A tool wear state monitoring method
By constructing a tool wear condition monitoring model and utilizing the ratio of characteristic values of multi-directional vibration signals, the problem of low detection accuracy caused by changes in cutting conditions in existing technologies has been solved, achieving higher accuracy in tool wear identification and production efficiency.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- HARBIN UNIV OF SCI & TECH
- Filing Date
- 2024-06-24
- Publication Date
- 2026-07-14
AI Technical Summary
Existing tool wear monitoring technologies suffer from low detection accuracy due to the increase in characteristic values under different cutting conditions.
A tool wear condition monitoring model is constructed. By collecting and dimensionality-reducing tool cutting data, the ratio of feature values of multi-directional vibration signals is extracted, interference from entry and exit signals is eliminated, and the recognition accuracy is improved.
It improves the accuracy of tool wear condition identification, reduces the error rate of judgment, is applicable to actual production and processing, and reduces costs.
Smart Images
Figure CN118559502B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of tool monitoring, and in particular relates to a method for monitoring tool wear condition. Background Technology
[0002] With the continuous improvement of intelligent manufacturing, tool wear monitoring is crucial in precision workpiece machining. Tool wear monitoring improves machining accuracy, reduces production costs, and significantly increases production efficiency. However, existing tool monitoring methods only consider features in a single direction as detection indicators. When the cutting amount increases, the signals in all three directions will increase, and judgment is made solely based on the feature value in one direction.
[0003] Therefore, existing technologies suffer from low detection accuracy due to the increase in characteristic values caused by different cutting conditions. Summary of the Invention
[0004] The purpose of this invention is to address the problem of low detection accuracy in existing technologies due to the increase in characteristic values caused by different cutting conditions. A method for monitoring tool wear condition includes:
[0005] I. Constructing a tool wear condition monitoring model;
[0006] 2. Collect the cutting data of the tool to be monitored, input it into the tool wear condition monitoring model, and obtain the tool wear condition to be monitored;
[0007] The first step involves constructing a tool wear condition monitoring model; the specific process is as follows:
[0008] S1: Obtain the tool wear dataset;
[0009] S2: Construct a tool wear condition monitoring model;
[0010] The specific process for obtaining the tool wear dataset in S1 is as follows:
[0011] N sets of raw data are collected, and dimensionality reduction extraction is performed on the N sets of raw data to obtain N sets of dimensionality-reduced data; these are used as the tool wear dataset, where N is a positive integer.
[0012] The specific process for obtaining the tool wear dataset in S1 is as follows:
[0013] N sets of raw data are collected, and dimensionality reduction extraction is performed on these N sets of raw data to obtain N sets of dimensionality-reduced data; these serve as the tool wear dataset, where N is a positive integer, and N is 5 in the test of this invention.
[0014] The raw data includes raw data of early tool wear signals, raw data of mid-tool wear signals, and raw data of late tool wear signals.
[0015] The beneficial effects of this invention are as follows:
[0016] First, by selecting feature values from different directions as ratios, this invention ensures that even if the directional signal increases due to other processing factors, their ratios will not change significantly, thereby improving the accuracy of tool wear condition identification.
[0017] Second, this method processes the data and determines a stable entry interval, which can exclude entry and exit signals from entering the model for judgment, prevent the influence of noisy entry and exit signals on the judgment, and improve the accuracy of tool wear condition identification. Attached Figure Description
[0018] Figure 1 This is a flowchart of the tool wear condition monitoring method with normalized monitoring data according to the present invention;
[0019] Figure 2 This is a schematic diagram of a line graph after dimensionality reduction of the original data in this invention;
[0020] Figure 3 This is a schematic diagram of the envelope scatter plot fitting curve of the present invention;
[0021] Figure 4 This is a schematic diagram of the first derivative of the envelope scatter plot fitting curve function of the present invention;
[0022] Figure 5 This is a schematic diagram of the root mean square eigenvalues of the vibration signal in the X direction according to the present invention;
[0023] Figure 6 This is a schematic diagram of the root mean square eigenvalue of the vibration signal in the Y direction according to the present invention.
[0024] Figure 7 This is a schematic diagram of the root mean square eigenvalue of the vibration signal in the Z direction according to the present invention;
[0025] Figure 8 This is a schematic diagram of the ratio characteristic value X / Y of the present invention;
[0026] Figure 9 This is a schematic diagram of the ratio characteristic value X / Z of the present invention;
[0027] Figure 10 This is a schematic diagram of the ratio characteristic value Y / Z of the present invention.
[0028] Figure 11 This is a schematic diagram of the fitted curve after adding the three ratios of the present invention and averaging them;
[0029] Figure 12 This is a schematic diagram of the first-order derivative of the mean-fit curve of the present invention.
[0030] Figure 13This is a schematic diagram of the XYZ coordinate system in which the cutting tool of this invention is located. Detailed Implementation
[0031] Specific implementation method one: Combining Figure 1 This invention describes the following:
[0032] S1: Obtain the tool wear dataset;
[0033] S2: Construct a tool wear condition monitoring model based on the tool wear dataset;
[0034] S3: Collect cutting data of the tool to be monitored;
[0035] S4: Input the collected cutting data of the tool to be monitored into the tool wear condition monitoring model to obtain the tool wear condition to be monitored.
[0036] The specific process for obtaining the tool wear dataset in S1 is as follows:
[0037] Collect N sets of raw data, and perform dimensionality reduction and extraction on the N sets of raw data to obtain N sets of dimensionality-reduced data; these will serve as the tool wear dataset, where N is a positive integer.
[0038] The specific process for collecting N sets of raw data is as follows:
[0039] Vibration signals from each cutting process throughout the tool's entire lifecycle are collected using an accelerometer to obtain the raw data for group A; 1≤A≤N;
[0040] Each cutting process is the process from when the tool enters the cutting area to when the tool exits the cutting area.
[0041] The dimensionality reduction and extraction of N sets of original data yields N sets of dimensionality-reduced data; the specific process is as follows:
[0042] S1.1: Divide the raw data of group A into X-direction signal, Y-direction signal and Z-direction signal;
[0043] The X-direction signal is the vibration signal of the tool in the X direction during each cutting process;
[0044] The Y-direction signal is the vibration signal in the Y direction of the tool during each cutting process;
[0045] The Z-direction signal is the vibration signal in the Z-direction during each cutting process of the tool;
[0046] The X direction is the radial depth of cut of the tool, the Y direction is the tool feed direction, and the Z direction is the axial depth of cut of the tool. The X, Y, and Z directions form a spatial rectangular coordinate system.
[0047] S1.2: Perform dimensionality reduction extraction on the X-direction signal to obtain the dimensionality-reduced X-direction signal;
[0048] The Y-direction signal is reduced in dimension to obtain the Y-direction signal after dimension reduction.
[0049] The Z-direction signal is reduced in dimension to obtain the dimension-reduced Z-direction signal.
[0050] S1.3: The X-direction signal, Y-direction signal, and Z-direction signal obtained from S1.2 after dimensionality reduction are used as the data after dimensionality reduction in group A;
[0051] In step S2, a tool wear condition monitoring model is constructed based on the tool wear dataset; the specific process is as follows:
[0052] S2.1: Extract features from the tool wear dataset to obtain tool wear feature values for each smooth cutting process.
[0053] S2.2: Based on the tool wear characteristic values obtained in S2.1 for each smooth cutting process, obtain the tool wear comparison characteristic values for each smooth cutting process.
[0054] S2.3: Sum the tool wear comparison feature values obtained in S2.2 for each smooth cutting process and calculate the average value to obtain the average comparison feature value for each smooth cutting process. Perform polynomial fitting on the obtained average comparison feature value for each smooth cutting process to obtain the tool wear detection curve.
[0055] S2.4: Divide the tool wear detection curve into the early stage of tool wear, the middle stage of tool wear, and the late stage of tool wear to obtain the tool wear state monitoring model.
[0056] Specific Implementation Method Two: The difference between this implementation method and Specific Implementation Method One is that...
[0057] The specific process of dimensionality reduction extraction in S1.2 is as follows:
[0058] S1.2.1: Divide the signal into F groups according to the acquisition time. Each group contains n data points, where F and n are both positive integers; n = f / 10, where f is the frequency at which the accelerometer collects data during each cutting process of the tool.
[0059] S1.2.2: Extract the maximum value from each set of data in S1.2.1, ultimately obtaining F extracted data points. These F extracted data points are used as the signal after dimensionality reduction extraction. Each cutting process refers to the process from tool entry to tool exit. The vibration signal during each cutting process throughout the tool's entire lifespan is collected using an accelerometer from the early stage of tool wear to the late stage of tool wear, yielding a set of raw data. The raw data includes raw data for the early stage of tool wear, the middle stage of tool wear, and the late stage of tool wear. Compared to other tool wear monitoring algorithms, this invention is more in line with practical applications. The monitoring algorithm is easier to write, has a lower error rate, does not require costly training of models with a large amount of experimental data, is more consistent with actual production and processing, and has a higher tool wear identification rate.
[0060] The other steps and parameters are the same as in Specific Implementation Method 1.
[0061] Specific Implementation Method Three: The difference between this implementation method and Specific Implementation Method One is that...
[0062] In step S2.1, feature extraction is performed on the tool wear dataset to obtain tool wear feature values during each smooth cutting process. The specific process is as follows:
[0063] S2.1.1: Obtain the best fitting function based on N sets of data from the tool wear dataset;
[0064] S2.1.2: Obtain the stable cutting interval during each cutting process based on the best fitting function;
[0065] S2.1.3: Calculate the tool wear characteristic value during each smooth cutting process based on the smooth cutting interval during each cutting process; the smooth cutting process is the cutting process that has not been affected by tool vibration or other reasons; other steps and parameters are the same as in one of the specific implementation methods one to two.
[0066] Specific Implementation Method Four: This implementation method differs from Specific Implementation Methods One through Four in that...
[0067] The best-fit functions in S2.1.1 include: the best-fit function in the X direction, the best-fit function in the Y direction, and the best-fit function in the Z direction;
[0068] The process of obtaining the best-fit curve based on N sets of data from the tool wear dataset is as follows:
[0069] S2.1.1.1: Draw N X-direction tool life cycle images based on the dimension-reduced X-direction signal data extracted from N sets of data in the tool wear dataset;
[0070] N Y-direction tool lifecycle images are drawn based on the Y-direction signal data extracted from N sets of data in the tool wear dataset after dimensionality reduction.
[0071] N Z-direction tool lifecycle images are drawn based on the dimension-reduced Z-direction signal data extracted from N sets of data in the tool wear dataset.
[0072] S2.1.1.2: Find N sets of envelope scatter points in the X direction based on N tool lifecycle images in the X direction;
[0073] Find N sets of envelope scatter points in the Y direction based on N tool life cycle images in the Y direction;
[0074] Find N sets of envelope scatter points in the Z direction based on N tool life cycle images in the Z direction;
[0075] S2.1.1.3: Perform polynomial fitting based on N sets of X-direction envelope scatter points to obtain the best-fit function in the X-direction;
[0076] The best-fit function in the Y direction is obtained by performing polynomial fitting based on N sets of Y-direction envelope scatter points.
[0077] The best-fitting function in the Z-direction is obtained by performing polynomial fitting based on N sets of Z-direction envelope scatter points.
[0078] The stable cutting intervals in each cutting process in S2.1.2 include: the stable cutting interval in the X direction, the stable cutting interval in the Y direction, and the stable cutting interval in the Z direction;
[0079] The specific process for obtaining the stable cutting range based on the best fitting function in S2.1.2 is as follows:
[0080] S2.1.2.1: Determine the zeros of the best-fit function in the X direction and obtain the zeros in all X directions;
[0081] Determine the zeros of the best-fit function in the Y direction to obtain all zeros in the Y direction;
[0082] Determine the zeros of the best-fit function in the Z direction to obtain the zeros in all Z directions;
[0083] S2.1.2.2: Determine the smooth entry point A in the X direction based on the zero points in all X directions. X Smooth tangent point B in the X direction X ;
[0084] Determine the smooth entry point A in the Y direction based on all zero points in the Y direction. Y Smooth tangent point B in the Y direction Y ;
[0085] Determine the smooth entry point A in the Z direction based on all the zero points in the Z direction.Z Smooth tangent point B in the Z direction Z ;
[0086] S2.1.2.3: Based on the stable entry point A in the X direction during each cutting process. X Smooth tangent point B in the X direction X Determine the stable cutting range in the X direction during each cutting process [A] X B X ]; where the stable cutting interval in the X direction during the i-th cutting process is [A Xi B Xi ];
[0087] Based on the stable entry point A in the Y direction during each cutting process Y Smooth tangent point B in the Y direction Y Determine the stable cutting zone in the Y direction during each cutting process [A] Y B Y ]; where the stable cutting interval in the X direction during the i-th cutting process is [A Yi B Yi ];
[0088] Based on the stable entry point A in the Z direction during each cutting process Z Smooth tangent point B in the Z direction Z Determine the smooth cutting range in the Z direction during each cutting process [A] Z B Z ]; where the stable cutting interval in the X direction during the i-th cutting process is [A Zi B Zi ];
[0089] The tool wear characteristic values in S2.1.3 during each smooth cutting process include: the characteristic value in the X direction, the characteristic value in the Y direction, and the characteristic value in the Z direction during each smooth cutting process.
[0090] The tool wear characteristic values during the i-th cutting process include: the X-direction characteristic value, the Y-direction characteristic value, and the Z-direction characteristic value during the i-th cutting process; other steps and parameters are the same as in one of the specific implementation methods one to three.
[0091] Specific Implementation Method Five: The difference between this implementation method and Specific Implementation Methods One to Four is that...
[0092] In step S2.1.1.3, a polynomial fitting is performed based on N sets of X-direction envelope scatter points to obtain the best-fit function in the X-direction; the specific process is as follows:
[0093] f X (x X ) = aX x X m +b X x X m-1 +……+c X x X +d X
[0094] In the formula: f X () denotes the best-fit function in the X direction, a², b², c², d² are the coefficients of the best-fit function in the X direction, m is the highest power of the best-fit function in the X direction, and x... X Indicates the time point for data acquisition in the X direction;
[0095] The best-fit function in the Y-direction is obtained by performing polynomial fitting based on N sets of Y-direction envelope scatter points; the specific process is as follows:
[0096] f Y (x Y ) = a Y x Y j +b Y x Y j-1 +……+c Y x Y +d Y
[0097] In the formula: f Y () represents the best-fit function in the Y direction, a Y b Y c Y d Y Here, x represents the coefficients of the best-fit function in the Y direction, j represents the highest power of the best-fit function in the Y direction, and x represents the coefficients of the best-fit function in the Y direction. Y Indicates the time point for data acquisition in the Y direction;
[0098] The process involves performing polynomial fitting based on N sets of Z-direction envelope scatter points to obtain the best-fit function in the Z-direction; the specific process is as follows:
[0099] f Z (x Z ) = a Z x Z q +b Z x Z q-1 +……+c Z x Z +d Z
[0100] In the formula: f Z () represents the best-fit function in the Z direction, a Z bZ c Z d Z Let x be the coefficient of the best-fit function in the Z direction, q be the highest power of the best-fit function in the Z direction, and x be the coefficient of the best-fit function in the Z direction. Z Indicates the time point of data acquisition in the Z direction.
[0101] The polynomial fitting process in this invention is as follows:
[0102] The data to be fitted is input into the polynomial fitting module in MATLAB, the highest fitting power is selected, and the fitted curve is output. The preferred fitting power in this invention is 8.
[0103] The advantages of taking the ratio first and then averaging for better fit:
[0104] 1. Reduce the impact of noise: Calculating the average can smooth the data and reduce the impact of noise, which helps the fitting process, especially when there are large random variations in the input data.
[0105] 2. Reduce data dimensionality: By comparing each pair of the three data points and then averaging them, the information from the three data points is effectively combined into a new composite feature. This helps reduce the dimensionality of the data and simplifies the model during the fitting process.
[0106] S2.4: Divide the tool wear detection curve into the early stage of tool wear, the middle stage of tool wear, and the late stage of tool wear to obtain a trained tool wear state monitoring model. Other steps and parameters are the same as in one of the specific implementation methods one to four.
[0107] Specific Implementation Method Six: The difference between this implementation method and Specific Implementation Methods One to Five is that...
[0108] The specific process for determining the zeros of the best-fit function in S2.1.2.1 is as follows:
[0109] The point in the best-fit function where the slope of the curve is zero and the slope of the curve changes before and after is taken as the zero point;
[0110] In S2.1.2.2, the smooth entry point and smooth exit point in the X direction are determined based on all zero points in the X direction. The specific process is as follows: starting from the first zero point in the X direction, if the slope of the point before the zero point is negative and the slope of the point after the zero point is positive, then the zero point is determined to be a smooth entry point in the X direction; if the slope of the point before the zero point is positive and the slope of the point after the zero point is negative, then the zero point is determined to be a smooth exit point in the X direction.
[0111] Finally, the stable entry point A in the X direction of the tool during each cutting process was determined. X Smooth tangent point B in the X direction X ;
[0112] The i-th stable cutting point in the X direction is taken as the stable cutting point A in the X direction during the i-th cutting process. Xi ;
[0113] The i-th stable cutting point in the X direction is taken as the stable cutting point in the X direction during the i-th cutting process, and is called B. Xi ;
[0114] In S2.1.2.2, the smooth entry point and smooth exit point in the Y direction are determined based on all zero points in the Y direction. The specific process is as follows: starting from the first zero point in the Y direction, if the slope of the point before the zero point is negative and the slope of the point after the zero point is positive, then the zero point is determined to be a smooth entry point in the Y direction; if the slope of the point before the zero point is positive and the slope of the point after the zero point is negative, then the zero point is determined to be a smooth exit point in the Y direction.
[0115] Finally, the stable entry point A in the Y direction of the tool during each cutting process was determined. Y Smooth tangent point B in the X direction Y ;
[0116] The i-th stable entry point is taken as the stable entry point A in the Y direction during the i-th cutting process. Yi ;
[0117] The i-th stable cut point is taken as the Y-direction stable cut point in the i-th cutting process, and is called B. Yi ;
[0118] In S2.1.2.2, the smooth entry point and smooth exit point in the Z direction are determined based on all the zero points in the Z direction. The specific process is as follows: starting from the first zero point in the Z direction, if the slope of the point before the zero point is negative and the slope of the point after the zero point is positive, then the zero point is determined to be a smooth entry point in the Z direction; if the slope of the point before the zero point is positive and the slope of the point after the zero point is negative, then the zero point is determined to be a smooth exit point in the Z direction.
[0119] Finally, the smooth entry point A in the Z direction of the tool during each cutting process was determined. Z Smooth tangent point B in the Z direction Z ;
[0120] The i-th stable entry point is taken as the Z-direction stable entry point A in the i-th cutting process. Zi ;
[0121] The i-th stable cut point is B, which is the stable cut point in the Z direction during the i-th cutting process. Zi ;
[0122] The characteristic values in the X direction during the i-th cutting process in S2.1.3 include: the mean value of the X direction during the i-th cutting process, the standard deviation of the X direction during the i-th cutting process, the skewness of the X direction during the i-th cutting process, the kurtosis of the X direction during the i-th cutting process, and the root mean square value of the X direction during the i-th cutting process.
[0123] The characteristic values in the Y direction during the i-th cutting process include: the mean value of the Y direction during the i-th cutting process, the standard deviation of the Y direction during the i-th cutting process, the skewness of the Y direction during the i-th cutting process, the kurtosis of the Y direction during the i-th cutting process, and the root mean square value of the Y direction during the i-th cutting process.
[0124] The Z-direction characteristic values during the i-th cutting process include: the mean value of the Z-direction during the i-th cutting process, the standard deviation of the Z-direction during the i-th cutting process, the skewness of the Z-direction during the i-th cutting process, the kurtosis of the Z-direction during the i-th cutting process, and the root mean square value of the Z-direction during the i-th cutting process.
[0125] The tool wear characteristic values used in calculating these characteristics are all well-known in the field. Taking the X direction as an example during the i-th cutting process, their calculation formulas are explained below. The calculation methods for the Y and Z directions are the same as those for the X direction.
[0126] The formula for calculating the average value in the X direction during the i-th cutting process is:
[0127] Assume that the stable cutting range in the X direction during the i-th cutting process is [A]. Xi B Xi There are T data acquisition points. Then, during the i-th cutting process, the average value X in the X direction is... mean The formula is:
[0128]
[0129] In the formula, T represents the stable cutting interval in the X direction during the i-th cutting process, which is [A]. Xi B Xi Number of time points collected in [x] X f represents the time point for data acquisition in the X direction. X () represents the best-fit function in the X direction.
[0130] The formula for calculating the standard deviation is:
[0131]
[0132] The formula for calculating skewness is:
[0133]
[0134] The formula for calculating kurtosis is:
[0135]
[0136] The formula for calculating the root mean square is:
[0137]
[0138] The other steps and parameters are the same as those in any of the specific implementation methods one to five.
[0139] Specific Implementation Method Seven: The difference between this implementation method and Specific Implementation Methods One through Six is that...
[0140] The tool wear comparison feature values in S2.2 for each smooth cutting process include: X / Y comparison feature values for each smooth cutting process, Y / Z comparison feature values for each smooth cutting process, and X / Z comparison feature values for each smooth cutting process.
[0141] Based on the tool wear characteristic values obtained in S2.2 for each cutting process, the tool wear comparison characteristic values for each cutting process are obtained. The specific process is as follows:
[0142] The specific process for obtaining the tool wear comparison feature value during the i-th cutting process based on the tool wear feature value during the i-th cutting process is as follows:
[0143] S2.2.1: Pair the X-direction feature value and the Y-direction feature value during the i-th cutting process to obtain the X / Y comparison feature value during the i-th cutting process;
[0144] S2.2.2: Pair the Y-direction feature value and the Z-direction feature value during the i-th cutting process to obtain the Y / Z comparison feature value during the i-th cutting process;
[0145] S2.2.3: Pair the X-direction feature value and the Z-direction feature value during the i-th cutting process to obtain the X / Z comparison feature value during the i-th cutting process; other steps and parameters are the same as in one of the specific implementation methods one to six.
[0146] Specific Implementation Method Eight: The difference between this implementation method and Specific Implementation Methods One to Seven is that...
[0147] The formula for the tool wear detection curve in S2.3 is as follows:
[0148] f1(x1)=a1x1 k +b1x1 k-1 +……+c1x1+d1
[0149] In the formula, f1() represents the tool wear detection function, a1, b1, c1, d1 represent the parameters of the tool wear detection function, k represents the highest power of the tool wear detection function, and x1 represents the number of cutting operations.
[0150] The other steps and parameters are the same as those in any of the specific implementation methods one to seven.
[0151] Specific Implementation Method Nine: The difference between this implementation method and Specific Implementation Methods One through Eight is that...
[0152] In step S2.4, the tool wear detection curve is divided into the early stage of tool wear, the middle stage of tool wear, and the late stage of tool wear to obtain the tool wear state monitoring model. The specific process is as follows:
[0153] S2.4.1: Calculate the slope of the tool wear detection curve and determine the zero point position and number of the tool wear detection curve;
[0154] S2.4.2: The first dividing point is obtained based on the slope and the zero point position of the tool wear detection curve;
[0155] S2.4.3: Take the zero point after the first dividing point as the second dividing point;
[0156] S2.4.4: The tool wear detection curve is divided into early tool wear, middle tool wear, and late tool wear according to the first and second division points; other steps and parameters are the same as those in one of the specific implementation methods one to eight.
[0157] Specific Implementation Method Ten: The difference between this implementation method and Specific Implementation Methods One through Nine is that...
[0158] In S2.4.4, the tool wear detection curve is divided into early tool wear, middle tool wear, and late tool wear based on the first and second dividing points. The specific process is as follows:
[0159] S2.4.4.1: The first dividing point and the area before the first dividing point are considered the early stage of tool wear;
[0160] S2.4.4.2: The tool wear is in the middle stage after the first dividing point, the second dividing point, and before the second dividing point;
[0161] S2.4.4.3: The period after the second dividing point is the later stage of tool wear;
[0162] In S2.4.2, the first dividing point is the zero point position of the first tool wear detection curve that meets the dividing conditions;
[0163] The segmentation condition is expressed by the formula:
[0164]
[0165] In the formula: h represents the slope of the tool wear detection curve after the zero point, and q represents the slope of the tool wear detection curve before the zero point. The specific process of collecting the tool cutting data to be monitored in the second step is as follows:
[0166] 21. Collect raw cutting data of the tool to be monitored.
[0167] 22. Dimensionality reduction extraction of the raw cutting data of the tool to be monitored is performed to obtain the tool wear extraction data to be monitored as the tool cutting data to be monitored. 21. The raw cutting data of the tool to be monitored is collected, and the specific process is as follows:
[0168] Vibration signals from the actual cutting process of the tool under test are collected using accelerometers to obtain raw data on tool wear. The specific number of samples collected depends on the tool's mass and the workpiece material.
[0169] The actual cutting process is the process from the entry of the monitored tool to the exit of the monitored tool.
[0170] The vibration signals of the tool to be monitored during each cutting process include: the vibration signal in the X direction; the vibration signal in the Y direction; and the vibration signal in the Z direction (XYZ are three directions in a spatial coordinate system, namely x, y, and z).
[0171] The second step involves dimensionality reduction extraction of the raw cutting data of the tool to be monitored to obtain the tool wear extraction data. The specific process is as follows:
[0172] 221: Divide the obtained raw data of tool wear to be monitored into G groups, each group containing n data points, where G and n are both positive integers; n = f / 10
[0173] In the formula, f is the frequency at which the accelerometer collects data from the tool during each cutting process;
[0174] 222: Extract the maximum value from each set of data, and use each extracted maximum value as the extracted data for that set of data, ultimately obtaining G extracted data;
[0175] 223: Use the G extracted data obtained from 222 as the tool wear extraction data to be monitored;
[0176] The other steps and parameters are the same as those in any of the specific implementation methods one to nine.
[0177] Combined with specific implementation methods one to ten and Figures 2 to 13 Simulation analysis explanation:
[0178] Combination Figure 13 The parameters for simulation analysis of this invention are:
[0179] The overall dimensions of the workpiece are 105mm × 200mm × 150mm.
[0180] The blade model is SDMT120512-GM
[0181] The spindle speed is 900 r / min;
[0182] The feed rate is 320 mm / min;
[0183] The cutting depth is 1 mm;
[0184] The cutting width is 45mm (for milling grooves);
[0185] like Figure 2 and Figure 3 To generate a full lifecycle image of a knife by finding its envelope scatter points in the dimensionality-reduced data of S2.1.1.3 and performing polynomial fitting on the envelope scatter points;
[0186] Figure 4 The best curve fitted to N knives in the dataset;
[0187] Figure 5 , 6 Figure 7 shows the root mean square eigenvalues of the X, Y, and Z vibration signals. It is difficult to determine the characteristics of tool wear from the signal eigenvalues in only one direction, and it is difficult to distinguish the state of tool wear. Now, we compare X with Y, X with Z, and Y with Z, and generate images of each ratio. By comparing the data eigenvalues in the X, Y, and Z directions pairwise, the differences between different eigenvalues can be eliminated, which helps to extract the key features of tool vibration and provides a more intuitive comparison and understanding of the milling cutter wear state. The ratio can also eliminate the influence of noise factors. Vibration signals contain many noise factors that affect the judgment of the tool, and the ratio can eliminate their influence.
[0188] Figure 8 , 9 The 10 represents the characteristic images of each smooth cutting process after comparing the characteristic values of the milling cutter vibration signal throughout the tool's entire life cycle (X / Y, X / Z, Y / Z).
[0189] Figure 11 , 12 The tool wear detection curves after summing and averaging, as well as the tool wear detection curves before, during, and after the process, are the trained tool wear state monitoring model.
[0190] The above description is merely of preferred embodiments of the present invention. It should be understood that the present invention is not limited to the specific embodiments described above. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent substitutions, and improvements made to the above embodiments without departing from the scope of the present invention, based on the technical essence of the present invention, and within the spirit and principles of the present invention, shall still fall within the protection scope of the present invention.
Claims
1. A method for monitoring tool wear condition, characterized in that, Includes the following steps: S1: Obtain the tool wear dataset; S2: Construct a tool wear condition monitoring model based on the tool wear dataset; S3: Collect cutting data of the tool to be monitored; S4: Input the collected cutting data of the tool to be monitored into the tool wear condition monitoring model to obtain the tool wear condition to be monitored. The specific process for obtaining the tool wear dataset in S1 is as follows: Collect N sets of raw data, and perform dimensionality reduction and extraction on the N sets of raw data to obtain N sets of dimensionality-reduced data, which are used as the tool wear dataset, where N is a positive integer; The specific process for collecting N sets of raw data is as follows: Vibration signals during each cutting process throughout the tool's entire lifecycle are collected using an accelerometer to obtain the raw data in group A. ; Each cutting process is the process from when the tool enters the cutting area to when the tool exits the cutting area. The dimensionality reduction and extraction of N sets of original data yields N sets of dimensionality-reduced data; the specific process is as follows: S1.1: Divide the raw data of group A into X-direction signal, Y-direction signal and Z-direction signal; The X-direction signal is the vibration signal of the tool in the X direction during each cutting process; The Y-direction signal is the vibration signal in the Y direction of the tool during each cutting process; The Z-direction signal is the vibration signal in the Z-direction during each cutting process of the tool; The X direction is the radial depth of cut of the tool, the Y direction is the tool feed direction, and the Z direction is the axial depth of cut of the tool. The X, Y, and Z directions form a spatial rectangular coordinate system. S1.2: Perform dimensionality reduction extraction on the X-direction signal to obtain the dimensionality-reduced X-direction signal; The Y-direction signal is reduced in dimension to obtain the Y-direction signal after dimension reduction. The Z-direction signal is reduced in dimension to obtain the dimension-reduced Z-direction signal. S1.3: The X-direction signal, Y-direction signal, and Z-direction signal obtained from S1.2 after dimensionality reduction are used as the data after dimensionality reduction in group A; In step S2, a tool wear condition monitoring model is constructed based on the tool wear dataset; the specific process is as follows: S2.1: Extract features from the tool wear dataset to obtain tool wear feature values for each smooth cutting process. S2.2: Based on the tool wear characteristic values obtained in each stable cutting process in S2.1, obtain the tool wear comparison characteristic values in each stable cutting process, including: X / Y comparison characteristic values, Y / Z comparison characteristic values and X / Z comparison characteristic values in each stable cutting process; S2.3: Sum the tool wear comparison feature values obtained in S2.2 for each smooth cutting process and calculate the average value to obtain the average comparison feature value for each smooth cutting process. Perform polynomial fitting on the obtained average comparison feature value for each smooth cutting process to obtain the tool wear detection curve. S2.4: Divide the tool wear detection curve into the early stage of tool wear, the middle stage of tool wear, and the late stage of tool wear to obtain the tool wear state monitoring model.
2. The method for monitoring tool wear condition according to claim 1, characterized in that, The specific process of dimensionality reduction extraction in S1.2 is as follows: S1.2.1: Divide the signal into F groups according to the acquisition time. Each group contains n data points, where F and n are both positive integers. In the formula, f is the frequency at which the accelerometer collects data from the tool during each cutting process; S1.2.2: Extract the maximum value from each group of data in S1.2.1, and finally obtain F extracted data. Use the F extracted data as the signal after dimensionality reduction extraction.
3. The method for monitoring tool wear condition according to claim 2, characterized in that, In step S2.1, feature extraction is performed on the tool wear dataset to obtain tool wear feature values during each smooth cutting process. The specific process is as follows: S2.1.1: Obtain the best fitting function based on N sets of data from the tool wear dataset; S2.1.2: Obtain the stable cutting interval during each cutting process based on the best fitting function; S2.1.3: Calculate the tool wear characteristic value during each smooth cutting process based on the smooth cutting interval during each cutting process.
4. The method for monitoring tool wear condition according to claim 3, characterized in that, The best-fit functions in S2.1.1 include: the best-fit function in the X direction, the best-fit function in the Y direction, and the best-fit function in the Z direction; The process of obtaining the best-fit curve based on N sets of data from the tool wear dataset is as follows: S2.1.1.1: Draw N X-direction tool life cycle images based on the dimension-reduced X-direction signal data extracted from N sets of data in the tool wear dataset; N Y-direction tool lifecycle images are drawn based on the Y-direction signal data extracted from N sets of data in the tool wear dataset after dimensionality reduction. N Z-direction tool lifecycle images are drawn based on the dimension-reduced Z-direction signal data extracted from N sets of data in the tool wear dataset. S2.1.1.2: Find N sets of envelope scatter points in the X direction based on N tool lifecycle images in the X direction; Find N sets of envelope scatter points in the Y direction based on N tool life cycle images in the Y direction; Find N sets of envelope scatter points in the Z direction based on N tool life cycle images in the Z direction; S2.1.1.3: Perform polynomial fitting based on N sets of X-direction envelope scatter points to obtain the best-fit function in the X-direction; The best-fit function in the Y direction is obtained by performing polynomial fitting based on N sets of Y-direction envelope scatter points. The best-fitting function in the Z-direction is obtained by performing polynomial fitting based on N sets of Z-direction envelope scatter points. The stable cutting intervals in each cutting process in S2.1.2 include: the stable cutting interval in the X direction, the stable cutting interval in the Y direction, and the stable cutting interval in the Z direction; The specific process for obtaining the stable cutting range based on the best fitting function in S2.1.2 is as follows: S2.1.2.1: Determine the zeros of the best-fit function in the X direction and obtain the zeros in all X directions; Determine the zeros of the best-fit function in the Y direction to obtain all zeros in the Y direction; Determine the zeros of the best-fit function in the Z direction to obtain the zeros in all Z directions; S2.1.2.2: Determine the smooth entry point in the X direction based on the zero points in all X directions. Smooth tangent point in the X direction ; Determine the smooth entry point in the Y direction based on all zero points in the Y direction. Smooth tangent point in the Y direction ; Determine the smooth entry point in the Z direction based on all zero points in the Z direction. Smooth tangent point in the Z direction ; S2.1.2.3: Based on the stable entry point in the X direction during each cutting process. Smooth tangent point in the X direction Determine the stable cutting range in the X direction during each cutting process. ; where the stable cutting interval in the X direction during the i-th cutting process is ; Based on the stable entry point in the Y direction during each cutting process Smooth tangent point in the Y direction Determine the stable cutting zone in the Y direction during each cutting process. ; where the stable cutting interval in the X direction during the i-th cutting process is ; Based on the stable entry point in the Z direction during each cutting process Smooth tangent point in the Z direction Determine the smooth cutting range in the Z-direction during each cutting process. ; where the stable cutting interval in the X direction during the i-th cutting process is ; The tool wear characteristic values in S2.1.3 during each smooth cutting process include: the characteristic value in the X direction, the characteristic value in the Y direction, and the characteristic value in the Z direction during each smooth cutting process. The tool wear characteristic values during the i-th cutting process include: the X-direction characteristic value, the Y-direction characteristic value, and the Z-direction characteristic value during the i-th cutting process.
5. The method for monitoring tool wear condition according to claim 4, characterized in that, In step S2.1.1.3, a polynomial fitting is performed based on N sets of X-direction envelope scatter points to obtain the best-fit function in the X-direction; the specific process is as follows: In the formula: This represents the best-fit function in the X direction. , , , Let be the coefficients of the best-fit function in the X direction, and m be the highest power of the best-fit function in the X direction. Indicates the time point for data acquisition in the X direction; The best-fit function in the Y-direction is obtained by performing polynomial fitting based on N sets of Y-direction envelope scatter points; the specific process is as follows: In the formula: This represents the best-fit function in the Y direction. , , , Let be the coefficients of the best-fit function in the Y direction, and j be the highest power of the best-fit function in the Y direction. Indicates the time point for data acquisition in the Y direction; The process involves performing polynomial fitting based on N sets of Z-direction envelope scatter points to obtain the best-fit function in the Z-direction; the specific process is as follows: In the formula: This represents the best-fit function in the Z direction. , , , Let be the coefficients of the best-fit function in the Z direction, and q be the highest power of the best-fit function in the Z direction. This indicates the time point for data acquisition in the Z direction.
6. The method for monitoring tool wear condition according to claim 5, characterized in that, The specific process for determining the zeros of the best-fit function in S2.1.2.1 is as follows: The point in the best-fit function where the slope of the curve is zero and the slope of the curve changes before and after is taken as the zero point; In S2.1.2.2, the smooth entry point and smooth exit point in the X direction are determined based on all zero points in the X direction. The specific process is as follows: starting from the first zero point in the X direction, if the slope of the point before the zero point is negative and the slope of the point after the zero point is positive, then the zero point is determined to be a smooth entry point in the X direction; if the slope of the point before the zero point is positive and the slope of the point after the zero point is negative, then the zero point is determined to be a smooth exit point in the X direction. Finally, determine the stable entry point in the X direction for each cutting process. Smooth tangent point in the X direction ; The i-th stable cutting point in the X direction is used as the stable cutting point in the X direction during the i-th cutting process. ; The i-th stable cutting point in the X direction is used as the stable cutting point in the X direction during the i-th cutting process. ; In S2.1.2.2, the smooth entry point and smooth exit point in the Y direction are determined based on all zero points in the Y direction. The specific process is as follows: starting from the first zero point in the Y direction, if the slope of the point before the zero point is negative and the slope of the point after the zero point is positive, then the zero point is determined to be a smooth entry point in the Y direction; if the slope of the point before the zero point is positive and the slope of the point after the zero point is negative, then the zero point is determined to be a smooth exit point in the Y direction. Finally, the stable entry point in the Y direction for each cutting process of the tool was determined. Smooth tangent point in the Y direction ; The i-th stable entry point is used as the stable entry point in the Y direction during the i-th cutting process. ; The i-th stable cut point is taken as the stable cut point in the Y direction during the i-th cutting process. ; In S2.1.2.2, the smooth entry point and smooth exit point in the Z direction are determined based on all the zero points in the Z direction. The specific process is as follows: starting from the first zero point in the Z direction, if the slope of the point before the zero point is negative and the slope of the point after the zero point is positive, then the zero point is determined to be a smooth entry point in the Z direction; if the slope of the point before the zero point is positive and the slope of the point after the zero point is negative, then the zero point is determined to be a smooth exit point in the Z direction. Finally, the stable Z-axis entry point of the tool during each cutting process was determined. Smooth tangent point in the Z direction ; The i-th stable entry point is used as the stable entry point in the Z direction during the i-th cutting process. ; The i-th stable cut point is taken as the stable cut point in the Z direction during the i-th cutting process. ; The characteristic values in the X direction during the i-th cutting process in S2.1.3 include: the mean value of the X direction during the i-th cutting process, the standard deviation of the X direction during the i-th cutting process, the skewness of the X direction during the i-th cutting process, the kurtosis of the X direction during the i-th cutting process, and the root mean square value of the X direction during the i-th cutting process. The characteristic values in the Y direction during the i-th cutting process include: the mean value of the Y direction during the i-th cutting process, the standard deviation of the Y direction during the i-th cutting process, the skewness of the Y direction during the i-th cutting process, the kurtosis of the Y direction during the i-th cutting process, and the root mean square value of the Y direction during the i-th cutting process. The Z-direction characteristic values during the i-th cutting process include: the mean value of the Z-direction during the i-th cutting process, the standard deviation of the Z-direction during the i-th cutting process, the skewness of the Z-direction during the i-th cutting process, the kurtosis of the Z-direction during the i-th cutting process, and the root mean square value of the Z-direction during the i-th cutting process.
7. The method for monitoring tool wear condition according to claim 6, characterized in that, Based on the tool wear characteristic values obtained in S2.2 for each cutting process, the tool wear comparison characteristic values for each cutting process are obtained. The specific process is as follows: The specific process for obtaining the tool wear comparison feature value during the i-th cutting process based on the tool wear feature value during the i-th cutting process is as follows: S2.2.1: Pair the X-direction feature value and the Y-direction feature value during the i-th cutting process to obtain the X / Y comparison feature value during the i-th cutting process; S2.2.2: Pair the Y-direction feature value and the Z-direction feature value during the i-th cutting process to obtain the Y / Z comparison feature value during the i-th cutting process; S2.2.3: Pair the X-direction feature value and the Z-direction feature value during the i-th cutting process to obtain the X / Z comparison feature value during the i-th cutting process.
8. The method for monitoring tool wear condition according to claim 7, characterized in that, The formula for the tool wear detection curve obtained in S2.3 is as follows: In the formula, This represents the tool wear detection function. , , , represents the parameters of the tool wear detection function, and k represents the highest power of the tool wear detection function. Indicates the number of cuts.
9. The method for monitoring tool wear condition according to claim 8, characterized in that, In step S2.4, the tool wear detection curve is divided into the early stage of tool wear, the middle stage of tool wear, and the late stage of tool wear to obtain the tool wear state monitoring model. The specific process is as follows: S2.4.1: Calculate the slope of the tool wear detection curve and determine the zero point position and number of the tool wear detection curve; S2.4.2: The first dividing point is obtained based on the slope and the zero point position of the tool wear detection curve; S2.4.3: Take the zero point after the first dividing point as the second dividing point; S2.4.4: Based on the first and second dividing points, the tool wear detection curve is divided into the early stage of tool wear, the middle stage of tool wear, and the late stage of tool wear.
10. A method for monitoring tool wear condition according to claim 9, characterized in that, In S2.4.4, the tool wear detection curve is divided into early tool wear, middle tool wear, and late tool wear based on the first and second dividing points. The specific process is as follows: S2.4.4.1: The first dividing point and the area before the first dividing point are considered the early stage of tool wear; S2.4.4.2: The tool wear is in the middle stage after the first dividing point, the second dividing point, and before the second dividing point; S2.4.4.3: The period after the second dividing point is the later stage of tool wear; In S2.4.2, the first dividing point is the zero point position of the first tool wear detection curve that meets the dividing conditions; The segmentation condition is expressed by the formula: In the formula: h represents the slope of the tool wear detection curve after the zero point, and q represents the slope of the tool wear detection curve before the zero point.