Liquor brewing retort feeding height prediction method based on laser radar point cloud data

Abstract

The invention discloses a white spirit brewing retort feeding height prediction method based on laser radar point cloud data, and the method comprises the steps: 1) carrying out the three-dimensional scanning of a retort feeding material through laser radar equipment, and obtaining the surface point cloud data of the material; 2) de-noising the point cloud data by using a straight-through filter, removing the side wall data of the trolley, and retaining the point cloud on the surface and the bottom surface of the material; 3) performing three-dimensional surface reconstruction on the denoised point cloud data, generating a convex hull model and calculating the material volume; 4) establishing a height prediction model based on a circular truncated cone volume formula and material expansion characteristics, and dynamically correcting an expansion coefficient and an offset through a least square method; and 5) outputting the predicted steaming height and adjusting the process parameters in real time. According to the method, the steaming height can be accurately and quantitatively predicted, the white spirit brewing efficiency and quality stability are improved, and meanwhile, the dependence of a traditional process on artificial experience is reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of liquor production, and in particular relates to a method for predicting the height of an upper steamer during the liquor brewing process by using laser radar point cloud data. Background Art

[0002] Baijiu brewing is a time-honored traditional craft, and the steaming chamber, a crucial step in the fermentation process, has a direct impact on the flavor and quality of the liquor. The steaming chamber, where the fermenting materials are stacked in a steamer barrel for steaming, significantly influences the steaming effect, heat conduction, and ultimately the liquor yield. Traditional baijiu brewing techniques rely heavily on empirically determined steaming chamber height, lacking quantitative monitoring methods. This reliance on experience often leads to inconsistent product quality and compromises brewing efficiency.

[0003] In recent years, with the development of intelligent manufacturing technology, laser detection and ranging (LiDAR) has been widely used in industrial measurement due to its high precision, high reliability, and non-contact measurement capabilities. LiDAR technology can acquire three-dimensional point cloud data of objects, providing a new solution for automated inspection in industrial production. However, in the baijiu brewing process, the application of LiDAR technology to predict the height of the upper steamer is still lacking. Summary of the Invention

[0004] The present invention aims to address the shortcomings of the above-mentioned existing technologies and proposes a method for predicting the height of the upper steamer in liquor brewing based on lidar point cloud data, in order to achieve accurate quantitative prediction of the height of the upper steamer through three-dimensional scanning, point cloud denoising, volume calculation and dynamic modeling, thereby improving the efficiency and quality stability of liquor brewing, and at the same time reducing the traditional process's dependence on manual experience, which is of great significance to the improvement of traditional processes.

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

[0006] The method for predicting the height of the upper steamer in liquor brewing based on laser radar point cloud data of the present invention is characterized in that it comprises the following steps:

[0007] Step 1: Install a laser radar device directly above the trolley track and collect point cloud data in real time during the trolley's operation. When the trolley reaches the specified position, collect point cloud data of the current batch of materials during the retorting process;

[0008] Step 2: De-noise the point cloud data of the current batch of materials through a straight-through filter, thereby removing the trolley side wall data, retaining the point clouds of the material surface and bottom, and obtaining the denoised point cloud data of the current batch of materials;

[0009] Step 3: Perform 3D surface reconstruction on the denoised point cloud data of the current batch of materials to generate a convex hull model of the current batch of materials, which is used to calculate the volume of the current batch of materials;

[0010] Step 4: Based on the volume of the current batch of materials, use the cone volume formula to predict the initial height of the upper steamer ;

[0011] Step 5: Dynamically adjust the initial height by the least squares method based on the expansion coefficient and offset in the historical library After making corrections, the corrected upper steamer height is obtained and used as the upper steamer height for the current batch of materials. Output;

[0012] Step 6: Compare the corrected upper steamer height with the actual upper steamer height to obtain the height difference, calculate the new expansion coefficient and the new offset and add them to the history library for the upper steamer height of the next batch of materials.

[0013] The method for predicting the height of the upper retort for liquor brewing according to the present invention is also characterized in that step 1 determines whether the trolley has reached the designated position by using spatial adjudication and a threshold method:

[0014] Step 1.1: Multiply the point cloud coordinate matrix of the point cloud data acquired in the LiDAR coordinate system by the rotation matrix to obtain the point cloud coordinate matrix in the observer coordinate system;

[0015] Step 1.2: Define the number of point cloud coordinates within the specified range as Num and initialize it to 0; mark the coordinates of any point P in the point cloud coordinate matrix under the observer coordinate system as (x, y, z);

[0016] Step 1.3: Traverse all point coordinates in the point cloud coordinate matrix. If < x < and < y < and < z < , then assign Num+1 to Num and execute step 1.4; otherwise, Num remains unchanged; where, is the minimum coordinate value of the car’s location, is the maximum coordinate value of the car’s location;

[0017] Step 1.4: If Num > M, the car reaches the specified position and stops moving; otherwise, the car continues moving.

[0018] Furthermore, the denoising condition of the straight-through filter in step 2 is:

[0019] If the coordinates (x, y, z) of P satisfy: , then retain P, otherwise, eliminate P; where d is the thickness of the car side wall.

[0020] Furthermore, in step 3, the boundary points of the bottom surface of the trolley are first supplemented by interpolation, and then the material volume corresponding to the convex hull model is calculated using the QuickHull algorithm:

[0021] Step 3.1: According to the four boundary conditions shown in equations (1) to (4), construct the same number of new points Q(x', y', z') on each boundary and insert them into the four boundary points of the bottom surface of the car in the point cloud coordinate matrix;

[0022] (1)

[0023] (2)

[0024] (3)

[0025] (4).

[0026] Furthermore, in step 4, the initial height of the upper steamer is obtained using formula (5) and formula (6). :

[0027] (5)

[0028] (6)

[0029] In formula (5) and formula (6), V is the volume of the current batch of materials, R is the upper surface radius of the current batch of materials, and r is the lower surface radius of the current batch of materials. The tilt angle of the side of the steamer pot cone;

[0030] Furthermore, in step 5, the upper steamer height of the current batch of materials is obtained using formula (7) :

[0031] Step 5: Combined with the material expansion characteristics, the final predicted height is:

[0032] H = kh + b (7)

[0033] In formula (7), k is the expansion coefficient in the history library, and b is the offset in the history library.

[0034] The electronic device of the present invention includes a memory and a processor, and is characterized in that the memory is used to store a program that supports the processor to execute the method for predicting the height of the upper steamer of white wine brewing, and the processor is configured to execute the program stored in the memory.

[0035] The present invention provides a computer-readable storage medium, and the computer-readable storage medium stores a computer program, which is characterized in that when the computer program is run by a processor, the steps of the method for predicting the height of the upper steamer in white wine brewing are executed.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] 1. The present invention installs a laser radar device directly above the trolley track to collect three-dimensional point cloud data of the material surface in real time and contactlessly. It then uses a direct-pass filter to accurately eliminate interference from the trolley sidewalls, and generates a convex hull model through three-dimensional surface reconstruction to accurately calculate the material volume. This series of steps completely changes the traditional method of relying on visual estimation by experienced craftsmen, providing objective and quantitative basic data for height prediction, thereby effectively solving the problem of inconsistent upper steamer height caused by subjective experience differences in traditional processes, achieving accurate quantitative prediction of the upper steamer height, significantly improving the quality stability of the liquor brewing process, significantly reducing reliance on manual experience, improving quality stability, and ensuring that the flavor and yield of each batch of products are consistent.

[0038] 2. This invention innovatively combines the preliminary height calculated using the frustum volume formula with the actual physical properties (expansion) of the material. Using the expansion coefficient k and offset b from a historical database, the preliminary height h is dynamically corrected using the least squares method to obtain the final predicted height H. This real-time iterative update and dynamic correction model, based on historical data and integrating material expansion characteristics, automatically learns and adapts to volumetric expansion variations caused by differences in moisture content, particle size, and fermentation status across batches of materials. This overcomes the limitation of a single geometric model that cannot accurately reflect complex material properties, significantly improves the accuracy and process adaptability of upper retort height prediction, and ensures that the prediction results are more aligned with actual process requirements.

[0039] 3. While outputting the predicted height H for the current batch, the present invention also compares this predicted height with the actual height to obtain the difference. Based on this difference, a new expansion coefficient and offset are calculated and updated to the historical database. This closed-loop "prediction-execution-verification-optimization" mechanism enables the prediction model to continuously absorb feedback from production practices, continuously self-correcting and improving, forming a closed-loop feedback optimization mechanism. This effectively addresses the difficulties traditional methods or static models face in adapting to minor equipment changes and material batch fluctuations during long-term production processes, ensuring the long-term reliability and robustness of the prediction system and significantly improving overall brewing efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is the flow chart of volume calculation and height prediction;

[0041] Figure 2 This is a diagram of the relevant parameters of the steamer pot. DETAILED DESCRIPTION

[0042] The specific technical solutions of the embodiments of the present invention are further described below with reference to the accompanying drawings.

[0043] In this embodiment, Livox_Mid-360 laser radar is used to collect point cloud data, and a method for predicting the upper steamer height based on the laser radar point cloud data is proposed. Figure 1 As shown, the following steps are included:

[0044] Step 1: Since the car will stay at a fixed position for a period of time before being lifted, and the overall point cloud number at that position is sparse when there is no car, we can determine whether the car has passed by by counting the number of point clouds in the space where the car has stayed. Set a threshold M (M is 50,000). When the number of point clouds exceeds the threshold, it can be determined that a car has stayed, and then proceed to subsequent processing.

[0045] Step 1.1: Multiply the coordinate matrix of the point cloud by the rotation matrix and transform it to the observer coordinate system.

[0046] Specifically, the angle between the radar installation bottom surface and the vertical surface The angle is fixed, so the rotation angle is , the rotation axis is the x-axis, and the rotation matrix is: .

[0047] Step 1.2: Define the number of point cloud coordinates within the specified range as Num and initialize it to 0; mark the coordinates of any point P in the point cloud coordinate matrix under the observer coordinate system as (x, y, z).

[0048] Step 1.3: Traverse all point coordinates in the point cloud coordinate matrix. If < x < and < y < and < z < , then assign Num+1 to Num and execute step 1.4; otherwise, Num remains unchanged; where, is the minimum coordinate value of the car’s location, is the maximum coordinate of the car's location.

[0049] in, is the extreme value of the x, y, and z coordinates of the point cloud. Since the car's stop position is determined, by measuring the car's own length, width, and height, and the height between the radar center point and the ground and the horizontal distance between the radar center point and the car, we can obtain .

[0050] Step 1.4: If Num > M, the car reaches the specified position and stops moving; otherwise, the car continues moving.

[0051] Step 2: De-noise the point cloud data of the current batch of materials through a straight-through filter, thereby eliminating the trolley side wall data, retaining the point clouds of the material surface and bottom, and obtaining the denoised point cloud data of the current batch of materials.

[0052] Specifically, traverse the point cloud coordinate matrix. If the coordinates (x, y, z) of P satisfy: , then retain P, otherwise, eliminate P; where d is the thickness of the car side wall.

[0053] Step 3.1: According to the four boundary conditions shown in equations (1) to (4), construct the same number of new points Q(x', y', z') on each boundary and insert them into the four boundary points of the bottom surface of the car in the point cloud coordinate matrix;

[0054] (1)

[0055] (2)

[0056] (3)

[0057] (4).

[0058] Step 3.2: Based on the processed point cloud coordinate matrix, call the QuickHull algorithm to calculate the volume V of the current batch of materials.

[0059] Step 4.1: If Figure 2 As shown, the initial height of the upper steamer is obtained using formula (5) and formula (6): :

[0060] (5)

[0061] (6)

[0062] In formula (5) and formula (6), R is the upper surface radius of the current batch of materials, r is the lower surface radius of the current batch of materials, The tilt angle of the side of the steamer pot cone;

[0063] Specifically, the simultaneous equations can be obtained , solve the cubic equation about h, and only keep the solutions with roots in the range of 600-1000 (the empirical value range of the upper steamer height) (it has been verified in practice that there is generally only one real root), recorded as h.

[0064] Step 5: Combined with the material expansion characteristics, the final predicted height H is:

[0065] H = kh + b (7)

[0066] In formula (7), k is the expansion coefficient in the history library, and b is the offset in the history library.

[0067] Step 6: Compare the corrected upper steamer height with the actual upper steamer height to obtain the height difference, calculate the new expansion coefficient and the new offset and add them to the history library for the upper steamer height of the next batch of materials.

[0068] Specifically, recursive least squares (RLS) is used to adjust k and b, and the steps are as follows:

[0069] (1) Initialize parameters and covariance matrix:

[0070] Initial parameters (k,b come from the least squares solution of historical data);

[0071] Initial covariance matrix , where H is the design matrix of historical data (each row );

[0072] (2) Iterative update: For each new data point :( is the true height);

[0073] Constructing a vector ;

[0074] Prediction error ,( is the predicted height obtained in step 5 );

[0075] Gain Vector ;

[0076] Update parameters ;

[0077] Update the covariance matrix ;

[0078] The new k and b are [0] and [1].

[0079] In this embodiment, an electronic device includes a memory and a processor, wherein the memory is used to store a program that supports the processor to execute the above method, and the processor is configured to execute the program stored in the memory.

[0080] In this embodiment, a computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above method are executed.

Claims

1. A method for predicting the height of the upper steamer in liquor brewing based on laser radar point cloud data, characterized in that: The following steps are involved: Step 1: Install a laser radar device directly above the trolley track and collect point cloud data in real time during the trolley's operation. When the trolley reaches the specified position, collect point cloud data of the current batch of materials during the retorting process; Step 2: De-noise the point cloud data of the current batch of materials through a straight-through filter, thereby removing the trolley side wall data, retaining the point clouds of the material surface and bottom, and obtaining the denoised point cloud data of the current batch of materials; Step 3: Perform 3D surface reconstruction on the denoised point cloud data of the current batch of materials to generate a convex hull model of the current batch of materials, which is used to calculate the volume of the current batch of materials; Step 4: Based on the volume of the current batch of materials, use the cone volume formula to predict the initial height of the upper steamer ; Step 5: Dynamically adjust the initial height by the least squares method based on the expansion coefficient and offset in the historical library After making corrections, the corrected upper steamer height is obtained and used as the upper steamer height for the current batch of materials. Output; Step 6: Compare the corrected upper steamer height with the actual upper steamer height to obtain the height difference, calculate the new expansion coefficient and the new offset and add them to the history library for the upper steamer height of the next batch of materials.

2. The method for predicting the height of the upper steamer in liquor brewing according to claim 1, wherein: Step 1 determines whether the car has reached the specified position by using spatial arbitration and threshold method: Step 1.1: Multiply the point cloud coordinate matrix of the point cloud data acquired in the LiDAR coordinate system by the rotation matrix to obtain the point cloud coordinate matrix in the observer coordinate system; Step 1.2: Define the number of point cloud coordinates within the specified range as Num and initialize it to 0; mark the coordinates of any point P in the point cloud coordinate matrix under the observer coordinate system as (x, y, z); Step 1.3: Traverse all point coordinates in the point cloud coordinate matrix. If < x < and < y < and <z < , then assign Num+1 to Num and execute step 1.4; Otherwise, Num remains unchanged; is the minimum coordinate value of the car’s location, is the maximum coordinate value of the car’s location; Step 1.4: If Num > M, the car reaches the specified position and stops moving; otherwise, the car continues moving.

3. The method for predicting the height of the upper steamer for brewing liquor according to claim 2, wherein: The denoising condition of the straight-through filter in step 2 is: If the coordinates (x, y, z) of P satisfy: , then retain P, otherwise, eliminate P; where d is the thickness of the car side wall.

4. The method for predicting the height of the upper steamer in liquor brewing according to claim 1, wherein: In step 3, the boundary points of the bottom surface of the trolley are first supplemented by interpolation, and then the volume of the material corresponding to the convex hull model is calculated using the QuickHull algorithm: Step 3.1: According to the four boundary conditions shown in equations (1) to (4), construct the same number of new points Q(x', y', z') on each boundary and insert them into the four boundary points of the bottom surface of the car in the point cloud coordinate matrix; (1) (2) (3) (4)。 5. The method for predicting the height of the upper steamer in liquor brewing according to claim 1, wherein: In step 4, the initial height of the upper steamer is obtained by using formula (5) and formula (6): : (5) (6) In formula (5) and formula (6), V is the volume of the current batch of materials, R is the upper surface radius of the current batch of materials, and r is the lower surface radius of the current batch of materials. It is the inclination angle of the side of the steamer pot cone.

6. The method for predicting the height of the upper steamer in liquor brewing according to claim 1, wherein: In step 5, the upper steamer height of the current batch of materials is obtained using formula (7): : Step 5: Combined with the material expansion characteristics, the final predicted height is: H = kh + b (7) In formula (7), k is the expansion coefficient in the history library, and b is the offset in the history library.

7. An electronic device comprising a memory and a processor, characterized in that: The memory is used to store a program that supports the processor to execute the method for predicting the height of the upper steamer for liquor brewing according to any one of claims 1 to 6, and the processor is configured to execute the program stored in the memory.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is run by a processor, the steps of the method for predicting the height of the upper steamer for brewing liquor according to any one of claims 1 to 6 are executed.

Images