Non-contact type coal car compartment weighing method and system based on laser radar

Abstract

The invention discloses a non-contact type coal car compartment weighing method and system based on a laser radar, and relates to the technical field of laser radar measurement, and the method comprises the steps: deploying a laser radar sensor above a track; periodically scanning and acquiring point cloud data of the top of the carriage through a laser radar; compartment gaps are detected through the collected point cloud data; splicing the point cloud information of each carriage, and restoring complete point cloud data of the top of the carriage; extracting a compartment edge contour, and obtaining depth information of the edge contour; according to the carriage type, generating 3D data of the goods; and according to the density of the cargo, carrying out mass estimation on the whole cargo. According to the invention, the weight of goods in the carriage is deduced by using the laser radar sensor through the obtained top point cloud data of each carriage, so that the transformation of a track is avoided, and the service life of the sensor is greatly prolonged by using a non-contact detection mode.

Description

Technical Field

[0001] The present invention relates to the field of laser radar measurement technology, and in particular to a laser radar-based non-contact coal carriage weighing method and system. Background Art

[0002] In coal preparation plants and power plants, open wagons are generally used as a means of transporting coal. Since the wagons are high, it is impossible to directly observe whether there is coal inside the wagons, and the weight of the coal transported in each wagon is also an important information, so it is usually necessary to weigh the vehicle's load. The existing methods all use gravity information directly or indirectly to calculate the vehicle's load, and then obtain the weight information of the coal in the wagons, but the weighing equipment is usually complex and the weighing efficiency is low.

[0003] The static weighing and collecting method and system based on the interlocking control of the tipping machine signal disclosed in Chinese patent CN118032102A uses the relevant signals on the tipping machine to determine the weight of the items in the carriage. This method relies on the tipping machine, which limits its promotion. In addition, since the vehicle is physically turned over, the force on its sensor is essentially the same as that on the ground scale. Another example is a railway vehicle weighing and checking system disclosed in Chinese patent CN220380596U, which directly weighs the object to be measured through the grabbing and releasing device, and its method is also the same as the principle of the ground scale. For example, a non-stop weighing system proposed in Chinese patent CN116929521A uses a weighing rail, which transmits the weight force of the carriage to the corresponding four weighing sensors. Based on the four weighing sensors, the weight of the corresponding carriage of the bogie can be directly read. This method is also a direct weighing method. It can be seen that this type of method basically requires the modification of the railway track. Since the mass of the object to be measured is large, the weighing equipment is very easy to be damaged. Summary of the invention

[0004] One of the purposes of the present invention is to provide a non-contact coal carriage weighing method based on laser radar, which estimates the weight of the coal transported in the carriage by collecting data through sensors, thereby avoiding the modification of the track.

[0005] In order to solve the above technical problems, the present invention adopts the following technical solution: a non-contact coal carriage weighing method based on laser radar, comprising the following steps:

[0006] (1) Deploy lidar sensors above the track;

[0007] (2) Periodically scan and obtain point cloud data on the top of the car through the laser radar;

[0008] (3) Detect the gap between carriages through the collected point cloud data;

[0009] (4) The point cloud information of each carriage is stitched together to restore the complete point cloud data of the top of the carriage;

[0010] (5) Extract the edge contour of the carriage and obtain the depth information of the edge contour;

[0011] (6) Generate 3D data of the cargo based on the carriage type;

[0012] (7) Estimate the overall mass of the cargo based on its density.

[0013] Preferably, in step (1), the detection direction of the laser radar sensor is perpendicular to the ground, and the X-axis of the detection plane is perpendicular to the track, and the Y-axis is parallel to the track.

[0014] More preferably, in step (2), the laser radar sends periodic laser pulses and receives signals, and uses the time-of-flight method to calculate the distance between each laser signal point and the detection point through the time difference between sending and receiving, and then generates point cloud data of the detected area through the combination of dense detection points, and the steps are as follows:

[0015] a. For each point X i , calculate the distance d between it and the remaining points ij =X i -X j ;

[0016] b. Screen out K mean X i The closest distance to the remaining points:

[0017] {d ij1 , d ij2 , …, d ijkmean};

[0018] c. Calculate the mean of the nearest distance set

[0019] d. Calculate K of all points mean Average distance di, and calculate the mean and variance of di:

[0020]

[0021] More preferably, after acquiring the point cloud data in step (2), preprocessing is performed on the point cloud data to filter out noise points, and the steps are as follows: set a filtering threshold threshold, threshold = mean + multi × stddev. Where multi is a parameter of the threshold:

[0022]

[0023] More preferably, assuming that after step (3), the point cloud data set of a carriage is {P1, P2, ..., Pn}, where the subscripts represent the temporal order of each sub-point cloud, then the step of splicing the point cloud information of each carriage in step (4) includes:

[0024] 1) Take two adjacent point clouds from the point cloud data set, record them as P and Q respectively, and use 3D feature extraction methods to extract feature points and feature descriptors for the two point cloud data respectively;

[0025] 2) Estimate an initial transformation from feature matching, including the rotation matrix R0 and the translation vector t0;

[0026] 3) Based on the rough matching results of R0 and t0, the ICP algorithm is used for iterative optimization;

[0027] 4) According to the calculated rotation matrix R and translation vector t, the point clouds of P and Q are spliced, and the synthesized results are put back into the collection;

[0028] 5) Repeat steps 1) to 4) until there is only one point cloud file left in the collection. This point cloud file is the point cloud data of the top of the entire carriage.

[0029] More preferably, in step (6), according to the type of the current carriage, relevant parameters matching it are set, including the shape of the carriage, the size of the carriage floor, and the height information of the carriage, and combined with the point cloud data of the top of the carriage, 3D data of the cargo in the carriage is generated, including the height of the cargo and the shape of the top of the cargo; wherein the height between the top of the cargo and the bottom of the carriage is calculated by the height difference between the top of the cargo and the edge of the carriage, and the height between the edge of the carriage and the bottom of the carriage; if the calculated height of the top of the cargo is the same as the height of the bottom of the carriage, it means that the carriage is empty.

[0030] More preferably, in step (7), the weight of the entire cargo is estimated by calculating the volume of the entire cargo based on the acquired 3D data of the cargo and combining the density of the cargo itself.

[0031] More preferably, the volume of the entire cargo is calculated by numerical integration: Assuming that the shape of the entire carriage floor is a rectangle of a×b, the entire cargo is divided into columns with a bottom of dx×dy. The height of the column is is the average height of the cargo in this area, and the volume of the entire cargo is:

[0032]

[0033] Assuming the mass of the cargo is ρ, the total mass of the cargo is:

[0034] Mass = ρ·V.

[0035] In addition, the present invention also provides a non-contact coal carriage weighing system based on laser radar, which comprises:

[0036] The data acquisition module includes a lidar sensor deployed above the track to periodically scan and obtain point cloud data on the top of the carriage;

[0037] The data processing module is used to detect the gap between carriages through the collected point cloud data, and to stitch the point cloud information of each carriage to restore the complete point cloud data of the top of the carriage;

[0038] The cargo data generation module is used to extract the edge contour of the carriage, obtain the depth information of the edge contour, and then generate 3D data of the cargo according to the carriage type;

[0039] The weighing estimation module is used to estimate the overall mass of the goods based on the density of the goods.

[0040] The system operates using the weighing method described above.

[0041] Compared with the prior art, the present invention uses a laser radar sensor to obtain point cloud data on the top of each carriage to infer the weight of the cargo in the carriage, thereby avoiding the modification of the track and using a non-contact detection method to greatly increase the service life of the sensor. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a schematic diagram of the method flow of the present invention;

[0043] Figure 2 A schematic diagram of a top view layout of the sensor installation in the embodiment;

[0044] Figure 3 A schematic diagram of the side view layout of the sensor installation in the embodiment;

[0045] Figure 4 Schematic diagram of detecting the edge of a carriage in an embodiment;

[0046] Figure 5 Schematic diagram of FPFH feature extraction in the embodiment;

[0047] Figure 6 A schematic diagram of the splicing of point cloud matching in the embodiment;

[0048] Figure 7 A global point cloud image of the cargo on the roof spliced ​​in the embodiment;

[0049] Figure 8 This is a schematic diagram of point cloud data of the edge of the carriage in the embodiment;

[0050] Fig. 9This is a schematic diagram of the 3D shape of the goods generated in the embodiment;

[0051] Fig.10 Schematic diagram of reference data in the cargo volume calculation method in the embodiment. DETAILED DESCRIPTION

[0052] In order to facilitate the understanding of those skilled in the art, the present invention is further described below in conjunction with embodiments and drawings. The contents mentioned in the implementation modes are not intended to limit the present invention.

[0053] like Figure 1 As shown, a non-contact coal carriage weighing method based on laser radar, the specific steps are as follows:

[0054] (1) Use sensors such as laser radar, which are installed on the top of the track using a gantry. The detection direction is perpendicular to the ground, and the x-axis of the detection plane is perpendicular to the track (e.g. Figure 2 and Figure 3 shown).

[0055] (2) The laser radar obtains point cloud data on the top of the car through periodic scanning.

[0056] LiDAR sends periodic laser pulses and receives signals, using the time-of-flight method (TOF) to calculate the distance between each laser signal point and the detection point through the time difference between sending and receiving. Through the combination of dense detection points, point cloud data of the detected area is generated.

[0057] After obtaining the point cloud data, preprocessing is first performed, including filtering out noise in the data and removing noise points in the original data. The process is as follows:

[0058] a. For each point X i , calculate the distance d between it and the remaining points ij =X i -X j ;

[0059] b. Screen out K mean X i The closest distance to the remaining points:

[0060] {d ij1 , d ij2 , …, d ijkmean};

[0061] c. Calculate the mean of the nearest distance set

[0062] d. Calculate K of all points mean Average distance di, and calculate the mean and variance of di:

[0063]

[0064]

[0065] e. Set the filter threshold, threshold = mean + multi × stddev. Where multi is the parameter of the threshold:

[0066]

[0067] (3) Use the point cloud model to detect the gap between carriages.

[0068] Since the transportation process involves multiple carriages, the collected point cloud sequence needs to be segmented and the point cloud information of the same carriage is grouped together. Due to the particularity of the open car structure, the gap between carriages (such as Figure 4 shown).

[0069] (4) Stitch the point cloud information of each carriage.

[0070] Because the length of the car is long, the multi-line laser radar cannot obtain the complete point cloud information of the top of the car in one acquisition, and the speed and acceleration of the vehicle when passing through the detection area are unknown or difficult to collect, so it is impossible to combine the local point cloud data in the time domain. Therefore, it is necessary to use an algorithm to splice the local point cloud data to restore the complete point cloud data of the top of the car.

[0071] Assuming that after step (3), the point cloud data set of a carriage is {P1, P2, ..., Pn}, where the subscripts represent the temporal order of each sub-point cloud, then the step of splicing the point cloud information of each carriage in step (4) includes:

[0072] 1) Take two adjacent point clouds from the point cloud data set, denoted as P and Q. For the two point cloud data, use 3D feature extraction methods (such as FPFH, SHOT, etc.) to extract feature points and feature descriptors. Among them, FPFH feature extraction is as follows: Figure 5 shown.

[0073] 2) Estimate an initial transformation from feature matching, including the rotation matrix R0 and the translation vector t0.

[0074] 3) Based on the rough matching results of R0 and t0, the ICP algorithm is used for iterative optimization, including:

[0075] ① Find the closest point pair, for each point p i ∈P, find the closest point q in Q j , forming the closest point pair (pi,qj).

[0076] ② Calculate the centroid and perform centralization, and calculate the centroid of point clouds P and Q respectively:

[0077]

[0078] ③Construct the covariance matrix and perform SVD decomposition.

[0079] The role of the covariance matrix in point cloud stitching is to help determine the best rotation matrix to achieve the best alignment of two point clouds.

[0080] The formula for calculating the covariance matrix is:

[0081]

[0082] For the convenience of subsequent calculations, the covariance matrix H is decomposed by SVD.

[0083] H=UΣV T ;

[0084] Where U is an m×m orthogonal matrix, V is an n×n orthogonal matrix, Σ is an m×n diagonal matrix, and the elements on the diagonal are called singular values.

[0085] ④Calculate the rotation matrix R and translation vector t.

[0086] In point cloud stitching, we need to find a rotation matrix R and a translation vector t that minimizes the alignment error between point clouds P and Q.

[0087] Using the results of the previous SVD decomposition, the rotation matrix R and translation vector t are calculated respectively.

[0088] R=UV T ;

[0089]

[0090] ⑤Update point cloud position.

[0091] Update the position of point cloud P according to the current rotation matrix R and translation vector t

[0092] p i ′=Rp i +t.

[0093] ⑤Evaluate the error and judge the convergence

[0094] The error is calculated as follows:

[0095]

[0096] Through the above steps, the covariance matrix and SVD decomposition can help find the optimal rotation and translation to minimize the alignment error of the two point clouds.

[0097] 4) Based on the calculated rotation matrix R and translation vector t, the point clouds of P and Q are concatenated, and the synthesized results are put back into the collection.

[0098] 5) Repeat steps 1) to 4) until there is only one point cloud file left in the collection. This point cloud file is the point cloud data of the top of the entire carriage (e.g. Figure 7 shown).

[0099] (5) Extract the edge contour of the car and obtain the depth information of the edge contour.

[0100] Considering that the track will sink slightly after the carriage is loaded, in order to improve the accuracy of the estimation, this method uses the edge information of the carriage as a reference for estimating the height of the cargo in the carriage. Figure 8 As shown in the figure, the geometric features of the edge of the carriage are quite obvious, and it is a relatively flat rectangular frame in the point cloud.

[0101] (6) Generate 3D data of the cargo based on the carriage type.

[0102] According to the type of the current carriage, set the relevant parameters to match it, including the shape of the carriage, the size of the carriage floor, the height of the carriage, etc., and combine the point cloud data of the top of the carriage to generate the 3D data of the cargo in the carriage, including the height of the cargo and the shape of the top of the cargo (such as Fig. 9 shown).

[0103] The height between the top of the cargo and the bottom of the compartment is calculated by combining the height difference between the top of the cargo and the edge of the compartment and the height between the edge of the compartment and the bottom of the compartment. If the calculated height of the top of the cargo is the same as the height of the bottom of the compartment, it means that the compartment is empty.

[0104] (7) Estimate the overall mass of the cargo based on its density.

[0105] Based on the acquired 3D information of the cargo, the weight of the entire cargo can be estimated by calculating the volume of the entire cargo and combining it with the density of the cargo itself.

[0106] like Fig.10 As shown in the figure, the volume of the entire cargo is calculated by numerical integration method: Assuming that the shape of the entire carriage floor is a rectangle of a×b, the entire cargo is divided into columns with a bottom of dx×dy. is the average height of the cargo in this area, and the volume of the entire cargo is:

[0107]

[0108] Assuming the mass of the cargo is ρ, the total mass of the cargo is:

[0109] Mass = ρ·V.

[0110] The non-contact coal carriage weighing method based on laser radar provided by the above embodiment can estimate the weight of coal in a carriage, avoiding the traditional method of measuring gravity to estimate the weight of cargo in the carriage. In addition, this method of obtaining point cloud information of cargo in the carriage by laser radar can also be used to detect whether the cargo is neatly placed and the position of the cargo in space. In addition to timely evaluating the transportation status of the cargo, it can also be used to guide the unloading equipment to automatically load and unload the cargo.

[0111] It can be seen that the present invention adopts a multi-line lidar sensor and an intelligent algorithm processing system to obtain point cloud data on the top of the coal carriage in a non-contact manner, uses point cloud stitching technology to restore the complete point cloud information on the top of the carriage, and estimates the weight of the coal in the carriage in combination with the carriage type and cargo density, thereby achieving high-precision, non-destructive weighing of coal carriages without track modification and direct gravity measurement, effectively improving the weighing efficiency and the service life of the sensor, while reducing the need for modification of track facilities and the risk of sensor damage.

[0112] In order to make it easier for ordinary technicians in the field to understand the improvements of the present invention over the prior art, some drawings and descriptions of the present invention have been simplified, and the above-mentioned embodiments are preferred implementation schemes of the present invention. In addition, the present invention can also be implemented in other ways. Any obvious replacement without departing from the concept of the present technical solution is within the protection scope of the present invention.

Claims

1. A non-contact coal carriage weighing method based on laser radar, characterized in that: The following steps are involved: (1) Deploy lidar sensors above the track; (2) Periodically scan and obtain point cloud data on the top of the car through the laser radar; (3) Detect the gap between carriages through the collected point cloud data; (4) The point cloud information of each carriage is stitched together to restore the complete point cloud data of the top of the carriage; (5) Extract the edge contour of the carriage and obtain the depth information of the edge contour; (6) Generate 3D data of the cargo based on the carriage type; (7) Estimate the overall mass of the cargo based on its density.

2. The non-contact coal carriage weighing method based on laser radar according to claim 1 is characterized in that: In step (1), the detection direction of the laser radar sensor is perpendicular to the ground, and the X-axis of the detection plane is perpendicular to the track, and the Y-axis is parallel to the track.

3. The non-contact coal carriage weighing method based on laser radar according to claim 1 is characterized in that: In step (2), the laser radar sends periodic laser pulses and receives signals. Using the time-of-flight method, the distance between each laser signal point and the detection point is calculated by the time difference between sending and receiving. Then, the point cloud data of the detected area is generated by combining dense detection points. The steps are as follows: a. For each point X i , calculate the distance d between it and the remaining points ij =X i -X j ; b. Screen out K mean X i The closest distance to the remaining points: {d ij1 ,d ij2 ,…,d ijkmean }; c. Calculate the mean of the nearest distance set d. Calculate K of all points mean Average distance di, and calculate the mean and variance of di:

4. The non-contact coal carriage weighing method based on laser radar according to claim 3 is characterized in that: After obtaining the point cloud data in step (2), preprocess it to filter out noise points. The steps are as follows: Set the filtering threshold threshold, threshold = mean + multi × stddev. Where multi is the parameter of the threshold:

5. The non-contact coal carriage weighing method based on laser radar according to claim 1 is characterized in that: Assuming that after step (3), the point cloud data set of a carriage is {P1, P2, ..., Pn}, where the subscripts represent the temporal order of each sub-point cloud, then the step of splicing the point cloud information of each carriage in step (4) includes: 1) Take two adjacent point clouds from the point cloud data set, record them as P and Q respectively, and use 3D feature extraction methods to extract feature points and feature descriptors for the two point cloud data respectively; 2) Estimate an initial transformation from feature matching, including the rotation matrix R0 and the translation vector t0; 3) Based on the rough matching results of R0 and t0, the ICP algorithm is used for iterative optimization; 4) According to the calculated rotation matrix R and translation vector t, the point clouds of P and Q are spliced, and the synthesized results are put back into the collection; 5) Repeat steps 1) to 4) until there is only one point cloud file left in the collection. This point cloud file is the point cloud data of the top of the entire carriage.

6. The non-contact coal carriage weighing method based on laser radar according to claim 1 is characterized in that: In step (6), according to the type of the current carriage, the relevant parameters matching it are set, including the shape of the carriage, the size of the carriage floor, and the height information of the carriage, and combined with the point cloud data of the top of the carriage, the 3D data of the cargo in the carriage is generated, including the height of the cargo and the shape of the top of the cargo; wherein the height between the top of the cargo and the bottom of the carriage is calculated by the height difference between the top of the cargo and the edge of the carriage, and the height between the edge of the carriage and the bottom of the carriage; if the calculated height of the top of the cargo is the same as the height of the bottom of the carriage, it means that the carriage is empty.

7. The non-contact coal carriage weighing method based on laser radar according to claim 1 is characterized in that: In step (7), the weight of the entire cargo is estimated by calculating the volume of the entire cargo based on the acquired 3D data of the cargo and combining it with the density of the cargo itself.

8. The non-contact coal carriage weighing method based on laser radar according to claim 7 is characterized in that: The volume of the entire cargo is calculated by numerical integration: Assuming that the shape of the entire carriage floor is a rectangle of a×b, the entire cargo is divided into columns with a bottom of dx×dy. The height of the column is is the average height of the cargo in this area, and the volume of the entire cargo is: Assuming the mass of the cargo is ρ, the total mass of the cargo is: Mass = ρ·V.

9. A non-contact coal carriage weighing system based on laser radar, characterized in that: include: The data acquisition module includes a lidar sensor deployed above the track to periodically scan and obtain point cloud data on the top of the carriage; The data processing module is used to detect the gap between carriages through the collected point cloud data, and to stitch the point cloud information of each carriage to restore the complete point cloud data of the top of the carriage; The cargo data generation module is used to extract the edge contour of the carriage, obtain the depth information of the edge contour, and then generate 3D data of the cargo according to the carriage type; The weighing estimation module is used to estimate the overall mass of the goods based on the density of the goods.

10. The non-contact coal carriage weighing system based on laser radar according to claim 9 is characterized in that: The weighing method according to any one of claims 1 to 8 is adopted for operation.

Images