A saw blade detection method and system based on multi-source data fusion

Abstract

This invention discloses a saw blade detection method and system based on multi-source data fusion, specifically relating to the field of saw blade detection technology. By simultaneously acquiring high-speed visual images of the saw blade's front and side views and spindle current signals, the initial tooth profile reference features of the saw blade are extracted. Furthermore, the system monitors in real time changes in the straightness of the projection line, the amount of cutting edge retraction, and abnormal changes in the current signal during the saw blade cutting process. By fusing visual and current data analysis, the normal wear and degradation trend of the saw blade and abnormal tooth loss characteristics are observed, thereby achieving accurate identification of saw blade wear or tooth loss conditions. A comprehensive evaluation is then performed based on the defect rate of sawn products, ultimately outputting an accurate and reliable saw blade replacement control signal. This invention effectively improves the accuracy of saw blade condition detection and the timeliness of replacement decisions, avoiding processing quality problems and ensuring production efficiency and processing safety.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of saw blade detection, and more particularly to a saw blade detection method and system based on multi-source data fusion. BACKGROUND

[0002] In the actual cutting process, the wear and failure state of the saw blade is usually judged by the experience of the operator or evaluated by a single visual image or current signal detection method. In the prior art, the visual detection method alone can only reflect the visible wear state of the saw blade surface, and cannot accurately reveal the load change and internal wear characteristics of the saw blade in the cutting process. Although the current signal monitoring method alone can indirectly reflect the cutting load change, it is easily affected by external disturbance factors in the machining process, and cannot accurately distinguish between normal wear and abnormal wear state. A single data source cannot fully reflect the actual running state of the saw blade, resulting in insufficient state detection accuracy, high false alarm rate, and further affecting the product machining quality and production efficiency.

[0003] To solve the above problems, a technical solution is provided. SUMMARY

[0004] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide a saw blade detection method and system based on multi-source data fusion to solve the problems raised in the background art.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical scheme: A saw blade detection method based on multi-source data fusion, comprising the following steps: S1: synchronously photographing the rake face, flank face, side edge and saw blade edge image of the saw blade, extracting the projection line straightness and edge initial position data corresponding to each tooth, and outputting the initial tooth profile reference feature of the saw blade; S2: taking the initial tooth profile reference feature of the saw blade as a reference, synchronously collecting the front and side high-speed image data of the saw blade when cutting a workpiece, analyzing the projection line straightness change and edge recession change, and outputting the visual wear feature data of the saw blade; S3: synchronously collecting the spindle current data when the saw blade cuts a workpiece, analyzing the increasing trend and variation characteristics of the current with the wear of the saw blade, and outputting the current wear feature data of the saw blade; S4: based on the visual wear feature data and the current wear feature data, analyzing the normal wear degradation trend and tooth loss abnormal change characteristics of the saw blade, and outputting the visual degradation trend data and the current abnormal change data; S5: based on the visual degradation trend data and the current abnormal change data, identifying the wear or tooth loss state of the saw blade, and outputting the comprehensive state evaluation result of the saw blade; S6: Based on the comprehensive condition assessment results of the saw blade and the non-conforming rate of the sawed products, output a saw blade replacement control signal.

[0006] In a preferred embodiment, S1 specifically refers to: Set the front camera and the side camera to trigger synchronously, drive the saw blade to rotate idling and collect images of the front cutting face, back cutting face, side cutting edge and saw blade cutting edge of the saw blade; Denoising, distortion correction and timing alignment are performed on the images of the front cutting face, back cutting face, side cutting edge and cutting edge of the saw blade to obtain the effective tooth shape image of each saw tooth; Based on the effective tooth shape image, the projection lines of each tooth are extracted and the straightness of the projection lines is calculated to locate the saw blade cutting edge and generate the initial position data of the cutting edge. The straightness of the projected lines and the initial position of the cutting edge are averaged to output the initial tooth profile reference features of the saw blade.

[0007] In a preferred embodiment, S2 specifically refers to: During the saw blade cutting workpiece, the front camera and side camera are triggered in real time to synchronously acquire images of the saw blade's front cutting face, back cutting face, side cutting edge and saw blade cutting edge based on the initial tooth profile reference features of the saw blade; Denoising, distortion correction, and timing alignment are performed on the images of the saw blade's front face, rear face, side edge, and cutting edge. Extract the straightness change data of the projection line and the position change data of the cutting edge, calculate the change in the straightness of the projection line and the change in the backlash of the cutting edge, and output the visual wear characteristic data of the saw blade.

[0008] In a preferred embodiment, S3 specifically refers to: During the saw blade cutting the workpiece, the spindle current data is collected synchronously, and the spindle current data is aligned according to the saw blade rotation cycle. Filtering and detrending processing are performed on the spindle current data to obtain a stable spindle current sequence; Based on the stable sequence of spindle current, mean features, growth slope features, variance features and peak features are extracted to form a current wear feature vector and output saw blade current wear feature data.

[0009] In a preferred embodiment, S4 specifically refers to: Based on the visual wear feature data of the saw blade, time series alignment processing is performed to analyze the continuous evolution relationship between the change in the straightness of the projected line and the change in the blade retraction, and visual wear degradation features are extracted to form visual degradation trend data. Based on the analysis of saw blade current wear characteristic data, the consistency and abrupt changes of mean characteristics, growth slope characteristics, variance characteristics and peak characteristics are analyzed to extract current abnormal change characteristics and form current abnormal change data.

[0010] In a preferred embodiment, S5 specifically refers to: Data synchronization processing is performed based on visual degradation trend data and abnormal current change data to establish a correlation mapping between visual wear and degradation characteristics and abnormal current change characteristics; Based on correlation mapping, it is determined whether visual degradation trend data and abnormal current change data change significantly at the same time, thus identifying the saw blade wear or tooth loss status and outputting the comprehensive condition assessment result of the saw blade.

[0011] In a preferred embodiment, S6 specifically refers to: Acquire quality inspection data of sawed products, collect the quality inspection data of sawed products according to a preset evaluation cycle, count the number of non-conforming products and the total number, and calculate the non-conforming rate of sawed product quality; The non-conformity rate of sawed products is compared with a preset trigger threshold, and the comparison result is jointly judged with the comprehensive condition evaluation result of the saw blade. When the joint determination meets the replacement conditions, a saw blade replacement control signal is generated.

[0012] On the other hand, the present invention provides a saw blade detection system based on multi-source data fusion, comprising: Synchronous shooting module: Synchronously shoots images of the front face, back face, side edge and saw blade cutting edge of the saw blade, extracts the straightness of the projection line corresponding to each tooth and the initial position data of the cutting edge, and outputs the initial tooth profile reference features of the saw blade. Visual wear module: Using the initial tooth profile reference features of the saw blade as a reference, it synchronously collects high-speed image data of the front and side of the saw blade when cutting the workpiece, analyzes the changes in the straightness of the projection line and the changes in the cutting edge retraction, and outputs visual wear feature data of the saw blade. Current Wear Module: Synchronously collects spindle current data when the saw blade is cutting the workpiece, analyzes the increasing trend and variation characteristics of current with saw blade wear, and outputs saw blade current wear characteristic data; Degradation Anomaly Module: Based on visual wear feature data and current wear feature data, analyze the normal wear degradation trend and abnormal tooth loss characteristics of the saw blade, and output visual degradation trend data and abnormal current change data; Condition assessment module: Based on visual degradation trend data and abnormal current change data, it identifies saw blade wear or tooth loss status and outputs a comprehensive condition assessment result of the saw blade; Replacement control module: Based on the comprehensive condition assessment results of the saw blade and the non-conforming rate of the sawed products, output a saw blade replacement control signal.

[0013] The technical effects and advantages of the saw blade detection method and system based on multi-source data fusion of this invention are as follows: By establishing initial tooth profile benchmark features for saw blades, a unified reference is provided for inspection, reducing the impact of batch and installation differences between saw blades on judgment. Quantitative analysis of changes in the straightness of the projected line and the amount of cutting edge retraction generates repeatable visual wear characteristic data, enabling continuous tracking of tooth profile degradation. Extracting the trend and variation characteristics of spindle current generates current wear characteristic data, enabling synchronous characterization of cutting load changes. Normal wear degradation trends and abnormal tooth loss characteristics are extracted separately, enhancing the ability to distinguish between different degradation mechanisms. Visual degradation trend data and abnormal current change data are fused and identified to output a comprehensive saw blade condition assessment result, improving the accuracy and stability of condition identification. The comprehensive saw blade condition assessment result is jointly judged with the defect rate of sawn products, and a saw blade replacement control signal is output, achieving objectivity and consistency in replacement timing. This reduces the risk of premature or delayed replacement due to misjudgment, improving processing quality stability and production continuity. Attached Figure Description

[0014] Figure 1 This is a schematic diagram of a saw blade detection method based on multi-source data fusion according to the present invention; Figure 2 This is a schematic diagram of the saw blade detection system based on multi-source data fusion according to the present invention. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention. Example

[0016] Figure 1 This invention presents a saw blade detection method based on multi-source data fusion, which includes the following steps: S1: Simultaneously capture images of the front face, back face, side edge, and cutting edge of the saw blade, extract the straightness of the projection line corresponding to each tooth and the initial position data of the cutting edge, and output the initial tooth profile reference features of the saw blade. S2: Using the initial tooth profile reference features of the saw blade as a reference, simultaneously collect high-speed image data of the front and side of the saw blade when cutting the workpiece, analyze the changes in the straightness of the projection line and the changes in the backlash of the cutting edge, and output the visual wear characteristic data of the saw blade. S3: Synchronously collect spindle current data when the saw blade is cutting the workpiece, analyze the increasing trend and variation characteristics of the current with saw blade wear, and output saw blade current wear characteristic data; S4: Based on visual wear characteristic data and current wear characteristic data, analyze the normal wear degradation trend and abnormal tooth loss characteristics of the saw blade, and output visual degradation trend data and abnormal current change data. S5: Based on visual degradation trend data and abnormal current change data, identify saw blade wear or tooth loss status, and output a comprehensive saw blade status assessment result. S6: Based on the comprehensive condition assessment results of the saw blade and the non-conforming rate of the sawed products, output a saw blade replacement control signal.

[0017] S1: Simultaneously capture images of the saw blade's front face, rear face, side edge, and cutting edge; extract the straightness of the projection lines corresponding to each tooth and the initial position data of the cutting edge; output the initial tooth profile reference features of the saw blade, including: Set the front camera and the side camera to trigger synchronously, drive the saw blade to rotate idling and collect images of the front cutting face, back cutting face, side cutting edge and saw blade cutting edge of the saw blade; The front camera and side camera are mounted directly in front of and to the side of the saw blade, respectively. The front camera captures visual images of the front and rear cutting faces of the saw blade, while the side camera captures visual images of the side and cutting edges of the saw blade. Synchronous triggering uses an external, uniform pulse trigger signal. This signal is generated by a signal generator and simultaneously transmitted to both the front and side cameras via a signal distributor, ensuring that both cameras initiate the exposure and image acquisition processes at the same time.

[0018] The saw blade is mounted on a controllable spindle and driven by a variable frequency speed-regulating motor to idle at a predetermined speed. The idle speed is selected based on the principle that it should not be lower than the maximum speed during actual processing; for example, the idle speed can be set to 120% of the maximum actual processing speed. Simultaneously, during the idle rotation of the saw blade, images of each saw tooth are continuously captured by synchronously triggered front and side cameras. These continuously captured images include the front cutting face, back cutting face, side cutting edge, and saw blade cutting edge images, until visual image data corresponding to all saw teeth is acquired.

[0019] Denoising, distortion correction and timing alignment are performed on the images of the front cutting face, back cutting face, side cutting edge and cutting edge of the saw blade to obtain the effective tooth shape image of each saw tooth; The noise reduction process is achieved through median filtering, which involves applying a set neighborhood size to each image, such as a three-pixel square neighborhood window, and performing median calculation pixel by pixel to eliminate random noise during image acquisition and obtain a clearer and more stable visual image of the saw blade.

[0020] Distortion correction is achieved using distortion correction parameters determined during camera calibration, including the calibration of the camera intrinsic matrix and distortion coefficients. Calibration images are acquired using a checkerboard calibration board, and the parameters are obtained according to Zhang's camera calibration method. Coordinate correction is then applied to each frame of the original image using the intrinsic matrix and distortion coefficients to remove the negative impact of lens distortion on the accuracy of visual measurement data.

[0021] Timing alignment processing uses the uniformly acquired synchronous trigger pulse signal as a reference to mark the timestamp frame by frame, ensuring the correspondence between each front camera image and the side camera image in the time dimension, and finally obtaining accurate and uniformly timed effective serrated images of each serration.

[0022] Based on the effective tooth shape image, the projection lines of each tooth are extracted and the straightness of the projection lines is calculated to locate the saw blade cutting edge and generate the initial position data of the cutting edge. The projection line extraction method combines edge detection and Hough line detection. First, the Canny edge detection algorithm is used to extract the edge contours of each saw tooth. Then, the Hough transform method is used to locate and fit the mathematical expression of the projection line of each saw tooth within the extracted edge contours. After obtaining the projection lines, the straightness of the projection lines is calculated. Based on the distance between the actual measurement point on the projection line and the fitted line, the ratio of the maximum deviation of the actual measurement point from the fitted line to the length of the fitted line is calculated and defined as the projection line straightness. The projection line straightness reflects the straightness characteristics of the saw face of each tooth in the initial state of the saw blade.

[0023] To generate initial cutting edge position data, the position of each saw blade cutting edge is accurately determined using a sub-pixel edge localization method. Specifically, a region of interest is selected along the saw blade cutting edge area in the side camera image. Preliminary edge detection is performed using the gray-level gradient method, and then the sub-pixel coordinates of the cutting edge position are determined using a curve fitting method. After the cutting edge position is determined, a two-dimensional Cartesian coordinate system is established with the spindle rotation center as a reference. The sub-pixel coordinates of the saw blade cutting edge are converted into real physical coordinates, forming unified initial cutting edge position data.

[0024] The straightness of the projected lines and the initial position of the cutting edge are averaged to output the initial tooth profile reference features of the saw blade. The straightness of all the saw tooth projection lines is summed and divided by the total number of saw teeth to obtain the average straightness of each saw tooth projection line, which serves as the baseline data for the straightness of the projection lines. Simultaneously, the initial position data of the cutting edge is averaged across two coordinate dimensions to obtain the average coordinate value of the cutting edge position of each saw tooth, which serves as the baseline data for the initial position of the cutting edge. The baseline characteristics of the saw blade's initial tooth profile include both the baseline data for the straightness of the projection lines and the baseline data for the initial position of the cutting edge.

[0025] S2: Using the initial tooth profile of the saw blade as a reference, simultaneously acquire high-speed image data of the front and side of the workpiece when the saw blade is cutting it, analyze the changes in the straightness of the projected lines and the changes in the cutting edge retraction, and output the visual wear characteristic data of the saw blade, including: During the saw blade cutting workpiece, the front camera and side camera are triggered in real time to synchronously acquire images of the saw blade's front cutting face, back cutting face, side cutting edge and saw blade cutting edge based on the initial tooth profile reference features of the saw blade; During actual sawing operations, the spindle speed of the saw blade is continuously monitored and recorded. A synchronized trigger signal matching the saw blade speed is generated in real time using a preset synchronized trigger signal generation method. The generation method involves measuring the spindle speed per minute and dividing it by the number of saw blade teeth per minute to obtain the time interval for each saw tooth to pass through the field of view. For example, when the spindle speed is 3000 revolutions per minute and the saw blade has 100 teeth, the interval for each tooth is 200 microseconds. Once the tooth passage interval is determined, a signal generator generates a trigger pulse signal with a period equal to the tooth passage interval in real time, which is then transmitted in real time to the front and side cameras via a signal distributor to ensure that the front and side cameras synchronously start image acquisition when each saw tooth passes through the shooting position.

[0026] Denoising, distortion correction, and timing alignment are performed on the images of the saw blade's front face, rear face, side edge, and cutting edge. The acquired images of the saw blade's front face, rear face, side edge, and cutting edge are processed one by one. Specifically, a square window of 3 pixels by 3 pixels is selected within each pixel of the image, centered on the current pixel. All pixel values ​​within the window are sorted, and the median value of the sorted values ​​replaces the original grayscale value of the center pixel. This removes random noise from the image and enhances its clarity and effectiveness. Distortion correction follows the same camera calibration parameters as in step S1. Specifically, using a pre-determined camera intrinsic parameter matrix and distortion coefficients, calibration parameters obtained through Zhang's calibration method are applied to each... After frame denoising, each image undergoes coordinate correction, including coordinate mapping transformation based on the intrinsic parameter matrix and distortion coefficients to eliminate radial and tangential distortions introduced by the camera lens, thereby obtaining a visual image with accurate geometric relationships. The temporal alignment process, while capturing each frame, uses the synchronous trigger signal corresponding to the image capture time as a reference to mark a unified timestamp for each frame, ensuring that the correspondence between each frame of image data from the front camera and the side camera on the time axis is completely consistent, thereby obtaining accurate and effective images of the saw blade's front face, back face, side edge, and saw blade edge.

[0027] Extract the straightness change data of the projection line and the position change data of the cutting edge, calculate the change in the straightness of the projection line and the change in the backlash of the cutting edge, and output the visual wear characteristic data of the saw blade. Following the projection line extraction method in step S1, the Canny edge detection algorithm is first used to extract edge features from the denoised and corrected front and rear face images of the saw blade in each frame. Then, the Hough line detection method is used to fit the projection lines of each saw tooth in the current real-time image from the extracted edge contours. After obtaining the real-time projection lines, the projection lines of each saw tooth are compared tooth by tooth with the linearity reference data of the projection lines in the initial tooth shape reference features of the saw blade. By calculating the least squares fitting error between the real-time image projection lines and the initial projection lines, the degree of linearity change of the projection lines is quantified, and the linearity change data of the projection lines is obtained. At the same time, in order to obtain the cutting edge position change data, the sub-pixel coordinates of the cutting edge position of each saw tooth in the real-time image are extracted. The real-time measured sub-pixel coordinates of the cutting edge are converted into real-time physical coordinate data through the spindle rotation center reference reference. Then, the position difference is calculated tooth by tooth with the initial position reference data of the cutting edge recorded in the initial tooth shape reference features of the saw blade. The difference is defined as the cutting edge position change data.

[0028] Statistical analysis was performed on the obtained projection line straightness variation data, including the calculation of the mean and standard deviation of the projection line straightness variation in multiple consecutive cutting cycles for each tooth, to quantify the trend of the projection line straightness variation of the saw blade's rake and flank faces during the cutting process. At the same time, statistical analysis was performed on the obtained cutting edge position variation data, including the statistical analysis of the movement distance of the cutting edge position variation data in multiple consecutive cutting cycles, and regression analysis was performed on the continuous change trend of the cutting edge position over time to calculate the average retraction rate and cumulative retraction amount of the cutting edge, so as to accurately quantify the degree of cutting edge retraction of the saw blade.

[0029] Finally, the output is the visual wear characteristic data of the saw blade, including the average value of the change in the straightness of the projection line corresponding to each saw tooth, the standard deviation of the change in the straightness of the projection line, the average value of the blade retraction, and the cumulative value of the blade retraction.

[0030] S3: Synchronously acquires spindle current data when the saw blade is cutting the workpiece, analyzes the increasing trend and variation characteristics of the current with saw blade wear, and outputs saw blade current wear characteristic data, including: During the saw blade cutting the workpiece, the spindle current data is collected synchronously, and the spindle current data is aligned according to the saw blade rotation cycle. While the saw blade is cutting the workpiece, a current sensor collects the current value of the spindle motor driving the saw blade in real time to assess the current wear characteristics of the saw blade. The selection of the current sensor is based on the rated power and rated current of the spindle motor. The range of the current sensor is set to 1.2 times the rated current of the spindle motor. For example, when the rated current of the spindle motor is 20 amps, the range of the current sensor can be set to 24 amps to ensure measurement accuracy and margin. The sampling frequency of the spindle current data is selected to be at least 10 times the product of the number of revolutions per second of the saw blade and the number of saw teeth. For example, when the saw blade rotates at 50 revolutions per second and has 100 teeth, the sampling frequency can be set to 50 revolutions / second × 100 teeth / revolution × 10 = 50,000 Hz, thus ensuring that each saw tooth obtains sufficient current data sampling points during the cutting process.

[0031] The timing alignment method for spindle current data is as follows: Based on the real-time measured rotation cycle of the saw blade, the continuously acquired spindle current data sequence is divided into multiple cycle segments. A single cycle segment is defined as the time length corresponding to the saw blade completing one full rotation. The calculation method is that the length of the cycle segment equals 60 seconds divided by the spindle speed per minute. For example, if the spindle speed is 3000 revolutions per minute, then the cycle segment length is 60 seconds / 3000 = 0.02 seconds, or 20 milliseconds. The start timestamp of each cycle segment is recorded by a real-time trigger signal. The spindle current data is then segmented segment by segment using these start timestamps, ensuring that the spindle current data contained in each cycle segment corresponds to one full rotation of the saw blade on the time axis, thus obtaining the timing alignment result.

[0032] Filtering and detrending processing are performed on the spindle current data to obtain a stable spindle current sequence; The filtering method specifically employs the Butterworth low-pass filter. The cutoff frequency of the Butterworth low-pass filter is set based on a multiple of the spindle rotation frequency; for example, the cutoff frequency can be set to a frequency value within the range of 2 to 5 times the spindle rotation frequency to eliminate high-frequency noise signals in the spindle current data while retaining the effective current change components caused by saw blade wear. Detrending processing is then performed on the filtered spindle current data using a linear regression method. This involves fitting a long-term trend line of the spindle current data over time using the least squares method, and then subtracting the corresponding trend line fitting value point by point from the filtered current data to obtain the detrended stable sequence data of the spindle current.

[0033] Based on the stable sequence of spindle current, mean features, growth slope features, variance features and peak features are extracted to form a current wear feature vector and output saw blade current wear feature data. The mean feature is defined as the arithmetic mean of all data points in the stable spindle current sequence within each saw blade rotation cycle, used to characterize the average energy consumption of saw blade cutting under stable current conditions; the growth slope feature is defined as the slope of the fitted line obtained after linear regression of the stable spindle current sequence data within each rotation cycle, characterizing the trend change characteristics of the spindle current data within a single cycle; the variance feature is defined as the average of the sum of squares of the deviations of the stable spindle current sequence data points from the mean within each cycle, used to quantitatively describe the magnitude of current fluctuations; the peak feature is defined as the difference between the maximum and minimum values ​​of the stable spindle current sequence data within each cycle, characterizing the maximum amplitude of current data fluctuations within a cycle.

[0034] The current wear feature vector, composed of mean feature, growth slope feature, variance feature, and peak feature, is organized in vector form. Specifically, within each saw blade rotation cycle, the current wear feature vector is recorded as [mean feature, growth slope feature, variance feature, peak feature]. Each feature value is calculated from the stable sequence data of the spindle current within the saw blade rotation cycle and expressed as a floating-point number. For example, the current wear feature vector within a single saw blade rotation cycle is [18.2350, 0.0152, 0.1124, 0.2500], representing the mean feature of the current data within the saw blade rotation cycle as 18.2350 amperes, the growth slope feature as 0.0152 amperes per millisecond, the variance feature as 0.1124 squared amperes, and the peak feature as 0.2500 amperes, respectively.

[0035] The final output saw blade current wear characteristic data is as follows: the current wear characteristic vectors calculated for each of the saw blade rotation cycles are arranged in a continuous time table to form a current wear characteristic data table. Each row corresponds to a saw blade rotation cycle, and each row records the mean characteristic, growth slope characteristic, variance characteristic and peak value characteristic within the saw blade rotation cycle.

[0036] S4: Based on visual wear characteristic data and current wear characteristic data, analyze the normal wear degradation trend and abnormal tooth loss characteristics of the saw blade, and output visual degradation trend data and abnormal current change data, including: Based on the visual wear feature data of the saw blade, time series alignment processing is performed to analyze the continuous evolution relationship between the change in the straightness of the projected line and the change in the blade retraction, and visual wear degradation features are extracted to form visual degradation trend data. The visual wear characteristic data of the saw blade, synchronously collected within each saw blade rotation cycle, are marked cycle by cycle. This includes four visual characteristic parameters: the average value of the change in the straightness of the projected line, the standard deviation of the change in the straightness of the projected line, the average value of the cutting edge retraction, and the cumulative value of the cutting edge retraction. Each cycle corresponds to a complete set of data. The visual characteristic parameters across multiple saw blade rotation cycles are marked with acquisition timestamps, forming a visual characteristic data sequence with continuous time identification. A sliding window alignment method is applied to the visual characteristic parameters of each consecutive cycle, segmenting the visual characteristic data sequence into segments with fixed-length time windows. The window length is determined based on the saw blade rotation cycle, for example, it can be set to 10 consecutive saw blade rotation cycles. The data sequence within each window obtained through this method accurately reflects the continuous evolution of visual wear characteristics over a continuous time period.

[0037] Using the visual feature data sequence within each window as the analysis unit, a periodic trend regression analysis is performed on the average value of the change in the straightness of the projected line and the average value of the edge retraction. The trend regression analysis employs the least squares method, which involves linearly or nonlinearly fitting the visual feature parameters corresponding to each period to obtain a trend model of the visual feature parameters changing over time. Linear fitting is suitable for representing linear degradation trends, while nonlinear fitting is suitable for representing nonlinear degradation trends. For example, when the visual feature parameters exhibit obvious linear changes, linear fitting is used to obtain the slope and intercept parameters of the trend line; when the visual feature parameters exhibit nonlinear changes, a polynomial fitting method is used to obtain a more accurate trend description. By comparing the trend models of the change in the straightness of the projected line and the trend models of the edge retraction, the correlation strength between the two visual wear features is quantified. For example, the Pearson correlation coefficient can be used to measure the strength of the continuous evolution relationship between the change in the straightness of the projected line and the change in the edge retraction.

[0038] The method for extracting visual wear and tear degradation features includes: obtaining the rate of change of the straightness of the projected line by analyzing the slope of the trend model of the straightness change of the projected line or the main coefficients in the higher-order polynomial model; obtaining the characteristics of the cutting edge retraction rate and the cumulative retraction by using the fitting parameters of the trend model of the cutting edge retraction; and extracting composite indicators reflecting the overall degradation state of visual wear and tear based on the analysis results of the two sets of trend model parameters and the Pearson correlation coefficient. For example, a weighted combination of the rate of change of the straightness of the projected line and the cutting edge retraction rate is used, with the weight coefficient determined according to the strength of the trend correlation, i.e., the higher the trend correlation, the larger the weight coefficient. The weight coefficient can be determined by normalizing the correlation coefficient value, for example, by dividing the correlation coefficient by the sum of the correlation coefficients of multiple visual feature parameters. This constructs visual degradation trend data, i.e., trend data containing degradation features is generated in each consecutive analysis window.

[0039] Based on the analysis of saw blade current wear characteristic data, the consistency and abrupt changes of mean characteristics, growth slope characteristics, variance characteristics and peak characteristics are analyzed to extract current abnormal change characteristics and form current abnormal change data. The current wear characteristic data for each cycle are normalized to ensure that the numerical ranges of the mean, growth slope, variance, and peak characteristics are consistent. The minimum value from the historical data of each characteristic parameter is subtracted from the minimum value, and then divided by the range of the historical data for that parameter. Based on the normalized current wear characteristic data, the changing trend of each characteristic parameter over multiple consecutive cycles is calculated. A sliding window method is used to fit the trend model cycle by cycle to extract the trend characteristics of each current characteristic parameter over consecutive cycles. For example, linear regression is used to obtain the slope parameter of the fitted straight line for the growth slope characteristic trend, thus characterizing the overall rate of change of the current characteristic.

[0040] The analysis method for consistency of changes involves calculating the sum of squared residuals of the trend models for each current characteristic parameter to evaluate the model fitting accuracy and thus measure the consistency of changes in each characteristic parameter over a continuous period. A smaller sum of squared residuals indicates higher consistency in the trend, while a larger sum of squared residuals indicates lower consistency and the possibility of abnormal fluctuations or abrupt changes. Simultaneously, to analyze the abrupt changes in the current characteristic parameters, abrupt change point detection analysis is performed on adjacent difference sequences of each parameter over a continuous period. The abrupt change point detection method can employ the moving standard deviation detection method, which calculates the standard deviation of adjacent windows within the difference sequence. When the standard deviation exceeds a preset threshold, an abrupt change is identified. The preset threshold is determined by statistically analyzing the average standard deviation and the standard deviation of the difference sequence under historical stable processing conditions. The preset threshold is set as the average standard deviation under historical stable processing conditions plus the standard deviation of three times the standard deviation. For example, when the average standard deviation of historical data is 0.05 and the standard deviation of the standard deviation is 0.02, the preset threshold can be set as 0.05 + 3 × 0.02 = 0.11. Exceeding the preset threshold indicates an abrupt change.

[0041] The characteristics of abnormal current changes include: the sum of squared residuals of the trend models for each current characteristic parameter; the number and magnitude of abrupt changes detected within a continuous period, with a higher number and magnitude indicating a more significant abnormal change. These characteristics combined form complete data on abnormal current changes.

[0042] S5: Based on visual degradation trend data and abnormal current change data, identify saw blade wear or tooth loss status, and output a comprehensive saw blade condition assessment result, including: Data synchronization processing is performed based on visual degradation trend data and abnormal current change data to establish a correlation mapping between visual wear and degradation characteristics and abnormal current change characteristics; Visual degradation trend data and current anomaly change data are labeled cycle-by-cycle according to the saw blade rotation cycle, using a unified cycle number as an identifier, ensuring a consistent correspondence between the two data over time. The characteristics of the projection line straightness change rate, blade retraction rate, and cumulative retraction amount for each cycle in the visual degradation trend data are recorded under the corresponding cycle number. Similarly, the characteristics of the mean residual sum of squares, growth slope residual sum of squares, variance residual sum of squares, peak residual sum of squares, number of abrupt change points, and abrupt change amplitude for each cycle in the current anomaly change data are also recorded one-to-one according to the cycle number. Through cycle labeling, a temporal correspondence is established between the visual degradation trend data and the current anomaly change data.

[0043] A composite index for visual wear degradation characteristics is constructed using the straightness change rate of the projected line and the blade retraction rate. This composite index of visual degradation trend data is arranged periodically to form a visual degradation trend sequence. Simultaneously, a composite index for current anomaly characteristics is formed using a weighted combination of the sum of squared residuals of each feature parameter in the current anomaly change data, as well as a weighted combination of the number of abrupt change points and the magnitude of the abrupt change. The weighting coefficients are determined based on the contribution of the sum of squared residuals of each feature parameter and the magnitude of the abrupt change to the sensitivity to anomaly changes. For example, the weighting coefficients are determined by the proportion of contribution of feature parameters obtained from historical data statistics to the sensitivity of saw blade anomaly identification. This composite index of current anomaly characteristics is then arranged periodically to form a current anomaly change sequence. The dynamic correlation between the visual degradation trend sequence and the current anomaly change sequence is calculated. The dynamic correlation calculation method uses a sliding window method combined with a dynamic Pearson correlation coefficient calculation method. Specifically, the Pearson correlation coefficient between the two sequences is calculated periodically with a preset window length to quantify the correlation strength between the two sequences within a continuous period and its changes over time. The length of the sliding window is determined by setting it based on the number of saw blade rotation cycles, for example, setting it to 20 consecutive cycles, to ensure that the data within the window length is sufficiently rich and statistically significant.

[0044] Based on correlation mapping, it is determined whether visual degradation trend data and abnormal current change data change significantly at the same time, thus identifying the saw blade wear or tooth loss status and outputting the comprehensive condition assessment result of the saw blade. The dynamic Pearson correlation coefficient calculated within each sliding window is used as the criterion. When the absolute value of the dynamic Pearson correlation coefficient exceeds a preset significant correlation threshold, it is determined that there is a simultaneous significant change between the visual degradation trend data and the abnormal current change data. The method for setting the significant correlation threshold is as follows: using the data set of saw blade operation under historical normal conditions, calculate the average and standard deviation of the Pearson correlation coefficients of the two sequences under historical conditions. The significant correlation threshold is determined to be the average of the historical Pearson correlation coefficients plus three times the standard deviation. For example, when the average value of the historical Pearson correlation coefficient is 0.25 and the standard deviation is 0.1, the significant correlation threshold can be set to 0.25 + 3 × 0.1 = 0.55. When it exceeds 0.55, it is considered that there is a simultaneous significant change.

[0045] Based on the numerical range and rate of change of the projection line straightness change rate, the blade retraction rate, and the cumulative retraction amount in the visual degradation trend data, and the number and magnitude of the residual sum of squares and abrupt change points in the abnormal current change data, a saw blade wear state judgment rule is constructed. The construction process of the judgment rule includes: for the normal wear state, using the visual and current characteristic data of the historical saw blade under normal operation as a reference, the numerical range of each characteristic parameter under the normal wear state is determined. Specifically, the range is the mean of the historical characteristic parameter data ± three times the standard deviation. Data within this range is considered to be in the normal wear state. For the abnormal tooth loss state, the abnormal thresholds of the blade retraction rate, cumulative retraction amount, and projection line straightness change rate in the visual degradation trend data are determined by statistical analysis of the historical tooth loss data set. At the same time, the abnormal thresholds of the residual sum of squares and abrupt change features in the abnormal current change data are also statistically determined. For example, if the projection line straightness change rate exceeds 1.5 times the upper limit of the normal range and the current residual sum of squares exceeds 2 times the upper limit of the normal range, it is judged as a tooth loss state.

[0046] Based on the wear condition judgment rules, the visual degradation trend data and current abnormal change data are comprehensively evaluated window by window. If the data in a certain sliding window simultaneously meets the abnormal thresholds for both the visual degradation trend data and the current abnormal change data, the saw blade is judged to be in an abnormal tooth loss state. If the abnormal thresholds are not met, but a significant change in the state persists and gradually approaches the upper limit of the abnormal threshold, it is judged to be in a critical state of severe wear. If a significant change in the state does not exist or gradually approaches the normal range, it is judged to be in a normal wear state. Finally, the judgment results of each window form a sequence of saw blade comprehensive state evaluation results.

[0047] To represent the comprehensive condition assessment results, the implementation method for outputting the comprehensive condition assessment results of the saw blade includes constructing a condition assessment output data table. Each row in the condition assessment output data table corresponds to a sliding window period segment, recording the composite index of visual degradation trend data, the composite index of abnormal current change data, the dynamic Pearson correlation coefficient, and the comprehensive condition judgment result within that period segment. For example, the record of a sliding window period segment in the condition assessment output data table can be represented as: [Window start period number, Window end period number, Visual composite index, Current composite index, Dynamic Pearson correlation coefficient, Comprehensive condition judgment result], such as [100, 119, 0.72, 0.85, 0.60, Tooth loss abnormality].

[0048] S6: Based on the overall condition assessment results of the saw blade and the defect rate of the sawn products, output a saw blade replacement control signal, including: Acquire quality inspection data of sawed products, collect the quality inspection data of sawed products according to a preset evaluation cycle, count the number of non-conforming products and the total number, and calculate the non-conforming rate of sawed product quality; Automated dimensional and surface quality inspection equipment is used to inspect the dimensions and surface quality of each sawn product after the sawing process. The automated dimensional inspection equipment specifically uses laser rangefinders or vision measurement sensors. The selection of the equipment is determined based on the accuracy requirements of the product being tested. For example, if the length accuracy requirement of the sawn product is within 0.05 mm, the laser rangefinder's measurement resolution must be better than 0.02 mm; if the surface quality requirement is no obvious scratches or cracks, the vision measurement sensor's resolution must be better than 0.01 mm. Each sawn product generates sawing product quality inspection data after passing through the inspection equipment, including dimensional deviation data and surface defect data. The dimensional deviation data is the deviation between the measured length or width and the design standard value, and the surface defect data records the number and location of scratches, cracks, or other abnormal features. Each product corresponds to unique sawing product quality inspection data.

[0049] After the quality inspection data of sawed products is generated, it is stored in the data management database according to the timestamp. The evaluation cycle is set by comprehensively considering the statistical significance of production cycle time and the amount of inspection data. For example, when the production cycle time is fast, and 20 products can be sawn per minute, the preset evaluation cycle can be set to 30 minutes, that is, data collection is completed every 30 minutes. All sawed product quality inspection data corresponding to each evaluation cycle are automatically extracted from the data management database according to the timestamp, and collected by marking the cycle number, forming a sawed product quality inspection data group indexed by the cycle number.

[0050] For each pre-set evaluation period, the quality inspection data sets of sawed products are collected, and their quality status is determined one by one based on pre-set product quality acceptance standards. The specific method for setting the product quality acceptance standards is determined by referring to relevant technical specifications and actual quality control objectives. For example, a product is considered acceptable if its dimensional deviation is within ±0.05 mm and the number of surface defects does not exceed 2; otherwise, it is considered unacceptable. Within each period's data set, all sawed product quality inspection data are individually judged as acceptable or unacceptable. The number of unacceptable products and the total number of all inspected products in that period are counted. Based on the number of unacceptable products and the total number, the unacceptability rate of the sawed products is calculated. The formula for calculating the unacceptability rate is: Unacceptability rate of sawed products = Number of unacceptable sawed products / Total number of all inspected sawed products. For example, when the number of unacceptable products is 5 and the total number is 200, the unacceptability rate of the sawed products in the evaluation period is 2.5%.

[0051] The non-conformity rate of sawed products is compared with a preset trigger threshold, and the comparison result is jointly judged with the comprehensive condition evaluation result of the saw blade. A preset trigger threshold is determined in advance based on actual production needs and quality control objectives. The preset trigger threshold is set using the average and standard deviation of the quality non-conformance rate obtained from historical production data. For example, the trigger threshold is determined by adding three times the standard deviation to the average non-conformance rate under normal historical production conditions. When the average non-conformance rate in historical production is 1% and the standard deviation is 0.3%, the preset trigger threshold is set to 1% + 3 × 0.3% = 1.9%. After each calculation of the non-conformance rate of sawed products, the value is compared with the preset trigger threshold. When the non-conformance rate of sawed products exceeds the preset trigger threshold, the comparison result is recorded as exceeding the threshold; otherwise, it is recorded as not exceeding the threshold. Comparisons are made for each evaluation cycle to determine the specific status of the sawed product quality within that cycle.

[0052] Extract the comprehensive condition assessment results of the saw blades that are completely consistent with the data collection period for the quality inspection of sawed products. The logical rule for joint judgment is as follows: if the comparison result of the defect rate of sawed products within the assessment period is above the threshold, and the comprehensive condition assessment result of the saw blades corresponding to the assessment period shows that the saw blades are in an abnormal state of tooth loss or a critical state of severe wear, then it is determined that the conditions for saw blade replacement are met; if the defect rate is not above the threshold, but the comprehensive condition assessment result of the saw blades shows an abnormal state of tooth loss, then it is determined that the conditions for saw blade replacement are met.

[0053] When the joint determination meets the replacement conditions, a saw blade replacement control signal is generated. When the joint judgment logic confirms that the sawing product quality non-conformity rate exceeds the threshold and the saw blade's overall condition assessment result is either a tooth loss abnormality or a critical state of severe wear within the current evaluation period, a saw blade replacement control signal is immediately triggered. The saw blade replacement control signal is implemented as a digital control signal; for example, a high logic level can be set to indicate that the saw blade replacement control signal is valid, and a low level to indicate invalidity. The control signal is transmitted to the automated control execution equipment or human-machine interface terminal via industrial control network communication protocol, and generates alarm or prompt information to notify on-site operators or automated tool changers to perform the saw blade replacement operation. For example, when the control signal is at a high level, the human-machine interface terminal automatically displays the prompt "Saw blade severe wear or tooth loss abnormality confirmed, please perform saw blade replacement immediately"; when the automated tool changer receives the high-level signal, it starts the automatic tool change program, stops the current sawing operation, and automatically replaces the saw blade, thereby ensuring the continuous and stable operation of the production line. Example

[0054] The difference between Embodiment 2 and Embodiment 1 is that this embodiment introduces a saw blade detection system based on multi-source data fusion.

[0055] Figure 2 A schematic diagram of a saw blade detection system based on multi-source data fusion is provided. The saw blade detection system based on multi-source data fusion includes: Synchronous shooting module: Synchronously shoots images of the front face, back face, side edge and saw blade cutting edge of the saw blade, extracts the straightness of the projection line corresponding to each tooth and the initial position data of the cutting edge, and outputs the initial tooth profile reference features of the saw blade. Visual wear module: Using the initial tooth profile reference features of the saw blade as a reference, it synchronously collects high-speed image data of the front and side of the saw blade when cutting the workpiece, analyzes the changes in the straightness of the projection line and the changes in the cutting edge retraction, and outputs visual wear feature data of the saw blade. Current Wear Module: Synchronously collects spindle current data when the saw blade is cutting the workpiece, analyzes the increasing trend and variation characteristics of current with saw blade wear, and outputs saw blade current wear characteristic data; Degradation Anomaly Module: Based on visual wear feature data and current wear feature data, analyze the normal wear degradation trend and abnormal tooth loss characteristics of the saw blade, and output visual degradation trend data and abnormal current change data; Condition assessment module: Based on visual degradation trend data and abnormal current change data, it identifies saw blade wear or tooth loss status and outputs a comprehensive condition assessment result of the saw blade; Replacement control module: Based on the comprehensive condition assessment results of the saw blade and the non-conforming rate of the sawed products, output a saw blade replacement control signal.

[0056] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0057] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0058] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0059] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0060] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0061] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0062] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0063] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0064] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A saw blade detection method based on multi-source data fusion, characterized in that, Includes the following steps: S1: Simultaneously capture images of the front face, back face, side edge, and cutting edge of the saw blade, extract the straightness of the projection line corresponding to each tooth and the initial position data of the cutting edge, and output the initial tooth profile reference features of the saw blade. S2: Using the initial tooth profile reference features of the saw blade as a reference, simultaneously collect high-speed image data of the front and side of the saw blade when cutting the workpiece, analyze the changes in the straightness of the projection line and the changes in the backlash of the cutting edge, and output the visual wear characteristic data of the saw blade. S3: Synchronously collect spindle current data when the saw blade is cutting the workpiece, analyze the increasing trend and variation characteristics of the current with saw blade wear, and output saw blade current wear characteristic data; S4: Based on visual wear characteristic data and current wear characteristic data, analyze the normal wear degradation trend and abnormal tooth loss characteristics of the saw blade, and output visual degradation trend data and abnormal current change data. S5: Based on visual degradation trend data and abnormal current change data, identify saw blade wear or tooth loss status, and output a comprehensive saw blade status assessment result. S6: Based on the comprehensive condition assessment results of the saw blade and the non-conforming rate of the sawed products, output a saw blade replacement control signal.

2. The saw blade detection method based on multi-source data fusion according to claim 1, characterized in that, S1, specifically: Set the front camera and the side camera to trigger synchronously, drive the saw blade to rotate idling and collect images of the front cutting face, back cutting face, side cutting edge and saw blade cutting edge of the saw blade; Denoising, distortion correction and timing alignment are performed on the images of the front cutting face, back cutting face, side cutting edge and cutting edge of the saw blade to obtain the effective tooth shape image of each saw tooth; Based on the effective tooth shape image, the projection lines of each tooth are extracted and the straightness of the projection lines is calculated to locate the saw blade cutting edge and generate the initial position data of the cutting edge. The straightness of the projected lines and the initial position of the cutting edge are averaged to output the initial tooth profile reference features of the saw blade.

3. The saw blade detection method based on multi-source data fusion according to claim 2, characterized in that, S2, specifically: During the saw blade cutting workpiece, the front camera and side camera are triggered in real time to synchronously acquire images of the saw blade's front cutting face, back cutting face, side cutting edge and saw blade cutting edge based on the initial tooth profile reference features of the saw blade; Denoising, distortion correction, and timing alignment are performed on the images of the saw blade's front face, rear face, side edge, and cutting edge. Extract the straightness change data of the projection line and the position change data of the cutting edge, calculate the change in the straightness of the projection line and the change in the backlash of the cutting edge, and output the visual wear characteristic data of the saw blade.

4. The saw blade detection method based on multi-source data fusion according to claim 3, characterized in that, S3, specifically: During the saw blade cutting the workpiece, the spindle current data is collected synchronously, and the spindle current data is aligned according to the saw blade rotation cycle. Filtering and detrending processing are performed on the spindle current data to obtain a stable spindle current sequence; Based on the stable sequence of spindle current, mean features, growth slope features, variance features and peak features are extracted to form a current wear feature vector and output saw blade current wear feature data.

5. The saw blade detection method based on multi-source data fusion according to claim 4, characterized in that, S4, specifically: Based on the visual wear feature data of the saw blade, time series alignment processing is performed to analyze the continuous evolution relationship between the change in the straightness of the projected line and the change in the blade retraction, and visual wear degradation features are extracted to form visual degradation trend data. Based on the analysis of saw blade current wear characteristic data, the consistency and abrupt changes of mean characteristics, growth slope characteristics, variance characteristics and peak characteristics are analyzed to extract current abnormal change characteristics and form current abnormal change data.

6. The saw blade detection method based on multi-source data fusion according to claim 5, characterized in that, S5, specifically: Data synchronization processing is performed based on visual degradation trend data and abnormal current change data to establish a correlation mapping between visual wear and degradation characteristics and abnormal current change characteristics; Based on correlation mapping, it is determined whether visual degradation trend data and abnormal current change data change significantly at the same time, thus identifying the saw blade wear or tooth loss status and outputting the comprehensive condition assessment result of the saw blade.

7. The saw blade detection method based on multi-source data fusion according to claim 6, characterized in that, S6, specifically: Acquire quality inspection data of sawed products, collect the quality inspection data of sawed products according to a preset evaluation cycle, count the number of non-conforming products and the total number, and calculate the non-conforming rate of sawed product quality; The non-conformity rate of sawed products is compared with a preset trigger threshold, and the comparison result is jointly judged with the comprehensive condition evaluation result of the saw blade. When the joint determination meets the replacement conditions, a saw blade replacement control signal is generated.

8. A saw blade detection system based on multi-source data fusion, used to implement the saw blade detection method based on multi-source data fusion as described in any one of claims 1-7, characterized in that, include: Synchronous shooting module: Synchronously shoots images of the front face, back face, side edge and saw blade cutting edge of the saw blade, extracts the straightness of the projection line corresponding to each tooth and the initial position data of the cutting edge, and outputs the initial tooth profile reference features of the saw blade. Visual wear module: Using the initial tooth profile reference features of the saw blade as a reference, it synchronously collects high-speed image data of the front and side of the saw blade when cutting the workpiece, analyzes the changes in the straightness of the projection line and the changes in the cutting edge retraction, and outputs visual wear feature data of the saw blade. Current Wear Module: Synchronously collects spindle current data when the saw blade is cutting the workpiece, analyzes the increasing trend and variation characteristics of current with saw blade wear, and outputs saw blade current wear characteristic data; Degradation Anomaly Module: Based on visual wear feature data and current wear feature data, analyze the normal wear degradation trend and abnormal tooth loss characteristics of the saw blade, and output visual degradation trend data and abnormal current change data; Condition assessment module: Based on visual degradation trend data and abnormal current change data, it identifies saw blade wear or tooth loss status and outputs a comprehensive condition assessment result of the saw blade; Replacement control module: Based on the comprehensive condition assessment results of the saw blade and the non-conforming rate of the sawed products, output a saw blade replacement control signal.

Images