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

Abstract

The application discloses a saw blade detection system based on multi-source data fusion, and particularly relates to the technical field of saw blade detection; the saw blade detection system is characterized in that: side surface image data of a saw blade is collected, tooth back surface projection line features and edge geometry features are extracted, initial visual feature data of the saw blade is established; side surface image data is continuously collected in actual processing, side edge projection line features and edge position changes are compared and analyzed, visual wear data is determined; main shaft current data is synchronously collected and current change trends and variability are analyzed, current wear data is determined; normal wear and tooth loss abnormal trends of the saw blade are analyzed based on the visual wear data and the current wear data, visual degradation features and current abnormal features are obtained; visual and current fusion feature data is calculated in combination with product processing unqualified rate, and a saw blade scrapping trigger threshold is established; finally, fusion feature data and the scrapping threshold are compared in real time, so that the saw blade scrapping or replacement time is determined; and the application improves the precision and reliability of saw blade state monitoring.

Description

Technical Field

[0001] This invention relates to the field of saw blade testing technology, and more specifically, to a saw blade testing system based on multi-source data fusion. Background Technology

[0002] In the field of machining, the wear condition of saw blades directly affects workpiece processing quality, processing efficiency, and equipment safety. Traditional saw blade wear detection methods typically rely on periodic manual inspections or single-parameter monitoring, such as simple current detection or basic visual inspection. Manual inspection suffers from subjectivity, lag, and low precision, making it difficult to achieve accurate real-time assessment. Relying solely on single detection methods like current monitoring or image monitoring cannot comprehensively reflect the saw blade's wear condition and abnormal states, potentially leading to missed detections or misjudgments. This results in the failure to promptly identify abnormal saw blade conditions, increasing the risk of decreased processing quality, increased product scrap rates, and even equipment damage.

[0003] To address the aforementioned problems, a technical solution is provided. Summary of the Invention

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

[0005] To achieve the above objectives, the present invention provides the following technical solution: A saw blade detection system based on multi-source data fusion includes: Image acquisition module: Acquires image data of the side of the saw blade, extracts the projection line features of the back face of the saw teeth and the geometric features of the cutting edge position, averages the feature parameters, and outputs the initial visual feature data of the saw blade. Visual wear module: Based on the initial visual feature data of the saw blade, it collects side images of the saw blade in real time during the processing, analyzes the changes in the projection line of the side edge and the amount of backlash of the cutting edge position, and determines the visual wear data of the saw blade. Current monitoring module: Synchronously collects spindle current data when the saw blade is processing the workpiece, analyzes the current change trend and variability, and determines the current wear data of the saw blade; Feature Analysis Module: Based on visual wear data and current wear data, analyzes the degradation trend of the saw blade under normal wear conditions and the abnormal trend under abnormal tooth loss conditions, and generates visual degradation features and current abnormal features. Fusion Threshold Module: Integrates visual degradation features and current anomaly features, calculates visual and current fusion feature data based on the product processing quality non-conformity rate, and establishes a saw blade scrapping trigger threshold; Scrapping determination module: Based on the fusion feature data and the saw blade scrapping trigger threshold, it identifies whether the saw blade has reached the scrapping conditions of wear or tooth loss, and generates a saw blade scrapping or replacement control signal.

[0006] In a preferred embodiment, image data of the side of the saw blade is acquired, and the projection line features of the back face of the saw teeth and the geometric features of the cutting edge position are extracted. The feature parameters are averaged to output the initial visual feature data of the saw blade, specifically: A high-speed camera is placed on the side of the saw blade to continuously capture images of the side of the saw teeth when the saw blade is idling, and obtain side image data including the back face and the cutting edge area of ​​the saw teeth. Extract the projection line contour corresponding to the serrated back face and the spatial position parameters of the cutting edge in the back face direction based on the side image data. The projection line contour parameters and cutting edge spatial position parameters corresponding to multiple saw teeth are statistically collected and averaged to form the initial visual feature data of the saw blade.

[0007] In a preferred embodiment, based on the initial visual feature data of the saw blade, side images of the saw blade are acquired in real time during the processing, and the changes in the projection lines of the side edges and the amount of retraction of the cutting edge position are analyzed to determine the visual wear data of the saw blade, specifically: During the saw blade's workpiece cutting process, the saw blade's side image data is continuously collected according to the saw tooth spatial position and imaging area corresponding to the initial visual feature data of the saw blade; Based on the side image data of the saw blade, the projection line contour of the saw tooth back face and the spatial position parameters of the cutting edge in the back face direction are extracted under the cutting state. The visual wear data of the saw blade is obtained by comparing and calculating the projection line contour parameters and cutting edge spatial position parameters obtained under the cutting state with the initial visual feature data of the saw blade.

[0008] In a preferred embodiment, the spindle current data during saw blade machining of the workpiece is collected synchronously, and the current variation trend and variability are analyzed to determine the current wear data of the saw blade. Specifically: The spindle current signal during the saw blade's participation in workpiece cutting is monitored synchronously. The saw tooth passing frequency is calculated based on the spindle speed. Periodic segmentation and phase calibration are performed on the continuous current sampling sequence. The instantaneous peak value, average value and variance in each segmentation period are calculated and a current statistical sequence is constructed in time order. Based on the monotonically increasing trend and abnormal abrupt change amplitude of the statistical sequence, the characteristic parameters of current change are extracted, and combined with the corresponding spatial position of the saw teeth in the saw blade visual wear data, the saw blade current wear data is generated.

[0009] In a preferred embodiment, based on visual wear data and current wear data, the degradation trend of the saw blade under normal wear conditions and the abnormal trend under abnormal tooth loss conditions are analyzed to generate visual degradation features and current abnormal features, specifically: Based on the changes in the projection line contour and the spatial position of the cutting edge in the visual wear data of the saw blade, the continuity characteristics of the wear evolution of the back face of the saw tooth are analyzed. Based on the statistical sequence changes in current in saw blade current wear data, the stability characteristics of saw tooth cutting load over time are analyzed. By correlating the geometric changes reflected in the visual wear data of the saw blade with the load changes reflected in the current wear data of the saw blade, visual degradation characteristics and current anomaly characteristics are formed.

[0010] In a preferred embodiment, visual degradation features and current anomaly features are integrated. Based on the product processing quality defect rate, visual and current fusion feature data are calculated, and a saw blade scrapping trigger threshold is established. Specifically: Based on the establishment of a corresponding mapping relationship between visual degradation features and current anomaly features, and combined with the product processing quality non-conformity rate obtained synchronously during the processing, the correlation interval of quality change under different combinations of visual degradation features and current anomaly features is statistically analyzed. Visual degradation feature parameters and current anomaly feature parameters within the quality change correlation interval are jointly collected and threshold boundary is calculated to form visual and current fusion feature data and saw blade scrapping trigger threshold.

[0011] In a preferred embodiment, based on the fused feature data and the saw blade scrapping trigger threshold, it is identified whether the saw blade has reached the scrapping condition of wear or tooth loss, and a saw blade scrapping or replacement control signal is generated, specifically: Based on the visual and current fusion feature data, the visual and current fusion feature data is compared with the saw blade scrapping trigger threshold in real time; When the visual and electrical fusion feature data reaches or exceeds the saw blade scrapping trigger threshold, a saw blade scrapping or replacement control signal is output. When the visual and current fusion feature data does not reach the saw blade scrapping trigger threshold, update the visual and current fusion feature data and repeat the comparison process.

[0012] The technical effects and advantages of the saw blade detection system based on multi-source data fusion of the present invention are as follows: This invention achieves consistent assessment of saw blade wear and tooth loss throughout the entire process by collaboratively acquiring and fusing saw blade side image data and spindle current data. It establishes initial visual feature data for the saw blade, creating a unified benchmark for the projection line characteristics of the saw tooth rear face and the geometric characteristics of the cutting edge position, reducing the impact of individual differences between saw blades on the assessment results. During processing, visual wear data is extracted from the initial visual feature data, making changes in the side edge projection line and the cutting edge retraction comparable and traceable. Simultaneously, current wear data is constructed, allowing current trends and variability to reflect changes in cutting load. Visual wear data and current wear data are analyzed into visual degradation features and current anomaly features, respectively, taking into account both normal wear degradation and tooth loss anomalies. A saw blade scrapping trigger threshold is established based on the product processing quality non-conformity rate, creating a closed-loop correlation between scrapping judgment and quality results. Finally, the fused feature data and threshold are compared in real-time to output a saw blade scrapping or replacement control signal, achieving stable consistency between saw blade condition assessment and disposal timing. Attached Figure Description

[0013] Figure 1 This is a schematic diagram of the saw blade detection system based on multi-source data fusion according to the present invention. Detailed Implementation

[0014] 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

[0015] Figure 1 This invention discloses a saw blade detection system based on multi-source data fusion, comprising: Image acquisition module: Acquires image data of the side of the saw blade, extracts the projection line features of the back face of the saw teeth and the geometric features of the cutting edge position, averages the feature parameters, and outputs the initial visual feature data of the saw blade. Visual wear module: Based on the initial visual feature data of the saw blade, it collects side images of the saw blade in real time during the processing, analyzes the changes in the projection line of the side edge and the amount of backlash of the cutting edge position, and determines the visual wear data of the saw blade. Current monitoring module: Synchronously collects spindle current data when the saw blade is processing the workpiece, analyzes the current change trend and variability, and determines the current wear data of the saw blade; Feature Analysis Module: Based on visual wear data and current wear data, analyzes the degradation trend of the saw blade under normal wear conditions and the abnormal trend under abnormal tooth loss conditions, and generates visual degradation features and current abnormal features. Fusion Threshold Module: Integrates visual degradation features and current anomaly features, calculates visual and current fusion feature data based on the product processing quality non-conformity rate, and establishes a saw blade scrapping trigger threshold; Scrapping determination module: Based on the fusion feature data and the saw blade scrapping trigger threshold, it identifies whether the saw blade has reached the scrapping conditions of wear or tooth loss, and generates a saw blade scrapping or replacement control signal.

[0016] Image data of the saw blade's side surface is acquired, and the projection line features of the saw tooth's rear face and the geometric features of the cutting edge position are extracted. The feature parameters are averaged, and the initial visual feature data of the saw blade is output, including: A high-speed camera is placed on the side of the saw blade to continuously capture images of the side of the saw teeth when the saw blade is idling, and obtain side image data including the back face and the cutting edge area of ​​the saw teeth. The implementation of placing a high-speed camera on the side of the saw blade is as follows: Based on the saw blade's diameter, tooth density, and the available space conditions, determine the camera's installation position. Ensure the camera's lens axis is perpendicular to the side of the saw blade, and adjust the lens focal length to cover the entire side feature area of ​​the saw teeth. The selected high-speed camera's imaging frequency should be higher than the frequency of the total number of saw teeth passing through the camera's field of view per second, calculated by multiplying the saw blade's rotational speed by the number of saw teeth. For example, based on actual processing conditions, measure the saw blade's maximum rotational speed and, based on the number of saw teeth on the saw blade's circumference, multiply the maximum rotational speed by the number of saw teeth to obtain the minimum image sampling frequency required per unit time. Then, based on this minimum image sampling frequency, select a higher high-speed camera imaging frequency, for example, two to five times the minimum image sampling frequency, to meet the needs of continuous imaging acquisition.

[0017] The method for continuous imaging acquisition of the saw tooth side in the idle state of the saw blade is as follows: Before the saw blade starts actual workpiece cutting, the high-speed camera is started at a predetermined imaging frequency to continuously capture images of the saw blade in the idle rotating state for multiple rotation cycles, obtaining multiple frames of side image data including the saw tooth back face area and the cutting edge area. The number of saw teeth contained in each frame of side image data must cover all saw teeth on the circumference of the saw blade to ensure that the side morphology information of each saw tooth can be completely acquired.

[0018] Extract the projection line contour corresponding to the serrated back face and the spatial position parameters of the cutting edge in the back face direction based on the side image data. Taking each frame of side image data as the processing object, the first step is to perform digital image preprocessing steps on the side image data, including but not limited to image grayscale processing, image noise reduction filtering, image sharpening, and image edge enhancement processing, to improve the accuracy of contour extraction. A projection line contour extraction algorithm based on image edge detection is then employed, using either the Canny operator or the Sobel operator to process the preprocessed side image data to identify and extract the projection line contour corresponding to the serrated back face. Using image coordinate calibration methods, the projection line contour data in the pixel coordinate system is transformed to a spatial coordinate system. For example, by pre-calibrating standard-sized marker points within the field of view of the high-speed camera, the conversion ratio between image pixel size and actual spatial size is determined to achieve accurate calibration of the spatial position of the projection line contour. Based on the spatial coordinate data of the projection line contour, the spatial position parameters of the cutting edge in the back face direction are calculated. These spatial position parameters include, for example, the spatial coordinate position of the cutting edge within the field of view of the high-speed camera or the distance value relative to a reference position.

[0019] Statistical collection and mean calculation are performed on the projection line contour parameters and cutting edge spatial position parameters corresponding to multiple saw teeth to form the initial visual feature data of the saw blade. The projection line contour parameters and cutting edge spatial position parameters of all saw teeth within the same frame of side image data are statistically classified. The classification criteria are, for example, based on the saw tooth's serial number on the saw blade circumference, and parameter data sets are established sequentially. Then, for all frames of side image data within multiple complete rotation cycles, the projection line contour parameters and cutting edge spatial position parameters of saw teeth with the same serial number are statistically calculated. For example, the arithmetic mean calculation method is used to average the data in each data set to obtain the mean of the projection line contour parameters and the mean of the cutting edge spatial position parameters corresponding to each saw tooth. Then, the mean of the projection line contour parameters and the mean of the cutting edge spatial position parameters corresponding to all saw teeth on the saw blade are statistically calculated again. For example, the arithmetic mean of all saw teeth is calculated again to obtain the initial visual feature data of the saw blade representing the initial state of the overall saw blade idling.

[0020] Based on the initial visual feature data of the saw blade, side images of the saw blade are acquired in real time during processing. The changes in the projection lines of the side edges and the amount of retraction of the cutting edge position are analyzed to determine the visual wear data of the saw blade, including: During the saw blade's workpiece cutting process, the saw blade's side image data is continuously collected according to the saw tooth spatial position and imaging area corresponding to the initial visual feature data of the saw blade; Throughout the entire process of the saw blade actually contacting the workpiece and beginning cutting, a high-speed camera positioned on the side of the saw blade operates continuously at a predetermined imaging frequency, uninterruptedly acquiring image data from the side of the saw blade to ensure that the side state of each saw tooth is completely recorded throughout the entire workpiece cutting process. The method for determining the imaging frequency of the high-speed camera is as follows: first, the real-time rotational speed of the saw blade under actual cutting conditions is measured, and the measured real-time rotational speed is multiplied by the total number of saw teeth on the saw blade to obtain the real-time passing frequency of the saw teeth through the camera's field of view per unit time. Based on the real-time passing frequency, the imaging frequency of the high-speed camera is set to an integer multiple of the real-time passing frequency, for example, it can be set to three or four times the real-time passing frequency, so that each saw tooth is continuously photographed at least three to four times under workpiece cutting conditions.

[0021] Based on the side image data of the saw blade, the projection line contour of the saw tooth back face and the spatial position parameters of the cutting edge in the back face direction are extracted under the cutting state. Each frame of continuously acquired saw blade side image data is used as an independent processing unit. First, digital image preprocessing is performed on each frame, including but not limited to image grayscale conversion, noise reduction filtering, sharpening enhancement, and edge refinement, to improve the image clarity and recognition accuracy of the saw tooth back face projection line contour and the spatial position of the cutting edge in each frame. The same image edge detection-based projection line contour extraction algorithm as used in the idling state image data processing, such as the Canny or Sobel operator, is applied to each preprocessed frame with the same edge detection algorithm parameter configuration to extract the projection line contour of each saw tooth in the cutting state. Based on the spatial coordinate transformation ratio determined by the image coordinate calibration method, the saw tooth projection line contour extracted from each frame is transformed from the image pixel coordinate system to the spatial coordinate system. The method for calculating the projection line profile in the spatial coordinate system is as follows: First, for the projection line profile of each saw tooth, the least squares method is used to perform straight line fitting to obtain the straight line fitting parameters, including but not limited to the slope and intercept of the fitted straight line; based on the fitted straight line parameters, the position parameters of the saw tooth cutting edge in the spatial coordinate system are calculated. For example, the spatial coordinate distance of the saw tooth cutting edge relative to the preset saw blade rotation center reference point is calculated as the spatial position parameter of the cutting edge in the cutting state.

[0022] The projection line contour parameters and cutting edge spatial position parameters obtained under the cutting state are compared and calculated with the initial visual feature data of the saw blade to obtain the visual wear data of the saw blade. The projection line contour parameters and cutting edge spatial position parameters extracted from each saw tooth in the current cutting state are compared one by one with the initial visual feature data of the saw blade obtained in the idle state for the corresponding saw tooth. First, for the comparison calculation of the projection line contour parameters, the difference of the projection line contour parameters of the saw teeth in the cutting state and the idle state is calculated respectively. Taking the contour parameters in the idle state as the reference value, the difference of the contour parameters in the current cutting state relative to the reference value is calculated for each saw tooth. For example, the difference of the slope and intercept parameters after the straight line fitting is calculated by direct subtraction, or the difference of the distance from each feature point of the contour to the fitted straight line, etc. For the comparative calculation of the cutting edge spatial position parameters, the cutting edge position parameters of each saw tooth in the initial visual feature data of the saw blade are used as the reference value. The offset of the cutting edge spatial position of the corresponding saw tooth in the actual cutting process relative to the reference value is calculated one by one. For example, by calculating the vector difference or distance difference of the spatial coordinate position, the change of the saw tooth cutting edge position in the cutting process can be obtained.

[0023] After statistical summarization of the difference in the projection line contour and the offset of the cutting edge position, the data is output in the form of a data table or data matrix to obtain the visual wear data of the saw blade.

[0024] Synchronously collect spindle current data during saw blade machining, analyze current variation trends and variability, and determine saw blade current wear data, including: The spindle current signal during the saw blade's participation in workpiece cutting is monitored synchronously. The saw tooth passing frequency is calculated based on the spindle speed. Periodic segmentation and phase calibration are performed on the continuous current sampling sequence. The instantaneous peak value, average value and variance in each segmentation period are calculated and a current statistical sequence is constructed in time order. Throughout the entire process of the saw blade actually contacting the workpiece and performing cutting, a current sensor is connected in series in the spindle drive motor circuit of the sawing machine to continuously collect the current of the spindle drive motor, thereby reflecting the cutting load state of the saw blade during processing in real time. The selection method for the current sensor is as follows: the range of the current sensor is determined based on the rated power and maximum load current value of the spindle drive motor. For example, when the rated current value of the spindle drive motor is 10 amps, a current sensor with a range of twice the rated current can be selected to ensure that the current signal of the spindle drive motor under peak load conditions can still be accurately captured. The signal output terminal of the current sensor is connected to a data acquisition device. The data acquisition device continuously collects the analog current signal output by the current sensor at a preset current signal sampling frequency and converts it into a digital current signal sequence. For example, the sampling frequency can be set to a multiple of the product of the spindle rotation speed and the number of saw teeth on the saw blade, ensuring that the current change of each saw tooth as it passes through the spindle rotation is fully recorded.

[0025] The calculation method for the saw tooth passing frequency based on the spindle speed is as follows: First, a rotational speed sensor is installed on the spindle rotation axis to measure the actual rotational speed of the spindle in real time. The rotational speed sensor can be, for example, a photoelectric encoder or a magnetoelectric speed sensor to ensure the accuracy and real-time performance of the speed measurement. The real-time measured spindle rotational speed is multiplied by the actual number of saw teeth installed on the circumference of the saw blade to obtain the frequency at which all saw teeth on the saw blade pass through the cutting position per unit time, which is the saw tooth passing frequency. The calculation method is to multiply the spindle rotational speed (in revolutions per second) by the total number of saw teeth on the circumference of the saw blade to obtain the saw tooth passing frequency in Hertz.

[0026] Periodic segmentation and phase calibration are performed on the continuous current sampling sequence. Specifically, the acquired continuous digital current signal sequence is segmented into segments based on the period of the sawtooth passage frequency. That is, the continuous current signal sequence is divided into several equal-length data segments according to each complete sawtooth passage period, with each data segment representing the current change process during the complete passage of a sawtooth at the cutting position. Phase calibration is performed on the current signal segments within each segmented period. The calibration method is to align the maximum current peak position of each current signal segment to a unified time reference point to achieve phase uniformity among all current segments. Phase calibration can be implemented using cross-correlation algorithms or peak detection algorithms. For example, a cross-correlation algorithm can be used to calculate the phase difference of the current signal within adjacent periods, and the current signal segments can be shifted and adjusted based on the phase difference to ensure the consistency of the current statistical sequence.

[0027] The instantaneous peak value, average value, and variance within each segmented period are calculated and a current statistical sequence is constructed in chronological order. Specifically, within each current signal segment after period segmentation and phase calibration, the instantaneous peak current, periodic average current, and periodic current variance are extracted. The instantaneous peak current is determined by the maximum current value in each period's current signal segment; the periodic average current is determined by calculating the arithmetic mean of the current values ​​at all sampling points within each period's current signal segment; and the periodic current variance is determined by first calculating the difference between the current value at each sampling point and the periodic average current value, then arithmetically averaging the squares of these differences. After the instantaneous peak current, periodic average current, and periodic current variance are calculated periodically, they are stored sequentially in time series format, ultimately yielding the current statistical sequence.

[0028] Based on the monotonically increasing trend and abnormal abrupt change amplitude of the statistical sequence, the characteristic parameters of current change are extracted, and combined with the corresponding saw tooth spatial position in the saw blade visual wear data, the saw blade current wear data is generated. The current variation characteristic parameters are extracted based on the monotonically increasing trend and anomalous abrupt change amplitude of the current statistical sequence. Specifically, for the periodic average current value and instantaneous peak current value in the current statistical sequence, differential calculation is performed periodically, that is, the current value of adjacent periods is subtracted to obtain the current variation trend characteristics. By judging the difference results, when the difference results of several consecutive periods are all positive, it can be determined that the current statistical sequence shows a monotonically increasing trend. If the difference value of the current value between periods is significantly greater than the average level of the difference values ​​of the previous several periods in a certain period, for example, when the current difference value is more than twice the average difference value of the previous ten periods, it is determined that there is an anomalous abrupt change amplitude in the period. In this way, the variation trend characteristics and anomalous abrupt change amplitude characteristics of the current statistical sequence are extracted as current variation characteristic parameters.

[0029] By combining the spatial positions of the saw teeth in the visual wear data of the saw blade, saw blade current wear data is generated. Specifically, based on the spatial position parameters corresponding to each saw tooth in the visual wear data of the saw blade, the current change characteristic parameters are correlated one-to-one with the spatial position parameters of the saw teeth to establish a spatial matching relationship between visual data and current data. Statistical analysis is performed on the spatial matching relationship, and the visual wear data of the saw teeth corresponding to the same spatial position and the current change characteristic parameters are normalized simultaneously to unify the order of magnitude of different parameters. Then, a weighted average algorithm or principal component analysis method is used to fuse the visual and current data to obtain a fused parameter that can simultaneously reflect the visual wear state and the current load state, which serves as the saw blade current wear data.

[0030] Based on visual wear data and current wear data, the degradation trend of saw blades under normal wear conditions and the abnormal trend under abnormal tooth loss conditions are analyzed, generating visual degradation features and current abnormal features, including: Based on the changes in the projection line contour and the spatial position of the cutting edge in the visual wear data of the saw blade, the continuity characteristics of the wear evolution of the back face of the saw tooth are analyzed. Using the difference in the projection line profile and the offset of the cutting edge spatial position in the visual wear data of each saw tooth as basic data, complete trend data sequences of the projection line profile difference and the cutting edge spatial position offset are established according to the time sequence of the saw blade throughout the entire workpiece cutting process. These are called the projection line profile change trend data sequence and the cutting edge spatial position change trend data sequence, respectively. The construction method of the projection line profile change trend data sequence is as follows: for the difference in the projection line profile of each saw tooth, data points collected at various moments during the actual cutting of the saw blade are sequentially sorted to form a time series. Regression analysis is used to fit the obtained projection line profile change trend data sequence. For example, linear regression or nonlinear regression models are used. If the projection line profile change trend data sequence shows a linear trend, the least squares linear regression method is used for fitting; if the change trend data sequence shows a nonlinear trend, multinomial regression analysis or exponential regression analysis is used for fitting. Through regression analysis, a trend function describing the change of the projection line profile is obtained, thereby quantitatively describing the continuous evolution characteristics of the saw tooth back face wear evolution throughout the entire cutting cycle.

[0031] Similarly, based on data points of the spatial position offset of the saw tooth cutting edge collected each time during the saw blade processing, a data sequence of the trend of the spatial position change of the cutting edge is established sequentially according to the time sequence of the processing process. Regression analysis and trend fitting are performed using the same method as the data sequence of the trend of the projection line contour change. Through regression analysis, a trend function describing the change of the spatial position of the cutting edge is obtained, thereby realizing the continuous description of the change characteristics of the spatial position of the saw tooth cutting edge with the cutting process, and fully reflecting the continuous characteristics of the geometric position change during the wear of the saw tooth back face.

[0032] Based on the statistical sequence changes in current in saw blade current wear data, the stability characteristics of saw tooth cutting load over time are analyzed. The current statistical series in the saw blade current wear data includes instantaneous peak current, periodic average current, and periodic current variance, representing the peak load, average load level, and load fluctuation degree of each saw tooth during the cutting process, respectively. Based on the instantaneous peak current, periodic average current, and periodic current variance, time series of instantaneous peak current, periodic average current, and periodic current variance are established. Time series analysis is performed on the instantaneous peak current time series, periodic average current time series, and periodic current variance time series, for example, using the sliding window statistical method to analyze the mean and variance changes of each series. A sliding window of fixed length is defined, and the length of the sliding window is set to several complete cutting cycles in the processing process, for example, 10 consecutive saw tooth passes through the cycle. Each time, the sliding window is advanced by one cycle, and the range of the sliding window is analyzed. The data of the instantaneous peak current time series, the periodic average current time series, and the periodic current variance time series were statistically calculated to obtain the mean and standard deviation of the instantaneous peak current, periodic average current, and periodic current variance within the sliding window. The changing trends of the mean and standard deviation over time were analyzed, and the difference between the mean and standard deviation after each window movement was calculated. If the difference did not show a continuous increase or drastic change within the processing cycle, it indicated that the sawtooth cutting load had stable characteristics within the processing cycle. If a continuous increase or large fluctuation occurred, it indicated that the load stability was reduced, and there might be abnormal sawtooth wear or tooth loss.

[0033] By correlating the geometric change characteristics reflected by the visual wear data of the saw blade with the load change characteristics reflected by the current wear data of the saw blade, visual degradation characteristics and current abnormality characteristics are formed. Establish the temporal and spatial correspondence between saw blade visual wear data and current wear data, that is, match data at the same processing time and the same saw tooth position one by one; normalize the fitting parameters of the projection line contour change trend function and the blade spatial position change trend function of the visual wear data, for example, by using the maximum-minimum normalization method or the Z-score standardization method, so that the visual wear feature parameters and the current wear data feature parameters are within the same numerical scale range; normalize the three parameters of the current wear data (instantaneous peak current, periodic average current, and periodic current variance) respectively.

[0034] After parameter normalization, statistical analysis methods, such as correlation analysis or multiple regression analysis, are used to perform a one-to-one correlation and matching analysis between the geometric change parameters of visual wear data and the load change parameters of current wear data. The correlation coefficient calculation method is used to calculate and analyze the relationship between visual wear characteristic parameters and current wear characteristic parameters. For example, when the absolute value of the correlation coefficient exceeds 0.8, it indicates that there is a strong correspondence between visual wear characteristic parameters and current wear characteristic parameters. Principal component analysis or cluster analysis is used to extract fused characteristic parameters that simultaneously reflect visual and current wear characteristics. The fused characteristic parameters are defined as visual degradation characteristics and current anomaly characteristic data.

[0035] By integrating visual degradation features and abnormal current features, and based on the product processing quality defect rate, the fused visual and current feature data is calculated, and a saw blade scrapping trigger threshold is established, including: Based on the establishment of a corresponding mapping relationship between visual degradation features and current anomaly features, and combined with the product processing quality non-conformity rate obtained synchronously during the processing, the correlation interval of quality change under different combinations of visual degradation features and current anomaly features is statistically analyzed. During the actual cutting process of the saw blade, real-time monitoring and recording of product processing quality are carried out simultaneously. The product processing quality monitoring method is as follows: using a quality inspection device for processed products, such as a high-precision laser rangefinder or surface roughness meter, the dimensional accuracy or surface quality parameters of each processed product are continuously measured; according to the actual processing requirements, the quality acceptance standards for dimensional deviation range or surface roughness are set. For example, the allowable deviation range for dimensional accuracy can be set to ±0.05 mm of the nominal size, and the allowable upper limit for surface roughness can be set to 1.6 micrometers; when the monitored dimensional deviation of the processed product exceeds the above allowable range or the surface roughness is higher than the allowable upper limit, the product processing quality is judged to be unqualified.

[0036] Based on the above methods, quality inspection is performed on all products within each processing cycle during the actual processing. The number of products with substandard processing quality and their proportion of the total number of processed products are recorded in each cycle, thus forming a product processing quality substandard rate data sequence. The product processing quality substandard rate data sequence is defined as: taking each complete saw blade cutting cycle as the time unit, the sequence formed by the proportion of substandard products in the total number of processed products within this cycle.

[0037] The product processing quality defect rate data sequence is correlated with the visual degradation feature and current abnormal feature data collected synchronously at the corresponding time. Specifically, the visual degradation feature data and current abnormal feature data are matched one-to-one at the collection time of each processing cycle to form a combined state data set of visual degradation feature and current abnormal feature. Based on the combined state data set, the processing quality defect rate is statistically analyzed. Within each processing cycle, a two-dimensional feature space is constructed using the normalized values ​​of visual degradation features and current anomaly features as coordinate axes. Feature state partitions are defined within this two-dimensional feature space. The determination of feature state partitions can be achieved through methods such as cluster analysis or grid partitioning. Using the K-means cluster analysis algorithm, all two-dimensional data points of visual degradation features and current anomaly features are clustered into several groups. The number of clusters can be determined using the elbow rule. For example, the sum of squared clustering errors under different numbers of clusters is first calculated, and the optimal number of clusters is determined by the number of clusters where the sum of squared errors significantly decreases. Typically, the number of clusters can be set to 3 to 6 to effectively distinguish different feature combination states. Based on the results of the cluster analysis, each cluster region is defined as a feature state partition within the two-dimensional feature space.

[0038] The defect rate values ​​within each processing cycle of the product processing quality defect rate data sequence are mapped to characteristic state partitions to determine the correlation intervals of product processing quality defect rate changes under different combinations of visual degradation characteristics and current anomaly characteristics. The statistical method for determining these correlation intervals is as follows: within each characteristic state partition, the product processing quality defect rate corresponding to all data points belonging to that partition is statistically analyzed. The average and variance of the product processing quality defect rate within each partition are calculated and used as statistical characteristic parameters of the product quality changes corresponding to that characteristic state partition. This establishes a quantitative correlation mapping relationship between visual degradation characteristics, current anomaly characteristics, and product processing quality defect rate.

[0039] The visual degradation feature parameters and current anomaly feature parameters within the quality change correlation interval are jointly collected and threshold boundary is calculated to form visual and current fusion feature data and saw blade scrapping trigger threshold. Within each characteristic state partition, all data points of visual degradation characteristic parameters and current anomaly characteristic parameters are jointly aggregated, including but not limited to calculating the mean and standard deviation of visual degradation characteristic parameters and current anomaly characteristic parameters within each partition, thereby describing the overall characteristic state of each partition; based on the target requirements for product processing quality non-conformity rate, the critical value of processing quality non-conformity rate required for saw blade scrapping or replacement is determined. For example, when the processing quality non-conformity rate remains at a certain high level for several consecutive processing cycles (e.g., the non-conformity rate exceeds 5%), the high level is set as the critical value of processing quality non-conformity rate.

[0040] After determining the critical value of the processing quality defect rate, the scrap threshold boundaries of visual degradation characteristic parameters and current anomaly characteristic parameters are determined based on the critical value of the processing quality defect rate using a reverse mapping analysis method. The method for determining the scrap threshold boundaries is as follows: In the two-dimensional feature space, the characteristic state partition boundary corresponding to the critical value of the processing quality defect rate of the corresponding product is found. By statistically analyzing the characteristic parameters of the data points at the critical value boundaries, the critical threshold ranges corresponding to the visual degradation characteristic parameters and current anomaly characteristic parameters are calculated respectively. After determining the critical threshold ranges corresponding to the visual degradation characteristic parameters and current anomaly characteristic parameters, a more granular analysis of the data within the critical threshold range is performed using linear interpolation or nonlinear regression analysis. Finally, a more accurate scrap trigger threshold for visual degradation characteristics and current anomaly characteristics is determined, thereby obtaining the fused feature data and the saw blade scrap trigger threshold.

[0041] Based on fused feature data and saw blade scrapping trigger thresholds, the system identifies whether the saw blade has reached the scrapping conditions of wear or tooth loss, and generates saw blade scrapping or replacement control signals, including: Based on the visual and current fusion feature data, the visual and current fusion feature data is compared with the saw blade scrapping trigger threshold in real time; When the visual and electrical fusion feature data reaches or exceeds the saw blade scrapping trigger threshold, a saw blade scrapping or replacement control signal is output. When the visual and current fusion feature data does not reach the saw blade scrapping trigger threshold, update the visual and current fusion feature data and repeat the comparison process. Specifically, during the actual processing of the saw blade, visual and current fusion feature data are continuously collected in real time. After each complete processing cycle, the current fusion feature data is compared with the pre-determined saw blade scrapping trigger threshold in real time. At the end of each processing cycle, based on the two-dimensional data points composed of the current visual degradation feature parameters and current abnormal feature parameters, it is determined whether the two-dimensional data points exceed the critical boundary of the scrapping trigger threshold. When the visual and current fusion feature data reaches or exceeds the saw blade scrapping trigger threshold, it is determined that the current saw blade has reached or exceeded the predetermined wear or tooth loss scrapping state.

[0042] If the combined data of current degradation features and abnormal current features exceeds a preset threshold, the data processing device will immediately generate a scrap or replacement control signal, such as an electrical output pulse signal or a digital logic signal. This control signal is sent directly to the saw control device or saw blade replacement actuator to stop the current processing and replace the saw blade in a timely manner.

[0043] If the fusion feature data at the end of the current processing cycle does not exceed the scrap threshold, the current visual degradation feature and current abnormal feature data are retained and updated in the data processing device, and real-time data acquisition continues in the next processing cycle. The fusion feature data is continuously collected and updated, and the real-time comparison process is repeated until the saw blade visual and current fusion feature data reaches or exceeds the scrap trigger threshold.

[0044] 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.

[0045] 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.

[0046] 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.

[0047] 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.

[0048] 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.

[0049] 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.

[0050] 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.

[0051] 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.

[0052] 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 system based on multi-source data fusion, characterized in that, include: Image acquisition module: Acquires image data of the side of the saw blade, extracts the projection line features of the back face of the saw teeth and the geometric features of the cutting edge position, averages the feature parameters, and outputs the initial visual feature data of the saw blade. Visual wear module: Based on the initial visual feature data of the saw blade, it collects side images of the saw blade in real time during the processing, analyzes the changes in the projection line of the side edge and the amount of backlash of the cutting edge position, and determines the visual wear data of the saw blade. Current monitoring module: Synchronously collects spindle current data when the saw blade is processing the workpiece, analyzes the current change trend and variability, and determines the current wear data of the saw blade; Feature Analysis Module: Based on visual wear data and current wear data, analyzes the degradation trend of the saw blade under normal wear conditions and the abnormal trend under abnormal tooth loss conditions, and generates visual degradation features and current abnormal features. Fusion Threshold Module: Integrates visual degradation features and current anomaly features, calculates visual and current fusion feature data based on the product processing quality non-conformity rate, and establishes a saw blade scrapping trigger threshold; Scrapping determination module: Based on the fusion feature data and the saw blade scrapping trigger threshold, it identifies whether the saw blade has reached the scrapping conditions of wear or tooth loss, and generates a saw blade scrapping or replacement control signal.

2. The saw blade detection system based on multi-source data fusion according to claim 1, characterized in that, Image data of the saw blade's side surface is acquired, and the projection line features of the saw tooth's rear face and the geometric features of the cutting edge position are extracted. The feature parameters are averaged, and the initial visual feature data of the saw blade is output, specifically: A high-speed camera is placed on the side of the saw blade to continuously capture images of the side of the saw teeth when the saw blade is idling, and obtain side image data including the back face and the cutting edge area of ​​the saw teeth. Extract the projection line contour corresponding to the serrated back face and the spatial position parameters of the cutting edge in the back face direction based on the side image data. The projection line contour parameters and cutting edge spatial position parameters corresponding to multiple saw teeth are statistically collected and averaged to form the initial visual feature data of the saw blade.

3. The saw blade detection system based on multi-source data fusion according to claim 2, characterized in that, Based on the initial visual feature data of the saw blade, side images of the saw blade are acquired in real time during the processing. The changes in the projection lines of the side edges and the amount of retraction of the cutting edge position are analyzed to determine the visual wear data of the saw blade, specifically: During the saw blade's workpiece cutting process, the saw blade's side image data is continuously collected according to the saw tooth spatial position and imaging area corresponding to the initial visual feature data of the saw blade; Based on the side image data of the saw blade, the projection line contour of the saw tooth back face and the spatial position parameters of the cutting edge in the back face direction are extracted under the cutting state. The visual wear data of the saw blade is obtained by comparing and calculating the projection line contour parameters and cutting edge spatial position parameters obtained under the cutting state with the initial visual feature data of the saw blade.

4. The saw blade detection system based on multi-source data fusion according to claim 3, characterized in that, Simultaneously collect spindle current data during saw blade machining, analyze current variation trends and variability, and determine saw blade current wear data, specifically: The spindle current signal during the saw blade's participation in workpiece cutting is monitored synchronously. The saw tooth passing frequency is calculated based on the spindle speed. Periodic segmentation and phase calibration are performed on the continuous current sampling sequence. The instantaneous peak value, average value and variance in each segmentation period are calculated and a current statistical sequence is constructed in time order. Based on the monotonically increasing trend and abnormal abrupt change amplitude of the statistical sequence, characteristic parameters of current change are extracted. Combined with the corresponding spatial position of the saw teeth in the visual wear data of the saw blade, saw blade current wear data is generated.

5. A saw blade detection system based on multi-source data fusion according to claim 4, characterized in that, Based on visual wear data and current wear data, the degradation trend of the saw blade under normal wear conditions and the abnormal trend under abnormal tooth loss conditions are analyzed, generating visual degradation features and current abnormal features, specifically: Based on the changes in the projection line contour and the spatial position of the cutting edge in the visual wear data of the saw blade, the continuity characteristics of the wear evolution of the back face of the saw tooth are analyzed. Based on the statistical sequence changes in current in saw blade current wear data, the stability characteristics of saw tooth cutting load over time are analyzed. By correlating the geometric changes reflected in the visual wear data of the saw blade with the load changes reflected in the current wear data of the saw blade, visual degradation characteristics and current anomaly characteristics are formed.

6. A saw blade detection system based on multi-source data fusion according to claim 5, characterized in that, By integrating visual degradation features and current anomaly features, and based on the product processing quality defect rate, the fused visual and current feature data is calculated, and a saw blade scrapping trigger threshold is established. Specifically: Based on the establishment of a corresponding mapping relationship between visual degradation features and current anomaly features, and combined with the product processing quality non-conformity rate obtained synchronously during the processing, the correlation interval of quality change under different combinations of visual degradation features and current anomaly features is statistically analyzed. Visual degradation feature parameters and current anomaly feature parameters within the quality change correlation interval are jointly collected and threshold boundary is calculated to form visual and current fusion feature data and saw blade scrapping trigger threshold.

7. A saw blade detection system based on multi-source data fusion according to claim 6, characterized in that, Based on fused feature data and saw blade scrapping trigger thresholds, the system identifies whether the saw blade has reached the scrapping conditions of wear or tooth loss, and generates a saw blade scrapping or replacement control signal, specifically: Based on the visual and current fusion feature data, the visual and current fusion feature data is compared with the saw blade scrapping trigger threshold in real time; When the visual and electrical fusion feature data reaches or exceeds the saw blade scrapping trigger threshold, a saw blade scrapping or replacement control signal is output. When the visual and current fusion feature data does not reach the saw blade scrapping trigger threshold, update the visual and current fusion feature data and repeat the comparison process.

Images