by virtue of the advantages of non-contact, high precision and high efficiency. The current reservoir slope cushion layer detection method based on three-dimensional laser scanning generally comprises the steps of acquiring point cloud data of the cushion layer surface through scanning equipment, performing splicing and filtering processing on the acquired point cloud data to construct a three-dimensional model reflecting the cushion layer surface morphology, and performing contrast analysis on the constructed three-dimensional model and a design reference so as to acquire detection data of geometric parameters such as cushion layer thickness, flatness, gradient and the like. The method can realize rapid evaluation of the construction quality of the cushion layer and can be initially applied to partial hydraulic engineering. However, in an actual engineering environment, the point cloud data collected in the scanning process often has the problem of sparse data or partial missing data due to factors such as environmental illumination, climatic conditions, construction interference and the like, so that the constructed three-dimensional model is difficult to completely present the real geometric form of the surface of the cushion layer. Deviation exists in thickness, flatness and gradient parameters calculated based on the incomplete data, so that a final detection result cannot accurately reflect the actual construction quality of the cushion layer. Therefore, the technical problem of insufficient accuracy of detection results caused by incomplete detection data exists in the prior art. Disclosure of Invention The application provides an intelligent detection system and method for reservoir slope cushions based on three-dimensional laser scanning, which are used for solving the problems of low precision and automatic level difference of reservoir slope cushion construction quality detection in the prior art. In order to solve the technical problems, in a first aspect, the application provides an intelligent detection method for a reservoir slope cushion layer based on three-dimensional laser scanning, which comprises the following steps: non-contact scanning is carried out on the surface of the slope cushion layer of the reservoir through three-dimensional laser scanning equipment, and high-precision point cloud data of the slope cushion layer are obtained; preprocessing the high-precision point cloud data to construct a three-dimensional point cloud model, and automatically identifying and extracting a target point cloud data set corresponding to a slope cushion distribution area from the three-dimensional point cloud model by utilizing a characteristic identification algorithm; amplifying the target point cloud data set by utilizing a condition generation countermeasure network to generate an enhanced point cloud sequence; carrying out plane fitting analysis on the enhanced point cloud sequence by utilizing a least square plane fitting algorithm so as to calculate fitting planes of all local areas on the surface of the slope cushion layer; and calculating thickness deviation data, flatness deviation data and gradient deviation data of the slope cushion layer based on the spatial position relation between the fitting plane and each data point in the enhanced point cloud sequence, and generating a comprehensive quality detection result based on a calculation result. Optionally, the calculating thickness deviation data, flatness deviation data and gradient deviation data of the slope pad layer based on the spatial position relationship between the fitting plane and each data point in the enhanced point cloud sequence, and generating a comprehensive quality detection result based on the calculation result includes: For each group of data in the enhanced point cloud sequence, calculating the vertical distance from each data point in each group of data to the corresponding fitting plane, screening the vertical distance of each group of data by using an outlier detection algori