Self-adaptive monitoring method for wear of rough machining cutter of titanium alloy aviation structural part

Abstract

The invention belongs to the technical field of cutter wear monitoring, and discloses a titanium alloy aviation structural member rough machining cutter wear self-adaptive monitoring method, which comprises the following steps: step 1, determining a cutter wear monitoring threshold T0 according to a machine tool spindle load torque value; 2, within the fixed period T1, a numerical control machine tool spindle load torque value set A in the rough machining process of a tool is collected; 3, performing low-pass filtering on the data in the set A to obtain a data set B; 4, performing kurtosis value analysis and evaluation processing on the set B to obtain a data set X; 5, performing data classification on the data set X by using a support vector machine, and obtaining a data set T by taking a set with the maximum number after classification; step 6, calculating a torque average value # imgabs0 # of the data set T; and step 7, performing cutter wear self-adaptive monitoring on the torque average value # imgabs1 # of each period according to a threshold value T0.

Description

Technical Field

[0001] The invention belongs to the technical field of tool wear monitoring, and in particular relates to a method for adaptively monitoring tool wear for rough machining of titanium alloy aviation structural parts. Background Art

[0002] Titanium alloy aviation structural parts are key structural products of aircraft. Their mechanical properties lead to large cutting forces of titanium alloy materials, excessive tool wear during processing, and low processing efficiency, which seriously affects the aircraft development cycle. In the traditional cutting process, the identification of tool wear status is usually based on the number of processed parts, the cumulative processing time, the surface quality of the parts, the abnormal sound during the tool cutting process, and the state of the chips. These judgment methods require operators to have rich production practice experience and have low accuracy. Therefore, exploring the adaptive monitoring mechanism of tool wear in the processing of aviation titanium alloy structural parts and forming a reliable and effective tool wear monitoring technology are of great significance to improving the processing efficiency of aviation titanium alloy structural parts, improving the processing quality of parts, reducing production costs, and extending the processing and operation life of machine tool systems.

[0003] At present, the functions of common monitoring systems are relatively simple. Most of them use preset alarm limits for monitoring, or monitor by learning and fitting the machine tool spindle torque curve in the time domain. Such methods have high requirements for product consistency and repeatability and are only applicable to mass production models. Aviation structural parts are a typical multi-variety, small-batch production model. The monitoring system does not match the production and manufacturing characteristics of complex aviation structural parts and cannot meet the needs of processing monitoring. Domestic and foreign scholars' research on the monitoring of CNC machining processes of multi-variety, small-batch aviation structural parts is mostly focused on adding vibration sensors and current sensors to the machine tools to collect high-frequency signals and monitor tool errors through frequency domain characteristics. This method has high requirements for matching machining features with signals, requires the machine tool to be modified, and requires additional sensors and computing equipment to be installed, which has a high modification cost. Summary of the invention

[0004] Purpose of the invention: In order to solve the above-mentioned problems, the present invention provides a method for adaptively monitoring tool wear during rough machining of titanium alloy aviation structural parts. By collecting and processing low-frequency machine tool torque data, the tool wear status of aviation titanium alloy structural parts during the rough machining stage is analyzed and monitored, and the tool status is determined to be abnormal and an alarm is issued. This monitoring method can effectively monitor tool breakage failures online.

[0006] An adaptive monitoring method for tool wear in rough machining of aviation titanium alloy structural parts, the steps are as follows:

[0007] Step 1: Determine the tool wear monitoring threshold T0 according to the machine tool spindle load torque value;

[0008] Step 2: Within a fixed period T1, a set A of load torque values ​​of the CNC machine tool spindle during rough machining is collected;

[0009] Step 3: Low-pass filter the data in set A to obtain data set B;

[0010] Step 4: Perform kurtosis analysis and evaluation on set B to obtain data set X;

[0011] Step 5: Use support vector machine to classify the data set X, and take the set with the largest number of classifications to obtain the data set T;

[0012] Step 6: Calculate the torque average of data set T

[0013] Step 7: Calculate the average torque value of each cycle according to the threshold value T0 Perform adaptive tool wear monitoring.

[0014] Furthermore, in step 1, within a fixed period T1, the spindle load torque value of the CNC machine tool when the tool is in the mid-term wear stage is collected through the CNC machine tool OPC UA server interface, and the average machine tool torque within this period is taken as the tool wear monitoring threshold T0.

[0015] Furthermore, in step 2, the set A = {a1, a2, ..., a i , …, a k}, k is the total amount of torque values ​​collected.

[0016] Further, in step three, the low-pass filtering process is as follows:

[0017] Determine each data a in data set A i Does it satisfy the following formula:

[0018]

[0019] In the formula, is the average value of data set A, K k is the Chauvelet coefficient, which is related to the number of data sets A and can be obtained from the Chauvelet coefficient table. σ is the load torque value a in data set A. i The mean square error of

[0020] If the above formula is satisfied, the corresponding data value is removed; if not, it is retained. All the retained data constitutes the data set B, where B = {b1, b2, ..., b j , ..., b p}, p is the number of torque values ​​in data set B.

[0021] Further, in step 4, the kurtosis value analysis and evaluation process is as follows:

[0022] Divide the data in dataset B into p / m groups;

[0023] Calculate the kurtosis value K of each set of data. The formula is as follows:

[0024]

[0025] Where m is the number of data in each group. If the kurtosis value K of a data group is greater than the threshold, the data group is cleared, and all data groups whose kurtosis values ​​do not exceed the threshold constitute the data set X.

[0026] Furthermore, in step 5, the support vector machine algorithm is based on binary classification and seeks a hyperplane (ω·x)+b=0 that meets the classification requirements, where ω is the normal vector, b is the bias vector, and x is the function, and the data is divided into two categories through the hyperplane.

[0027] Further, in step five, the specific process is as follows:

[0028] (1) Dataset X = {x1, x2, ..., x j , …, x n} is divided into two categories by the hyperplane (ω·x)+b=0. The sample data set {(x j ,y i ), i = 1, 2, ..., n} consists of two categories, if x i Belongs to category 1, then marked as positive (y j =+1); if x i Belongs to the second category, then marked as negative (y i =-1), expressed as:

[0029]

[0030] (2) Normalize the classification equation (ω·x)+b=0, and the sample set (x i ,y i ), y j ∈{+1,-1}, satisfying y j [(ω·x j )+b]-1≥0;

[0031] (3) Any point x i The distance d to the hyperplane (ω·x)+b=0 is expressed as

[0032]

[0033] (4) When the classification interval d is the largest, the classification plane (ω·x)+b=0 is the optimal plane. The optimal solution is obtained by solving the following optimization problem:

[0034]

[0035] The Lagrange multiplier method is used to solve this quadratic programming problem with linear constraints as follows:

[0036]

[0037] where α i is a Lagrange multiplier and satisfies 0≤α j ≤C, C is the penalty factor, j = 1, 2, ..., n;

[0038] (5) Input sample data (x i ,y i ), solve the above quadratic programming problem function and obtain the Lagrange multiplier vector

[0039] (6) α * Bring in Get ω * ;

[0040] (7) Take α * A positive component of Calculate b * =y j -ω * x j , and get b * ;

[0041] (8) Bring in ω * , b * , we get the decision function: f(x)=sgn[(ω * ·x i )+b * ];

[0042] (9) For a given unknown type sample x i , calculate f(x)=sgn[(ω * ·x j )+b * ], it can be determined which category the sample belongs to;

[0043] (10) After using the function to divide the data set X into two categories, we take the set with larger data volume as the tool stable cutting data set T.

[0044] Further, in step seven, if the torque average value is within the range of threshold value T0±10%T0, the tool wear is normal, otherwise the machine is stopped for inspection.

[0045] The beneficial effects of this application are:

[0046] The present invention proposes an adaptive monitoring method for tool wear in rough machining of titanium alloy aviation structural parts, which can calculate and monitor the tool wear state of aviation titanium alloy structural parts in the rough machining stage, set the tool wear monitoring threshold to monitor the change in the torque value of the machine tool, determine whether the tool state is abnormal and alarm, and the monitoring method can effectively monitor tool breakage failures online. The use of the adaptive monitoring method for rough machining tool wear can effectively avoid damage to parts and machine tools caused by abnormal tool conditions, improve the stability of the machining process, reduce the scrap rate of parts and production risks, and has the advantages of good versatility, low data requirements, simple algorithm, fast response speed, and zero-cost transformation. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 It is a flow chart of the adaptive monitoring method for rough machining tool wear provided by the present invention.

[0048] Figure 2 This is the torque data diagram of the spindle load torque variable A after low-pass filtering.

[0049] Figure 3 It is the torque data graph after the torque variable B is processed by the kurtosis value.

[0050] Figure 4 It is the torque data diagram after the torque variable X is classified by the support vector machine. DETAILED DESCRIPTION

[0051] The following will be combined with the accompanying drawings in the embodiments of the present invention to clearly describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.

[0052] The terms "first", "second", etc. in the specification and claims of the present invention are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first", "second", etc. are generally of the same type, and the number of objects is not limited. For example, the first object can be one or more. In addition, "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0053] An adaptive monitoring method for tool wear in rough machining of aviation titanium alloy structural parts comprises the following steps:

[0054] Step 1: within a fixed period T1, collect the spindle load torque value of the machine tool when the tool is in the mid-term wear stage, and take the average torque value of the machine tool within this period as the tool wear monitoring threshold T0;

[0055] Step 2: within a fixed period T1, collect the set of CNC machine tool spindle load torque values ​​A during rough machining, where A = {a1, a2, ..., a i , …, a k}, k is the total amount of torque value collected;

[0056] Step 3: Filter the data set A to obtain the data set B, where B = {b1, b2, ..., b j , …, b p}, p is the number of torque values ​​after filtering;

[0057] Step 4: Perform kurtosis analysis and evaluation on B to obtain a data set X, where X = {x1, x2, ..., x j , …, x n}, n is the number of torque values ​​after kurtosis value processing;

[0058] Step 5: classify X using support vector machine to obtain data set T, where T = {t1, t2, ..., t j ,…,t s}, s is the number of torque values ​​after data classification, and s<n<p<k;

[0059] Step 6: Calculate the average torque after data classification

[0060] Step 7: calculate the average torque value of each cycle according to the adaptive tool wear monitoring threshold T0 Perform adaptive tool wear monitoring.

[0061] In step 1, the CNC machine tool spindle load torque value is collected through the CNC machine tool OPC UA server interface.

[0062] In step 3, the spindle load torque value set A is filtered to weaken and eliminate the influence of noise and improve the measurement accuracy of the system. The filtering process uses the Chauvinist criterion to eliminate abnormal values ​​of the machine tool torque. The actual filter algorithm is as follows:

[0063]

[0064] In the formula, is the average value of data set A, K k The Chauvelet coefficient is related to the number of data sets A (obtained from the Chauvelet coefficient table), and σ is the load torque value a in data set A.i The mean square error of .

[0065] Eliminate the set A that satisfies the Chauville criterion The sampling value is obtained, and the storage unit is cleared, and the data set B is obtained by judging one by one until all obvious interference items are eliminated. After filtering, the sampling data is initially stabilized to avoid monitoring errors caused by abnormal values.

[0066] In step 4, the kurtosis value processing is performed by analyzing and evaluating the kurtosis of the data set. As the tool wear increases, the increase in cutting force caused by the gradual blunting of the tool changes slowly over time. The tool cutting amount in the part increases and decreases sharply when the tool cuts in, cuts out, and turns at special positions, which is manifested as a large kurtosis value. At this time, the data is useless for judging the tool wear state. Therefore, by analyzing and evaluating the data kurtosis, large kurtosis data is removed to retain a stable cutting state. The p data in data set B are divided into p / m groups, and the kurtosis value processing calculation method is as follows:

[0067] Kurtosis value K:

[0068]

[0069] Where m refers to the number of torques in the data set to be processed. If the kurtosis value K of the data set is greater than 1.6, the data set is cleared, and the data set is judged group by group until all the groups with large kurtosis values ​​are removed, and finally the data set X is obtained.

[0070] In step 5, in the fixed cycle time, the tool feed and retract process is shorter than the stable cutting process, so the data set X needs to be divided into two categories, and the set with the largest amount of data is taken as the tool stable cutting data set. The support vector machine algorithm is based on binary classification and finds a hyperplane (ω·x)+b=0 (where ω is the normal vector, b is the bias vector, and x is the function) that meets the classification requirements, so that the hyperplane divides the data into two categories.

[0071] Furthermore, the support vector machine data classification method is as follows:

[0072] (1) Dataset X = {x1, x2, ..., x j , …, x n} is divided into two categories by the hyperplane (ω·x)+b=0, that is, the sample data set {(x j ,y j ), i = 1, 2, ..., n} consists of two categories, if x i Belongs to category 1, then marked as positive (y i =+1); if x i Belongs to the second category, then marked as negative (y i =-1), which can be expressed as:

[0073]

[0074] (2) Normalize the classification equation (ω·x)+b=0, and the sample set (x i ,y i ), y j ∈{+1,-1}, satisfying y j [(ω·x j )+b]-1≥0.

[0075] (3) Any point x i The distance d to the hyperplane (ω·x)+b=0 can be expressed as

[0076]

[0077] (4) When the classification interval d is the largest, the classification plane (ω·x)+b=0 is the optimal plane, that is, the optimal solution is obtained by solving the following optimization problem:

[0078]

[0079] The Lagrange multiplier method is used to solve this quadratic programming problem with linear constraints, namely

[0080]

[0081] where α i is a Lagrange multiplier and satisfies 0≤α j ≤C, C is the penalty factor, j = 1, 2, …, n.

[0082] (5) Input sample data (x j ,y j ), solve the above quadratic programming problem function and obtain the Lagrange multiplier vector

[0083] (6) α * Bring in Get ω * ;

[0084] (7) Take α * A positive component of Calculate b * =y j -ω * x j , and get b * .

[0085] (8) Bring in ω * , b * , and get the decision function:

[0086] f(x)=sgn[(ω * ·x i )+b * ]

[0087] (9) For a given unknown type sample x i , we only need to calculate f(x)=sgn[(ω * ·x i )+b * ], you can determine the category to which the sample belongs;

[0088] (10) After using the function to divide the data set X into two categories, we take the set with larger data volume as the tool stable cutting data set T.

[0089] Example

[0090] Taking the rough machining process of the door swing rocker arm part of a titanium alloy aviation structure on a five-axis linkage PV2216 CNC machine tool as an example, the specific steps of the titanium alloy aviation structure rough machining tool wear adaptive monitoring method proposed in the present invention are as follows:

[0091] 1) The five-axis linkage PV2216 CNC machine tool uses the OPC UA server interface to read the machine tool spindle load torque variable. In this embodiment, 5 minutes is a fixed time period, and about 5,000 torque data are collected in each time period.

[0092] 2) Use the Chauvinist criterion to eliminate abnormal values ​​of machine tool torque. The actual filter algorithm is as follows:

[0093]

[0094] Elimination by Chauville criterion The sample value X i , that is, if Then clear the storage unit and judge one by one until all obvious interference items are eliminated. Figure 2 shown.

[0095] 3) Analyze and evaluate the data kurtosis, remove the data with large kurtosis, and retain the stable cutting state.

[0096] The kurtosis value processing calculation method is as follows:

[0097] K>1.6 In this embodiment, in order to make the result as accurate as possible, m=10, that is, the number of data in each group is 10, and the kurtosis value K of each group of data is calculated. If the kurtosis value K of the group of data is greater than 1.6, the group is cleared, and the group is judged group by group until all the groups with large kurtosis values ​​are removed. The data processing results are as follows: Figure 3 shown.

[0098] 4) The support vector machine divides the data into two sets according to consistency and obtains the optimal classification function:

[0099] f(x)=0.081x j -235.2

[0100] Through the above data classification, the largest set is taken, which is the tool stable cutting data, such as Figure 4 shown.

[0101] 5) According to experimental results and processing experience, when the tool cutting time is between 15 and 20 minutes, all types of rough machining tools are in the mid-term wear stage. The average machine tool torque at this time is taken as the tool wear monitoring reference value T0. In this embodiment, T0=10.81.

[0102] a) Calculate the average value of the torque data after processing

[0103]

[0104] Where s is the number of torque value data.

[0105] b) Based on on-site processing experience, the processing torque fluctuation does not exceed the reference value by 10%.

[0106] c) The calculation formula for the adaptive tool wear monitoring threshold is:

[0107]

[0108] The monitoring threshold serves as the tolerance range boundary of the machining curve. In subsequent machining, the machining process is monitored based on the threshold and tolerance range boundary. If abnormal conditions such as tool missing, tool breakage, blade breakage, and tool burnt occur during the rough machining process, the machining process curve will definitely have a sudden change or a large difference compared to the normal machining process. When the monitored torque value exceeds the preset monitoring threshold, the monitoring system automatically identifies this sudden change and difference, and then issues an alarm and controls the machine tool to pause.

[0109] 6) During the monitoring time of the embodiment, no false alarms occurred. After 60 minutes of rough machining, in order to verify the monitoring accuracy, the machine was stopped to check the tool status. After observation, some areas of the back blade changed color, and one blade corner was slightly damaged. The tool was in the normal wear stage, and the monitoring system did not show wear characteristics, so the rough machining continued. When the system alarmed at 105 minutes of rough machining, the tool was in a damaged state. Two of the four blades were slightly damaged. The damaged area was about 0.6 square millimeters and 0.2 square millimeters. At this time, there was no abnormality on the machined surface of the part, and the machine tool cut smoothly.

[0110] It should be noted that, in this article, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device including the element. In addition, it should be pointed out that the scope of the method and device in the embodiment of the present invention is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0111] Through the description of the above implementation modes, those skilled in the art can clearly understand that the above implementation modes can be implemented by means of software plus a necessary general hardware platform, and of course by means of hardware, but in many cases the former is a better implementation mode. The above describes the embodiments of the present invention in conjunction with the accompanying drawings, but the present invention is not limited to the above specific implementation modes, which are merely illustrative and not restrictive. Under the enlightenment of the present invention, those skilled in the art can also make many forms without departing from the scope of protection of the present invention and the claims, all of which are within the protection of the present invention.

Claims

1. An adaptive monitoring method for tool wear in rough machining of aviation titanium alloy structural parts, characterized by: Here are the steps: Step 1: Determine the tool wear monitoring threshold T0 according to the machine tool spindle load torque value; Step 2: Within a fixed period T1, a set A of load torque values ​​of the CNC machine tool spindle during rough machining is collected; Step 3: Low-pass filter the data in set A to obtain data set B; Step 4: Perform kurtosis analysis and evaluation on set B to obtain data set X; Step 5: Use support vector machine to classify the data set X, and take the set with the largest number of classifications to obtain the data set T; Step 6: Calculate the torque average of data set T Step 7: Calculate the average torque value of each cycle according to the threshold value T0 Perform adaptive tool wear monitoring.

2. The method according to claim 1, characterized in that: In step 1, within a fixed period T1, the spindle load torque value of the CNC machine tool when the tool is in the mid-term wear stage is collected through the CNC machine tool OPC UA server interface, and the average machine tool torque within this period is taken as the tool wear monitoring threshold T0.

3. The method according to claim 2, characterized in that: In step 2, the set A = {a1, a2, ..., a i , …, a k }, k is the total amount of torque values ​​collected.

4. The method according to claim 3, characterized in that: In step three, the low-pass filtering process is as follows: Determine whether each data ai in data set A satisfies the following formula: In the formula, is the average value of data set A, K k is the Chauvelet coefficient, which is related to the number of data sets A and can be obtained from the Chauvelet coefficient table. σ is the load torque value a in data set A. i The mean square error of If the above formula is satisfied, the corresponding data value is removed; if not, it is retained. All the retained data constitute the data set B, where B = {b1, b2, ..., b j , …, b p ), p is the number of torque values ​​in data set B.

5. The method according to claim 4, characterized in that: In step 4, the kurtosis value analysis and evaluation process is as follows: Divide the data in dataset B into p / m groups; Calculate the kurtosis value K of each set of data. The formula is as follows: Where m is the number of data in each group. If the kurtosis value K of a data group is greater than the threshold, the data group is cleared, and all data groups whose kurtosis values ​​do not exceed the threshold constitute the data set X.

6. The method according to claim 5, characterized in that: In step 5, the support vector machine algorithm is based on binary classification and looks for a hyperplane (ω·x)+b=0 that meets the classification requirements, where ω is the normal vector, b is the bias vector, and x is the function. The data is divided into two categories through the hyperplane.

7. The method according to claim 6, characterized in that: In step 5, the specific process is as follows: (1) Dataset X = {x1, x2, ..., x j , …, x n } is divided into two categories by the hyperplane (ω·x)+b=0. The sample data set {(x i ,y i ), i = 1, 2, ..., n} consists of two categories, if x i Belongs to category 1, then marked as positive (y i =+1); if x i Belongs to the second category, then marked as negative (y i =-1), expressed as: (2) Normalize the classification equation (ω·x)+b=0, and the sample set (x i ,y i ), y i ∈{+1,-1), satisfying y j [(ω·x j )+b]-1≥0; (3) Any point x i The distance d to the hyperplane (ω·x)+b=0 is expressed as (4) When the classification interval d is the largest, the classification plane (ω·x)+b=0 is the optimal plane. The optimal solution is obtained by solving the following optimization problem: The Lagrange multiplier method is used to solve this quadratic programming problem with linear constraints as follows: where α i is a Lagrange multiplier and satisfies 0≤α j ≤C, C is the penalty factor, j = 1, 2, ..., n; (5) Input sample data (x j ,y j ), solve the above quadratic programming problem function and obtain the Lagrange multiplier vector (6) α 8 Bring in Get ω 8 ; (7) Take α 8 A positive component of Calculate b * =y j -ω * x j Get b 8 ; (8) Bring in ω 8 、b 8 , we get the decision function: f(x)=sgn[(ω 8 ·x i )+b 8 ]; (9) For a given unknown type sample x i , calculate f(x)=sgn[(ω 8 ·x j )+b 8 ], it can be determined which category the sample belongs to; (10) After using the function to divide the data set X into two categories, we take the set with larger data volume as the tool stable cutting data set T.

8. The method according to claim 7, characterized in that: In step 7, if the average torque value is within the threshold value T0±10%T0, the tool wear is normal, otherwise the machine is stopped for inspection.

Images