A Fault Identification Method for the Boring and Milling Process of a Horizontal CNC Boring and Milling Machine

Abstract

The present invention relates to the technical field of boring and milling fault identification, and specifically relates to a method for identifying faults in the boring and milling process of a horizontal CNC boring and milling machine. The method includes: installing a strain-type three-dimensional sensor at the connection between the spindle housing and the tool of the boring and milling machine to obtain vibration force data segments in three directions; analyzing the vibration force data segments to obtain the residual terms of the vibration force data segments; respectively obtaining the outlier data within each vibration force data segment according to the distribution of the residual terms of each vibration force data segment in the residual plot, and further obtaining the outlier data points; performing clustering in a spatial rectangular coordinate system to obtain clusters, and further obtaining the mean points; obtaining the outlier abnormality degree of the outlier data points within a cluster based on the distribution of the outlier data points within a cluster and the connection line between the cluster center and the mean point of the cluster; and judging whether a fault occurs in the boring and milling process according to the outlier abnormality degree of the outlier data points. The present invention can accurately identify clamping looseness or spindle faults in the boring and milling process.

Description

Technical Field

[0001] The present invention relates to the technical field of boring and milling fault identification, and particularly to a method for identifying faults in the boring and milling process of a horizontal CNC boring and milling machine. Background Art

[0002] A horizontal CNC boring and milling machine is a high-precision and high-efficiency CNC machine tool, mainly used for various machining operations such as boring, milling, drilling, and tapping of large or medium-sized parts. It is widely used in industries such as mold manufacturing, machining, heavy equipment, and energy equipment. During the boring and milling process of a horizontal CNC boring and milling machine, due to its complex structure and high machining accuracy requirements, some problems that affect machining quality and equipment safety are likely to occur. Therefore, real-time monitoring is required to ensure the normal progress of the boring and milling process.

[0003] The force generated by vibration is one of the main data monitored by a boring and milling machine. Existing technologies usually use data analysis algorithms for the force generated by the vibration of a boring and milling machine to capture potential abnormal boring and milling. Since periodic tremors will form during the boring and milling chip removal process, the STL time series decomposition algorithm can usually be used to capture non-periodic and non-trending data points as abnormal trend points. However, due to the complex reasons for the vibration of a boring and milling machine, which may be caused by actual faults or the material of the cutting, directly using the STL time series decomposition to obtain the outlier residuals as abnormal data points has a low accuracy. Summary of the Invention

[0004] In order to solve the above technical problems, the purpose of the present invention is to provide a method for identifying faults in the boring and milling process of a horizontal CNC boring and milling machine, and the specific technical solution adopted is as follows:

[0005] An embodiment of the present invention provides a method for identifying faults in the boring and milling process of a horizontal CNC boring and milling machine, and the method includes:

[0006] Install a strain-type three-dimensional sensor at the connection between the spindle housing and the tool of the boring and milling machine, obtain the forces in three directions of a spatial rectangular coordinate system within a preset time period and respectively form vibration force data segments; analyze the vibration force data segments to obtain the residual terms of the vibration force data segments;

[0007] Obtain the outlier data within each vibration force data segment respectively according to the distribution of the residual terms of each vibration force data segment in the residual plot; obtain the outlier data points according to the outlier data within each vibration force data segment;

[0008] Map the outlier data points into the spatial rectangular coordinate system and perform clustering to obtain clusters; obtain the mean points according to the forces corresponding to the moments when the forces in three directions within the vibration force data segments are not outlier data;

[0009] In a three-dimensional Cartesian coordinate system, the degree of outlier anomaly of the outlier data points within a cluster is obtained based on the distribution of the outlier data points within the cluster and the line connecting the cluster center and the mean point of the cluster.

[0010] Based on the degree of outlier anomaly of the outlier data points, it is determined whether the outlier data points are abnormal outlier points; based on the number of abnormal outlier points within a preset time period, it is determined whether a fault occurs during the boring and milling process.

[0011] Preferably, the forces in the three directions of the three-dimensional Cartesian coordinate system are the forces in the three directions of the X-axis, Y-axis, and Z-axis respectively.

[0012] Preferably, analyzing the vibration force data segment to obtain the residual term of the vibration force data segment includes:

[0013] Performing STL time series decomposition on the vibration force data segment in one direction to obtain the residual term in the vibration force data segment. The residual term includes different residual points, and one residual point corresponds to one data in the vibration force data segment.

[0014] Preferably, obtaining the outlier data within each vibration force data segment according to the distribution of the residual terms of each vibration force data segment in the residual plot includes:

[0015] In the residual plot of the residual terms of a vibration force data segment, obtain a preset number of residual points adjacent to the left and a preset number of residual points adjacent to the right of a residual point in the residual plot, and record them as the neighbor points of the residual point; obtain the distance from the residual point to the residual median axis in the residual plot and normalize it to obtain the distance feature term; arrange the neighbor points and the residual point in the order of the horizontal axis to obtain the local point set, multiply the sum of the distances between each neighbor point and the residual point by the standard deviation of the distances between every two adjacent points in the local point set to obtain the local continuous distribution feature term; multiply the distance feature term by the normalized local continuous distribution feature term to obtain the outlier degree of the residual point; based on the outlier degree of the residual points corresponding to each data in each vibration force data segment, obtain the outlier data; the residual median axis is the straight line of y = 0 in the residual plot.

[0016] Preferably, obtaining the outlier data based on the outlier degree of the residual points corresponding to each data in each vibration force data segment includes:

[0017] When the outlier degree of a residual point is greater than or equal to the first threshold, the residual point is an outlier residual point; the data in the vibration force data segment corresponding to the outlier residual point is the outlier data.

[0018] Preferably, obtaining the outlier data points according to the outlier data within each vibration force data segment includes:

[0019] For an outlier data, the forces in three directions corresponding to the time of the outlier data are combined into a data point, which is denoted as the outlier data point.

[0020] Preferably, the mean point is obtained according to the forces corresponding to the time when the forces in three directions within the vibration force data segment are not outlier data, including:

[0021] The forces in three directions corresponding to a time when the forces in three directions within the vibration force data segment are not outlier data are combined into a normal data point, and the means of the forces in the three directions included in all normal data points are respectively obtained to get the X-axis mean, Y-axis mean and Z-axis mean, which form the mean point.

[0022] Preferably, based on the distribution of outlier data points within a cluster and the connection line between the cluster center of the cluster and the mean point, the outlier degree of the outlier data points within the cluster is obtained, including:

[0023] Obtain the minimum value of the Euclidean distances between an outlier data point within a cluster and other outlier data points, which is denoted as the minimum distance; multiply the mean and standard deviation of the minimum distances corresponding to each outlier data point within the cluster and normalize to obtain the intra-cluster distribution characteristic term; obtain the connection line between the cluster center of the cluster and the mean point, and get the angles between the connection line and the three axes of the space rectangular coordinate system. The smallest angle is denoted as the minimum angle, and the other two angles are respectively denoted as the first angle and the second angle; respectively obtain the angle differences between the minimum angle and the first angle and the second angle, take the reciprocal of the sum of the two angle differences and normalize to obtain the orientation characteristic term of the cluster; the sum of the intra-cluster distribution characteristic term and the orientation characteristic term is the outlier degree of the outlier data points within the cluster.

[0024] Preferably, according to the outlier degree of the outlier data points, it is judged whether the outlier data points are abnormal outlier points, including:

[0025] Obtain the standard deviation of the outlier degree of each normalized outlier data point. If the standard deviation is greater than the third threshold, obtain the outlier data points whose normalized outlier degree is greater than the second threshold, which are abnormal outlier points.

[0026] Preferably, based on the number of abnormal outlier points within a preset time period, it is judged whether a fault occurs in the boring and milling process, including:

[0027] If the number of abnormal outlier points within the preset time period is greater than or equal to the quantity threshold, an alarm is given to notify the relevant operation and maintenance personnel.

[0028] The embodiments of the present invention have at least the following beneficial effects: In this application, forces in three directions of a spatial rectangular coordinate system are acquired within a preset time period, and vibration force data segments are respectively formed. Then, the residual terms of each vibration force data segment are obtained by using STL time series decomposition. Outlier data is obtained according to the distribution of the residual terms in the residual plot, and then outlier data points are obtained. Then, the outlier data points are mapped into the spatial rectangular coordinate system for clustering analysis, and the characteristics of the forces in three dimensions are jointly analyzed, making the subsequent analysis more accurate. Further, the mean points are obtained through normal data, and the clusters formed by clustering are analyzed based on the mean points. Combining the vibration occurrence situation in the actual operation process, the outlier abnormality degree of the outlier data points in each cluster is obtained, and the outlier data points caused by uneven materials are distinguished and excluded, avoiding the influence of various vibration causes, accurately obtaining the abnormal outlier points caused by clamping looseness or spindle failure, improving the accuracy of fault identification in the boring and milling process, and ensuring the quality of boring and milling. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0030] Figure 1 It is a method flow chart of a method for fault identification in the boring and milling process of a horizontal CNC boring and milling machine provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0031] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the intended invention purpose, the following will, in combination with the drawings and preferred embodiments, detail the specific implementation manners, structures, features, and effects of a method for fault identification in the boring and milling process of a horizontal CNC boring and milling machine proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0032] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs.

[0033] The following will specifically describe the specific solution of a method for fault identification in the boring and milling process of a horizontal CNC boring and milling machine provided by the present invention with reference to the drawings.

[0034] Embodiment:

[0035] The main application scenario of the present invention is as follows: during the operation of a boring and milling machine, a force sensor is used to monitor the vibration data of boring and milling, and based on the STL time series decomposition algorithm, the residuals of historical time series data are decomposed to determine whether there is a fault risk during the boring and milling process.

[0036] Please refer to Figure 1 , which shows a method flow chart of a method for fault identification during the boring and milling process of a horizontal CNC boring and milling machine provided by an embodiment of the present invention. The method includes the following steps:

[0037] Step S1, install a strain type three-dimensional sensor at the connection between the spindle housing and the tool of the boring and milling machine, obtain the forces in three directions of a spatial rectangular coordinate system within a preset time period and respectively form vibration force data segments; analyze the vibration force data segments to obtain the residual terms of the vibration force data segments.

[0038] The machining centers of horizontal CNC boring and milling machines are generally divided into the X-axis, Y-axis, Z-axis, and B-axis. This application mainly analyzes the directions of the main three axes, that is, the X, Y, and Z axes. Further, a strain type three-dimensional sensor is used for data acquisition. A strain type three-dimensional sensor is installed at the connection between the spindle housing and the tool of the boring and milling machine to respectively collect the , , forces in three directions. The forces in three directions are collected at one moment, and the vibration situation during the boring and milling process is reflected by the continuous forces. The acquisition frequency is 100 Hz, and the monitored data is transmitted to the numerical control terminal of the boring and milling machine in real time for data processing at the numerical control terminal.

[0039] In this application, the forces in three directions within a preset time period are respectively formed into vibration force data segments, and one vibration force data segment corresponds to one direction; the preset time period is the time period with a preset time length before the monitoring moment, that is, the data of the 10 seconds before the monitoring moment.

[0040] Further, it is necessary to decompose and analyze each vibration force data segment. The vibration force data segment in one direction is respectively projected onto its corresponding two-dimensional plane coordinate. The horizontal axis of the coordinate is the time series, and the vertical axis is the force value. Then, the STL time series decomposition is performed on the vibration force data segment in this direction to separate the periodic term and trend term of the vibration force data segment in this direction, and obtain the residual term of the vibration force data segment in this direction. The residual term includes multiple residual points, and one residual point corresponds to one data in the vibration force data segment. Among them, the residual term exists alone in the residual diagram (residual coordinate system). There is a horizontal line with a y value of 0 in the residual diagram (residual coordinate system) as the residual mid-axis. The distance between the residual point and the residual mid-axis represents the degree of fit of the data point to the periodic term and trend term. The closer the residual point is to the mid-axis, the more the residual point conforms to the periodic term and trend term, and the farther away, the more the residual point does not conform to the periodic term and trend term, and the more abnormal it is.

[0041] Step S2: Obtain the outlier data within each vibration force data segment respectively according to the distribution of the residual terms of each vibration force data segment in the residual plot; obtain the outlier data points based on the outlier data within each vibration force data segment.

[0042] Obtain the relatively outlier data within the vibration force data segment according to the residual distribution characteristics in the residual plot. Here, an analysis is carried out based on the distance between the residual point and the mid-axis of the residual. The farther the distance between the residual point and the mid-axis of the residual, the more outlier it is. Since the deviation of the period and trend may also be a short-term deviation of the normal boring and milling pressure, it is necessary to combine the continuity between the residual point and the neighboring residual points at the same time to obtain the outlier data within the vibration force data segment. If the residual point is outlier but continuous with the neighboring residual points, it indicates a deviation of the normal boring and milling pressure. If there is a discontinuous and sudden outlier, it is the outlier data caused by abnormal vibration.

[0043] Here, the outlier data is evaluated through the distance between the residual point and the mid-axis of the residual and the neighborhood continuity of the residual point. The neighborhood continuity of the residual point is obtained through the Euclidean distance between the residual point and multiple adjacent nearest residual points (in this solution, the number of selected adjacent residual points is 6, that is, 3 on the left and 3 on the right, which can reflect the neighborhood distribution of the residual point). If the Euclidean distance between a residual point and multiple adjacent residual points is smaller, and the difference degree of the Euclidean distance between every two adjacent residual points including this residual point is smaller, it means that the distance from the adjacent residual points of a residual point to this residual point is closer, and the change degree between points is more uniformly continuous, and it is more likely to be the force change of connectivity caused by the deviation of the normal boring and milling pressure; otherwise, if the Euclidean distance between a residual point and multiple adjacent residual points is larger, and the difference degree of the Euclidean distance between every two adjacent residual points including this residual point is larger, it means that the distance from the adjacent residual points of a residual point to this residual point is farther, and the change degree between points is more sudden, and it is more likely to be the sudden change caused by abnormal vibration.

[0044] Obtain the outlier data in each vibration force data segment according to the distribution of the residual terms in each vibration force data segment within the residual plot. Specifically, in the residual plot of the residual terms of a vibration force data segment, obtain a preset number of residual points adjacent to the left and a preset number of residual points adjacent to the right of a residual point in the residual plot, and denote them as the neighbor points of this residual point; obtain the distance of this residual point from the residual median axis of the residual plot and normalize it to obtain the distance feature term; arrange the neighbor points and this residual point in the order of the horizontal axis to obtain a local point set, multiply the sum of the distances between each neighbor point and this residual point by the standard deviation of the distances between every two adjacent points in the local point set to obtain the local continuous distribution feature term; multiply the distance feature term by the normalized local continuous distribution feature term to obtain the outlier degree of this residual point; obtain the outlier data based on the outlier degree of the residual points corresponding to each data in each vibration force data segment; the residual median axis is the straight line y = 0 in the residual plot.

[0045] The calculation model for the outlier degree of the residual point is:

[0046] ,

[0047] where, is the outlier degree of the q-th residual point among the residual points corresponding to each data in a vibration force data segment, represents the distance between the q-th residual point and the residual median axis. The farther the distance, the more likely the data corresponding to this residual point is an outlier. norm represents normalization, represents the distance feature term; represents the number of neighbor points, which is 6 in this application, and the preset number is 3, represents the Euclidean distance between the i-th neighbor point and this residual point. When the cumulative sum of the Euclidean distances between the adjacent residual points (neighbor points) on both sides of this residual point and this residual point is larger, it indicates that the distances from the adjacent residual points of this residual point to this residual point are farther, represents the standard deviation of the distances between every two adjacent points in the local point set. The larger this standard deviation, the more abrupt the change degree between points, and the more likely it is a mutational change caused by abnormal vibration. The greater the probability of mutation of the data corresponding to this residual point, represents the local continuous distribution feature term. The larger this value, the greater the probability that the mutation of the data corresponding to this residual point is caused by a fault.

[0048] Furthermore, outlier data is obtained based on the degree of outlier of each residual point corresponding to each data in each vibration force data segment. Specifically, a first threshold is set. Preferably, the value of the first threshold is 0.5, which can be adjusted according to the specific requirements of the scenario. When the degree of outlier of a residual point is greater than or equal to the first threshold, the residual point is an outlier residual point, and the data in the vibration force data segment corresponding to the outlier residual point is outlier data. Thus, the outlier data in each vibration data segment can be obtained.

[0049] Furthermore, in order to perform a joint analysis on the forces that may fail in three directions in a three-dimensional space (space rectangular coordinate system) subsequently, it is necessary to extract the data points containing outlier data, and obtain the outlier data points according to the outlier data in each vibration force data segment. Specifically, for an outlier data, the forces in three directions at the moment corresponding to the outlier data are combined into a data point, which is denoted as an outlier data point. Thus, the outlier data points corresponding to each outlier data can be obtained.

[0050] Step S3: Map the outlier data points into the space rectangular coordinate system and perform clustering to obtain clusters; obtain the mean points according to the forces corresponding to the moments when the forces in three directions in the vibration force data segment are not outlier data.

[0051] There are usually various reasons for generating vibrations. Here, the most important several reasons in the boring and milling process are analyzed. For example, the vibrations caused by faults are usually clamping looseness or spindle faults, which will cause vibrations in boring and milling cutting, have a greater impact on the quality of boring and milling products, and have a greater safety risk. In the normal boring and milling process, if the material of the boring and milling object changes during the boring and milling cutting process (that is, a bored and milled object contains different materials), there will also be a large change in the vibration of the monitored force, and the change of material during the boring and milling cutting process is also a common phenomenon. In order to obtain accurate fault monitoring results and exclude the influence of material changes, the characteristics of the vibrations caused by clamping looseness or spindle faults and the vibrations affected by materials are analyzed here.

[0052] Furthermore, all the outlier data points are mapped into the space rectangular coordinate system, and the directions of the three axes of the space rectangular coordinate system are the directions when collecting forces. Thus, the correlation characteristics between the outlier data in each dimension and the data in the dimensions corresponding to the forces in other directions can be analyzed in the space rectangular coordinate system.

[0053] Next, the kmeans clustering algorithm is used to cluster all the outlier data points in the three-dimensional Cartesian coordinate system. The value of k is obtained by the elbow method, and the cluster centers are randomly selected. In the clustering results of kmeans obtained, similar data will be clustered into one cluster. Due to the different vibration amplitudes caused by material influence and those caused by clamping looseness or spindle failure, the outlier data points caused by material influence will be separated from those caused by clamping looseness or spindle failure into multiple clusters. However, since the vibration amplitude corresponding to each case cannot be absolutely determined as a certain value, it is necessary to combine the cluster distribution characteristics here to identify whether the cluster is a vibration cluster caused by material change or a fault vibration cluster.

[0054] Affected by the regular texture of each material, the vibration amplitudes caused by material influence have high similarity, and the concentration of force values for each vibration is relatively high; while after clamping looseness or spindle failure, random oscillations deviating from the axis will occur under the high-speed rotation of the spindle, and the boring and milling vibration amplitudes will randomly float within a certain range of vibration amplitudes, being relatively discrete. On the other hand, due to the cutting force fluctuations caused by material changes, since the boring and milling tool usually enters from one material area of an object into another material area in a fixed direction. For example, in boring, when the tool feeds radially (X direction) and encounters transverse sand inclusion, it will cause sudden fluctuations; in milling, when the spindle feeds axially (Z direction) and encounters surface hard particles, it will cause sudden changes. Therefore, during the boring and milling process, due to material transitions, it usually shows fluctuating changes in a certain direction; while clamping looseness or spindle failure directly affects the overall cutting of the tool, so forces in multiple directions may fluctuate.

[0055] Furthermore, it is necessary to obtain all the data points without anomalies to construct a basis for subsequent analysis. The mean point is obtained according to the forces corresponding to the moments when the forces in the three directions within the vibration force data segment are not outlier data. Specifically, a normal data point is composed of the forces in the three directions corresponding to a moment when the forces in the three directions within the vibration force data segment are not outlier data. The means of the forces in the three directions included in all the normal data points are respectively calculated to obtain the mean value of the X-axis, the mean value of the Y-axis, and the mean value of the Z-axis, which together form the mean point.

[0056] The mean point represents the average position of the force before vibration occurs. In the three-dimensional Cartesian coordinate system, if the vibration is caused by material change, the cluster center of the generated cluster should only change in one direction compared with the mean point, showing a vertical distribution characteristic; if the vibration is caused by a fault, the cluster center of the generated cluster may change in multiple directions compared with the mean point, not showing a vertical distribution characteristic.

[0057] Step S4. In the space rectangular coordinate system, based on the distribution of the outlier data points within a cluster and the line connecting the cluster center and the mean point of the cluster, obtain the outlier degree of the outlier data points within the cluster.

[0058] Here, obtain the outlier degree of the outlier data points based on the cluster characteristics where the outlier data points are located, which consists of the in-cluster distribution characteristics and the cluster azimuth distribution characteristics. Obtain the uniformity of the aggregation of the outlier data points within the cluster where the outlier data points are located as the in-cluster distribution characteristics. When the outlier data points within the cluster where the outlier data points are located are more aggregated and the interval distances are more uniform, it is more in line with the characteristics of the material change vibration cluster; otherwise, when the outlier data points within the cluster where the outlier data points are located are more discrete and the interval distances are more chaotic, it is more in line with the characteristics of the fault vibration cluster.

[0059] For the analysis of the cluster azimuth distribution characteristics, the mean point was obtained in step S3. Using this as the basis, analyze the cluster azimuth distribution characteristics. Connect the cluster center and the mean point to obtain a connecting line. Obtain the included angles between the connecting line of the cluster center and the mean point and the three direction axes in the space rectangular coordinate system. If the cluster only produces a change in one direction compared with the mean point, the included angle between the connecting line and the changing direction will be very small, and the included angles between the connecting line and the non-changing direction axes will be very large. (For example, if the cluster center is in the positive X-axis direction of the mean point, the included angle between the connecting line and the X-axis is very small, and the included angles between the connecting line and the Y-axis and Z-axis are very large); in contrast, if the cluster produces changes in multiple directions compared with the mean point, the included angles between the connecting line and the three direction axes will not show an extreme perpendicular distribution. Therefore, take the difference between the smallest angle among the included angles with the three axes and the other two angles as the cluster azimuth distribution characteristics. The greater the difference between the smallest angle and the other two angles, the more in line with the characteristics of the material change vibration cluster; otherwise, the smaller the difference, it indicates that there are changes in all three directions, and it is more in line with the characteristics of the fault vibration cluster.

[0060] Based on the distribution of the outlier data points within a cluster and the line connecting the cluster center and the mean point of the cluster, obtain the outlier degree of the outlier data points within the cluster. Specifically, obtain the minimum value of the Euclidean distances between an outlier data point within a cluster and other outlier data points, denoted as the minimum distance; multiply the mean and standard deviation of the minimum distances corresponding to each outlier data point within the cluster and normalize to obtain the in-cluster distribution characteristic term; obtain the line connecting the cluster center and the mean point of the cluster, and obtain the included angles between the connecting line and the three axes of the space rectangular coordinate system. Denote the smallest included angle as the minimum angle, and the other two angles as the first angle and the second angle respectively; calculate the angle differences between the minimum angle and the first angle and the second angle respectively, take the reciprocal of the sum of the two angle differences and normalize to obtain the azimuth characteristic term of the cluster; the sum of the in-cluster distribution characteristic term and the azimuth characteristic term is the outlier degree of the outlier data points within the cluster.

[0061] The calculation model for the outlier degree is:

[0062] ,

[0063] wherein, represents the outlier degree of the outlier data point within the p-th cluster, represents the minimum distance corresponding to the i-th outlier data point within the cluster, that is, the Euclidean distance between the i-th outlier data point and the outlier data point closest to it, is the mean value of the minimum distances corresponding to all outlier data points within the cluster. The larger this mean value, the farther the distances between all outlier data points within the cluster and adjacent outlier data points are, the greater the overall dispersion degree, and the more it conforms to the fault anomaly; represents the standard deviation of the minimum distances corresponding to all outlier data points within the cluster. The larger this standard deviation, the more uneven the distribution of outlier data points within the cluster, and the more it conforms to the fault anomaly; , and respectively represent the minimum angle value among the angles between the line connecting the cluster center and the mean point of the cluster and the three axes in the spatial rectangular coordinate system, and the angle values of the other two angles, that is, the minimum angle, the first angle, and the second angle, is the angle difference between the minimum angle and the remaining first angle, is the angle difference between the minimum angle and the remaining second angle. The larger these two angle differences are, the more the connection line has the vertical distribution characteristic, and the more likely the distribution orientation of the cluster is the orientation with only one direction change, and the smaller the fault anomaly degree; the smaller the angle difference, the more likely the distribution orientation of the cluster is the orientation with multiple direction changes, and the greater the fault anomaly degree. The norm linear normalization unifies the magnitudes of the two parts.

[0064] Thus, the outlier degree of the outlier data points within each cluster can be obtained.

[0065] Step S5: Determine whether the outlier data point is an abnormal outlier based on the outlier degree of the outlier data point; determine whether a fault occurs in the boring and milling process based on the number of abnormal outlier points within a preset time period.

[0066] According to the distribution of the outlier degrees of all outlier data points, determine whether there are abnormal outlier points. Normalize the outlier degrees of all outlier data points. After normalization, if the overall data distribution is uniform, it indicates that the outlier degrees of all outlier data points are similar, and there are no abnormal outlier points. All outlier data points are vibration outlier points caused by material influence. If the overall data distribution shows a polarized characteristic, it indicates that there are abnormal outlier points caused by faults. At this time, due to the polarized characteristic between the outlier data points caused by faults and those caused by material influence, take the midpoint 0.5 as the second threshold. If the outlier degree of an outlier data point is greater than the second threshold, then the outlier data point is an abnormal outlier point, that is, an outlier data point caused by a fault.

[0067] It should be noted that when analyzing whether the distribution of the outlier degrees of outlier data points tends to be polarized, it can be analyzed according to the outlier degrees of each outlier data point after normalization. Specifically, obtain the standard deviation of the outlier degrees of each outlier data point after normalization. If the standard deviation is greater than the third threshold, it is considered that the distribution of the outlier degrees of outlier data points tends to be polarized. At this time, there are large differences in the outlier degrees of each outlier data point, and screening of abnormal outlier points is required. When the standard deviation is less than or equal to the third threshold, it is considered that there are no abnormal outlier points at this time, and there is no need to use the second threshold to screen abnormal outlier points.

[0068] After obtaining the abnormal outlier points, it is considered that there is a fault risk in the current boring and milling process. Moreover, the more abnormal outlier points there are, the greater the risk of faults in the current boring and milling machine. Obtain the number of abnormal outlier points within a preset time period, and take 1% of the number of all data points within the preset time period as the quantity threshold (a data point consists of forces in three directions at one moment). If the number of abnormal outlier points within the preset time period is greater than or equal to the quantity threshold, then promptly alarm and notify relevant maintenance personnel, and automatically stop the operation of the boring and milling machine to ensure the safety of boring and milling. In this application, the number of data points within the preset time period is 1000, and the quantity threshold is 10.

[0069] To sum up, through STL time series decomposition of the forces in the three monitored directions, analyze the residual distribution characteristics, and further analyze the clustering characteristics in the spatial rectangular coordinate system combined with the three-axis monitoring data, so as to more accurately obtain the possible abnormal outlier points in the boring and milling process. After discovering the abnormal outlier points, promptly stop the boring and milling process, thereby realizing a safer boring and milling method.

[0070] It should be noted that the above order of the embodiments of the present invention is only for description and does not represent the superiority or inferiority of the embodiments. In addition, the above specific embodiments of this specification have been described. Further, the processes depicted in the drawings do not necessarily require the specific order or sequential order shown to achieve the desired result. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0071] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other, and the key point of each embodiment is to illustrate the differences from other embodiments.

[0072] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for fault identification during the boring and milling process of a horizontal CNC boring and milling machine, characterized in that, The method includes: Installing a strain - type three - dimensional sensor at the connection between the spindle housing and the tool of a boring and milling machine to obtain the forces in three directions of a spatial rectangular coordinate system within a preset period and respectively form vibration force data segments; analyzing the vibration force data segments to obtain the residual terms of the vibration force data segments; Respectively obtaining the outlier data within each vibration force data segment according to the distribution of the residual terms of each vibration force data segment in the residual plot; obtaining outlier data points according to the outlier data within each vibration force data segment; Mapping the outlier data points into the spatial rectangular coordinate system and performing clustering to obtain clusters; obtaining mean points according to the forces corresponding to the moments when the forces in the three directions within the vibration force data segment are not outlier data; Within the spatial rectangular coordinate system, obtaining the outlier anomaly degree of the outlier data points within a cluster based on the distribution of the outlier data points within a cluster and the connection line between the cluster center and the mean point of the cluster; Judging whether the outlier data points are abnormal outlier points according to the outlier anomaly degree of the outlier data points; judging whether a fault occurs in the boring and milling process based on the number of abnormal outlier points within a preset period; The analyzing the vibration force data segments to obtain the residual terms of the vibration force data segments includes: Performing STL time - series decomposition on the vibration force data segment in one direction to obtain the residual terms in the vibration force data segment, where the residual terms include different residual points, and one residual point corresponds to one data in the vibration force data segment; The respectively obtaining the outlier data within each vibration force data segment according to the distribution of the residual terms of each vibration force data segment in the residual plot includes: In the residual plot of the residual terms of a vibration force data segment, obtaining a preset number of residual points adjacent to the left and a preset number of residual points adjacent to the right of a residual point in the residual plot, and recording them as the neighbor points of the residual point; obtaining the distance of the residual point from the residual mid - axis in the residual plot and normalizing it to obtain a distance feature term; arranging the neighbor points and the residual point in the order of the horizontal axis to obtain a local point set, multiplying the sum of the distances between each neighbor point and the residual point by the standard deviation of the distances between every two adjacent points in the local point set to obtain a local continuous distribution feature term; multiplying the distance feature term and the normalized local continuous distribution feature term to obtain the outlier degree of the residual point; obtaining outlier data based on the outlier degree of the residual points corresponding to each data in each vibration force data segment; the residual mid - axis is the straight line of y = 0 in the residual plot; The obtaining the outlier data points according to the outlier data within each vibration force data segment includes: For an outlier data, forming a data point by the forces in the three directions at the moment corresponding to the outlier data, and recording it as an outlier data point; The obtaining the mean points according to the forces corresponding to the moments when the forces in the three directions within the vibration force data segment are not outlier data includes: Forming normal data points by the forces in the three directions at a moment when the forces in the three directions within the vibration force data segment are not outlier data, respectively obtaining the means of the forces in the three directions included in all normal data points to obtain the X - axis mean, Y - axis mean, and Z - axis mean, and forming a mean point; Obtaining the outlier anomaly degree of the outlier data points within a cluster based on the distribution of the outlier data points within the cluster and the line connecting the cluster center and the mean point of the cluster, includes: Obtaining the minimum value of the Euclidean distances between an outlier data point within a cluster and other outlier data points, denoted as the minimum distance; multiplying the mean and standard deviation of the minimum distances corresponding to each outlier data point within the cluster and normalizing to obtain the intra-cluster distribution feature term; obtaining the line connecting the cluster center and the mean point of the cluster, and obtaining the angles between the line and the three axes of the spatial rectangular coordinate system, the smallest angle is denoted as the minimum angle, and the other two angles are denoted as the first angle and the second angle respectively; calculating the angle differences between the minimum angle and the first angle and the second angle respectively, taking the reciprocal of the sum of the two angle differences and normalizing to obtain the azimuth feature term of the cluster; the sum of the intra-cluster distribution feature term and the azimuth feature term is the outlier anomaly degree of the outlier data points within the cluster.

2. The fault identification method for the boring and milling process of a horizontal CNC boring and milling machine according to claim 1, characterized in that, The forces in the three directions of the spatial rectangular coordinate system are the forces in the X-axis, Y-axis, and Z-axis directions respectively.

3. A fault identification method for the boring and milling process of a horizontal CNC boring and milling machine according to claim 1, characterized in that, Obtaining outlier data based on the outlier degree of the residual points corresponding to each data in each vibration force data segment, includes: When the outlier degree of a residual point is greater than or equal to the first threshold, the residual point is an outlier residual point; the data in the vibration force data segment corresponding to the outlier residual point is outlier data.

4. A method for identifying faults in the boring and milling process of a horizontal CNC boring and milling machine according to claim 1, characterized in that, Judging whether an outlier data point is an abnormal outlier point according to the outlier anomaly degree of the outlier data point, includes: Obtaining the standard deviation of the outlier anomaly degree of each normalized outlier data point, if the standard deviation is greater than the third threshold, obtaining the outlier data points with the outlier anomaly degree greater than the second threshold after normalization, which are abnormal outlier points.

5. The fault identification method for the boring and milling process of a horizontal CNC boring and milling machine according to claim 1, characterized in that, Judging whether a boring and milling process fails based on the number of abnormal outlier points within a preset time period, includes: If the number of abnormal outlier points within the preset time period is greater than or equal to the number threshold, then an alarm is given to notify the relevant maintenance personnel.

Images