Unmanned aerial vehicle high-precision target deformation monitoring method

Abstract

The invention discloses a high-precision target deformation monitoring method for an unmanned aerial vehicle. According to the invention, an image identification target is arranged on a monitoring target, and according to an RTK and PPK real-time differential dynamic positioning technology, cloud data acquisition of different dimensions and angles of laser point cloud is carried out on the image identification target through an unmanned aerial vehicle laser radar; and normal distribution screening is carried out on different-dimension measurement cloud data of the image recognition target so as to obtain a high-precision value of a monitoring coordinate of the image recognition target, and finally the purpose of high-precision deformation monitoring is achieved. Real-time differential dynamic positioning and laser radar high-precision technologies are utilized, and high-precision, fast and efficient deformation safety monitoring of the structure is realized through the high-speed maneuvering capability of the unmanned aerial vehicle and a cloud data normal screening post-processing method. The method is simple and reliable in principle and clear in treatment process, meets the current specification requirements, and is particularly suitable for deformation safety monitoring of seawall and dike long-distance structures.

Description

Technical Field

[0001] The present invention relates to the field of deformation monitoring, and in particular to a method for high-precision deformation monitoring of a target of an unmanned aerial vehicle. Background Art

[0002] With the rapid development of drones and the digital age, drone-based RTK and PPK real-time differential dynamic positioning technology has been widely used in surveying and mapping projects. Compared with traditional measurement methods, it has many advantages such as high timeliness, low cost, low risk, and high mobility. However, due to its measurement accuracy, drones are difficult to apply in the field of "high-precision" deformation monitoring.

[0003] The field of engineering deformation monitoring has strict regulatory requirements for the layout, measurement accuracy, and frequency of deformation monitoring points, and sets fixed deformation monitoring points to reflect the overall deformation of the structure. Especially for long-distance structures (embankments, seawalls), the traditional manual deformation monitoring method is difficult to arrange the benchmark points, resulting in cumbersome measurements and low precision, while the conventional UAV and RTK measurement accuracy is low. Although the fixed static RTK monitoring technology has improved the measurement accuracy, the cost is relatively high, and it is difficult to apply it on a large scale in long-distance structures (embankments, seawalls). In short, at this stage, traditional measurement technology, static RTK, and conventional UAV measurement are difficult to meet the development requirements of the new era of low-cost, high-precision, fast, and digital deformation monitoring. Summary of the invention

[0004] In order to solve the problems existing in the prior art, the present invention provides a high-precision target deformation monitoring method for UAV, which can realize efficient, high-precision and low-cost target deformation monitoring of UAV, and greatly improve the efficiency and accuracy of deformation monitoring while meeting the requirements of specifications.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] A method for high-precision target deformation monitoring of an unmanned aerial vehicle comprises the following steps:

[0007] S1. Set up image recognition targets on open surfaces: Carry out the setting of deformation monitoring points for structures, select open and unobstructed structures to set up deformation monitoring points outside, and set up highly recognizable image recognition targets on their surfaces;

[0008] S2. Set up drones and RTK or PPK base stations: Set up drones with lidar mapping functions, and set up RTK or PPK base stations on an open, unobstructed and stable surface around the structure;

[0009] S3. Control the UAV to collect multi-dimensional and multi-angle approach cloud data of the deformation monitoring image recognition target: Control or set up the UAV with laser radar mapping function to carry out multi-dimensional and multi-angle near-field flight on the deformation monitoring image recognition target, and collect multi-dimensional and multi-angle near-field cloud data of the deformation monitoring image recognition target through the UAV with laser radar mapping function and RTK or PPK base station;

[0010] S4. Extracting coordinate values ​​of multi-dimensional cloud map data: Post-processing the multi-dimensional laser cloud map of deformation monitoring image recognition targets collected in the near field by the UAV with laser radar mapping function, and extracting the coordinate values ​​of the deformation monitoring image recognition targets in each flight dimension and angle;

[0011] S5. Perform normal distribution screening of multi-dimensional cloud map coordinate value data samples and select standard values ​​of low confidence zone sample intervals: perform normal distribution sample data screening of multi-dimensional and angular coordinate values ​​collected by deformation monitoring image recognition targets, select low confidence zone data values, and select their standard coordinate values ​​as the coordinate values ​​of this measurement;

[0012] S6, repeat S1 to S5 to obtain the measured coordinate values ​​of the deformation monitoring image recognition target in multiple time periods, and obtain the displacement change through the change of the coordinate values ​​of the deformation monitoring image recognition target in different time periods.

[0013] Furthermore, in step S5, the sample data of the normal distribution is screened, and the low confidence zone data values ​​are selected as follows:

[0014] A normal Laida quasi-distribution analysis is performed on the collected multi-dimensional and angular coordinate values ​​to compare whether the deviation of each coordinate value from the average value is obvious. If it deviates significantly from the average value, it is an outlier. The outliers are eliminated to obtain standard data values, and a normal Laida quasi-distribution analysis is performed. Data samples in the low confidence zone [u-σ, u+σ] are selected. The probability that the value is distributed in [u-σ, u+σ] is 68.27%, and the standard value of the data sample in [u-σ, u+σ] is selected as the true value of the coordinate value of this measurement, where σ is the standard deviation and u is the mean.

[0015] The beneficial effects of the present invention are:

[0016] The present invention makes full use of the advantages of UAV measurement such as high timeliness, low cost, low risk, and high mobility, combines UAV RTK and PPK real-time differential dynamic positioning technology, and further improves the measurement accuracy based on multi-data post-processing technical means, greatly improving the efficiency and accuracy of deformation monitoring while meeting the requirements of specifications. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a flow chart of the method of the present invention;

[0018] Figure 2 It is a schematic diagram of the working principle of the present invention;

[0019] Figure 3 It is a normal Laida quasi-distribution analysis diagram of sample data of the present invention.

[0020] In the figure: 1-UAV; 2-reference station; 3-image recognition target. DETAILED DESCRIPTION

[0021] The present invention will be further described below in conjunction with the accompanying drawings.

[0022] like Figure 1 As shown, the present invention provides a high-precision target deformation monitoring method for an unmanned aerial vehicle, comprising the following steps:

[0023] S1. Setting image recognition targets 3 on open surfaces: Carry out the setting of deformation monitoring points of structures, select the deformation monitoring points on the outside of open and unobstructed structures, and set highly recognizable image recognition targets 3 on their surfaces;

[0024] S2. Set up UAV 1 and RTK or PPK base station 2: Set up UAV 1 with laser radar mapping function, and set up RTK or PPK base station 2 on an open, unobstructed and stable surface around the structure;

[0025] S3. Control the UAV 1 with a laser radar mapping function to perform multi-dimensional and multi-angle approach cloud data collection on the deformation monitoring image recognition target 3: Control the UAV 1 with a laser radar mapping function to make it perform multi-dimensional and multi-angle near-field flight on the deformation monitoring image recognition target 3, and perform multi-dimensional and multi-angle near-field cloud data collection on the deformation monitoring image recognition target 3 through the UAV 1 with a laser radar mapping function and the RTK or PPK base station 2;

[0026] S4. Extracting coordinate values ​​of multi-dimensional cloud map data: Post-processing the multi-dimensional laser cloud map of deformation monitoring image recognition target 3 collected in the near field of UAV 1 with laser radar mapping function, extracting the coordinate values ​​of deformation monitoring image recognition target 3 of each flight dimension and angle {(x1, y1, z1), (x2, y2, z2), (x3, y3, z3) ... (x n ,y n 、z n )};

[0027] S5. Perform normal distribution screening of multi-dimensional cloud map coordinate value data samples and select standard values ​​of low confidence zone sample intervals: by using the multi-dimensional and angular coordinate values ​​{(x1, y1, z1), (x2, y2, z2), (x3, y3, z3) ... (x n ,yn 、z n )} to screen the sample data of normal distribution and select the data value of low confidence zone {(x a ,y a 、z a )、(x b ,y b 、z b )、(x c ,y c 、z c )……(x an ,y an 、z an )}, where a, b... are the data number identifiers in the low confidence area, and their standard coordinate values ​​are selected as the coordinate values ​​of this measurement ( x 1. y 1. z 1);

[0028] S6, repeat S1 to S5, and obtain the measurement coordinate value of the multi-period deformation monitoring image recognition target 3 {( x 1. y 1. z 1) ( x 2. y 2. z 2)……( x n , y n , z n )}, and identify the change of target 3 coordinate value through deformation monitoring images at different time periods {( x n , y n , z n )-( x 1. y 1. z 1)}, and obtain the displacement change.

[0029] In step S5, the low confidence region data value is selected as:

[0030] The normal Laida quasi-distribution analysis is performed on the multi-dimensional and angular coordinate values ​​collected, and the deviation of each coordinate value from the average value is compared. If it deviates significantly from the average value, it is an outlier. The outlier is eliminated to obtain the standard data value {(x1, y1, z1), (x2, y2, z2), (x3, y3, z3)...(x n ,y n 、z n )}, perform normal Laida quasi-distribution analysis, select data samples {(x a,y a 、z a )、(x b ,y b 、z b )、(x c ,y c 、z c )……(x an ,y an 、z an )}, the probability that the value is distributed in [u-σ,u+σ] is 68.27%, and the standard value of the data sample in [u-σ,u+σ] is selected ( x 1. y 1. z 1) is used as the true value of the coordinates of this measurement, where σ is the standard deviation and u is the mean.

[0031] like Figure 2 As shown, the working principle of the present invention is to use the high maneuverability of the UAV to achieve convenient and fast monitoring; use the UAV 1 with a laser radar mapping function and the RTK base station 2 set on the surface to maximize the measurement accuracy limit of the UAV; near-field multi-dimensional and multi-angle flight around the image recognition target 3 is to collect cloud data coordinates and provide reliable data samples for subsequent normal distribution screening; finally, the measurement accuracy is further improved through normal data screening to achieve the purpose of monitoring accuracy.

[0032] like Figure 3 As shown, the normal Laida quasi-distribution analysis obtains the screening data sample values ​​{(x1, y1, z1), (x2, y2, z2), (x3, y3, z3)...(x n ,y n 、z n )}, perform normal Laida quasi-distribution analysis, select data samples {(x a ,y a 、z a )、(x b ,y b 、z b )、(x c ,y c 、z c )……(x an ,y an 、z an )}, the probability that the value is distributed in [u-σ,u+σ] is 68.27%, and the standard value of the data sample in [u-σ,u+σ] is selected ( x 1. y 1. z 1) is used as the true value of the coordinates of this measurement, where σ is the standard deviation and u is the mean.

[0033] The above embodiments are used in conjunction with the accompanying drawings to illustrate the present invention, but they should not be understood as limiting the patent scope of the present invention. It should be pointed out that a person skilled in the art can make several modifications and improvements without departing from the concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. A high-precision target deformation monitoring method for an unmanned aerial vehicle, characterized in that: It includes the following steps: S1. Set up image recognition targets on open surfaces: Carry out the setting of deformation monitoring points for structures, select open and unobstructed structures to set up deformation monitoring points outside, and set up highly recognizable image recognition targets on their surfaces; S2. Set up drones and RTK or PPK base stations: Set up drones with lidar mapping functions, and set up RTK or PPK base stations on an open, unobstructed and stable surface around the structure; S3. Control the UAV to collect multi-dimensional and multi-angle approach cloud data of the deformation monitoring image recognition target: Control or set up the UAV with laser radar mapping function to carry out multi-dimensional and multi-angle near-field flight on the deformation monitoring image recognition target, and collect multi-dimensional and multi-angle near-field cloud data of the deformation monitoring image recognition target through the UAV with laser radar mapping function and RTK or PPK base station; S4. Extracting coordinate values ​​of multi-dimensional cloud map data: Post-processing the multi-dimensional laser cloud map of deformation monitoring image recognition targets collected in the near field by the UAV with laser radar mapping function, and extracting the coordinate values ​​of the deformation monitoring image recognition targets in each flight dimension and angle; S5. Perform normal distribution screening of multi-dimensional cloud map coordinate value data samples and select standard values ​​of low confidence zone sample intervals: perform normal distribution sample data screening of multi-dimensional and angular coordinate values ​​collected by deformation monitoring image recognition targets, select low confidence zone data values, and select their standard coordinate values ​​as the coordinate values ​​of this measurement; S6, repeat S1 to S5 to obtain the measured coordinate values ​​of the deformation monitoring image recognition target in multiple time periods, and obtain the displacement change through the change of the coordinate values ​​of the deformation monitoring image recognition target in different time periods.

2. The method for high-precision target deformation monitoring of an unmanned aerial vehicle according to claim 1, characterized in that: In step S5, the low confidence region data value is selected as follows: A normal Laida quasi-distribution analysis is performed on the collected multi-dimensional and angular coordinate values ​​to compare whether the deviation of each coordinate value from the average value is obvious. If it deviates significantly from the average value, it is an outlier. The outliers are eliminated to obtain standard data values, and a normal Laida quasi-distribution analysis is performed. Data samples in the low confidence zone [u-σ, u+σ] are selected. The probability that the value is distributed in [u-σ, u+σ] is 68.27%, and the standard value of the data sample in [u-σ, u+σ] is selected as the true value of the coordinate value of this measurement, where σ is the standard deviation and u is the mean.

Images