A method and system for analyzing and managing data of a wood industry based on machine learning

Abstract

The application discloses a kind of based on machine learning's timber industry data analysis management method and system, it is related to timber industry data management field, the present application is adjusted according to the edge range of the defect type and the defect existing in the internal of the log to be managed, the coefficient of the plane equation of the cut surface located at the both sides of defect when sawing to the log to be managed and the parameter data when sawing are adjusted, so that the difference between the strength of the wood section after sawing and the strength of the wood section before sawing is less than the preset threshold, to avoid the quality of the log in the sawing process will have greater influence, to ensure the quality of wood processing;Wherein, by clustering the parameter data of several use scenarios of wood, it provides a basis for determining whether the use scenario of the defect part of the log to be managed is reasonable, to ensure that the defect part of the log to be managed is applied to the most suitable use scenario, to avoid safety hazards in the use process.

Description

A Machine Learning-Based Data Analysis and Management Method and System for the Timber Industry Technical Field

[0001] This invention belongs to the field of timber industry data management, and specifically relates to a timber industry data analysis and management method and system based on machine learning. Background Technology

[0002] In the timber industry, sawing is an essential operation. However, some timber has defects such as knots, decay, and cracks. If the sawing position or parameters, such as the cutting speed, are not appropriate when sawing the timber, it may affect the structural integrity and strength of the timber, leading to safety hazards during use, such as cracks, warping, or decay, which could threaten the lives of users. Summary of the Invention

[0003] In response to the problems in related technologies, this invention proposes a data analysis and management method and system for the timber industry based on machine learning, so as to overcome the aforementioned technical problems existing in the existing related technologies.

[0004] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution:

[0005] This invention relates to a data analysis and management method for the timber industry based on machine learning, comprising the following steps:

[0006] S1. Collect X-ray scan images from multiple angles at multiple image acquisition areas on the surface of the log to be managed to obtain the original scan image data matrix of the log to be managed;

[0007] S2. Preprocess the original scanned image data matrix to be managed to obtain the final scanned image data matrix to be managed.

[0008] S3. Compare the final scanned image data matrix to be managed with the standard defect image data to obtain the final scanned image defect type data matrix to be managed; collect parameter data of several usage scenarios corresponding to each defect type and calculate the corresponding cluster center data to obtain the historical defect scenario feature parameter center data matrix.

[0009] S4. Collect plane equation coefficient data, defect edge coordinate data, sawing parameter data, and average strength difference data of the normal part before and after sawing of multiple test logs with single defects. Construct a strength difference mapping model for knot defects, a strength difference mapping model for decay defects, and a strength difference mapping model for crack defects.

[0010] S5. Based on the mapping models of knot defect strength difference, decay defect strength difference, and crack defect strength difference, the initial coefficient data, initial sawing parameter data, and defect edge coordinate data of the two plane equations of the logs to be managed during sawing are mapped to obtain the initial strength difference data to be managed.

[0011] S6. Based on the initial strength difference data to be managed, adjust the initial coefficient data of the plane equation of the two cut surfaces when sawing the logs to be managed, and the initial sawing parameter data to obtain the final cut surface equation coefficient dataset and the final sawing parameter dataset to be managed; then adjust and replace the usage scenario of the defective parts of the logs to be managed.

[0012] Preferably, step S1 includes the following steps:

[0013] S11. Set the log to be managed and set that the log to be managed has only a single type of defect; divide the surface of the log to be managed into multiple image acquisition areas to obtain the current image acquisition area set;

[0014] S12. In conjunction with the current image acquisition area set, use an X-ray scanning image acquisition system to acquire X-ray scan images from multiple angles at each image acquisition area on the log to be managed, and obtain the original scan image data matrix to be managed.

[0015] Preferably, step S2 includes the following steps:

[0016] S21. Perform grayscale conversion on each original scanned image data in the original scanned image data matrix to be managed to obtain the grayscale converted scanned image data matrix to be managed.

[0017] S22. The median filter and the Gaussian filter are used to perform frequency domain filtering and spatial domain filtering operations on each scan image data in the grayscale-converted scan image data matrix to be managed, respectively, to obtain the filtered scan image data matrix to be managed.

[0018] S23. Perform mathematical morphological processing on the filtered scan image data matrix to be managed to obtain the final scan image data matrix to be managed.

[0019] Preferably, step S3 includes the following steps:

[0020] S31. Set several types of wood defects and several corresponding standard wood defect images to obtain a set of wood defect types and a wood defect image data matrix; compare the final scan image data matrix to be managed with the multiple wood defect image data corresponding to each type of wood defect in the wood defect image data matrix and obtain the corresponding defect type to obtain the final scan image defect type data matrix to be managed.

[0021] S32. Set several types of characteristic parameters for timber use scenarios to obtain a set of characteristic parameters for timber use scenarios; collect characteristic parameter data of multiple timber use scenarios corresponding to each timber defect type from the historical timber defect type set according to the set of characteristic parameters for timber use scenarios to obtain a set of historical defect scenario characteristic parameter data matrix.

[0022] S33. Calculate the cluster center data of each historical defect scene feature parameter data matrix in the historical defect scene feature parameter data matrix set to obtain the historical defect scene feature parameter center data matrix.

[0023] Preferably, step S33 includes the following steps:

[0024] S331. Construct a cluster of tunicate species based on scene defect feature parameters; set the maximum number of iterations for the cluster of tunicate species based on scene defect feature parameters to be c~1 and the current number of iterations to be c~2, which are respectively denoted as the maximum number of iterations for parameter clustering and the current number of iterations for parameter clustering.

[0025] S332. Randomly select one row of data from each historical defect scene feature parameter data matrix in the historical defect scene feature parameter data matrix set multiple times as the initial position of each salver in the scene defect feature parameter clustering salver population, and obtain the first initial position matrix set.

[0026] S333. Calculate the Euclidean distance between each row of data in the historical defect scene feature parameter data matrix set and the corresponding row in the initial position matrix of the kth salver in the scene defect feature parameter clustering salver population, and obtain the Euclidean distance data matrix.

[0027] Based on the Euclidean distance data matrix, construct the fitness function of the kth salver in the salver population by clustering scene defect feature parameters;

[0028] S334. Start the iteration. Before the iteration, set the current iteration number of the parameter clustering to 1. During the first iteration, use the fitness function of the kth salps in the salps population to calculate the fitness value of the initial position matrix of each salps in the first initial position matrix set, and obtain the first fitness value set. Take the largest fitness value in the first fitness value set and the corresponding initial position matrix of the salps as the first global best fitness and the first global best position, respectively. Update the initial position matrix of each salps in the first initial position matrix set according to the first global best fitness and the first global best position. After the update is completed, increment the current iteration number of the parameter clustering by 1 and enter the next iteration.

[0029] In each iteration, the fitness function of the kth salver in the clustered salver population based on scene defect feature parameters is used to calculate the fitness value of the position matrix of each salver in the clustered salver population based on scene defect feature parameters updated in the previous iteration, thus obtaining a second fitness value set. The maximum fitness value in the second fitness value set and the corresponding salver position matrix are taken as the second global best fitness and the second global best position, respectively. The position matrix of each salver in the clustered salver population based on scene defect feature parameters updated in the previous iteration is updated according to the second global best fitness and the second global best position. After the update is completed, the current iteration number of the parameter clustering is incremented by 1 and the next iteration is started.

[0030] S335, when If the first optimal position is found, stop iterating and obtain the first final global optimal position; otherwise, continue iterating until... Up to this point; the first final global optimal position is used as the center data matrix of historical defect scene feature parameters;

[0031] The salps optimization algorithm searches for the optimal solution by simulating the reproduction, filtering, feeding, and migration behaviors of salps. It can maintain good performance and exhibit excellent robustness in complex and ever-changing environments. Based on the above advantages, this scheme uses the salps optimization algorithm to iteratively adjust the cluster centers of each historical defect scene feature parameter data matrix in the historical defect scene feature parameter data matrix set. The overall dispersion of the historical defect scene feature parameter data matrix set is used as the fitness function. As the iteration progresses, the overall dispersion of the historical defect scene feature parameter data matrix set becomes smaller and smaller.

[0032] Preferably, S4 includes the following guarantees:

[0033] S41. Based on the set of wood defect types, set up multiple test logs with a single corresponding defect, and divide multiple image acquisition areas on their surface. Then, use the X-ray scanning image acquisition system in S12 to acquire corresponding scanning image data at each image acquisition area on each test log to be sawed. Process each acquired scanning image data in S21, S22 and S23 to obtain a matrix of scan image datasets of test logs to be sawed.

[0034] S42. When sawing each test log to be sawed, a three-dimensional coordinate system is constructed for each test log to be sawed, and a test three-dimensional coordinate system matrix is ​​obtained; in conjunction with the test three-dimensional coordinate system matrix and according to the scan image dataset matrix of the test log to be sawed, several coordinate acquisition points are set at the defect edges in each test log to be sawed and the corresponding coordinate data is collected, and a test defect edge coordinate dataset matrix is ​​obtained; then, in conjunction with the test three-dimensional coordinate system matrix, the plane equation coefficients corresponding to the two cut surfaces during sawing are collected, and a test cut surface equation coefficient data matrix is ​​obtained.

[0035] S43. Set several sawing parameter types to obtain a sawing parameter type set; the sawing parameter type set includes cutter speed and cutter displacement speed, etc.; collect parameter data when sawing each test log to be sawed according to the sawing parameter type set to obtain a test sawing parameter dataset matrix.

[0036] Then, the average strength difference data of the normal part before and after sawing of each log to be sawn is collected to obtain the average data matrix of test strength difference.

[0037] S44. Construct a node defect strength difference mapping model, a decay defect strength difference mapping model, and a crack defect strength difference mapping model based on the test defect edge coordinate dataset matrix, the test section equation coefficient data matrix, the test sawing parameter dataset matrix, and the test strength difference average data matrix.

[0038] Preferably, S44 includes the following steps:

[0039] S441. Construct a first initial SVM model, a second initial SVM model, and a third initial SVM model, and set a first training data ratio, a second training data ratio, and a third training data ratio; according to the first training data ratio, the second training data ratio, and the third training data ratio, divide the data corresponding to the nodule defect type, decay defect type, and crack defect type in the test defect edge coordinate dataset matrix, the test section equation coefficient data matrix, the test sawing parameter dataset matrix, and the test strength difference average data matrix, respectively, to obtain the test defect edge coordinate training dataset matrix, the test section equation coefficient training dataset matrix, the test sawing parameter training dataset matrix, the test strength difference average training data matrix, the test defect edge coordinate test dataset matrix, the test section equation coefficient test data matrix, the test sawing parameter test dataset matrix, and the test strength difference average test data matrix.

[0040] S442. Using the data corresponding to the nodule defect type, decay defect type, and crack defect type in the test defect edge coordinate training dataset matrix, test section equation coefficient training dataset matrix, test sawing parameter training dataset matrix, and test strength difference average training dataset matrix, respectively, train and test the first initial SVM model, the second initial SVM model, and the third initial SVM model; after training and testing, obtain the nodule defect strength difference mapping model, the decay defect strength difference mapping model, and the crack defect strength difference mapping model.

[0041] Preferably, step S5 includes the following steps:

[0042] S51. When sawing the log to be managed, a three-dimensional coordinate system of the log to be managed is constructed to obtain the three-dimensional coordinate system to be managed; according to the defect type data matrix of the final scanned image to be managed, several coordinate points are set at the edge of the defect in the log to be managed to obtain the defect edge coordinate dataset to be managed; in conjunction with the sawing parameter type set, the initial coefficient data of the plane equation of the two cut surfaces when sawing the log to be managed and the initial sawing parameter data are randomly set to obtain the initial cut surface equation coefficient dataset and the initial sawing parameter dataset to be managed;

[0043] S52. Determine the defect type of the log to be managed based on the final scanned image defect type data matrix, and the first defect type of the log to be managed;

[0044] When the first log defect type to be managed is a knot defect, the dataset of edge coordinates of the defect to be managed, the dataset of initial cross-sectional equation coefficients to be managed, and the dataset of initial sawing parameters to be managed are input into the knot defect strength difference mapping model for mapping; when the first log defect type to be managed is a decay defect, the dataset of edge coordinates of the defect to be managed, the dataset of initial cross-sectional equation coefficients to be managed, and the dataset of initial sawing parameters to be managed are input into the decay defect strength difference mapping model for mapping; when the first log defect type to be managed is a crack defect, the dataset of edge coordinates of the defect to be managed, the dataset of initial cross-sectional equation coefficients to be managed, and the dataset of initial sawing parameters to be managed are input into the crack defect strength difference mapping model for mapping; after mapping, the initial strength difference data to be managed is obtained.

[0045] Preferably, step S6 includes the following steps:

[0046] S61. Set a first intensity difference threshold, a second intensity difference threshold, and a third intensity difference threshold;

[0047] S62. When the defect type of the first log to be managed is a knot defect and the initial strength difference data is greater than or equal to the first strength difference threshold, or when the defect type of the first log to be managed is a decay defect and the initial strength difference data is greater than or equal to the second strength difference threshold, or when the defect type of the first log to be managed is a crack defect and the initial strength difference data is greater than or equal to the third strength difference threshold, the initial cross-sectional equation coefficient dataset and the initial sawing parameter dataset are adjusted until the first log to be managed... The final cross-sectional equation coefficient dataset and the final sawing parameter dataset are obtained when the defect type is a knot defect and the initial strength difference data to be managed is less than the first strength difference threshold, or when the first log defect type to be managed is a decay defect and the initial strength difference data to be managed is less than the second strength difference threshold, or when the first log defect type to be managed is a crack defect and the initial strength difference data to be managed is less than the third strength difference threshold; otherwise, no adjustment is needed to the initial cross-sectional equation coefficient dataset and the initial sawing parameter dataset to be managed.

[0048] S63. After adjusting the initial cross-section equation coefficient dataset and the initial sawing parameter dataset to be managed, the logs to be managed are sawn in conjunction with the final cross-section equation coefficient dataset and the final sawing parameter dataset to be managed.

[0049] By combining the set of characteristic parameters of timber usage scenarios with the data of characteristic parameters of the defective parts of the logs to be managed after processing, a dataset of characteristic parameters of usage scenarios to be managed is obtained.

[0050] S64. Calculate the Euclidean distance between the feature parameter dataset of the usage scenario to be managed and each row of data in the central data matrix of the feature parameter of the historical defect scenario to obtain the first Euclidean distance dataset; take the defect type corresponding to the smallest Euclidean distance data in the first Euclidean distance dataset as the second log defect type to be managed.

[0051] When the defect type of the second log to be managed is inconsistent with the defect type of the first log to be managed, the usage scenario of the defective part of the log to be managed after processing shall be changed; otherwise, it is not necessary to change the usage scenario of the defective part of the log to be managed after processing.

[0052] A machine learning-based data analysis and management system for the timber industry includes a module for acquiring scanned images of logs to be managed, an image preprocessing module, a module for acquiring defect type data of logs to be managed, a module for setting scene parameter types, a module for acquiring historical scene parameters, a module for calculating cluster centers, a module for acquiring sawing test data, a module for constructing a mapping model, a mapping module, a module for adjusting sawing data of logs to be managed, and a module for adjusting usage scenarios.

[0053] The present invention has the following beneficial effects:

[0054] 1. In this invention, based on the type of defects present inside the log to be managed and the edge range of the defects, the coefficients of the plane equations of the cut surfaces on both sides of the defects and the parameter data during sawing are adjusted so that the difference between the strength of the sawn wood segment and the strength of the wood segment before sawing is less than a preset threshold, thereby avoiding a significant impact on the quality of the log during sawing and ensuring the quality of wood processing.

[0055] 2. In this invention, the parameter data of several usage scenarios of wood corresponding to different types of wood defects are clustered by using the sea squirt optimization algorithm. This provides a basis for determining whether the usage scenarios of the defective parts of the logs to be managed are reasonable, thereby ensuring that the defective parts of the logs to be managed are applied to the most suitable usage scenarios and avoiding safety hazards during use.

[0056] 3. In this invention, grayscale conversion, frequency domain filtering, spatial domain filtering, and mathematical morphology processing are performed on the original scanned image data of the logs to be managed, providing high-quality data for subsequent determination of the defect type of the logs to be managed based on the scanned image data.

[0057] 4. In this invention, by constructing a strength difference mapping model for knot defects, a strength difference mapping model for decay defects, and a strength difference mapping model for crack defects for different defect types, a judgment model is provided to facilitate the subsequent sawing of logs with a certain defect type to determine whether the sawing parameters and the coefficients of the section equation are reasonable.

[0058] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0059] To more clearly illustrate the technical solutions of the embodiments of the invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For those skilled in the art, the drawings can be obtained from these drawings without creative effort.

[0060] Figure 1 is a flowchart illustrating a machine learning-based data analysis and management method for the timber industry according to the present invention. Detailed Implementation

[0061] The technical solutions of the embodiments of the invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the invention, and not all embodiments. Based on the embodiments of the invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the invention.

[0062] Example 1

[0063] Please refer to Figure 1. This embodiment illustrates a data analysis and management method for the timber industry based on machine learning, comprising the following steps:

[0064] S1. Collect X-ray scan images from multiple angles at multiple image acquisition areas on the surface of the log to be managed to obtain the original scan image data matrix of the log to be managed;

[0065] S1 includes the following steps:

[0066] S11. Define the log to be managed and set that the log to be managed has only a single type of defect; divide the surface of the log to be managed into multiple image acquisition areas to obtain the current image acquisition area set a1 = {a 11 ,...,a 1i ,...,a 1a′}, a 1i represents the i-th image acquisition region divided on the log to be managed, and a′ represents the total number of image acquisition regions divided on the log to be managed;

[0067] S12. In conjunction with the current image acquisition area set, use an X-ray scanning image acquisition system to acquire X-ray scan images from multiple angles at each image acquisition area on the log to be managed, obtaining the original scan image data matrix a2 to be managed; as follows.

[0068]

[0069] Among them, a 2ij This represents the j-th raw scan image data of the i-th image acquisition area on the log to be managed. This represents the total number of raw scanned image data collected at each image acquisition area on the log to be managed;

[0070] S2. Preprocess the original scanned image data matrix to be managed to obtain the final scanned image data matrix to be managed.

[0071] S2 includes the following steps:

[0072] S21. Perform grayscale conversion on each original scanned image data in the original scanned image data matrix to be managed to obtain the grayscale converted scanned image data matrix to be managed.

[0073] S22. The median filter and the Gaussian filter are used to perform frequency domain filtering and spatial domain filtering operations on each scan image data in the grayscale-converted scan image data matrix to be managed, respectively, to obtain the filtered scan image data matrix to be managed.

[0074] S23. Perform mathematical morphological processing on the filtered scan image data matrix to be managed to obtain the final scan image data matrix a3 to be managed.

[0075]

[0076] Among them, a 3ij Indicates that for a 2ij The final scanned image data to be managed is obtained after grayscale conversion, filtering, and mathematical morphology processing.

[0077] S3. Compare the final scanned image data matrix to be managed with the standard defect image data to obtain the final scanned image defect type data matrix to be managed; collect parameter data of several usage scenarios corresponding to each defect type and calculate the corresponding cluster center data to obtain the historical defect scenario feature parameter center data matrix.

[0078] S3 includes the following steps:

[0079] S31. Define several types of wood defects and corresponding standard wood defect images to obtain a set of wood defect types. And a matrix of wood defect image data; These represent the types of defects: knots, decay, cracks, and non-defects, with values ​​of 1, 2, 3, and 4, respectively. The final managed scan image data matrix is ​​compared with multiple wood defect image data corresponding to each wood defect type in the wood defect image data matrix to obtain the corresponding defect type, resulting in the final managed scan image defect type data matrix a4; as follows.

[0080]

[0081] Among them, a 4ij Indicates a 3ij Corresponding image defect type data;

[0082] S32. Define several types of characteristic parameters for different timber use scenarios to obtain a set of timber use scenario characteristic parameter types b = {b1,...,b...} i ,...,b b′}, b i Let represent the set characteristic parameter type for the i-th type of timber use scenario, and b′ represent the total number of set characteristic parameter types for timber use scenarios. The set of characteristic parameter types for timber use scenarios includes requirements for strength, load-bearing capacity, and decay characteristics, such as the creation of decay-style artworks. Based on the set of characteristic parameter types for timber use scenarios, characteristic parameter data for multiple timber use scenarios corresponding to each timber defect type are collected from the historical timber defect type set to obtain a historical defect scenario characteristic parameter data matrix set. These represent the historical defect scene feature parameter data matrices corresponding to knot defect type, decay defect type, and crack defect type in the wood defect type set, respectively, as follows:

[0083]

[0084]

[0085] in, These represent the feature parameter data for the j-th type of the i-th timber in the usage scenario corresponding to the knot defect type, decay defect type, and crack defect type in the timber defect type set. This represents the total number of timbers corresponding to the knot defect type, decay defect type, and crack defect type in the collected timber defect type set.

[0086] S33. Calculate the cluster center data for each historical defect scene feature parameter data matrix in the historical defect scene feature parameter data matrix set, to obtain the historical defect scene feature parameter center data matrix c; as follows.

[0087]

[0088] Among them, c 1i c 2i c 3i These represent the feature parameter data of the i-th type of the cluster center dataset corresponding to the historical defect scene feature parameter data matrix for knot defect type, decay defect type and crack defect type in the wood defect type set, respectively.

[0089] S33 includes the following steps:

[0090] S331. Constructing a cluster of *Sulphurus salicifolius* populations based on scene defect feature parameters. c1′i This indicates the i-th salver in the salver population clustered by the defect feature parameters of the described scene. This indicates the size of the clustered tunicate population based on the scene defect feature parameters; the maximum number of iterations for clustering the tunicate population based on the scene defect feature parameters is set to [value missing]. And the current iteration number is These are denoted as the maximum number of iterations for parameter clustering and the current number of iterations for parameter clustering, respectively; the search space dimension for the tunicate population in the scene defect feature parameter clustering is 3·b′;

[0091] S332. Randomly select one row of data from each historical defect scene feature parameter data matrix in the historical defect scene feature parameter data matrix set multiple times as the initial position of each salver in the scene defect feature parameter clustering salver population, to obtain the first initial position matrix set. d 1k The matrix representing the initial position of the k-th salver in the clustered salver population representing the scene defect feature parameters is as follows:

[0092]

[0093] Where, d 1k1j d 1k2j d 1k3j These represent the positional components of the initial position of the kth salver in the clustered salver population of the scene defect feature parameters, respectively, on the feature parameter data dimension of the j-th type of the cluster center dataset corresponding to the historical defect scene feature parameter data matrix corresponding to the knot defect type, decay defect type, and crack defect type in the wood defect type set.

[0094] S333. Calculate the Euclidean distance between each row of data in the historical defect scene feature parameter data matrix set and the corresponding row in the initial position matrix of the kth tunicate in the scene defect feature parameter clustering tunicate population, to obtain the Euclidean distance data matrix d. k ';as follows,

[0095]

[0096] Where, d k ′ 1i d k ′ 2i d k ′ 3iThis represents the Euclidean distance between the i-th row of the 1st, 2nd, and 3rd historical defect scene feature parameter data matrices in the historical defect scene feature parameter data matrix set and the 1st, 2nd, and 3rd rows of the initial position matrix of the k-th salver in the scene defect feature parameter clustering salver population; the calculation formulas are as follows.

[0097]

[0098]

[0099] Based on the Euclidean distance data matrix, a fitness function for the k-th salver in the salver population is constructed using scene defect feature parameters clustering. as follows,

[0100]

[0101] S334. Begin iteration. Before iteration, set the current iteration number of the parameter clustering to 1. During the first iteration, use the scene defect feature parameters to cluster the fitness function of the kth salver in the salver population. Calculate the fitness value of the initial position matrix of each salps in the first initial position matrix set to obtain the first fitness value set; take the largest fitness value in the first fitness value set and the corresponding initial position matrix of the salps as the first global best fitness and the first global best position, respectively; update the initial position matrix of each salps in the first initial position matrix set according to the first global best fitness and the first global best position; after the update is completed, increment the current iteration number of the parameter clustering by 1 and enter the next iteration.

[0102] In each iteration, the fitness function of the kth salver in the salver population is used to cluster the scene defect feature parameters. Calculate the fitness value of the position matrix of each salver in the clustered salver population obtained in the previous iteration, and obtain the second fitness value set; take the largest fitness value in the second fitness value set and the corresponding salver position matrix as the second global best fitness and the second global best position, respectively; update the position matrix of each salver in the clustered salver population obtained in the previous iteration according to the second global best fitness and the second global best position; after the update is completed, increment the current iteration number of the parameter clustering by 1 and enter the next iteration.

[0103] S335, when If the first optimal position is found, stop iterating and obtain the first final global optimal position; otherwise, continue iterating until... Up to this point; the first final global optimal position is used as the center data matrix of historical defect scene feature parameters;

[0104] S4. Collect plane equation coefficient data, defect edge coordinate data, sawing parameter data, and average strength difference data of the normal part before and after sawing of multiple test logs with single defects. Construct a strength difference mapping model for knot defects, a strength difference mapping model for decay defects, and a strength difference mapping model for crack defects.

[0105] S4 includes the following guarantees:

[0106] S41. Based on the set of wood defect types, set up multiple test logs with a single corresponding defect, and divide their surfaces into multiple image acquisition areas. Then, use the X-ray scanning image acquisition system in S12 to acquire corresponding scan image data for each image acquisition area on each test log. Process each acquired scan image data using methods S21, S22, and S23 to obtain a matrix of scan image datasets for the test logs to be sawed. as follows,

[0107]

[0108] in, e1 represents the scanned image dataset of the i-th test log to be sawn for the type of knot defect, decay defect, and crack defect, which has been collected and processed. e1 represents the total number of test logs to be sawn for each type of wood defect.

[0109] They represent The scanned image data corresponding to the j-th image acquisition area in the image, e2 represents The total amount of scanned image data;

[0110] S42. When sawing each test log to be sawn, a three-dimensional coordinate system is constructed for each test log to be sawn, and a test three-dimensional coordinate system matrix is ​​obtained; in conjunction with the test three-dimensional coordinate system matrix and the scanned image dataset matrix of the test log to be sawn. Several coordinate acquisition points were set at the defect edges of each log to be sawed for testing, and the corresponding coordinate data were collected to obtain the test defect edge coordinate dataset matrix e1′; then, in conjunction with the test three-dimensional coordinate system matrix, the plane equation coefficients corresponding to the two cut surfaces during sawing were collected to obtain the test cut surface equation coefficient data matrix e′2; e1′ and e′2 are as follows.

[0111]

[0112] Among them, e1′ 1i e1′ 2i e1′ 2i Let e′ represent the dataset of defect edge coordinates of the i-th test log to be sawn, corresponding to the knot defect type, decay defect type, and crack defect type, respectively. 21i1 、e′ 22i1 、e′ 23i1 Let e′ represent the coefficient set of the first section equation of the i-th log to be sawn for the defects of knot, decay, and crack, respectively. 21i2 、e′ 22i2 、e′ 23i3 e1′ represents the set of coefficients for the second section equation of the i-th log to be sawn, corresponding to the knot defect type, decay defect type, and crack defect type, respectively. 1i e1′ 2i e1′ 3i 、e′ 21i1 、e′ 22i1 、e′ 23i1 、e′ 21i2 、e′ 22i2 、e′ 23i3 They are as follows:

[0113]

[0114]

[0115]

[0116] Among them, e1′ 1ij1 e1′ 1ij2 e1′ 1ij3 e1′ 2ij1 e1′ 2ij2 e1′ 2ij3 e1′ 3ij1 e1′ 3ij2 e1′ 3ij3 They represent e1′ respectively 1i e1′ 2i e1′ 3i The 1D, 2D, and 3D coordinate data of the j-th defect edge point. This represents the total number of coordinate collection points set at the defect edge of each log to be sawed for testing;

[0117] S43. Set several sawing parameter types to obtain a sawing parameter type set; the sawing parameter type set includes cutter speed and cutter displacement speed, etc.; collect parameter data when sawing each test log to be sawed according to the sawing parameter type set to obtain the test sawing parameter dataset matrix e3′; as follows.

[0118]

[0119] Among them, e3′ 1i e3′ 2i e3′ 2i These represent the sawing parameter datasets when sawing the i-th test log corresponding to the knot defect type, decay defect type, and crack defect type, respectively. e3′ 1ij e3′ 2ij e3′ 3ij They represent e3′ respectively 1i e3′ 2i e3′ 2i The j-th type of sawing parameter data;

[0120] Next, the average strength difference data of the normal parts before and after sawing of each log to be sawn was collected, resulting in the average strength difference data matrix e′4, as shown below.

[0121]

[0122] Among them, e′ 41i 、e′ 42i 、e′ 42i These represent the average strength difference data of the normal part before and after sawing the i-th test log corresponding to the knot defect type, decay defect type, and crack defect type, respectively.

[0123] S44. Construct a node defect strength difference mapping model, a decay defect strength difference mapping model, and a crack defect strength difference mapping model based on the test defect edge coordinate dataset matrix, the test section equation coefficient data matrix, the test sawing parameter dataset matrix, and the test strength difference average data matrix.

[0124] S44 includes the following steps:

[0125] S441. Construct a first initial SVM model, a second initial SVM model, and a third initial SVM model, and set a first training data ratio, a second training data ratio, and a third training data ratio; according to the first training data ratio, the second training data ratio, and the third training data ratio, divide the data corresponding to the nodule defect type, decay defect type, and crack defect type in the test defect edge coordinate dataset matrix, the test section equation coefficient data matrix, the test sawing parameter dataset matrix, and the test strength difference average data matrix, respectively, to obtain the test defect edge coordinate training dataset matrix, the test section equation coefficient training dataset matrix, the test sawing parameter training dataset matrix, the test strength difference average training data matrix, the test defect edge coordinate test dataset matrix, the test section equation coefficient test data matrix, the test sawing parameter test dataset matrix, and the test strength difference average test data matrix.

[0126] S442. Using the data corresponding to the nodule defect type, decay defect type, and crack defect type in the test defect edge coordinate training dataset matrix, test section equation coefficient training dataset matrix, test sawing parameter training dataset matrix, and test strength difference average training dataset matrix, respectively, train and test the first initial SVM model, the second initial SVM model, and the third initial SVM model; after training and testing, obtain the nodule defect strength difference mapping model, the decay defect strength difference mapping model, and the crack defect strength difference mapping model.

[0127] S5. Based on the mapping models of knot defect strength difference, decay defect strength difference, and crack defect strength difference, the initial coefficient data, initial sawing parameter data, and defect edge coordinate data of the two plane equations of the logs to be managed during sawing are mapped to obtain the initial strength difference data to be managed.

[0128] S5 includes the following steps:

[0129] S51. When sawing the logs to be managed, a three-dimensional coordinate system of the logs to be managed is constructed to obtain the three-dimensional coordinate system to be managed; according to the defect type data matrix of the final scanned image to be managed, several coordinate points are set at the edges of the defects in the logs to be managed to obtain the defect edge coordinate dataset to be managed. as follows,

[0130]

[0131] in, The first, second, and third dimensions represent the 1st, second, and third dimensions of the coordinate point i-th at the edge of the defect in the log to be managed; combined with the initial coefficient data of the plane equations of the two cut surfaces when sawing the log to be managed, and the initial sawing parameter data, the initial cut surface equation coefficient dataset is obtained by randomly setting the initial coefficient data of the log to be managed with respect to the sawing parameter type set. and the initial sawing parameter dataset to be managed Let x, y, z, and constants represent the coefficients of the independent variable x, y, and z of the plane equation for the first cut surface when the log to be managed is sawn. The coefficients of the x-variable, y-variable, and z-variable, as well as the constants, are respectively represented by the plane equation of the second cut surface when the log to be managed is sawn. This represents the initial parameter data for the i-th type of sawing when sawing the logs to be managed;

[0132] S52. Determine the defect type of the log to be managed based on the final scanned image defect type data matrix, and the first defect type of the log to be managed;

[0133] When the first log defect type to be managed is a knot defect, the dataset of edge coordinates of the defect to be managed, the dataset of initial cross-sectional equation coefficients to be managed, and the dataset of initial sawing parameters to be managed are input into the knot defect strength difference mapping model for mapping; when the first log defect type to be managed is a decay defect, the dataset of edge coordinates of the defect to be managed, the dataset of initial cross-sectional equation coefficients to be managed, and the dataset of initial sawing parameters to be managed are input into the decay defect strength difference mapping model for mapping; when the first log defect type to be managed is a crack defect, the dataset of edge coordinates of the defect to be managed, the dataset of initial cross-sectional equation coefficients to be managed, and the dataset of initial sawing parameters to be managed are input into the crack defect strength difference mapping model for mapping; after mapping, the initial strength difference data to be managed is obtained.

[0134] S6. Based on the initial strength difference data to be managed, adjust the initial coefficient data of the plane equation of the two cut surfaces when sawing the logs to be managed, and the initial sawing parameter data to obtain the final cut surface equation coefficient dataset and the final sawing parameter dataset to be managed; then adjust and replace the usage scenario of the defective parts of the logs to be managed.

[0135] S6 includes the following steps:

[0136] S61. Set a first intensity difference threshold, a second intensity difference threshold, and a third intensity difference threshold;

[0137] S62. When the defect type of the first log to be managed is a knot defect and the initial strength difference data is greater than or equal to the first strength difference threshold, or when the defect type of the first log to be managed is a decay defect and the initial strength difference data is greater than or equal to the second strength difference threshold, or when the defect type of the first log to be managed is a crack defect and the initial strength difference data is greater than or equal to the third strength difference threshold, the initial cross-sectional equation coefficient dataset and the initial sawing parameter dataset are adjusted until the first log to be managed... The final cross-sectional equation coefficient dataset and the final sawing parameter dataset are obtained when the defect type is a knot defect and the initial strength difference data to be managed is less than the first strength difference threshold, or when the first log defect type to be managed is a decay defect and the initial strength difference data to be managed is less than the second strength difference threshold, or when the first log defect type to be managed is a crack defect and the initial strength difference data to be managed is less than the third strength difference threshold; otherwise, no adjustment is needed to the initial cross-sectional equation coefficient dataset and the initial sawing parameter dataset to be managed.

[0138] S63. After adjusting the initial cross-section equation coefficient dataset and the initial sawing parameter dataset to be managed, the logs to be managed are sawn in conjunction with the final cross-section equation coefficient dataset and the final sawing parameter dataset to be managed.

[0139] By combining the aforementioned set of timber usage scenario characteristic parameters with the usage scenario characteristic parameter data of the defective parts of the logs to be managed after processing, a dataset of usage scenario characteristic parameters to be managed, f = {f1,...,f...}, is obtained. i ,...,f b′}, f i This represents the i-th type of parameter data representing the usage scenario after the logs to be managed have been processed.

[0140] S64. Calculate the Euclidean distance between the feature parameter dataset of the usage scenario to be managed and each row of data in the central data matrix of the feature parameter of the historical defect scenario to obtain the first Euclidean distance dataset; take the defect type corresponding to the smallest Euclidean distance data in the first Euclidean distance dataset as the second log defect type to be managed.

[0141] When the defect type of the second log to be managed is inconsistent with the defect type of the first log to be managed, the usage scenario of the defective part of the log to be managed after processing shall be changed; otherwise, it is not necessary to change the usage scenario of the defective part of the log to be managed after processing.

[0142] Example 2

[0143] This embodiment discloses a data analysis and management system for the timber industry based on machine learning. The system can implement the methods of the above embodiment, including a log scan image data acquisition module, an image preprocessing module, a log defect type data acquisition module, a scene parameter type setting module, a historical scene parameter acquisition module, a cluster center calculation module, a sawing test data acquisition module, a mapping model construction module, a mapping module, a log sawing data adjustment module, and a usage scenario adjustment module.

[0144] The log scanning image data acquisition module is used to acquire X-ray scanning images from multiple angles at multiple image acquisition areas on the surface of the log to be managed, and obtain the original scanning image data matrix of the log to be managed.

[0145] The image preprocessing module is used to perform grayscale conversion, frequency domain filtering, spatial domain filtering, and mathematical morphology processing on the original scanned image data matrix to be managed, so as to obtain the final scanned image data matrix to be managed.

[0146] The log defect type data acquisition module is used to compare the final scanned image data matrix to be managed with the standard defect image data to obtain the final scanned image defect type data matrix to be managed.

[0147] The scenario parameter type setting module is used to set several types of timber use scenario feature parameters to obtain a set of timber use scenario feature parameter types.

[0148] The historical scene parameter acquisition module is used to acquire parameter data of several usage scenarios corresponding to each defect type according to the timber usage scenario feature parameter type set, and obtain a historical defect scene feature parameter data matrix set.

[0149] The cluster center calculation module is used to calculate the cluster center data of each historical defect scene feature parameter data matrix in the historical defect scene feature parameter data matrix set, so as to obtain the historical defect scene feature parameter center data matrix.

[0150] The sawing test data acquisition module is used to collect the plane equation coefficient data, defect edge coordinate data, sawing parameter data, and the average strength difference data of the normal part before and after sawing when sawing multiple logs with single defects. The module is used to obtain the test defect edge coordinate dataset matrix, the test cut surface equation coefficient data matrix, the test sawing parameter dataset matrix, and the test strength difference average data matrix.

[0151] The mapping model construction module is used to construct a node defect strength difference mapping model, a decay defect strength difference mapping model, and a crack defect strength difference mapping model using a test defect edge coordinate dataset matrix, a test section equation coefficient data matrix, a test sawing parameter dataset matrix, and a test strength difference average data matrix.

[0152] The mapping module is used to map the initial coefficient data, initial sawing parameter data, and defect edge coordinate data of the two plane equations of the logs to be managed when sawing them according to the knot defect strength difference mapping model, decay defect strength difference mapping model, and crack defect strength difference mapping model, so as to obtain the initial strength difference data to be managed.

[0153] The log sawing data adjustment module is used to adjust the initial coefficient data of the plane equation of the two cut surfaces when sawing the logs under management based on the initial strength difference data, and the initial sawing parameter data, to obtain the final cut surface equation coefficient dataset and the final sawing parameter dataset.

[0154] The usage scenario adjustment module is used to adjust and replace the usage scenarios of the defective parts of the logs to be managed, in conjunction with the central data matrix of historical defect scenario feature parameters.

[0155] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0156] The preferred embodiments of the invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.

Claims

1. A data analysis and management method for the timber industry based on machine learning, characterized in that, Includes the following steps: S1. Collect X-ray scan images from multiple angles at multiple image acquisition areas on the surface of the log to be managed to obtain the original scan image data matrix of the log to be managed; S2. Preprocess the original scanned image data matrix to be managed to obtain the final scanned image data matrix to be managed; S3. Compare the final scanned image data matrix to be managed with the standard defect image data to obtain the final scanned image defect type data matrix to be managed; collect parameter data for several usage scenarios corresponding to each defect type and calculate the corresponding cluster center data to obtain the historical defect scenario feature parameter center data matrix; S4. Collect the plane equation coefficient data, defect edge coordinate data, sawing parameter data, and the average strength difference data of the normal part before and after sawing of the cut surface of multiple logs to be sawed with single defects, and construct the knot defect strength... S5. Based on the strength difference mapping models for knot defects, decay defects, and crack defects, the initial coefficient data, initial sawing parameters, and defect edge coordinate data of the two plane equations of the logs to be managed during sawing are mapped to obtain the initial strength difference data to be managed; S6. Based on the initial strength difference data to be managed, the initial coefficient data and initial sawing parameters of the two plane equations of the logs to be managed during sawing are adjusted to obtain the final plane equation coefficient dataset and the final sawing parameter dataset to be managed. The usage scenarios for the defective parts of the logs to be managed will be adjusted and replaced.

2. The data analysis and management method for the timber industry based on machine learning according to claim 1, characterized in that, S1 includes the following steps: S11, setting the log to be managed and setting that the log to be managed has only a single type of defect; dividing the surface of the log to be managed into multiple image acquisition areas to obtain a current image acquisition area set; S12, in conjunction with the current image acquisition area set, using an X-ray scanning image acquisition system to acquire X-ray scanning images at multiple angles at each image acquisition area on the log to be managed to obtain the original scanning image data matrix to be managed.

3. The data analysis and management method for the timber industry based on machine learning according to claim 2, characterized in that, S2 includes the following steps: S21, performing grayscale conversion on each original scanned image data in the original scanned image data matrix to be managed, to obtain a grayscale-converted scanned image data matrix to be managed; S22, performing frequency domain filtering and spatial domain filtering operations on each scanned image data in the grayscale-converted scanned image data matrix to be managed using a median filter and a Gaussian filter, respectively, to obtain a filtered scanned image data matrix to be managed; S23, performing mathematical morphological processing on the filtered scanned image data matrix to be managed, to obtain the final scanned image data matrix to be managed.

4. The data analysis and management method for the timber industry based on machine learning according to claim 3, characterized in that, S3 includes the following steps: S31, comparing the final scanned image data matrix to be managed with the corresponding multiple standard wood defect image data and obtaining the corresponding defect types to obtain the final scanned image defect type data matrix; S32, collecting feature parameter data of multiple wood usage scenarios corresponding to each wood defect type in the historical wood defect type set to obtain the historical defect scenario feature parameter data matrix set; S33, calculating the cluster center data of each historical defect scenario feature parameter data matrix in the historical defect scenario feature parameter data matrix set to obtain the historical defect scenario feature parameter center data matrix.

5. The data analysis and management method for the timber industry based on machine learning according to claim 4, characterized in that, S33 includes the following steps: S331, constructing a cluster of *Triplophysa salina* populations based on scene defect feature parameters; setting the maximum number of iterations for the cluster of *Triplophysa salina* populations based on scene defect feature parameters. And the current iteration number is Let S331 be the maximum number of iterations for parameter clustering and the current number of iterations for parameter clustering, respectively; S332, randomly select one row of data from each historical defect scene feature parameter data matrix in the historical defect scene feature parameter data matrix set multiple times as the initial position of each salver in the scene defect feature parameter clustering salver population, to obtain the first initial position matrix set; S333, calculate the corresponding row in the initial position matrix of the kth salver in the scene defect feature parameter clustering salver population for each row of data in the historical defect scene feature parameter data matrix set. The Euclidean distance between them is used to obtain the Euclidean distance data matrix; the fitness function of the kth salver in the scene defect feature parameter clustering salver population is constructed based on the Euclidean distance data matrix; S334, start the iteration; in each iteration, the fitness value of the position matrix of each salver in the scene defect feature parameter clustering salver population updated in the previous iteration is calculated using the fitness function of the kth salver in the scene defect feature parameter clustering salver population, and the position matrix of each salver in the scene defect feature parameter clustering salver population updated in the previous iteration is updated; S335, when If the first optimal position is found, stop iterating and obtain the first final global optimal position; otherwise, continue iterating until... Until then, the first final global optimal position is used as the center data matrix of historical defect scene feature parameters.

6. The data analysis and management method for the timber industry based on machine learning according to claim 5, characterized in that, S4 includes the following guarantees: S41, according to the wood defect type set, set multiple test logs to be sawed with a single corresponding defect and collect scan image data at each corresponding image acquisition area, and process each scan image data obtained by means of S21, S22 and S23 to obtain a test log scan image dataset matrix; S42, construct a three-dimensional coordinate system for each test log to be sawed to obtain a test three-dimensional coordinate system matrix; Using the test three-dimensional coordinate system matrix and based on the scanned image dataset matrix of the test log to be sawed, several coordinate acquisition points are set at the defect edge in each test log to be sawed and the corresponding coordinate data is collected to obtain the test defect edge coordinate dataset matrix; then, using the test three-dimensional coordinate system matrix, the plane equation coefficients corresponding to the two cut surfaces during sawing are collected to obtain the test cut surface equation coefficient data matrix. S43. Set several sawing parameter types to obtain a sawing parameter type set; the sawing parameter type set includes cutter speed and cutter displacement speed; Based on the sawing parameter type set, parameter data were collected when sawing each test log to be sawed, and a test sawing parameter dataset matrix was obtained. Then, the average strength difference data of the normal part before and after sawing of each log to be sawn is collected to obtain the average data matrix of test strength difference. S44. Construct a node defect strength difference mapping model, a decay defect strength difference mapping model, and a crack defect strength difference mapping model based on the test defect edge coordinate dataset matrix, the test section equation coefficient data matrix, the test sawing parameter dataset matrix, and the test strength difference average data matrix.

7. The data analysis and management method for the timber industry based on machine learning according to claim 6, characterized in that: The strength difference mapping models for joint defects, decay defects, and crack defects all employ the SVM model.

8. The data analysis and management method for the timber industry based on machine learning according to claim 7, characterized in that, S5 includes the following steps: S51, constructing a three-dimensional coordinate system for the log to be managed during sawing, obtaining the three-dimensional coordinate system to be managed; setting several coordinate points at the edges of defects in the log to be managed according to the final defect type data matrix of the scanned image to be managed, obtaining the defect edge coordinate dataset to be managed; randomly setting the initial coefficient data of the plane equations of the two cut surfaces during sawing of the log to be managed and the initial sawing parameter data in conjunction with the sawing parameter type set, obtaining the initial cut surface equation coefficient dataset and the initial sawing parameter dataset to be managed; S52, determining the defect type of the log to be managed according to the final defect type data matrix of the scanned image to be managed, and the first defect type of the log to be managed; when the first defect type of the log to be managed is a node... When the first log defect type is a decay defect, the dataset of edge coordinates of the defect to be managed, the dataset of initial cross-sectional equation coefficients to be managed, and the dataset of initial sawing parameters to be managed are input into the sub-defect strength difference mapping model for mapping. When the first log defect type to be managed is a crack defect, the dataset of edge coordinates of the defect to be managed, the dataset of initial cross-sectional equation coefficients to be managed, and the dataset of initial sawing parameters to be managed are input into the crack defect strength difference mapping model for mapping. After mapping, the initial strength difference data to be managed is obtained.

9. A data analysis and management method for the timber industry based on machine learning according to claim 8, characterized in that, S6 includes the following steps: S61, setting a first strength difference threshold, a second strength difference threshold, and a third strength difference threshold; S62, when the first log defect type to be managed is a knot defect type and the initial strength difference data to be managed is greater than or equal to the first strength difference threshold, or when the first log defect type to be managed is a decay defect type and the initial strength difference data to be managed is greater than or equal to the second strength difference threshold, or when the first log defect type to be managed is a crack defect type and the initial strength difference data to be managed is greater than or equal to the third strength difference threshold. When setting the difference threshold, the dataset of initial cross-section equation coefficients and the dataset of initial sawing parameters to be managed are adjusted until the following conditions are met: the first log defect type is a knot defect and the initial strength difference data is less than the first strength difference threshold; the first log defect type is a decay defect and the initial strength difference data is less than the second strength difference threshold; or the first log defect type is a crack defect and the initial strength difference data is less than the third strength difference threshold. This process yields the final cross-section equation system to be managed. The dataset includes the initial cross section equation coefficient dataset and the initial sawing parameter dataset to be managed; otherwise, no adjustment is needed for the initial cross section equation coefficient dataset and the initial sawing parameter dataset to be managed; S63, after adjusting the initial cross section equation coefficient dataset and the initial sawing parameter dataset to be managed, the logs to be managed are sawn in conjunction with the final cross section equation coefficient dataset and the final sawing parameter dataset to be managed; the usage scenario feature parameter data of the defective part of the logs to be managed after processing is collected to obtain the usage scenario feature parameter dataset to be managed; S64, the Euclidean distance between the usage scenario feature parameter dataset to be managed and each row of data in the historical defect scenario feature parameter center data matrix is ​​calculated to obtain the first Euclidean distance dataset; the defect type corresponding to the smallest Euclidean distance data in the first Euclidean distance dataset is taken as the second log defect type to be managed; when the second log defect type to be managed is inconsistent with the first log defect type to be managed, the usage scenario of the defective part of the logs to be managed after processing is changed; otherwise, no change is needed for the usage scenario of the defective part of the logs to be managed after processing.

10. A system for implementing a machine learning-based data analysis and management method for the timber industry as described in any one of claims 1-9, characterized in that: It includes a log scanning image data acquisition module, an image preprocessing module, a log defect type data acquisition module, a scene parameter type setting module, a historical scene parameter acquisition module, a cluster center calculation module, a sawing test data acquisition module, a mapping model construction module, a mapping module, a log sawing data adjustment module, and a usage scenario adjustment module. The log scanning image data acquisition module is used to acquire X-ray scanning images from multiple angles at multiple image acquisition areas on the surface of the log to be managed, to obtain the original scanning image data matrix to be managed; the image preprocessing module is used to perform grayscale conversion, frequency domain filtering, spatial domain filtering and mathematical morphology processing on the original scanning image data matrix to be managed, to obtain the final scanning image data matrix to be managed. The log defect type data acquisition module is used to compare the final scanned image data matrix to be managed with the standard defect image data to obtain the final scanned image defect type data matrix to be managed. The scenario parameter type setting module is used to set several types of timber use scenario characteristic parameters to obtain a timber use scenario characteristic parameter type set; the historical scenario parameter acquisition module is used to acquire parameter data of several use scenarios corresponding to each defect type according to the timber use scenario characteristic parameter type set to obtain a historical defect scenario characteristic parameter data matrix set; the cluster center calculation module is used to calculate the cluster center data of each historical defect scenario characteristic parameter data matrix in the historical defect scenario characteristic parameter data matrix set to obtain a historical defect scenario characteristic parameter center data matrix; the sawing test data acquisition module is used to acquire the plane equation coefficient data, defect edge coordinate data, sawing parameter data, and the average strength difference data of the normal part before and after sawing of multiple logs to be sawed with single defects, to obtain the test defect edge coordinate dataset matrix, the test cut surface equation coefficient data matrix, the test sawing parameter dataset matrix, and the test strength difference average data matrix. The mapping model construction module is used to construct a knot defect strength difference mapping model, a decay defect strength difference mapping model, and a crack defect strength difference mapping model using a test defect edge coordinate dataset matrix, a test section equation coefficient data matrix, a test sawing parameter dataset matrix, and a test strength difference average data matrix. The mapping module is used to map the initial coefficient data, initial sawing parameter data, and defect edge coordinate data of the two sections of the log to be managed during sawing based on the knot defect strength difference mapping model, the decay defect strength difference mapping model, and the crack defect strength difference mapping model, to obtain the initial strength difference data to be managed. The log sawing data adjustment module is used to adjust the initial coefficient data and initial sawing parameter data of the two sections of the log to be managed based on the initial strength difference data to obtain the final section equation coefficient dataset and the final sawing parameter dataset to be managed. The usage scenario adjustment module is used to adjust and change the usage scenario of the defective parts of the log to be managed in conjunction with the historical defect scenario feature parameter center data matrix.

Images