An automatic balance detection method for resin grinding wheels

Abstract

The present invention belongs to the technical field of balance detection. The present invention discloses an automatic balance detection method for resin grinding wheels, including: for the resin grinding wheel before use, obtaining radial symmetry data and thickness uniformity data, and then calculating static balance index data by using the membership weighted average method; setting a static balance threshold, comparing the static balance index data with the static balance threshold to determine whether the resin grinding wheel before use is balanced; for the balanced resin grinding wheel, collecting and preprocessing dynamic monitoring data, environmental data and resin hardness data; based on the preprocessed dynamic monitoring data, using the AdaBoost model to predict dynamic balance index data; setting an adaptive dynamic balance threshold, comparing the dynamic balance index data with the adaptive dynamic balance threshold to determine whether the resin grinding wheel after use is balanced; significantly improving the accuracy and efficiency of resin grinding wheel balance detection, prolonging the service life, and reducing the risk of equipment damage caused by imbalance.

Description

Technical Field

[0001] The present invention relates to the technical field of balance detection, and specifically to an automatic balance detection method for resin grinding wheels. Background Art

[0002] Existing resin grinding wheel balance detection methods have many problems. Existing balance detection only performs balance detection on resin grinding wheels before use or only on resin grinding wheels after use, without considering that resin grinding wheels balanced before installation may become unbalanced after installation and use;

[0003] Among them, for balance detection of resin grinding wheels before use, the resin grinding wheel is suspended at a support point, and its inclination degree is observed to judge static balance. It is impossible to effectively evaluate the small imbalance of the grinding wheel, and non-contact technology is not used for static balance judgment;

[0004] Among them, for balance detection of resin grinding wheels after use, first, the balanced resin grinding wheel is directly used without considering whether the resin grinding wheel after use is balanced;

[0005] Secondly, the influence of the use environment of the resin grinding wheel is not considered, such as temperature, humidity, electromagnetic environment, etc. These factors will affect the physical properties of the grinding wheel and thus affect its balance. Therefore, failure to comprehensively consider environmental factors may lead to inaccurate detection results or inability to adapt to changes in different working environments;

[0006] In addition, the setting of the dynamic balance threshold is usually fixed, making it difficult to adapt to resin grinding wheels of different types and working conditions, and other factors are not used to design the dynamic balance threshold.

[0007] In view of this, the present invention proposes an automatic balance detection method for resin grinding wheels to solve the above problems. Summary of the Invention

[0008] In order to overcome the above-mentioned defects of the prior art and to achieve the above object, the present invention provides the following technical solution. An automatic balance detection method for resin grinding wheels includes:

[0009] For resin grinding wheels before use, 3D laser scanning technology is used to obtain radial symmetry data and thickness uniformity data; based on the radial symmetry data and thickness uniformity data, the membership weighted average method is used to calculate the static balance index data;

[0010] A static balance threshold is set, and the static balance index data is compared with the static balance threshold to judge whether the resin grinding wheel before use is balanced;

[0011] For the use of a balanced resin grinding wheel, dynamic monitoring data, environmental data, and resin hardness data are collected and preprocessed to obtain preprocessed dynamic monitoring data, preprocessed environmental data, and preprocessed resin hardness data;

[0012] Based on the preprocessed dynamic monitoring data, the AdaBoost model is used to predict the dynamic balance index data;

[0013] The preprocessed environmental data and preprocessed resin hardness data are used to set an adaptive dynamic balance threshold, and the dynamic balance index data is compared with the adaptive dynamic balance threshold to determine whether the resin grinding wheel after use is balanced.

[0014] Furthermore, for the resin grinding wheel before use, the specific methods for obtaining radial symmetry data and thickness uniformity data using 3D laser scanning technology include:

[0015] Use 3D laser scanning technology to obtain the point cloud data of the resin grinding wheel, use MATLAB for point cloud visualization and radial edge extraction, and for each radial edge point cloud data, calculate the centroid of the resin grinding wheel, , , where, and are respectively the abscissa and ordinate of the th radial edge point cloud data, is the total number of radial edge point cloud data, is the index of the radial edge point cloud data, and are respectively the abscissa and ordinate of the centroid, and calculate the radial distance from each edge point cloud data on the radial direction to the centroid of the resin grinding wheel. The formula is: , is the radial distance of the th edge point cloud data;

[0016] Calculate the radial symmetry data through the radial distance. The formula is: , where, is the radial symmetry data, is the average radial distance;

[0017] The resin grinding wheel is divided into sections in the vertical coordinate direction. For each section, calculate the difference between the maximum height value and the minimum height value according to the vertical coordinates of the point cloud data to obtain the thickness of the section. The formula is: , and are the maximum height value and the minimum height value of the th section, is the vertical coordinate value of the th section, is the cross-section index, is the thickness of the

[0018] Thickness uniformity data is calculated through the thickness of the cross-section: , where is the thickness uniformity data, is the average value of the cross-section thickness.

[0019] Furthermore, the specific method for calculating the static balance index data using the membership weighted average method based on the radial symmetry data and the thickness uniformity data includes:

[0020] The static balance index data is calculated by the membership weighted average method, and the formula is: , where is the static balance index data, and are the weight coefficients of the radial symmetry data and the thickness uniformity data respectively, and the weight coefficients are obtained by calculating through the triangular membership function.

[0021] Furthermore, the method for obtaining the weight coefficients by calculating through the triangular membership function includes:

[0022] For the radial symmetry data and the thickness uniformity data, the maximum-minimum normalization method is used to define the data within the range of 0 to 1, the triangular membership function is used to calculate the membership values of the radial symmetry data and the thickness uniformity data, and the weight coefficients of the radial symmetry data and the thickness uniformity data are calculated by the membership value weighting method;

[0023] The formula of the triangular membership function is: , where is the data value, is the lowest value of the maximum-minimum normalized radial symmetry data or thickness uniformity data, is the highest value of the maximum-minimum normalized radial symmetry data or thickness uniformity data, is the median value between the lowest value and the highest value, is the data value of the membership value;

[0024] The membership values of the radial symmetry data and the thickness uniformity data are calculated through the triangular membership function, and the weight coefficients of the radial symmetry data and the thickness uniformity data are calculated according to the membership values. The formula is: , , where is the membership value of the radial symmetry data, is the membership value of the thickness uniformity data.

[0025] Further, the specific method for setting the static balance threshold and comparing the static balance index data with the static balance threshold to determine whether the resin grinding wheel before use is balanced includes:

[0026] Based on the historical static balance index data, calculate the average value and standard deviation, and set the static balance threshold; when the static balance index data is greater than or equal to the static balance threshold, it indicates that the resin grinding wheel is unbalanced;

[0027] When the static balance index data is less than the static balance threshold, it indicates that the resin grinding wheel is balanced;

[0028] Install the balanced resin grinding wheel and detect whether it is balanced during use; for the unbalanced resin grinding wheel, remove the material in the unbalanced area or add weights to the unbalanced area, then recalculate the static balance index data and compare it with the static balance threshold to determine whether it is balanced.

[0029] Further, the specific method for performing preprocessing to obtain the preprocessed dynamic monitoring data, preprocessed environmental data, and preprocessed resin hardness data includes:

[0030] The dynamic monitoring data includes vibration acceleration data, vibration frequency data, rotational speed data, axial acceleration data, radial acceleration data, torque data, resin grinding wheel temperature data, displacement data, and impact signal data;

[0031] The environmental data includes environmental temperature data, humidity data, air pressure data, and electromagnetic environment data;

[0032] Perform missing value, outlier, unified timestamp, and normalization processing on the dynamic monitoring data, environmental data, and resin hardness data to obtain the preprocessed dynamic monitoring data, preprocessed environmental data, and preprocessed resin hardness data.

[0033] Further, the acquisition method of the rotational speed data includes:

[0034] Use the laser Doppler effect to obtain the surface speed of the resin grinding wheel during movement. The formula is: , where is the frequency change, is the laser wavelength compensated by the current temperature data, is the frequency of the laser source, is the surface speed of the resin grinding wheel; the laser Doppler velocimeter is equipped with a frequency analyzer or a digital signal processor to monitor and calculate the frequency change of the reflected light in real time;

[0035] Calculate the rotational speed data of the resin grinding wheel using the surface speed of the resin grinding wheel. The formula is: , where is the outer radius of the resin grinding wheel, obtained from the resin grinding wheel instruction manual, is the rotational speed data of the resin grinding wheel;

[0036] Among them, the formula for the laser wavelength compensated by the current temperature data is: , where, is the laser wavelength at the reference temperature, is the temperature coefficient, is the current ambient temperature, is the reference temperature;

[0037] Use the current temperature data compensation method to compensate the laser wavelength at the reference temperature to obtain the laser wavelength compensated by the current ambient temperature. The change in temperature will affect the laser wavelength, thereby affecting the surface speed of the resin grinding wheel. Therefore, in order to improve the measurement accuracy, compensate for the change in ambient temperature.

[0038] Furthermore, the acquisition method of the resin hardness data includes:

[0039] For the resin grinding wheel in use, install an ultrasonic transmitter and an ultrasonic receiver at the dg of the resin grinding wheel, and calculate the propagation time of the ultrasonic wave. The formula is: , where, is the propagation speed of the ultrasonic wave, SJ is the propagation time of the ultrasonic wave, represents the distance between the positions where the ultrasonic transmitter and the ultrasonic receiver are installed and the resin grinding wheel in use. Among them, the ultrasonic transmitter and the ultrasonic receiver are installed on the same side of the resin grinding wheel, and the distances from the installation positions to the resin grinding wheel are the same;

[0040] Calculate the ultrasonic attenuation coefficient. The formula is: , where, is the ultrasonic attenuation coefficient, is the emission signal intensity, is the receiver signal intensity;

[0041] Calculate the resin hardness data through a linear regression model. The formula is: , where, , and are the regression coefficients of the ultrasonic propagation time, the ultrasonic attenuation coefficient, and the propagation speed of the ultrasonic wave respectively. The regression coefficients are obtained by the least squares method, is the error term.

[0042] Furthermore, the specific method of using the AdaBoost model to predict the dynamic balance index data based on the preprocessed dynamic monitoring data includes:

[0043] Step A1: The input sample set includes G_M groups of samples. Each group of samples includes a set of preprocessed dynamic monitoring data and corresponding dynamic balance index data. Among them, the preprocessed dynamic monitoring data is the feature variable, and the dynamic balance index data is the target variable;

[0044] Step A2: Set the number of iterations and initialize the weights ;

[0045] Step A3: In the first round of iteration, use the initialized weights to train the weak learner. The formula of the weak learner is: , where is the bias term of the weak learner, is the initial dynamic balance index data predicted by the weak learner for the th group of samples in the first round of iteration, represents the th group of samples of the preprocessed dynamic monitoring data, is the sample index;

[0046] Step A4: Calculate the weighted error of the weak learner. The formula is: , where represents the weighted error in the first round of iteration, represents the th group of samples of the dynamic balance index data;

[0047] Step A5: Use the weighted error of the weak learner to calculate the weight of the weak learner. The formula is: , where represents the weight of the weak learner in the first round of iteration;

[0048] Based on the weighted error of the weak learner, update the weight of each group of samples. The formula is: , represents the updated weight of the th group of samples in the first round;

[0049] Step A6: Repeat Step A3 to Step A5 until the set number of iterations is reached and stop;

[0050] Step A7: Perform weighted combination on the initial dynamic balance index data predicted by the weak learner to construct a strong learner. The formula is: , is the strong learner, represents the predicted dynamic balance index data, represents the index of the number of iterations, is the number of iterations, represents the th weight of the weak learner in the represents the th iteration of the weak learner for the Initial dynamic balance index data predicted by the group samples.

[0051] Furthermore, the specific method of using the preprocessed environmental data and preprocessed resin hardness data to set the adaptive dynamic balance threshold, comparing the dynamic balance index data with the adaptive dynamic balance threshold, and determining whether the used resin grinding wheel is balanced includes:

[0052] Based on the historical dynamic balance index data, calculate the average value and standard deviation, set the preliminary dynamic balance threshold, and use the preprocessed environmental data and preprocessed resin hardness data to correct the preliminary dynamic balance threshold to obtain the adaptive dynamic balance threshold;

[0053] The formula for the preliminary dynamic balance threshold is: , where is the average value of the historical dynamic balance index data, is the standard deviation of the historical dynamic balance index data, is the adjustment coefficient, , is the preliminary dynamic balance threshold;

[0054] The formula for the adaptive dynamic balance threshold is: ; where is the adaptive dynamic balance threshold, is the environmental coefficient, is the preprocessed comprehensive environmental data, is the resin coefficient, is the preprocessed resin hardness data;

[0055] When the dynamic balance index data is less than or equal to the adaptive dynamic balance threshold, it indicates that the used resin grinding wheel is balanced;

[0056] When the dynamic balance index data is greater than the adaptive dynamic balance threshold, it indicates that the used resin grinding wheel is unbalanced;

[0057] Among them, the preprocessed comprehensive environmental data is obtained from the preprocessed environmental data, and the formula is: , where , , and are the preprocessed temperature data, humidity data, air pressure data, and electromagnetic environment data respectively, , , and are the weights of the preprocessed temperature data, humidity data, air pressure data, and electromagnetic environment data respectively, each set to 0.25.

[0058] Technical effects and advantages of an automatic balance detection method for a resin grinding wheel according to the present invention:

[0059] The present invention obtains the radial symmetry data and thickness uniformity data of the resin grinding wheel through advanced 3D laser scanning technology, enabling high-precision static balance detection; the static balance detection calculates the balance index data using the membership weighted average method and sets the static balance threshold based on historical data, ensuring accurate judgment of the balance state of the grinding wheel before use;

[0060] For dynamic balance detection, the AdaBoost model is used to train multi-dimensional data, predict the dynamic balance index, and combine with an adaptive threshold adjustment mechanism to optimize the balance standard according to the real-time environment and material hardness, thus ensuring high-precision judgment of the balance of the grinding wheel after use. The overall technical solution comprehensively covers the balance detection process before and after the use of the grinding wheel, effectively guaranteeing the use performance and reliability of the grinding wheel;

[0061] This method has extremely high detection accuracy and intelligent level. The static balance detection uses 3D point cloud data and the membership function weighting method to deeply analyze the geometric characteristics and thickness uniformity of the grinding wheel, ensuring accurate detection results; the dynamic balance detection combines multi-dimensional dynamic monitoring data, uses a strong learner model to improve the prediction accuracy, and further enhances the adaptability of the method to complex environments and material changes through adaptive threshold dynamic adjustment;

[0062] In addition, this method reduces the need for manual intervention, significantly improves the detection efficiency and consistency through a fully automated detection process, provides an intelligent and reliable balance detection tool for industrial production, and has broad application value and promotion potential. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 It is a schematic diagram of an automatic balance detection method for a resin grinding wheel according to the present invention;

[0064] Figure 2 It is a schematic diagram of an automatic balance detection system for a resin grinding wheel according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0065] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0066] Example 1, please refer to Figure 1 As shown, the automatic balance detection method for a resin grinding wheel described in this embodiment includes:

[0067] For a resin grinding wheel before use, 3D laser scanning technology is used to obtain radial symmetry data and thickness uniformity data; based on the radial symmetry data and thickness uniformity data, the membership weighted average method is used to calculate the static balance index data;

[0068] A static balance threshold is set, and the static balance index data is compared with the static balance threshold to determine whether the resin grinding wheel before use is balanced;

[0069] For a balanced resin grinding wheel in use, dynamic monitoring data, environmental data, and resin hardness data are collected and preprocessed to obtain preprocessed dynamic monitoring data, preprocessed environmental data, and preprocessed resin hardness data;

[0070] Based on the preprocessed dynamic monitoring data, the AdaBoost model is used to predict the dynamic balance index data;

[0071] The preprocessed environmental data and preprocessed resin hardness data are used to set an adaptive dynamic balance threshold, and the dynamic balance index data is compared with the adaptive dynamic balance threshold to determine whether the resin grinding wheel after use is balanced.

[0072] Resin grinding wheels are generally installed on various grinding machines and processing equipment for use. Before use, the static balance detection method is used to detect whether the resin grinding wheel is balanced;

[0073] Regarding using the tilt method to determine whether a resin grinding wheel is balanced, the grinding wheel is suspended on a fixed fulcrum to make it in a free state. If the mass distribution of the grinding wheel is uneven, it will cause tilting; by observing the tilt angle and direction, the balance of the grinding wheel can be judged, and the tilt angle can be recorded through tools such as an angle gauge or a level, and the degree of imbalance can be further analyzed;

[0074] As a static balance test method, although the tilt method is simple to operate and low in cost, its accuracy is low, it is difficult to detect small imbalances, this method relies on manual observation and simple tools, and is easily affected by environmental factors such as temperature, wind speed, etc. and the stability of the equipment. Therefore, its application is limited in occasions with high precision requirements or where detailed data analysis is needed.

[0075] For a resin grinding wheel before use, the specific ways to obtain radial symmetry data and thickness uniformity data using 3D laser scanning technology include:

[0076] Use 3D laser scanning technology to obtain the point cloud data of the resin grinding wheel, use MATLAB for point cloud visualization and radial edge extraction, and for each radial edge point cloud data, calculate the centroid of the resin grinding wheel, , , where, and are respectively the The abscissa and ordinate of the radial edge point cloud data, is the total number of the radial edge point cloud data, is the index of the radial edge point cloud data, and are the abscissa and ordinate of the centroid respectively, and calculate the radial distance from each edge point cloud data on the radial direction to the centroid of the resin grinding wheel. The formula is: , is the th radial distance of the edge point cloud data;

[0077] Calculate the radial symmetry data through the radial distance. The formula is: , where is the radial symmetry data, is the average radial distance;

[0078] In the geometric structure of the resin grinding wheel, the radial direction refers to the direction extending along the radius of the grinding wheel from the center of the grinding wheel; therefore, the radial direction is the direction from the center to the outside, that is, the direction from the center of the grinding wheel to the edge of the grinding wheel;

[0079] Using the standard deviation of the radial distance to represent the radial symmetry is because the standard deviation measures the dispersion degree of the radial distances from each edge point cloud data in the point cloud to the centroid. If the grinding wheel is symmetric, the radial distances of all points will be close to a fixed value, and the standard deviation is small, indicating good radial symmetry; if the distribution is uneven, the differences in the radial distances are large, and the standard deviation increases, indicating poor radial symmetry; therefore, the standard deviation can be defined as the radial symmetry data;

[0080] Divide the resin grinding wheel into sections through the vertical coordinate direction. For each section, calculate the difference between the maximum height value and the minimum height value according to the vertical coordinates of the point cloud data to obtain the thickness of the section. The formula is: , and are the maximum height value and the minimum height value on the th section, is the th vertical coordinate value on the section, is the section index, is the th section thickness;

[0081] Calculate the thickness uniformity data through the thickness of the section: , where is the thickness uniformity data, is the average value of the section thickness;

[0082] The thickness uniformity data of the grinding wheel is defined by calculating the standard deviation of the cross-sectional thickness, because the standard deviation can reflect the degree of dispersion between the cross-sectional thicknesses.

[0083] The radial symmetry data and thickness uniformity data of the resin grinding wheel are obtained using 3D laser scanning technology; laser scanning can provide very high measurement accuracy and can obtain data without contacting the grinding wheel, avoiding damage or change to the shape of the grinding wheel.

[0084] Based on the radial symmetry data and thickness uniformity data, the specific method of calculating the static balance index data using the membership weighted average method includes:

[0085] The static balance index data is calculated by the membership weighted average method, and the formula is: , where is the static balance index data, and are the weight coefficients of the radial symmetry data and thickness uniformity data respectively, and the weight coefficients are obtained by calculating using the triangular membership function;

[0086] The membership weighted average method measures the importance of each data point through the membership function, and the membership function can reflect the uncertainty and ambiguity of the data; in practical applications, the membership weighted average method can more accurately quantify these uncertainties through the membership degree, thus avoiding excessive weight bias when dealing with incomplete or inaccurate data.

[0087] The way of obtaining the weight coefficients by calculating using the triangular membership function includes:

[0088] For the radial symmetry data and thickness uniformity data, the maximum-minimum normalization method is used to define the data within the range of 0 to 1, the triangular membership function is used to calculate the membership degree values of the radial symmetry data and thickness uniformity data, and the weight coefficients of the radial symmetry data and thickness uniformity data are calculated by the membership degree value weighting method;

[0089] The formula of the triangular membership function is: , where is the data value, that is, the specific data value of the radial symmetry data or thickness uniformity data after maximum-minimum normalization; when calculating the membership degree value of the radial symmetry data, the specific data value of the radial symmetry data after maximum-minimum normalization is substituted into the formula of the triangular membership function, and at this time represents the data value of the radial symmetry data after maximum-minimum normalization; when calculating the membership degree value of the thickness uniformity data, the specific data value of the thickness uniformity data after maximum-minimum normalization is substituted into the formula of the triangular membership function, and at this time The data value representing the thickness uniformity data after maximum - minimum normalization;

[0090] is the lowest value of the radially symmetric data or thickness uniformity data after maximum - minimum normalization, is the highest value of the radially symmetric data or thickness uniformity data after maximum - minimum normalization, where the radially symmetric data or thickness uniformity data of HK resin grinding wheels are collected and maximum - minimum normalization is performed, and by comparing the HK radially symmetric data or thickness uniformity data after maximum - minimum normalization, the lowest value and the highest value are obtained; when calculating the membership degree value of the radially symmetric data, and represent the lowest value and the highest value of the radially symmetric data after maximum - minimum normalization; when calculating the membership degree value of the thickness uniformity data, and represent the lowest value and the highest value of the thickness uniformity data after maximum - minimum normalization;

[0091] is the median value of the lowest value and the highest value, is the data value 's membership degree value;

[0092] The membership degree values of the radially symmetric data and the thickness uniformity data are calculated through the triangular membership function, and the weight coefficients of the radially symmetric data and the thickness uniformity data are calculated according to the membership degree values. The formula is: , , where, is the membership degree value of the radially symmetric data, is the membership degree value of the thickness uniformity data;

[0093] The maximum - minimum normalization method is a method of scaling data to a specified minimum and maximum value, usually between 0 and 1;

[0094] Using the triangular membership function to set the weight coefficient can dynamically adjust the weight according to the relative quality and reliability of the data. Considering the fuzziness and uncertainty of the data, after normalizing the data through the maximum - minimum normalization method, the triangular membership function can accurately reflect the membership degree of the data within a specific interval and assign a higher weight to more reliable data.

[0095] Set the static balance threshold, and compare the static balance index data with the static balance threshold. The specific methods for judging whether the resin grinding wheel before use is balanced include:

[0096] Based on historical static balance index data, calculate the mean and standard deviation, and set the static balance threshold; when the static balance index data is greater than or equal to the static balance threshold, it indicates that the resin grinding wheel is unbalanced;

[0097] When the static balance index data is less than the static balance threshold, it indicates that the resin grinding wheel is balanced;

[0098] Install the balanced resin grinding wheel and detect whether it is balanced during use; for the unbalanced resin grinding wheel, remove the material in the unbalanced area or add weights to the unbalanced area, then recalculate the static balance index data and compare it with the static balance threshold to determine whether it is balanced.

[0099] The specific methods for preprocessing to obtain preprocessed dynamic monitoring data, preprocessed environmental data, and preprocessed resin hardness data include:

[0100] The dynamic monitoring data includes vibration acceleration data, vibration frequency data, rotational speed data, axial acceleration data, radial acceleration data, torque data, resin grinding wheel temperature data, displacement data, and impact signal data;

[0101] The environmental data includes environmental temperature data, humidity data, air pressure data, and electromagnetic environment data;

[0102] Perform missing value, outlier, unified timestamp, and normalization processing on the dynamic monitoring data, environmental data, and resin hardness data to obtain preprocessed dynamic monitoring data, preprocessed environmental data, and preprocessed resin hardness data;

[0103] Among them, by installing acceleration sensors on the spindle of the machine tool or other equipment where the resin grinding wheel is installed, the support frame, or the area near the grinding wheel mounting seat, vibration acceleration data, vibration frequency data, axial acceleration data, and radial acceleration data are obtained in real time;

[0104] By installing a torque sensor around the resin grinding wheel shaft, torque data is obtained in real time;

[0105] By installing a temperature sensor around the resin grinding wheel shaft, the resin grinding wheel temperature data is obtained in real time, and this temperature data is inconsistent with the environmental temperature data in the environment;

[0106] By installing an eddy current sensor around the resin grinding wheel shaft, displacement data is obtained in real time;

[0107] By installing impact response sensor data on the spindle of the machine tool or other equipment, the support frame, or the area near the grinding wheel mounting seat, impact signal data is obtained;

[0108] By installing temperature sensors, humidity sensors, air pressure sensors, and electromagnetic field sensors in the environment where the resin grinding wheel works, environmental temperature data, humidity data, air pressure data, and electromagnetic environment data are obtained in real time.

[0109] The methods for obtaining rotational speed data include:

[0110] Obtaining rotational speed data during the movement of the resin grinding wheel usually relies on contact sensors, such as methods like encoders or tachometer wheels. Although these methods can provide rotational speed data, they have certain defects. For example, contact sensors may be affected by friction, wear, or installation errors, resulting in unstable or inaccurate measurement results; in addition, contact methods may cause damage to the surface of the grinding wheel, especially in high-speed operation or high-temperature environments, affecting the reliability of measurement and the durability for long-term use; therefore, these traditional methods may have limitations in terms of accuracy, reliability, and applicability;

[0111] Using the laser Doppler effect to obtain the surface speed of the resin grinding wheel during movement, the formula is: , where is the frequency change, is the laser wavelength compensated by the current temperature data, is the frequency of the laser source, is the surface speed of the resin grinding wheel; the laser Doppler velocimeter is equipped with a frequency analyzer or a digital signal processor to monitor and calculate the frequency change of the reflected light and the frequency of the laser source in real time;

[0112] Calculating the rotational speed data of the resin grinding wheel using the surface speed of the resin grinding wheel, the formula is: , where is the outer radius of the resin grinding wheel, obtained from the resin grinding wheel instruction manual, is the rotational speed data of the resin grinding wheel;

[0113] Among them, the formula for the laser wavelength compensated by the current temperature data is: , where is the laser wavelength at the reference temperature, is the temperature coefficient, obtained by referring to reference materials, is the current ambient temperature, is the reference temperature;

[0114] The advantage of using the laser Doppler effect to obtain the surface speed of the resin grinding wheel lies in its non-contact, high-precision, and real-time nature. The laser Doppler velocimetry technology can accurately calculate the surface speed of the grinding wheel through the frequency change of the reflected light without contacting the surface of the resin grinding wheel, avoiding problems such as friction, wear, or damage that may be introduced by traditional contact measurement methods, and calculating the rotational speed data of the resin grinding wheel during movement through the surface speed;

[0115] The laser wavelength changes with temperature and is affected by the ambient temperature, which may lead to deviations in measurement results. By introducing a temperature data compensation mechanism to adjust the laser wavelength according to the current ambient temperature, the error caused by temperature fluctuations can be eliminated, ensuring the accuracy of the speed measurement data. This is particularly significant in working environments with large temperature variations.

[0116] The methods for obtaining resin hardness data include:

[0117] Generally, a hardness tester is used to obtain the resin hardness data in a static state. When the hardness tester is installed on the resin grinding wheel, since it is a contact measurement device, it will be affected by high-speed rotation, vibration, and surface temperature changes of the grinding wheel, resulting in unstable or inaccurate measurement results. High-speed rotation may cause uneven contact between the tester and the grinding wheel surface, generating errors. In addition, frictional heat may be generated on the grinding wheel surface during movement, changing the hardness of the resin material and further interfering with the test results.

[0118] For the resin grinding wheel in use, an ultrasonic transmitter and an ultrasonic receiver are installed at the dg position of the resin grinding wheel, and the propagation time of the ultrasonic wave is calculated. The formula is: , where is the propagation speed of the ultrasonic wave. The average value of the emission speed obtained by the ultrasonic transmitter and the reception speed obtained by the ultrasonic receiver is defined as the propagation speed of the ultrasonic wave. SJ is the propagation time of the ultrasonic wave, represents the distance between the positions where the ultrasonic transmitter and the ultrasonic receiver are installed and the resin grinding wheel in use. Among them, the ultrasonic transmitter and the ultrasonic receiver are installed on the same side of the resin grinding wheel, and the distances from the installation positions to the resin grinding wheel are the same;

[0119] The ultrasonic attenuation coefficient is calculated. The formula is: , where is the ultrasonic attenuation coefficient, is the emission signal intensity, is the receiver signal intensity. The emission signal intensity and the receiver signal intensity are directly obtained on the ultrasonic transmitter and the ultrasonic receiver;

[0120] The resin hardness data is predicted through a linear regression model. The formula is: , where is the predicted resin hardness data, , and are the regression coefficients of the ultrasonic propagation time, the ultrasonic attenuation coefficient, and the propagation speed of the ultrasonic wave respectively. The regression coefficients are calculated by the least squares method, is the error term;

[0121] By minimizing the objective function of the sum of squared errors between the actual resin hardness data and the model-predicted resin hardness data, the optimal solution of the regression coefficients can be obtained by solving the partial derivatives of this objective function and setting them to zero. The optimal solution of the regression coefficients is the regression coefficients of the ultrasonic propagation time, ultrasonic attenuation coefficient, and ultrasonic propagation speed.

[0122] The advantage of using an ultrasonic transmitter and receiver to obtain hardness data during the movement of a resin grinding wheel is that it can achieve non-contact, real-time, and dynamic hardness monitoring. Ultrasonic technology can effectively reflect the hardness changes of resin materials by measuring the propagation time and attenuation coefficient, without being disturbed by factors such as movement, friction, or temperature fluctuations. This method can perform continuous monitoring under the condition of the high-speed movement of the resin grinding wheel.

[0123] It avoids the instability caused by friction, temperature, and surface changes of the hardness tester, providing higher-precision and more reliable dynamic hardness data.

[0124] By associating ultrasonic data with hardness values through a linear regression model, the accuracy and real-time performance of hardness evaluation can be further improved, which is applicable to real-time quality control in high-efficiency production environments.

[0125] Based on the preprocessed dynamic monitoring data, the specific ways to use the AdaBoost model to predict the dynamic balance index data include:

[0126] Step A1: The input sample set includes G_M groups of samples. Each group of samples includes a set of preprocessed dynamic monitoring data and the corresponding dynamic balance index data, where the preprocessed dynamic monitoring data is the feature variable and the dynamic balance index data is the target variable.

[0127] Step A2: Set the number of iterations and initialize the weights ;

[0128] Step A3: In the first round of iteration, use the initialized weights to train the weak learner. The formula of the weak learner is: , where, is the bias term of the weak learner, is the initial dynamic balance index data predicted by the weak learner for the th group of samples in the first round of iteration, represents the preprocessed dynamic monitoring data of the th group of samples, is the sample index;

[0129] Step A4: Calculate the weighted error of the weak learner. The formula is: , where, represents the weighted error in the first round of iteration, represents the Dynamic balance index data in the group of samples;

[0130] Step A5: Calculate the weight of the weak learner using the weighted error of the weak learner. The formula is: , where represents the weight of the weak learner in the first round of iteration;

[0131] Update the weight of each group of samples based on the weighted error of the weak learner. The formula is: , represents the updated weight of the th group of samples in the first round;

[0132] Step A6: Repeat Step A3 to Step A5 until the set number of iterations is reached and stop;

[0133] Step A7: Construct a strong learner by weighted combination of the initial dynamic balance index data predicted by the weak learner. The formula is: , is the strong learner, represents the predicted dynamic balance index data, represents the index of the number of iterations, is the number of iterations, represents the th weight of the weak learner in the th iteration, represents the initial dynamic balance index data predicted by the weak learner for the th group of samples in the

[0134] The advantage of using the AdaBoost model to predict dynamic balance index data is that it can significantly improve the prediction accuracy and robustness of the model by integrating multiple weak learners; AdaBoost effectively reduces errors by continuously adjusting the weights of the data and focusing on samples that are difficult to predict; in addition, AdaBoost has strong adaptability and automatic feature selection ability, and can dynamically adjust the model according to changes in the data, making it perform well in real-time monitoring and dynamic balance analysis and providing more accurate and stable prediction results.

[0135] Use the preprocessed environmental data and preprocessed resin hardness data to set the adaptive dynamic balance threshold, and compare the dynamic balance index data with the adaptive dynamic balance threshold. The specific methods for judging whether the used resin grinding wheel is balanced include:

[0136] Based on the historical dynamic balance index data, calculate the mean and standard deviation, set the preliminary dynamic balance threshold, and use the preprocessed environmental data and preprocessed resin hardness data to correct the preliminary dynamic balance threshold to obtain the adaptive dynamic balance threshold;

[0137] The formula for the preliminary dynamic balance threshold is: , where is the average value of the historical dynamic balance index data, is the standard deviation of the historical dynamic balance index data, is the adjustment coefficient, , is the preliminary dynamic balance threshold;

[0138] The formula for the adaptive dynamic balance threshold is: ; where is the adaptive dynamic balance threshold, is the preprocessed comprehensive environment data, and are the environment coefficient and the resin coefficient respectively, set by the rule of thumb, is the preprocessed resin hardness data;

[0139] When the dynamic balance index data is less than or equal to the adaptive dynamic balance threshold, it indicates that the resin grinding wheel after use is balanced;

[0140] When the dynamic balance index data is greater than the adaptive dynamic balance threshold, it indicates that the resin grinding wheel after use is unbalanced;

[0141] For the unbalanced resin grinding wheel, remove the material in the unbalanced area or add weights to the unbalanced area, then recalculate the dynamic balance index data and compare it with the adaptive dynamic balance threshold to determine whether it is balanced;

[0142] Among them, the preprocessed comprehensive environment data is obtained from the preprocessed environment data, and the formula is: , where , , and are the preprocessed temperature data, humidity data, air pressure data and electromagnetic environment data respectively, , , and are the weights of the preprocessed temperature data, humidity data, air pressure data and electromagnetic environment data respectively, each set to 0.25;

[0143] The advantage of designing the adaptive dynamic balance threshold is that it can dynamically adjust the dynamic balance standard according to the real-time environmental changes and the fluctuations of the resin hardness, thus improving the accuracy and adaptability of the balance detection; by using the average value and standard deviation of the historical data as the preliminary dynamic balance threshold and introducing the environmental data and resin hardness data as correction factors, the influence of environmental factors and material changes on the dynamic balance detection results can be effectively eliminated, ensuring that the dynamic balance state of the resin grinding wheel can still be accurately judged under different working conditions; this adaptive adjustment mechanism makes the dynamic balance threshold more flexible and reliable, which helps to improve the real-time monitoring and quality control in the production process.

[0144] In this embodiment, the radial symmetry data and thickness uniformity data of the resin grinding wheel are obtained through advanced 3D laser scanning technology, enabling high-precision static balance detection; the static balance detection uses the membership weighted average method to calculate the balance index data, and sets the static balance threshold based on historical data, ensuring accurate judgment of the balance state of the grinding wheel before use;

[0145] For dynamic balance detection, the AdaBoost model is used to train multi-dimensional data, predict the dynamic balance index, and combine with an adaptive threshold adjustment mechanism to optimize the balance standard according to the real-time environment and material hardness, thereby ensuring high-precision judgment of the balance of the grinding wheel after use. The overall technical solution comprehensively covers the balance detection process before and after the use of the grinding wheel, and can effectively guarantee the use performance and reliability of the grinding wheel;

[0146] This method has extremely high detection accuracy and intelligent level. The static balance detection uses 3D point cloud data and the membership function weighting method to deeply analyze the geometric characteristics and thickness uniformity of the grinding wheel, ensuring accurate detection results; the dynamic balance detection combines multi-dimensional dynamic monitoring data, uses a strong learner model to improve the prediction accuracy, and further enhances the adaptability of the method to complex environments and material changes through adaptive threshold dynamic adjustment;

[0147] In addition, this method reduces the need for manual intervention, significantly improves the detection efficiency and consistency through a fully automated detection process, provides an intelligent and reliable balance detection tool for industrial production, and has broad application value and promotion potential.

[0148] Example 2, please refer to Figure 2 As shown, for the parts not described in detail in this embodiment, refer to the description content of Embodiment 1. A resin grinding wheel automatic balance detection system is provided, including:

[0149] Static balance detection module: For the resin grinding wheel before use, use 3D laser scanning technology to obtain radial symmetry data and thickness uniformity data; based on the radial symmetry data and thickness uniformity data, use the membership weighted average method to calculate the static balance index data;

[0150] Static balance judgment module: Set the static balance threshold, compare the static balance index data with the static balance threshold, and judge whether the resin grinding wheel before use is balanced;

[0151] Dynamic balance data processing module: For the balanced resin grinding wheel, collect dynamic monitoring data, environmental data, and resin hardness data, and perform preprocessing to obtain preprocessed dynamic monitoring data, preprocessed environmental data, and preprocessed resin hardness data;

[0152] Dynamic balance detection module: Based on the preprocessed dynamic monitoring data, use the AdaBoost model to predict the dynamic balance index data;

[0153] Dynamic balance judgment module: Use the preprocessed environmental data and resin hardness data to set an adaptive dynamic balance threshold, compare the dynamic balance index data with the adaptive dynamic balance threshold, and judge whether the used resin grinding wheel is balanced.

[0154] Embodiment 3. This embodiment publicly provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it implements the operation mode of the above-provided method for automatically detecting the balance of a resin grinding wheel.

[0155] Since the electronic device introduced in this embodiment is the electronic device used to implement the method for automatically detecting the balance of a resin grinding wheel in the embodiments of the present application, based on the method for automatically detecting the balance of a resin grinding wheel introduced in the embodiments of the present application, those skilled in the art can understand the specific implementation manners and various variations of the electronic device in this embodiment. Therefore, the specific implementation of how this electronic device implements the method in the embodiments of the present application will not be described in detail here. As long as those skilled in the art implement the electronic device used in the method for automatically detecting the balance of a resin grinding wheel in the embodiments of the present application, it falls within the scope of protection of the present application.

[0156] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to obtain a formula that is closest to the actual situation. The preset parameters and threshold selection in the formulas are set by those skilled in the art according to the actual situation.

[0157] The above are only the preferred embodiments of the present invention. The protection scope of the present invention is not limited to the above embodiments. All technical solutions within the idea of the present invention belong to the protection scope of the present invention. It should be noted that for ordinary technical users in the technical field, several improvements and refinements made without departing from the principle of the present invention should also be regarded as within the protection scope of the present invention.

Claims

1. A resin grinding wheel automatic balancing detection method, characterized in that: include: Step S1: using 3D laser scanning technology to obtain radial symmetry data and thickness uniformity data of the resin grinding wheel before use; Based on the radial symmetry data and thickness uniformity data, the static balance index data is calculated using the membership weighted average method; Step S2, based on the historical static balance index data, the average value and the standard deviation are calculated, the static balance threshold is set, and the static balance index data is compared with the static balance threshold to determine whether the resin grinding wheel is balanced before use; Step S3, using the balanced resin grinding wheel, collecting dynamic monitoring data, environmental data and resin hardness data, and preprocessing to obtain preprocessed dynamic monitoring data, preprocessed environmental data and preprocessed resin hardness data; Step S4: Based on the pre-processed dynamic monitoring data, use the AdaBoost model to predict the dynamic balance index data; Step S5, based on the historical dynamic balancing index data, calculate the average value and standard deviation, set the preliminary dynamic balancing threshold, use the pretreatment environment data and the pretreatment resin hardness data to correct the preliminary dynamic balancing threshold, obtain the adaptive dynamic balancing threshold, compare the dynamic balancing index data with the adaptive dynamic balancing threshold, and determine whether the resin grinding wheel is balanced after use.

2. The automatic balancing detection method for a resin grinding wheel according to claim 1, characterized in that: The specific method of using 3D laser scanning technology to obtain radial symmetry data and thickness uniformity data of the resin grinding wheel before use includes: Use 3D laser scanning technology to obtain the point cloud data of the resin grinding wheel, use MATLAB to visualize the point cloud and extract the radial edge, and calculate the center of mass of the resin grinding wheel for each radial edge point cloud data. , ,in, and Respectively The horizontal and vertical coordinates of the radial edge point cloud data, is the total number of radial edge point cloud data, is the index of radial edge point cloud data, and are the horizontal and vertical coordinates of the center of mass respectively, and the radial distance from each edge point cloud data in the radial direction to the center of mass of the resin grinding wheel is calculated. The formula is: , For the The radial distance of edge point cloud data; The radial symmetry data is calculated by radial distance, the formula is: ,in, is the radial symmetry data, is the average radial distance; The resin grinding wheel is divided into For each section, the difference between the maximum height value and the minimum height value is calculated according to the vertical coordinate of the point cloud data to obtain the thickness of the section. The formula is: , and For the The maximum and minimum height values ​​on each section, For the The vertical coordinate value on the cross section, is the cross-section index, For the The thickness of each section; The thickness uniformity data is obtained by calculating the thickness of the cross section: ,in, is the thickness uniformity data, is the average value of the cross-section thickness.

3. The automatic balancing detection method for a resin grinding wheel according to claim 2, characterized in that: The specific method of calculating the static balance index data based on the radial symmetry data and the thickness uniformity data using the membership weighted average method includes: The static balance index data is calculated by the membership weighted average method, and the formula is: ,in, is the static balance index data, and are the weight coefficients of radial symmetry data and thickness uniformity data respectively, and the weight coefficients are obtained by calculating the triangle membership function.

4. The automatic balancing detection method for a resin grinding wheel according to claim 3, characterized in that: The weight coefficient is obtained by calculating the triangle membership function in the following manner: For radial symmetry data and thickness uniformity data, the maximum-minimum normalization method is used to define the data in the range of 0 to 1, the triangle membership function is used to calculate the membership values ​​of radial symmetry data and thickness uniformity data, and the weight coefficients of radial symmetry data and thickness uniformity data are calculated by the membership value weighting method; The formula for the triangle membership function is: ,in, is the data value, is the lowest value of the radial symmetry data or thickness uniformity data after maximum-minimum normalization. It is the highest value of the radial symmetry data or thickness uniformity data after maximum-minimum normalization. is the median value between the lowest and highest values, For data value The membership value of The membership values ​​of radial symmetry data and thickness uniformity data are calculated by the triangle membership function, and the weight coefficients of radial symmetry data and thickness uniformity data are calculated according to the membership values. The formula is: , ,in, is the membership value of radial symmetry data, is the membership value of thickness uniformity data.

5. The automatic balancing detection method for a resin grinding wheel according to claim 4, characterized in that: The specific method of determining whether the resin grinding wheel is balanced before use by comparing the static balance index data with the static balance threshold based on the historical static balance index data and calculating the average value and standard deviation includes: When the static balance index data is greater than or equal to the static balance threshold, it means that the resin grinding wheel is unbalanced; When the static balance index data is less than the static balance threshold, it means that the resin grinding wheel is balanced.

6. The automatic balancing detection method for a resin grinding wheel according to claim 5, characterized in that: The specific method of performing preprocessing to obtain preprocessing dynamic monitoring data, preprocessing environment data and preprocessing resin hardness data includes: Dynamic monitoring data includes vibration acceleration data, vibration frequency data, rotation speed data, axial acceleration data, radial acceleration data, torque data, resin grinding wheel temperature data, displacement data and impact signal data; Environmental data includes ambient temperature data, humidity data, air pressure data, and electromagnetic environment data; The dynamic monitoring data, environmental data and resin hardness data are processed with missing values, abnormal values, unified timestamps and normalization to obtain preprocessed dynamic monitoring data, preprocessed environmental data and preprocessed resin hardness data.

7. The automatic balancing detection method for a resin grinding wheel according to claim 6, characterized in that: The method for obtaining the speed data includes: The surface velocity of the resin grinding wheel is obtained by using the laser Doppler effect. The formula is: ,in, is the frequency change, is the laser wavelength after current temperature data compensation, is the frequency of the laser source, is the surface speed of the resin grinding wheel; The surface speed of the resin grinding wheel is used to calculate the rotation speed data of the resin grinding wheel. The formula is: ,in, is the outer radius of the resin grinding wheel, is the rotation speed data of the resin grinding wheel; Among them, the formula for the laser wavelength after current temperature data compensation is: ,in, is the laser wavelength at the reference temperature, is the temperature coefficient, is the current ambient temperature, is the base temperature.

8. The automatic balancing detection method for a resin grinding wheel according to claim 7, characterized in that: The resin hardness data is obtained by: For the resin grinding wheel in use, the propagation time of ultrasonic waves is calculated using the formula: ,in, is the propagation speed of ultrasound, is the ultrasonic wave propagation time, Indicates the distance between the installation position of the ultrasonic transmitter and the ultrasonic receiver and the resin grinding wheel in use; Calculate the ultrasonic attenuation coefficient using the formula: ,in, is the ultrasonic attenuation coefficient, is the transmission signal strength, is the receiver signal strength; The resin hardness data is calculated by a linear regression model, and the formula is: ,in, , and are the regression coefficients of ultrasonic propagation time, ultrasonic attenuation coefficient and ultrasonic propagation velocity, respectively. is the error term.

9. The automatic balancing detection method for a resin grinding wheel according to claim 8, characterized in that: The specific method of using the AdaBoost model to predict dynamic balancing index data based on pre-processed dynamic monitoring data includes: Step A1, the input sample set includes G_M groups of samples, each group of samples includes a group of pre-processed dynamic monitoring data and corresponding dynamic balancing index data, wherein the pre-processed dynamic monitoring data is a characteristic variable, and the dynamic balancing index data is a target variable; Step A2: Set the number of iterations and initialize the weights ; Step A3: In the first iteration, the initial weights are used to train the weak learner. The weak learner formula is: ,in, is the bias term of the weak learner, is the weak learner pair in the first iteration The initial dynamic balance indicator data of the group sample prediction, Indicates Preprocessing dynamic monitoring data of group samples, is the sample index; Step A4: Calculate the weighted error of the weak learner. The formula is: ,in, represents the weighted error in the first iteration, Indicates Dynamic balance index data in group samples; Step A5: Calculate the weight of the weak learner using the weighted error of the weak learner. The formula is: ,in, represents the weight of the weak learner in the first iteration; Based on the weighted error of the weak learner, the weight of each group of samples is updated. The formula is: , Indicates the first round Update weights of group samples; Step A6, repeating steps A3 to A5 until the set number of iterations is reached; Step A7: Perform weighted combination on the initial dynamic balance index data predicted by the weak learner to construct a strong learner. The formula is: , is a strong learner, representing the predicted dynamic balance index data, The index representing the number of iterations, is the number of iterations, Indicates The weights of the weak learners in the iteration, Indicates In the iteration, the weak learner The initial dynamic balance indicator data for group sample prediction.

10. The automatic balancing detection method for a resin grinding wheel according to claim 9, characterized in that: The specific method of calculating the average value and standard deviation based on the historical dynamic balance index data, setting the preliminary dynamic balance threshold, using the pre-processing environment data and the pre-processing resin hardness data to correct the preliminary dynamic balance threshold, obtaining the adaptive dynamic balance threshold, comparing the dynamic balance index data with the adaptive dynamic balance threshold, and judging whether the resin grinding wheel after use is balanced includes: The formula for the initial dynamic balance threshold is: ,in, is the average value of the historical dynamic balance index data, is the standard deviation of the historical dynamic balance indicator data, is the adjustment coefficient, , is the initial dynamic balance threshold; The formula for adaptive dynamic balancing threshold is: ;in, is the adaptive dynamic balancing threshold, is the environmental factor, To preprocess the synthetic environment data, is the resin coefficient, To pre-process the resin hardness data; When the dynamic balance index data is less than or equal to the adaptive dynamic balance threshold, it means that the resin grinding wheel is balanced after use; When the dynamic balance index data is greater than the adaptive dynamic balance threshold, it means that the resin grinding wheel is unbalanced after use.

Images