A control method for a log intelligent processing device

Abstract

The present application relates to the field of log processing equipment, and in particular to a control method for intelligent log processing equipment, comprising the following steps: 1) spraying an arrow symbol on the cross section of the log to obtain an identification feature; 2) collecting a log cross section image to obtain log cross section contour information; 3) using an optimization algorithm to add multiple production constraints based on the log cross section contour information to obtain the optimal cutting angle and cutting path for the log; 4) obtaining a real-time log image through a camera, and using a deep learning algorithm in the processing terminal to obtain the optimal cutting angle formed between the log cross section on the sawing table and the horizontal plane based on the sprayed arrow symbol; 5) detecting the zero point position of the log end face through a zero point measuring device and sending the zero point position of the log end face to the processing terminal; and 6) realizing automatic processing and cutting in the processing terminal based on the optimal cutting angle and the zero point position of the log end face relative to the log on the processing equipment.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of log processing equipment, in particular to a control method for intelligent log processing equipment. BACKGROUND

[0002] In recent years, the annual import volume of logs in China has exceeded 50 million cubic meters, most of which are processed into wood squares. Almost all processing plants use manual sawing tables. Due to the irregular shape of logs and the uneven level of operators, the yield of manual sawing tables is extremely unstable, generally between 55% and 65%, with low efficiency, low yield, high labor cost, and also causing raw material waste.

[0003] The existing sawing table processing equipment mainly relies on manual operation, and the use of the equipment is completely dependent on the quality and working state of the operator. The reason why the sawing table processing equipment has a very low degree of automation is that there are two main problems. The first problem is that the end face of the log is irregular in shape, and the optimal cutting angle is calculated by the optimization algorithm, but it is difficult for the equipment to control the log to rotate to the optimal angle. The second problem is that due to the existence of irregular shapes and knots in logs, the optimal cutting path is calculated by the optimization algorithm, but the end face zero point cannot be obtained, so the equipment cannot be controlled to process and cut according to the plan. SUMMARY

[0004] The purpose of the present application is to provide a control method for intelligent log processing equipment, which realizes the automatic processing of logs and overcomes the many problems existing in the process of manually sawing logs.

[0005] The technical solution adopted by the present application to achieve the above purpose is as follows: a control method for intelligent log processing equipment, comprising the following steps:

[0006] 1) Spray arrow symbols on the cross section of the log to provide identification features for obtaining the real-time angle of log processing relative to the optimal cutting state angle of the log;

[0007] 2) Obtain the caliper diameter of the end face of the log by using a handheld caliper diameter measuring device, at the same time, collect the image of the cross section of the log and send it to the processing terminal for processing to obtain the cross section contour information of the log;

[0008] 3) According to the cross section contour information of the log, an optimization algorithm is used to add multiple production constraints to obtain the optimal cutting angle and cutting path for the log;

[0009] 4) Obtain real-time log images by using a camera and upload them to the processing terminal. The processing terminal uses a deep learning algorithm to obtain the optimal cutting angle formed between the cross section of the log on the sawing table and the horizontal plane according to the sprayed arrow symbols;

[0010] 5) The zero position of the end face of the log is detected by the zero measurement device and sent to the processing terminal;

[0011] 6) The processing terminal controls the processing equipment to cut at the optimal cutting angle when the log rotates to the zero position according to the optimal cutting angle of the log on the processing equipment and the zero position of the end face of the log, and the controller of the processing equipment controls the movement of the trolley relative to the band saw to realize automatic processing and cutting.

[0012] In step 2), the cross-sectional image of the log is collected and sent to the processing terminal for processing to obtain the cross-sectional profile information of the log, specifically:

[0013] 2-1) The cross-sectional image of the log collected by the handheld log diameter measurement device is sent to the processing terminal for light plane calibration;

[0014] 2-2) The processing terminal performs gray scale conversion and filter denoising operations on the collected cross-sectional image of the log to obtain a preprocessed cross-sectional image of the log;

[0015] 2-3) A deep learning instance segmentation algorithm is used to extract the cross-sectional profile of the log and the pose of the light plane of the handheld log diameter measurement device;

[0016] 2-4) According to the pose of the light plane of the handheld log diameter measurement device, the profile coordinates of the cross-sectional image of the log are spatially transformed to obtain the true size of the cross-sectional profile of the log, i.e., the cross-sectional profile information of the log.

[0017] The step 2-3) is specifically:

[0018] a. For the cross-sectional profile of the log, an ellipse is fitted through the end face profile points of the wood;

[0019] b. Extract the laser line in the cross-sectional image of the log, perform coordinate transformation, transform the image coordinates of the laser line into 3D coordinates in the camera coordinate system, and fit the plane where the laser line is located; then perform a coordinate system transformation to obtain the final pose of the light plane.

[0020] The step 3) is specifically:

[0021] A dynamic programming algorithm is used to plan the cutting sequence of the log, i.e.:

[0022] All cutting angles are traversed to obtain the maximum value, the cutting path is recorded while the value is calculated, the path corresponding to the maximum value is the optimal cutting path, and the angle corresponding to the maximum value is the optimal cutting angle, then:

[0023] CurValue=Num*UnitValue

[0024] value(n, degree) = value(n-step, degree) + CurValue

[0025] value = max(value(n, degree))

[0026] wherein Num represents the number of final billets contained in the current cut wood block, UnitValue represents the value of a billet, CurValue represents the value of the current cut wood block, MaxWidth represents the maximum width of the profile in the horizontal direction, value(n-step, degree) represents the total value obtained by cutting to the n-step position, value(n, degree) represents the total value obtained by cutting to the n position, n is the index position of the dynamic programming algorithm in the horizontal direction, n = 0, 1, 2, …, MaxWidth, and degree is the angle between the arrow and the horizontal direction, with an angle step of 10, degree = 0, 10, 20, …, 170.

[0027] The plurality of production constraints are specifically:

[0028] (1) If only horizontal cutting can be performed at the specified position, then when horizontal cutting is performed, step is equal to the billet width to meet the processing requirements;

[0029] (2) Since the shape of the log is irregular, the cross section cannot represent the overall shape of the log, so when vertical cutting is performed, step is equal to the billet thickness.

[0030] (3) The dynamic programming algorithm plans the cutting of two specifications of products, including main material wood and auxiliary material wood, wherein the main material is set to account for more than 90%; that is, before calculating the maximum value, the cutting path with a main material ratio of less than 90% is excluded, and then the optimization algorithm is used for optimization operation.

[0031] The deep learning algorithm is an improved YOLOv8 algorithm;

[0032] The improved YOLOv8 algorithm is: adding an angle classification loss to the original YOLOv8 structure to realize the detection of rotating targets by yolov8, introducing an attention mechanism to improve the accuracy of the recognition model, and completing the identification of the arrow angle of the log cross section.

[0033] The processing terminal uses a deep learning algorithm to obtain the real-time angle of log processing relative to the best cutting state angle of the log according to the sprayed arrow symbol, specifically:

[0034] The real-time image of the log is processed using an improved YOLOv8 detection algorithm, including the following steps:

[0035] 3-1) Load real-time log images, and pre-process the log images, including scaling, cropping, normalization, edge padding, and encapsulation operations, to generate standard format data, i.e., pre-processed images;

[0036] 3-2) Extract feature information in the pre-processed images through a CSPDarkNet network;

[0037] 3-3) Perform feature fusion on the feature information in the images using a PAN-FPN network to obtain fused data;

[0038] 3-4) Judge the class of the target and output the predicted arrow target position by performing judgment on the fused data through a Decoupled Head network;

[0039] 3-5) Obtain the rotation bounding rectangle of the arrow, i.e., the angle of the arrow, by performing threshold screening and non-maximum suppression on the arrow target, and then obtain the optimal cutting angle between the log cross section on the saw bench and the horizontal plane.

[0040] The zero point measuring device comprises a V-shaped baffle, a controller, a linear slide, a photoelectric sensor, a cylinder and a pressure sensor connected to the controller.

[0041] The photoelectric sensor is arranged above the moving track of the sliding block of the linear slide and is used to detect the arrival signal triggered after the sliding block moves to the position.

[0042] The cylinder is fixed to the sliding block, the output shaft of the cylinder is provided with a V-shaped baffle, one side of the V-shaped baffle is provided with a pressure sensor, and the other side abuts against the log.

[0043] The controller is used to control the driving motor of the linear slide to move and drive the sliding block to displace on the lead screw shaft, and to acquire the arrival trigger signal sent by the photoelectric sensor, and to control the cylinder to start according to the received arrival trigger signal.

[0044] In step 5), the zero point measuring device detects the zero point position of the end surface of the log, specifically as follows:

[0045] 4-1) The controller controls the sliding block of the linear slide to move to the trigger position of the photoelectric sensor, and the photoelectric sensor generates an arrival trigger signal and sends it to the controller.

[0046] 4-2) The controller controls the cylinder movement, and the cylinder shaft extends forward to drive the V-shaped baffle into contact with the log. After the V-shaped baffle is closed, the pressure sensor arranged on the V-shaped baffle triggers and generates a pressure signal, and sends the pressure signal to the controller, and the controller controls the cylinder to stop moving;

[0047] 4-3) The controller detects the extension distance of the cylinder shaft of the current cylinder, i.e. the zero position of the log cross section surface, and sends the zero position of the log cross section surface to the control terminal.

[0048] The present application has the following advantages and benefits:

[0049] 1. The control method of the log intelligent processing equipment provided by the present application realizes intelligent processing of logs. By spraying arrow symbols, collecting images by handheld devices, calculating the optimal angle and path by algorithms, rotating to the optimal angle, measuring the end face zero point, and performing processing and cutting, the intelligent cutting process of logs is realized.

[0050] 2. Since the color and texture of the log end face are uncertain and cannot be identified by the characteristics of the log end face itself, the present application sprays arrows to identify the direction by identifying the arrows. Even if the arrows are sprayed, when the log end face is moldy and the color is black, it is difficult to obtain good stability through traditional image processing methods, so the present application uses the most advanced RTMDet (a high-efficiency real-time target detector) to identify the arrow direction, and optimizes the model structure to make the detector more suitable for fixed scene and single target detection, while ensuring accuracy and good real-time performance.

[0051] 3. The current sawing table equipment estimates the zero point by the human eye and selects the opening position, which has great randomness and seriously affects the stability of the equipment. The log end face zero point measurement algorithm proposed by the present application can accurately measure the log end face zero position, so as to control the car to move to the opening position, eliminate human interference, and greatly improve the stability of the equipment.

[0052] 4. The current sawing table processing equipment is planned to cut on site by workers, which is difficult to obtain the best cutting scheme, resulting in a large amount of wood waste. The optimization algorithm proposed by the present application first measures the accurate profile based on the self-developed handheld device, then calculates the angle, opening, cutting sequence and other data by using the dynamic programming algorithm according to the profile information, and finally controls the sawing table equipment to complete the log cutting.

[0053] 5. The log sawing table processing equipment using the intelligent control method has a theoretical yield of more than 67%, and an actual yield of more than 65%. The yield and equipment stability are greatly improved, the labor cost and the dependence of the yield on the labor level are reduced, the utilization rate of raw materials is improved, and the use amount of logs is reduced, thereby realizing energy saving and emission reduction. BRIEF DESCRIPTION OF DRAWINGS

[0054] Figure 1 The control method flow chart of the log intelligent processing equipment of the present application;

[0055] Figure 2 The control method flow chart of the log intelligent processing equipment of the present application;

[0056] Figure 3 The structural schematic diagram of the zero point measuring device of the present application;

[0057] Figure 4 The optimization algorithm of the present embodiment for log cutting sequence planning diagram;

[0058] Wherein, 1 is a straight line sliding table, 2 is a photoelectric sensor, 3 is a cylinder, 4 is a V-shaped baffle, 5 is a pressure sensor, 6 is a log, 7 is a spray arrow. DETAILED DESCRIPTION

[0059] The present application will be further described in detail below in combination with the drawings and embodiments.

[0060] As shown in the control method flow chart of the log intelligent processing equipment of the present application, the present application is a control method for log intelligent processing equipment, which comprises the following steps: Figure 2 1) Spray arrow symbols on the cross section of the log, which provides identification features for the acquisition of the real-time angle of log processing relative to the best cutting state angle of the log;

[0061] 2) Obtain the caliper of the end face of the log by a handheld caliper measuring device, at the same time, collect the cross section image of the log and send it to the processing terminal for processing to obtain the cross section contour information of the log;

[0062] 3) According to the cross section contour information of the log, an optimization algorithm is used to add multiple production constraints to obtain the best cutting angle and cutting path for the log;

[0063] 4) Obtain the real-time log image by the camera and upload it to the processing terminal, and the processing terminal uses a deep learning algorithm to obtain the best cutting angle formed between the cross section of the log on the sawing table and the horizontal plane according to the spray arrow symbol;

[0064] 5) Detect the zero point position of the end face of the log by the zero point measuring device and send the zero point position of the end face of the log to the processing terminal; as shown in the structural schematic diagram of the zero point measuring device of the present application, the working principle of the log end face zero point measuring device of the present application is that:

[0065] Figure 3

[0066] ​​a. The controller controls the movement of the linear slide 1 to the position where the photoelectric sensor 2 triggers, and the photoelectric sensor 2 generates a position triggering signal and sends it to the controller;

[0067] b. The controller controls the movement of the air cylinder 3, and the air cylinder extends forward to drive the V-shaped baffle 4 to contact the log 6. After the V-shaped baffle 4 is closed, the pressure sensor 5 arranged on the V-shaped baffle 4 triggers and generates a pressure signal, and sends the pressure signal to the controller. The controller controls the air cylinder 3 to stop moving;

[0068] c. The controller detects the extension distance of the air cylinder shaft of the air cylinder 3, that is, the zero position of the log cross-section surface, and sends the zero position of the log cross-section surface to the processing terminal.

[0069] 6) The processing terminal controls the processing equipment to cut at the optimal cutting angle of the log rotating to the zero position according to the optimal cutting angle of the log on the processing equipment and the zero position of the log end surface. The controller of the processing equipment controls the movement of the trolley relative to the band saw to realize automatic processing and cutting.

[0070] (1) In step 2), the log cross-section image is collected and sent to the processing terminal for processing to obtain the log cross-section contour information, specifically:

[0071] 2-1) The log cross-section image collected by the handheld log caliper measuring device is sent to the processing terminal for light plane calibration;

[0072] 2-2) The processing terminal performs gray scale conversion and filter denoising operation on the collected log cross-section image to obtain the preprocessed log cross-section image;

[0073] 2-3) A deep learning instance segmentation algorithm is used to extract the log cross-section contour and the pose of the light plane of the handheld log caliper measuring device;

[0074] a. For the log cross-section contour, an ellipse is fitted by the wood end surface contour points;

[0075] b. The laser line in the log cross-section image is extracted, the image coordinates of the laser line are converted into 3D coordinates in the camera coordinate system, and the plane where the laser line is located is fitted. Then, a coordinate system conversion is performed to obtain the final pose of the light plane.

[0076] 2-4) According to the pose of the light plane of the handheld log caliper measuring device, the contour coordinates of the log cross-section image are spatially transformed to obtain the true size of the log cross-section contour, that is, the log cross-section contour information is obtained.

[0077] (2) As Figure 4As shown, the optimization algorithm of the present embodiment plans the log cutting sequence diagram, and the optimization algorithm in step 3) is specifically:

[0078] The dynamic programming algorithm is used to plan the cutting sequence of the log, that is:

[0079] All cutting angles are traversed to obtain the maximum value, and the cutting path is recorded while calculating the value. The path corresponding to the maximum value is the optimal cutting path, and the angle corresponding to the maximum value is the optimal cutting angle. Then:

[0080] CurValue=Num*UnitValue

[0081] value(n,degree)=value(n-step,degree)+CurValue

[0082] value=max(value(n,degree))

[0083] Where Num represents the number of final blanks contained in the current cut wood, UnitValue represents the value of a piece of blank, CurValue represents the value of the current cut wood, MaxWidth represents the maximum width of the outline in the horizontal direction, value(n-step,degree) represents the total value obtained by cutting to the n-step position, value(n,degree) represents the total value obtained by cutting to the n position, n is the index position of the dynamic programming algorithm in the horizontal direction, n=0,1,2,...,MaxWidth, degree is the angle between the arrow and the horizontal direction, the angle step is set to 10, degree=0,10,20,...,170.

[0084] Add multiple production constraints, specifically:

[0085] (1) If only horizontal cutting can be performed at the specified position, set step equal to the blank width when horizontal cutting to meet the processing requirements;

[0086] (2) Since the shape of the log is irregular, the cross section cannot represent the overall shape of the log, so set step equal to the blank thickness when vertical cutting.

[0087] (3) The dynamic programming algorithm plans the cutting of two specifications of products, including main material wood and auxiliary material wood, wherein the main material is set to account for more than 90%; That is, before calculating the maximum value, exclude the cutting path with a main material ratio of less than 90%, and then perform optimization algorithm optimization operation.

[0088] (Three) The deep learning algorithm mentioned in step 4) is an improved YOLOv8 algorithm;

[0089] The improved YOLOv8 algorithm is: adding an angle classification loss to the original YOLOv8 structure to realize the detection of the rotating target by yolov8, introducing an attention mechanism to improve the accuracy of the recognition model, and completing the recognition of the arrow angle of the log cross section. The processing terminal uses a deep learning algorithm to obtain the real-time angle of the log processing relative to the best cutting state angle of the log according to the sprayed arrow symbol, specifically:

[0090] The real-time image of the log is processed by using the improved YOLOv8 detection algorithm, including the following steps:

[0091] 3-1) Load the real-time log image, and pre-process the log image, including scaling, cropping, normalizing, padding the edge, and encapsulation operation, to generate standard format data, i.e. the pre-processed image;

[0092] 3-2) Extract the feature information in the pre-processed image through the CSPDarkNet network;

[0093] 3-3) The feature information in the image is fused by using the PAN-FPN network to obtain the fused data;

[0094] 3-4) The fused data is judged by using the Decoupled Head network to judge the category of the target, and the predicted arrow target position is output;

[0095] 3-5) The arrow target is screened by threshold and non-maximum suppression method, and the rotating bounding rectangle of the arrow is obtained, i.e. the arrow angle is obtained; and then the best cutting angle between the log cross section on the saw bench and the horizontal plane is obtained.

[0096] As shown in Figure 3 The zero point measuring device includes a V-shaped baffle 4, a controller, a straight line sliding table 1 connected with the controller, a photoelectric sensor 2, a gas cylinder 3 and a pressure sensor 5.

[0097] The photoelectric sensor 2 is arranged above the sliding track of the sliding block of the straight line sliding table 1, and is used to detect the arrival signal triggered after the sliding block moves to the position;

[0098] The gas cylinder 3 is fixedly arranged on the sliding block, the output shaft of the gas cylinder 3 is provided with the V-shaped baffle 4, one side of the V-shaped baffle 4 is provided with the pressure sensor 5, and the other side is in abutment with the log 6, so that the V-shaped baffle 4 is closed after contacting the log 6, and the pressure sensor 5 generates a pressure signal;

[0099] The controller is used to control the movement of the drive motor of the linear slide 1, which drives the slider to move on the lead screw shaft, and to acquire the position trigger signal sent by the photoelectric sensor 2. Based on the received position trigger signal, the controller controls the cylinder to start. At the same time, the controller acquires the pressure signal sent by the pressure sensor 5 in real time, and based on the pressure signal, it collects the current cylinder shaft extension distance of the cylinder 3 and uploads it to the processing terminal.

[0100] Example:

[0101] like Figure 1 As shown, the handheld log measuring device and the host computer (all-in-one machine) are connected via a local area network (Wi-Fi). The handheld log measuring device sends the acquired images, measured contours, and measuring diameter data to the all-in-one machine via MODBUS TCP. The all-in-one machine and the PLC controller are connected via a network cable. The all-in-one machine runs an optimization algorithm to calculate the optimal angle and path, runs a deep learning model to calculate the current angle of the log, and then sends the calculation results to the PLC via MODBUS TCP. The PLC controller first controls the log to rotate to the specified position based on the calculated current angle, then measures the zero point position of the log end face, and finally controls the saw table to automatically process and cut the log based on the zero point position.

[0102] like Figure 2 As shown, in order to perform angle recognition, arrow symbols are first sprayed, and then an optimization algorithm is used to obtain the optimal cutting angle and cutting path of the log. The angle of the arrow is then identified by a deep learning algorithm (i.e., the improved YOLOv8 algorithm in this invention). Finally, the most advanced rotating target detection model is used to obtain the most accurate arrow angle, i.e., the optimal cutting angle formed between the cross-section of the log on the saw table and the horizontal plane, while ensuring real-time performance.

[0103] Combination Figure 3 To obtain the zero-point position of the log, firstly, log 6 is loaded, and then the controller controls the linear slide 1 to move to the position that triggers the photoelectric sensor 2. The cooperation between the linear slide 1 and the photoelectric sensor 2 solves the problem of large positional movement before and after loading. Next, the controller controls the cylinder 3 to move. The cylinder 3 extends forward, and the V-shaped baffle 4 of the two baffles closes, triggering the pressure sensor 5. The cylinder 3 stops moving. The cooperation between the cylinder 3 and the pressure sensor 5 solves the problem of interference from log burrs and knots. Finally, the zero-point position of the log end face is measured.

[0104] The processing terminal controls the processing equipment to cut at the optimal cutting angle when the log rotates to the zero point position, based on the optimal cutting angle of the log relative to the processing equipment and the zero point position of the log end face. The controller of the processing equipment controls the movement of the carriage relative to the band saw to achieve automated processing and cutting.

[0105] The above merely illustrates the embodiments of the present application, but should not be used to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, extension, etc. within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A control method for intelligent log processing equipment, characterized in that, Includes the following steps: 1) Spray arrow symbols on the cross-section of the log to provide identification features for obtaining the real-time angle of log processing relative to the optimal cutting state angle of the log; 2) Obtain the measuring diameter of the log end face using a handheld measuring device. At the same time, acquire the cross-sectional image of the log and send it to the processing terminal for processing to obtain the cross-sectional contour information of the log. The process of acquiring cross-sectional images of logs and sending them to a processing terminal for processing to obtain the log cross-sectional contour information specifically involves: 2-1) Images of the log cross-sections are collected using a handheld log diameter measuring device and sent to a processing terminal for calibration of the light plane; 2-2) The processing terminal performs grayscale conversion and filtering noise reduction on the acquired log cross-section image to obtain the preprocessed log cross-section image; 2-3) Using a deep learning instance segmentation algorithm, extract the cross-sectional contour of the log and the optical plane pose of the handheld log measuring device; 2-4) Based on the optical plane pose of the handheld log diameter measuring device, the contour coordinates of the log cross-section image are spatially transformed to obtain the true size of the log cross-section contour, that is, to obtain the log cross-section contour information. 3) Based on the cross-sectional contour information of the log, an optimization algorithm is used to add multiple production constraints to obtain the optimal cutting angle and cutting path for the log; The aforementioned production constraints are as follows: (1) If only cross-cutting can be performed at the specified position, then when cross-cutting, step is set to be equal to the width of the cutting material to meet the processing requirements; (2) Since the shape of the log is irregular, the cross section cannot represent the overall shape of the log. Therefore, when setting the vertical cut, step is equal to the thickness of the cut material. (3) Dynamic programming algorithm plans the two specifications of products at the cutting point, including: main material wood and secondary material wood, of which the main material is set to account for more than 90%; that is, before calculating the maximum value, the cutting path with the main material accounting for less than 90% is excluded, and then the optimization algorithm is used to perform the optimization operation. 4) Real-time images of logs are obtained through a camera and uploaded to the processing terminal. The processing terminal uses a deep learning algorithm to obtain the optimal cutting angle between the cross-section of the log and the horizontal plane on the sawing table based on the sprayed arrow symbol. The deep learning algorithm is the improved YOLOv8 algorithm; The improved YOLOv8 algorithm is as follows: an angle classification loss is added to the original YOLOv8 structure to achieve YOLOv8 detection of rotating targets. An attention mechanism is introduced to improve the accuracy of the recognition model and complete the recognition of the arrow angle of the log cross section. 5) The zero-point position of the log end face is detected by the zero-point measuring device and the zero-point position of the log end face is sent to the processing terminal. 6) The processing terminal controls the processing equipment to cut at the optimal cutting angle when the log is rotated to the zero point position based on the optimal cutting angle of the log relative to the processing equipment and the zero point position of the log end face. The controller of the processing equipment controls the movement of the carriage relative to the band saw to realize automated processing and cutting. The processing terminal uses a deep learning algorithm to obtain the real-time angle of log processing relative to the optimal cutting state of the log based on the sprayed arrow symbol. Specifically: The real-time image of the log is processed using an improved YOLOv8 detection algorithm, including the following steps: 3-1) Load the real-time log image, preprocess the log image, including scaling, cropping, normalization, edge filling, and encapsulation operations, and generate standard format data, i.e., the preprocessed image; 3-2) Extract feature information from the preprocessed image using the CSPDarkNet network; 3-3) The feature information in the image is fused using a PAN-FPN network to obtain the fused data; 3-4) The fused data is processed through a Decoupled Head network to determine the category of the target and output the predicted arrow target location; 3-5) By applying threshold filtering and non-maximum suppression methods to the arrow target, the circumscribed rectangle of the arrow is obtained, which is the arrow angle; then the optimal cutting angle formed between the log cross-section on the saw table and the horizontal plane is obtained.

2. The control method for intelligent log processing equipment according to claim 1, characterized in that, Steps 2-3) are specifically as follows: a. For the cross-sectional profile of the log, fit an ellipse using the profile points on the end face of the log; b. Extract the laser line from the cross-sectional image of the log, perform coordinate transformation to convert the image coordinates of the laser line into 3D coordinates in the camera coordinate system, and fit the plane where the laser line is located; then perform another coordinate transformation to obtain the final light plane pose.

3. The control method for intelligent log processing equipment according to claim 1, characterized in that, Step 3) specifically includes: The cutting sequence of logs is planned using a dynamic programming algorithm, namely: By iterating through all cutting angles, the maximum value is obtained. While calculating the value, the cutting path is recorded. The path corresponding to the maximum value is the optimal cutting path, and the angle corresponding to the maximum value is the optimal cutting angle. Therefore: CurValue = Num * UnitValue value(n,degree)=value(n-step,degree)+CurValue value = max(value(n, degree)) Where Num represents the number of final cut pieces in the currently cut wood block, UnitValue represents the value of one cut piece, CurValue represents the value of the currently cut wood block, MaxWidth represents the maximum width of the outline in the horizontal direction, value(n-step,degree) represents the total value obtained by cutting to position n-step, and value(n,degree) represents the total value obtained by cutting to position n. n is the index position in the horizontal direction of the dynamic programming algorithm, n=0,1,2,...,MaxWidth, and degree is the angle between the arrow and the horizontal direction, with the angle step size set to 10, degree=0,10,20,...,170.

4. The control method for intelligent log processing equipment according to claim 1, characterized in that, The zero-point measuring device includes: a V-shaped baffle (4), a controller and a linear slide (1) connected to the controller, a photoelectric sensor (2), a cylinder (3) and a pressure sensor (5); A photoelectric sensor (2) is provided above the slider movement trajectory of the linear slide (1) to detect when the slider moves to the position and triggers a positioning signal; The cylinder (3) is fixed on the slider. The output shaft of the cylinder (3) is provided with a V-shaped baffle (4). A pressure sensor (5) is provided on one side of the V-shaped baffle (4), and the other side abuts against the log (6). After contacting the log (6), the V-shaped baffle (4) closes, triggering the pressure sensor (5) to generate a pressure signal. The controller is used to control the movement of the drive motor of the linear slide (1), drive the slider to move on the lead screw shaft, and obtain the position trigger signal sent by the photoelectric sensor (2). It controls the cylinder to start according to the received position trigger signal. At the same time, it obtains the pressure signal sent by the pressure sensor (5) in real time, and collects the cylinder shaft extension distance of the current cylinder (3) according to the pressure signal, and uploads it to the processing terminal.

5. The control method for intelligent log processing equipment according to claim 1, characterized in that, In step 5), the zero-point measuring device detects the zero-point position of the log end face. Specifically, 4-1) the controller controls the slider of the linear slide (1) to move to the trigger position of the photoelectric sensor (2), and the photoelectric sensor (2) generates a position trigger signal and sends it to the controller. 4-2) The controller controls the movement of the cylinder (3). The cylinder extends forward and drives the V-shaped baffle (4) to contact the log (6). After the V-shaped baffle (4) closes, the pressure sensor (5) set on the V-shaped baffle (4) is triggered and generates a pressure signal, and sends the pressure signal to the controller. The controller controls the cylinder (3) to stop moving. 4-3) The controller detects the extension distance of the cylinder shaft of the current cylinder (3), that is, the zero point position of the cross-sectional surface of the log, and sends the zero point position of the cross-sectional surface of the log to the control terminal.

Images