A wood sorting system and its sorting method based on intelligent defect recognition

Abstract

An embodiment of the present application discloses a wood sorting system and a sorting method based on intelligent defect recognition, including a program control system, and further including: a main conveyor line and a plurality of sub-conveyor lines, the main conveyor line being capable of conveying along the x-axis direction; a detection mechanism, the detection mechanism straddling above the main conveyor line, suitable for detecting the surface roughness of the object to be measured, obtaining result data characterizing the surface roughness, and sending the result data to the program control system; a sorting mechanism, the sorting mechanism connecting the main conveyor line and the plurality of sub-conveyor lines, the sorting mechanism receiving the judgment result of the program control system and conveying the object to be measured to the corresponding sub-conveyor line according to the judgment result. It has the advantages of simple calculation method, simple self-learning process, and high judgment accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of wood sorting, and specifically to a wood sorting system based on intelligent defect recognition. The present invention also relates to a sorting method for such a wood sorting system. Background Art

[0002] The first process in solid wood floor production is the on-line inspection process of boards, and the inspection items include moisture content sorting, wood color sorting, defect screening, etc.

[0003] Specifically, moisture content sorting is to select boards with moisture content test values within a specified range. Boards with moisture content higher or lower than the specified range should be used for other purposes or rebalanced to adjust the moisture content. Among them, the specified range is an empirical range value determined according to factors such as the use environment (south or north geographically, household, commercial or underfloor heating occasions, etc.) and tree species. Wood color sorting is to select boards with similar colors. Generally, first, select the darkest acceptable sample board, the lightest sample board, and the sample board with the largest color difference (two wood colors such as dark and light or pink and light yellow existing simultaneously on the same board). Then, based on the three color boards, select boards with colors between the darkest sample board and the lightest sample board and with a color difference smaller than that of the sample board with the largest color difference. Boards with wood colors outside the three sample boards should be used for other purposes, such as for making solid wood floors with antique process, color rubbing process, etc. Defect screening is to select boards without wood defects such as surface cracks, knots (knots), discoloration, decay, and insect holes. Boards with the above defects should be repaired, re-sized or used as waste. Of course, the above inspections are carried out simultaneously on the same board.

[0004] The wood color of wood is formed by factors such as color, texture, and light and shade. Therefore, it is difficult to form a data-based standard for the judgment of wood color. In actual production, the sorting of wood color mostly uses manual sorting, and subjectively judges whether the color of a board is within the range defined by the three sample boards. Similarly, there are many types of wood defects, and the degree of each defect on a board is also different. Therefore, defect screening is also judged manually. These factors make the on-line inspection process of boards rely on a large number of operators.

[0005] In order to achieve the intelligent sorting of wood panels, a great deal of research and practice has been carried out in the prior art. In the Chinese patent database, the invention patent application with the publication number CN101767094A and the title "Method and Equipment for Sorting Wood According to Color and Wood Grain" discloses the following technical solution: extracting information such as the color, the inclination angle of the wood grain, and the color plate ratio from the wood image obtained by photography and generating wood image feature data, and comparing the feature data with the database to obtain which tree species the measured wood belongs to. This solution requires a large amount of manpower to collect data to establish a database, and can only identify which tree species the wood belongs to without being able to make more detailed judgments. A more serious problem is that even if two measured wood panels belong to different tree species, it is possible to have the same (or similar) image features, resulting in the system being unable to identify. The same situation appears in the paper "Research on Wood Classification and Sorting Algorithm Based on Image Multi-Feature Pattern Recognition" (author Luo Wei, Northeast Forestry University). The author collected samples of 5 common tree species in Northeast China, made wood images of 2 types of cut surfaces (radial cut, tangential cut) of these 5 tree species, 100 for each tree species, a total of 1000 images, including 104 dead knots, 40 live knots, 72 insect damages, and 92 crack defect images, to construct a sample library. Using the main color feature extraction method based on non-uniform quantization of the color space to perform feature matching on the wood images for comparative retrieval and discrimination of tree species, and using the BP neural network model, the SVM support vector machine classification model, and the CNN convolutional neural network model to classify the defects in the collected wood images to determine what kind of defects the measured wood panel has. These technical solutions are all very academic, but obviously, they are not very operable in actual production. They are all based on precise instruments and elaborate algorithms, which are very difficult to manage and maintain at the actual production site, and the cost is often too high.

[0006] The invention patent with the publication number CN111862028B and the title "Wood Defect Detection and Sorting Device and Method Based on Depth Camera and Deep Learning" discloses the following technical solution: obtaining the RGB image and depth information of the wood, processing the RGBD color depth information into a four-dimensional feature with data normalization and inputting it into a convolutional neural network, and comparing the data of the measured wood panel with the standard data to determine whether the measured wood panel has defects and what kind of defects it belongs to. Obviously, this method can only screen out the wood panels with defects, and the function is relatively single. The same problem also exists in the utility model patent with the publication number CN201331499Y and the title "Automatic Imaging and Positioning Instrument for Wood Defects", and the utility model patent with the publication number CN212093295 and the title "Wood Defect Detection and Sorting Device Based on Ranger Series 3D Camera".

[0007] In summary, in the prior art, there is no system that uses relatively simple instruments, has relatively high algorithm compatibility, and can simultaneously detect the wood color and defects of wood panels. Summary of the Invention

[0008] The object of the present invention is to overcome the above technical problems and provide a wood sorting system based on intelligent defect recognition; the present invention also provides a sorting method for the wood sorting system based on intelligent defect recognition.

[0009] To achieve the above object, an embodiment of the present invention provides a wood sorting system based on intelligent defect recognition, including a program control system, and further including:

[0010] A main conveyor line and a plurality of sub-conveyor lines, the main conveyor line can convey along the x-axis direction;

[0011] A detection mechanism, the detection mechanism straddles above the main conveyor line, is suitable for detecting the surface roughness of the object to be measured, obtaining result data representing the surface roughness, and sending the result data to the program control system;

[0012] A sorting mechanism, the sorting mechanism connects the main conveyor line and the plurality of sub-conveyor lines, the sorting mechanism receives the judgment result of the program control system and conveys the object to be measured to the corresponding sub-conveyor line according to the judgment result.

[0013] Preferably, the detection mechanism includes a plurality of contact type roughness detectors.

[0014] Preferably, the plurality of roughness detectors are linearly arrayed at intervals along the y-axis direction.

[0015] Preferably, a sanding machine connected to it is provided at the front end of the main conveyor line.

[0016] Preferably, a water spraying device is provided at the front end of the sanding machine.

[0017] To achieve the above object, another embodiment of the present invention provides a sorting method for a wood sorting system based on intelligent defect recognition, including the following steps: a standard value input step, a sampling step, a defect judgment step, and a diversion step;

[0018] In the standard value input step, sampling points of a standard template selected manually are sampled, standard sampling point data is obtained and an evaluation numerical range is calculated;

[0019] In the sampling step, the detection mechanism takes points from the wood board to be measured that is conveyed by the main conveyor line and passes under it, and takes (4 - 6)×(4 - 6) points at equal intervals along the x-axis and y-axis directions respectively, obtains the measured sampling point data, the interval between adjacent measured sampling points is 0.2s - 1.0s, the sampling length of each measured sampling point is 0.1mm - 0.8mm, and then sends the measured sampling point data to the judgment mechanism;

[0020] In the step of defect judgment, the program control system sends the judgment result to the sorting mechanism according to whether the measured sampling point data belongs to the evaluation numerical range, is higher than the evaluation numerical range, or is lower than the evaluation numerical range;

[0021] In the step of shunting, the sorting mechanism conveys the measured objects to the corresponding sub-conveyor lines respectively according to the judgment result of the program control system.

[0022] Preferably, in the step of standard value input, at least for the sampling points of the lightest allowed sample plate and the darkest allowed sample plate, (4-6)×(4-6) points are taken at equal intervals along the x-axis and y-axis directions respectively to obtain standard sampling point data;

[0023] Set the maximum value of the maximum profile height (Rz) of the lightest allowed sample plate as the upper limit, denoted as Rzmax, and set the minimum value of the maximum profile height (Rz) of the darkest allowed sample plate as the lower limit, denoted as Rzmin. The numerical range of (Rzmin~Rzmax) constitutes the first evaluation numerical range;

[0024] Calculate the mean value of the arithmetic mean deviation of the profile (Ra) of all sampling points of the lightest allowed sample plate, denoted as Ramax, and calculate the mean value of the arithmetic mean deviation of the profile (Ra) of all sampling points of the darkest allowed sample plate, denoted as Ramin. The numerical range of (Ramin~Ramax) constitutes the second evaluation numerical range.

[0025] Preferably, in the step of sampling, the measured sampling point data includes the maximum profile height (Rz) and the arithmetic mean deviation of the profile (Ra) of the measured sampling point, denoted as Rztest and Ratest respectively.

[0026] Preferably, in the step of defect judgment, calculate the average value of Rztest of each column of sampling point data, denoted as Rz’test, and compare the Rz’test of each column with the first evaluation numerical range. If all Rz’test of the measured object belong to the first evaluation numerical range, it is judged as qualified; if all Rz’test of the measured object are higher than the first evaluation numerical range, it is judged as a light-colored plate; if all Rz’test of the measured object are lower than the first evaluation numerical range, it is judged as a dark-colored plate; if some Rz’test of the measured object belong to the first evaluation numerical range and some are lower than or higher than the first evaluation numerical range, it is judged as a two-color plate; if some Rz’test of the measured object are lower than the first evaluation numerical range and some are higher than the first evaluation numerical range, it is judged as a two-color plate;

[0027] Calculate the average value of Ratest for all sampled data points, denoted as Ra’test, and compare Ra’test with the second evaluation numerical range for auxiliary judgment.

[0028] If there are cases where some of the Rztest and Ratest values of the sampled data points are too small or continuous sampling cannot be carried out, it is judged as a board with split node defects. Among them, the case of too small values means that Rztest and Ratest are less than 20% of the lower limit of the first evaluation numerical range and the second evaluation numerical range.

[0029] If there are cases where some or all of the Rztest and Ratest values of the sampled data points are too large, it is judged as a mildew and rot defective board. Among them, the case of too large values means that Rztest and Ratest are greater than 5 times the upper limit of the first evaluation numerical range and the second evaluation numerical range.

[0030] Preferably, before sampling, the surface of the object to be measured is roughly sanded.

[0031] In summary, compared with the prior art, the beneficial effects of the present invention are:

[0032] The wood sorting system based on intelligent defect recognition of the present application uses the surface planing or sanding performance of wood-based panels, that is, the surface flatness after planing or sanding, as the detection index to predict the unqualified situations of light color, dark color, and double color in the wood color category of wood-based panels and the unqualified situations of cracking, knot, mildew, blue stain, and decay in the wood property category, and finds that the surface planing or sanding performance of wood-based panels can better characterize the wood color and wood property characteristics of wood-based panels.

[0033] The wood sorting system based on intelligent defect recognition of the present application judges the type of defects of the object to be measured based on the principle that the surface planing or sanding performance of wood-based panels characterizes the wood color and wood property characteristics. Compared with the prior art's judgment method based on the visualization principle, it has the advantages of simple calculation method, simple self-learning process, and high judgment accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present application. For those skilled in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0035] Figure 1 It is a schematic structural diagram of the wood sorting system based on intelligent defect recognition in Embodiment 1 of the present application.

[0036] Figure 2Schematic diagram of sampling points of the standard template for Embodiment 1 of the present application.

[0037] Figure 3 Schematic diagram of the structure of the wood sorting system based on intelligent defect recognition for Embodiment 2 of the present application.

[0038] Figure 4 Schematic diagram of the structure of the sander for Embodiment 3 of the present application.

[0039] Figure 5 Schematic diagram of the structure of the wood sorting system based on intelligent defect recognition for Embodiment 4 of the present application.

[0040] In the figure: A, the object to be measured, 10, the program control system, 20, the main conveyor line, 31, 32, 33, the sub-conveyor lines, 40, the detection mechanism, 41, 42, 43, 44, 45, the roughness detectors, 50, the sorting mechanism, 51, the roller conveyor, 52, the first material deflecting unit, 53, the second material deflecting unit, 60, the sander, 70, the water spraying device, 80, the camera, 90, the display. Detailed implementation manners

[0041] In order to enable those skilled in the art to better understand the technical solutions in the present application, the following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative efforts shall fall within the protection scope of the present application. Embodiment

[0042] Refer to Figure 1 A wood sorting system based on intelligent defect recognition as shown, including a program control system 10, a main conveyor line 20, a plurality of sub-conveyor lines 31, 32, 33, a detection mechanism 40, and a sorting mechanism 50. The program control system 10 is an industrial control computer capable of data processing in the prior art.

[0043] The main conveyor line 20 is a belt conveyor in the prior art and can convey along the x-axis direction. The plurality of sub-conveyor lines 31, 32, 33 are all belt conveyors in the prior art. The first sub-conveyor line 31 can convey along the x-axis direction and is used to receive and continue to convey qualified objects to be measured. The second sub-conveyor line 32 can convey along the y-axis direction and is used to receive and continue to convey the objects to be measured that are unqualified in terms of wood color; the unqualified situations in terms of wood color include too light wood color, too dark wood color, and double-color boards. The third sub-conveyor line 33 can convey along the y-axis direction and is used to receive and continue to convey the objects to be measured that are unqualified in terms of wood properties; the unqualified situations in terms of wood properties include split knots (cracks and knots), mildew and rot (blue stain and decay).

[0044] The inspection mechanism 40 is spanned above the main conveyor line 20, and is applicable to inspect the surface roughness of the object to be measured A, obtain the result data characterizing the surface roughness, and send the result data to the program control system 10. In this embodiment, the inspection mechanism 40 includes 5 contact type roughness detectors 41, 42, 43, 44, 45 arranged at intervals in the y-axis direction, such as roughness detectors produced in Japan (model: E-35A). The 5 roughness detectors 41, 42, 43, 44, 45 are respectively communicatively connected to the program control system 10.

[0045] The sorting mechanism 50 is connected to the main conveyor line 20 and multiple sub-conveyor lines 31, 32, 33. The sorting mechanism 50 receives the judgment result of the program control system 10 and conveys the object to be measured to the corresponding sub-conveyor line according to the judgment result. In this embodiment, the sorting mechanism 50 includes a roller conveyor 51 connecting the main conveyor line 20 and the first sub-conveyor line 31, and a first material pushing unit 52 and a second material pushing unit 53 respectively arranged on both sides of the roller conveyor 51. The first material pushing unit 52 and the second material pushing unit 53 are both pneumatic material pushing units. For example, each of them includes a pneumatic rod fixing seat fixedly installed on the frame of the roller conveyor 51, a pneumatic push rod installed through the pneumatic rod fixing seat, and a push plate screwed to the free end of the pneumatic push rod. The first material pushing unit 52 can push the object to be measured A towards the third sub-conveyor line 33, and the second material pushing unit 53 can push the object to be measured A towards the second sub-conveyor line 32. In other words, when the program control system 10 gives a qualified indication, only the roller conveyor 51 operates to enable the object to be measured A to enter the first sub-conveyor line 31 from the main conveyor line 20, and the object to be measured A is continuously conveyed by the first sub-conveyor line 31; when the program control system 10 gives an indication of Class 1 non-conformance (the case of non-conformance in material color), the pneumatic push rod of the second material pushing unit 53 is pushed out to push the object to be measured A towards the second sub-conveyor line 32, and the object to be measured A is continuously conveyed by the second sub-conveyor line 32; when the program control system 10 gives an indication of Class 2 non-conformance (the case of non-conformance in material properties), the pneumatic push rod of the first material pushing unit 52 is pushed out to push the object to be measured A towards the third sub-conveyor line 33, and the object to be measured A is continuously conveyed by the third sub-conveyor line 33.

[0046] In this embodiment, the object to be measured A is a Parashorea chinensis solid wood floor blank after four-side planing, with specifications of 925mm×123mm×18.5mm and a moisture content of 12±1%. The screening target of this embodiment is to select qualified oak solid wood floor blanks with neutral material color, no cracks, no knots, no mildew (including blue stain), and no decay from 50 Parashorea chinensis solid wood floor blanks using the above sorting system.

[0047] This embodiment completes the sorting work according to the above requirements through steps of standard value input, sampling, defect judgment, and diversion.

[0048] First, in the step of standard value input, select the standard template according to Table 1.

[0049] Table 1. Standard template selection rules

[0050]

[0051] In actual production, the Pometia wood-based panels are mainly quarter-sawn panels. Moreover, in the preliminary correction experiment, the inventor selected quarter-sawn Pometia panels and rift-sawn panels with similar wood colors and detected the maximum height of the average profile (Rz) respectively. It was found that although the numerical results of the quarter-sawn panels and rift-sawn panels were different, they were much smaller than the difference between the heartwood and sapwood (the main reason for the board to have too dark, too light or two-color). Therefore, the cutting direction of the measured board can be ignored.

[0052] Place the lightest allowed template and the darkest allowed template in the standard templates on the main conveyor line 20 one by one. As the standard templates are conveyed on the main conveyor line 20, each roughness detector 41, 42, 43, 44, 45 takes 5 standard sampling points at the corresponding width of the standard template, which is equivalent to taking 5×5 points at equal intervals along the x-axis and y-axis directions respectively. The interval between adjacent standard sampling points is 0.2s to 1.0s, and the sampling length of each standard sampling point is 0.1mm to 0.8mm (for example, 0.5mm). Of course, those of ordinary skill in the art can understand that the sampling time interval and sampling length between adjacent sampling points should be adapted to the actual width of the standard template and the actual conveying speed of the main conveyor line 20. In this embodiment, the actual width of the standard template is 125mm, the conveying speed of the main conveyor line 20 is 35m / s, the sampling length of each standard sampling point is 0.5mm, and the interval between adjacent standard sampling points is 0.6s. The array diagram of the standard sampling points on the standard template is referred to Figure 2 as shown.

[0053] The roughness detectors 41, 42, 43, 44, 45 obtain and get the values of each standard sampling point. The values include the maximum height of the profile (Rz) and the arithmetic mean deviation of the profile (Ra). Send the values of the 25 sampling points of each standard template and the corresponding relationship between the values and the sampling points to the program control system 10, and the program control system 10 performs the following self-learning:

[0054] Extract the maximum value of the maximum height of the profile (Rz) among the values of 25×3 standard sampling points of the lightest allowed template, and use this value as the upper limit, denoted as Rz max ; extract the minimum value of the maximum height of the profile (Rz) among the values of 25×3 standard sampling points of the darkest allowed template, and use this value as the lower limit, denoted as Rz min ; with (Rz min ~Rz max) The numerical range constitutes the first evaluation numerical range.

[0055] Calculate the mean value of the profile arithmetic mean deviation (Ra) of 25×3 standard sampling points of the lightest color sample board allowed, denoted as Ra max ; Calculate the mean value of the profile arithmetic mean deviation (Ra) of 25×3 standard sampling points of the darkest color sample board allowed, denoted as Ra min ; With (Ra min ~ Ra max ) The numerical range constitutes the second evaluation numerical range. Of course, in the calculation process of the above average value, the maximum value and the minimum value should be excluded.

[0056] Place the sample boards with live knots or dead knots, the sample boards with mildew / blue stain, and the sample boards with decay in the standard sample boards one by one on the main conveyor line 20, and select a certain roughness detector 41, 42, 43, 44, 45 corresponding to the defect location of the above standard sample board to obtain the numerical values at the defect, and the numerical values include the maximum profile height (Rz) and the profile arithmetic mean deviation (Ra). Send the numerical values of 1 sampling point of each standard sample board and the corresponding relationship between the numerical values and the sampling points to the program control system 10, and the program control system 10 performs the following self-learning:

[0057] Calculate the mean values of the maximum profile height (Rz) and the profile arithmetic mean deviation (Ra) at 3 live knots in the sample board with live knots, denoted as Rz 1ave 、Ra 1ave ;

[0058] Calculate the mean values of the maximum profile height (Rz) and the profile arithmetic mean deviation (Ra) at 3 dead knots in the sample board with dead knots, denoted as Rz 2ave 、Ra 2ave ; However, in this embodiment, the dead knots have cracked and partially fallen off, resulting in no valid numerical values being readable at the knot;

[0059] Calculate the mean values of the maximum profile height (Rz) and the profile arithmetic mean deviation (Ra) at 3 mildew spots in the sample board with mildew / blue stain, denoted as Rz 3ave 、Ra 3ave ;

[0060] Calculate the mean values of the maximum profile height (Rz) and the profile arithmetic mean deviation (Ra) at 3 decay spots in the sample board with decay, denoted as Rz 4ave 、Ra 4ave ;

[0061] With the Rz of 1ave to Rz 4ave 、Ra 1ave to Ra 4aveThe numerical values are used as reference data.

[0062] The evaluation numerical range of the standard template is shown in Table 2.

[0063] Table 2. Evaluation Numerical Range of Standard Template

[0064]

[0065] Subsequently, in the sampling step, the testing agency 40 takes points on the tested sheet passing under it while being conveyed by the acceptor conveyor line 20. Similarly, 5×5 points are taken at equal intervals along the x-axis and y-axis directions respectively to obtain the data of the tested sampling points. The array diagram of the tested sampling points on the tested object A is referred to Figure 2 as shown. The interval between each tested sampling point is 0.2 s to 1.0 s, and the sampling length of each tested sampling point is 0.1 mm to 0.8 mm. Subsequently, the data of the tested sampling points are sent to the program control system 10. Similarly, the sampling time interval and sampling length between adjacent sampling points should be adapted to the actual width of the tested object A and the actual conveying speed of the main conveyor line 20. In this embodiment, the actual width of the tested object A is 125 mm, the conveying speed of the main conveyor line 20 is 35 m / s, the sampling length of each standard sampling point is 0.5 mm, and the interval between adjacent standard sampling points is 0.6 s. The data of the tested sampling points are respectively denoted as Rz test and Ra test .

[0066] The following processing is performed on the numerical values of the 25 sampling points on each tested object A: 1. Calculate the average value of Rztest of the sampling point data in each column, denoted as Rz’ test . For example, calculate the average value of the tested sampling point data in the column from No. 411 to No. 451, denoted as Rz’ 1test ; calculate the average value of the tested sampling point data in the column from No. 412 to No. 452, denoted as Rz’ 2test ; and so on. 2. Calculate the average value of Ra test of all the sampling point data, denoted as Ra’ test .

[0067] Then, in the defect judgment step, the program control system 10 sends the judgment result to the shunting mechanism according to whether the data of the tested sampling points belong to the evaluation numerical range, are higher than the evaluation numerical range or are lower than the evaluation numerical range. The specific judgment rules are as follows:

[0068] 1. Compare the values of Rz’ test (Rz’ 1test to Rz’ 5test ) in each column with the first evaluation numerical range. If the values of all 5 Rz’ test belong to the first evaluation numerical range and Ra’ testIf the value belongs to the second evaluation value range, it is judged as qualified;

[0069] 2. Compare the values of Rz’ test (Rz’ 1test to Rz’ 5test ) in each column with the first evaluation value range. If the values of all 5 Rz’ test are lower than the first evaluation value range and the value of Ra’ test is also lower than the second evaluation value range, it is judged as a dark-colored plate, belonging to non-conformance of Class 1.1;

[0070] 3. Compare the values of Rz’ test (Rz’ 1test to Rz’ 5test ) in each column with the first evaluation value range. If the values of all 5 Rz’ test are higher than the first evaluation value range and the value of Ra’ test is also higher than the second evaluation value range, it is judged as a light-colored plate, belonging to non-conformance of Class 1.2;

[0071] 4. Compare the values of Rz’ test (Rz’ 1test to Rz’ 5test ) in each column with the first evaluation value range. If some of the values of 5 Rz’ test belong to the first evaluation value range and at least 2 Rz’ test are lower than or higher than the first evaluation value range, it is judged as a two-color plate, belonging to non-conformance of Class 1.3;

[0072] 5. Compare the values of Rz’ test (Rz’ 1test to Rz’ 5test ) in each column with the first evaluation value range. If some of the values of 5 Rz’ test are lower than the first evaluation value range and some are higher than the first evaluation value range, it is judged as a two-color plate, belonging to non-conformance of Class 1.3;

[0073] 6. If there is a situation where the comparison results of Rz’ test and Ra’ test are inconsistent in Rules 1 and 2, it shall be transferred to manual determination;

[0074] 7. If there is a situation where some of the Rz test , Ra test values at the sampling points are too small or continuous sampling cannot be carried out, it is judged as a plate with crack and node defects. Among them, the situation where the values are too small means that Rz test , Ra test are less than 20% of the lower limit of the first evaluation value range and the second evaluation value range, belonging to non-conformance of Class 2.1;

[0075] 8. If some or all of the Rztest and Ratest values of the sampling point data are too large, it is determined as a mildew and rot defective board. Among them, the situation where the values are too large means that Rztest and Ratest are greater than 5 times the upper limits of the first evaluation value range and the second evaluation value range, belonging to non-conformities of Class 2.2.

[0076] Finally, in the shunting step, the sorting mechanism 50 sends the object A to be measured to the sub-conveyor lines 31, 32, and 33 according to the judgment result of the program control system 10. The working method of the sorting mechanism 50 is as described above.

[0077] The test results are shown in Table 3.

[0078] Table 3. Test Results

[0079]

[0080] In the above results, the manual inspection uses the traditional manual inspection method. Before the inspection mechanism 40, the object A to be measured is first inspected, judged, and marked on the side of the object A.

[0081] By analyzing the results in Table 3, it can be found that the sorting system of this embodiment is basically consistent with the results of manual inspection in the judgment and sorting of wood color, and only makes mistakes in the judgment of non-conformities of Class 1.3 (two-color board) and Class 2.1 (live knot). After observation, it is found that this defective board belongs to small-area local two-color, and the program control system 10 misjudges it as a live knot. However, the sorting system of this embodiment is superior to manual screening and selects the object A with hidden crack defects. As is well known, because hidden cracks are not easy to find, the screening of hidden crack boards has always been a major difficulty in manual sorting. But in roughness detection, hidden cracks will be detected as extremely large profile valley depths, making the measured Rz value at the location with hidden cracks much larger than the average value, so that hidden cracks are easier to find. text and the value is much larger than the average value, making hidden cracks easier to be found.

[0082] Obviously, the sorting system of this embodiment can greatly reduce the dependence on manual labor for the sorting of solid wood boards, and at least can achieve semi-automatic sorting of solid wood boards. Embodiment

[0083] The difference between Embodiment 2 and Embodiment 1 is that, as shown in reference Figure 3 a sanding machine 60 connected to it is provided at the front end of the main conveyor line 20. This sanding machine 60 is a heavy-duty sanding machine in the prior art, and the preferably used sand belt models are 60 # 、80 # or 100 #Through comparative experiments, the inventor found that, compared with the surface properties of wood-based panels treated by polishing, the surface properties of wood-based panels treated by coarse sanding can more accurately reflect the color characteristics and material properties of wood-based panels. At the same time, the sanding process of the heavy-duty sander is an essential process between the sorting of panels and the processing of tongue-and-groove or painting. Therefore, in this embodiment, only this step is advanced to before sorting, which can improve the accuracy of intelligent sorting without complicating the production process. Embodiment

[0084] The difference between Embodiment 3 and Embodiment 2 is that, as shown in Figure 4 a water spraying device 70 is provided at the front end of the sander 60. Preferably, the water spraying device 70 includes a water pipe arranged perpendicular to the feeding direction of the sander 60, a plurality of atomizing water spray heads arranged at intervals on the water pipe, and a water source connected to the water pipe. Through comparative experiments, the inventor found that the coarse sanding treatment of a wet surface can more accurately reflect the color characteristics and material properties of wood-based panels, especially in the judgment of two unqualified situations: two-color boards and hidden cracks. At the same time, short-time spraying of atomized water on the surface of the object A to be measured will not affect the moisture content of the board and will not affect the detection accuracy of the contact-type moisture detector.

[0085] Preferably, the atomizing water spray heads are inclined towards the side of the sander 60 to spray atomized water, and the vertical distance between it and the conveying working surface of the sander 60 is 50 cm to 80 cm. Embodiment

[0086] The difference between Embodiment 4 and Embodiment 1 is that, as shown in Figure 5 roughness detectors 41, 42, 43, 44, 45 obtain and get the values of each standard sampling point, and the values include the maximum profile height (Rz) and the arithmetic mean deviation of the profile (Ra). The values of 25 sampling points of each standard sample plate and the corresponding relationship between the values and the sampling points are sent to the program control system 10, and the program control system 10 performs the following self-learning:

[0087] 1. Extract the maximum value of the maximum profile height (Rz) from the values of 25×3 standard sampling points of the lightest color sample plate allowed, and use this value as the upper limit, denoted as Rz max ; extract the minimum value of the maximum profile height (Rz) from the values of 25×3 standard sampling points of the darkest color sample plate allowed, and use this value as the lower limit, denoted as Rz min ; use the numerical range of (Rz min ~Rz max ) to form the first evaluation numerical range.

[0088] 2. Calculate the mean value of the arithmetic mean deviation of the profile (Ra) of 25×3 standard sampling points of the lightest color sample plate allowed, denoted as Ramax ; Calculate the average value of the arithmetic mean deviation of the profile (Ra) of 25×3 standard sampling points of the darkest sample plate allowed, denoted as Ra min ; With (Ra min ~Ra max ) to form the second evaluation value range. Of course, during the calculation process of the above average value, the maximum value and the minimum value should be excluded.

[0089] The first sub-conveyor line 31 can be conveyed along the x-axis direction and is used to receive and continue to convey qualified objects to be measured. The second sub-conveyor line 32 can be conveyed along the y-axis direction and is used to receive and continue to convey unqualified objects to be measured.

[0090] The sorting mechanism 50 is connected to the main conveyor line 20 and the sub-conveyor lines 31, 32. The sorting mechanism 50 receives the judgment result of the program control system 10 and conveys the object to be measured to the corresponding sub-conveyor line according to the judgment result. In this embodiment, the sorting mechanism 50 includes a roller conveyor 51 connecting the main conveyor line 20 and the first sub-conveyor line 31, and a second material pushing unit 53 arranged on one side of the roller conveyor 51. The second material pushing unit 53 is a pneumatic material pushing unit, for example, including a pneumatic rod fixed seat fixedly installed on the frame of the roller conveyor 51, a pneumatic push rod installed through the pneumatic rod fixed seat, and a push plate screwed to the free end of the pneumatic push rod. The second material pushing unit 53 can push the object to be measured A towards the second sub-conveyor line 32. In other words, when the program control system 10 gives a qualified indication, only the roller conveyor 51 operates to enable the object to be measured A to enter the first sub-conveyor line 31 from the main conveyor line 20, and the object to be measured A is continuously conveyed by the first sub-conveyor line 31; when the program control system 10 gives an unqualified indication, the pneumatic push rod of the second material pushing unit 53 is pushed out to push the object to be measured A towards the second sub-conveyor line 32, and the object to be measured A is continuously conveyed by the second sub-conveyor line 32. The operator observes the object to be measured A passing through the second sub-conveyor line 32 to manually screen specific unqualified types, which is a sorting method combining the system and manual work. In other words, the sorting system of this embodiment is responsible for selecting qualified plates, and the operator is responsible for secondary sorting of unqualified plates to classify defects. Generally, the proportion of unqualified plates is 10 - 20%. Therefore, this method can greatly reduce the number of workers and the labor intensity of the operator in the plate sorting process, and can also give full play to the advantages of the high efficiency of the sorting system and the flexibility of manual sorting at the same time.

[0091] In some preferred embodiments, a camera 80 is mounted on the second sub-conveyor line 32 via a cross beam. The end of the second sub-conveyor line 32 is connected to a multi-layer offline conveyor line through a sub-sorting mechanism. The multiple offline conveyor lines are all belt conveyor lines and are stacked on the same rack. The sub-sorting mechanism is a roller conveyor section that can be lifted. During operation, the camera 80 is communicatively connected to a display 90 in the machine room, and is used to transmit real-time images of the object A on the second sub-conveyor line 32 to the display 90. The sub-sorting mechanism is communicatively connected to the program control system 10 in the machine room. The operator views the images on the display 90 and manipulates the roller conveyor section of the sub-sorting mechanism to connect to the corresponding offline conveyor line with defects through the program control system 10, so as to classify and convey the object A judged manually.

[0092] The test results are shown in Table 4.

[0093] Table 4. Test Results

[0094]

[0095] By analyzing the results in Table 4, it can be found that since the sorting system selects the boards with hidden crack defects, the operator (different operators were selected in the two tests) found the hidden cracks on the boards through careful comparison; at the same time, by manually subdividing the defects, the boards with local two-color were correctly classified. Obviously, the sorting method combining the system and manual work improves the sorting accuracy. Example

[0096] The difference between Example 5 and Example 1 is that the object A is an oak solid wood floor blank that has been planed on four sides, with a specification of 925mm×123mm×18.5mm and a moisture content of 14 - 16%. The screening target of this example is to select qualified oak solid wood floor blanks with neutral wood color, no cracks, no knots, no mildew (including blue stain), and no decay from 50 oak solid wood floor blanks using the above sorting system. The test results are shown in Table 5.

[0097] Table 5. Test Results

[0098]

[0099] Compared with the relatively uniform structure of the keruing boards, the oak boards have the characteristics of relatively obvious two colors, relatively more live knots, and relatively smaller live knot diameters. Therefore, by analyzing the results in Table 5, it can be known that: 1. Since the two colors of the oak boards are relatively obvious and generally no local two-color boards appear, the sorting system correctly selects the boards with two-color defects; 2. Since the live knot diameters of the oak are relatively small and the density of the measured sampling points is insufficient, all the boards with knot defects are not found; 3. The sorting system can relatively stably select the boards with hidden crack defects.

[0100] The foregoing description is for purposes of illustration and not limitation. Many embodiments and many applications other than the examples provided will be apparent to those of ordinary skill in the art upon reading the foregoing description. Therefore, the scope of the present teachings should not be determined with reference to the foregoing description, but should instead be determined with reference to the appended claims and the full scope of equivalents to which such claims are entitled. For the sake of completeness, all articles and references, including patent applications and publications, are incorporated herein by reference. Omitting any aspect of the subject matter disclosed herein from the foregoing claims is not intended to abandon such subject matter nor should it be considered that the applicant has not considered such subject matter to be a part of the disclosed subject matter of the application.

Claims

1. A sorting method for a wood sorting system based on intelligent defect recognition, characterized in that, It includes the following steps: The steps of standard value input, sampling, defect judgment, and shunting; In the step of standard value input, for the sampling points of the manually selected standard templates, at least for the sampling points of the lightest allowed template and the darkest allowed template, (4 - 6)×(4 - 6) points are equally spaced along the x-axis and y-axis directions respectively to obtain standard sampling point data; set the maximum value of the profile maximum height (Rz) of the lightest allowed template as the upper limit, denoted as Rzmax, set the minimum value of the profile maximum height (Rz) of the darkest allowed template as the lower limit, denoted as Rzmin, and form the first evaluation value range with the numerical range of (Rzmin - Rzmax); calculate the mean value of the profile arithmetic mean deviation (Ra) of all sampling points of the lightest allowed template, denoted as Ramax, calculate the mean value of the profile arithmetic mean deviation (Ra) of all sampling points of the darkest allowed template, denoted as Ramin, and form the second evaluation value range with the numerical range of (Ramin - Ramax); In the step of sampling, the testing agency takes points on the object to be measured that is conveyed by the acceptor conveyor line and passes under it, (4 - 6)×(4 - 6) points are equally spaced along the x-axis and y-axis directions respectively to obtain the measured sampling point data, the interval between adjacent measured sampling points is 0.2s - 1.0s, the sampling length of each measured sampling point is 0.1mm - 0.8mm, and then the measured sampling point data is sent to the judgment agency; the measured sampling point data includes the profile maximum height (Rz) and the profile arithmetic mean deviation (Ra) of the measured sampling point, denoted as Rztest and Ratest respectively; In the step of defect judgment, the program control system sends the judgment result to the sorting agency according to whether the measured sampling point data belongs to, is higher than, or is lower than the evaluation value range; Calculate the average value of Rztest of each column of sampling point data, denoted as Rz’test, compare each column of Rz’test with the first evaluation value range, if all Rz’test of the object to be measured belong to the first evaluation value range, it is judged as qualified, if all Rz’test of the object to be measured are higher than the first evaluation value range, it is judged as a light color plate, if all Rz’test of the object to be measured are lower than the first evaluation value range, it is judged as a dark color plate, if some Rz’test of the object to be measured belong to the first evaluation value range and some are lower or higher than the first evaluation value range, it is judged as a two-color plate, if some Rz’test of the object to be measured are lower than the first evaluation value range and some are higher than the first evaluation value range, it is judged as a two-color plate; Calculate the average value of Ratest of all sampling point data, denoted as Ra’test, and compare Ra’test with the second evaluation value range for auxiliary judgment; If there are cases where some of the Rztest and Ratest values in the sampling point data are too small or continuous sampling points cannot be obtained, it is determined as a board with a splitting defect. Among them, the case of too small values means that Rztest and Ratest are less than 20% of the lower limit of the first evaluation value range and the second evaluation value range. If there are cases where some or all of the Rztest and Ratest values in the sampling point data are too large, it is determined as a mildew and rot defect board. Among them, the case of too large values means that Rztest and Ratest are greater than 5 times the upper limit of the first evaluation value range and the second evaluation value range; in the step of shunting, the sorting mechanism conveys the object to be measured to the corresponding sub-conveyor line according to the judgment result of the program control system.

2. The sorting method according to claim 1, characterized in that, Before sampling, the surface of the object to be measured is roughly sanded.

Images