CN114675598A - Method and system for predicting tool tip modal parameters of different numerical control machines based on transfer learning - Google Patents
Method and system for predicting tool tip modal parameters of different numerical control machines based on transfer learning Download PDFInfo
- Publication number
- CN114675598A CN114675598A CN202210319572.0A CN202210319572A CN114675598A CN 114675598 A CN114675598 A CN 114675598A CN 202210319572 A CN202210319572 A CN 202210319572A CN 114675598 A CN114675598 A CN 114675598A
- Authority
- CN
- China
- Prior art keywords
- tool
- machine tool
- nose
- modal
- modal parameters
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 55
- 238000013526 transfer learning Methods 0.000 title claims abstract description 27
- 238000002474 experimental method Methods 0.000 claims abstract description 37
- 238000003801 milling Methods 0.000 claims abstract description 31
- 238000005520 cutting process Methods 0.000 claims abstract description 20
- 230000006870 function Effects 0.000 claims description 37
- 238000009826 distribution Methods 0.000 claims description 36
- 238000013508 migration Methods 0.000 claims description 20
- 230000005012 migration Effects 0.000 claims description 20
- 230000003044 adaptive effect Effects 0.000 claims description 13
- 238000012549 training Methods 0.000 claims description 13
- 238000012360 testing method Methods 0.000 claims description 10
- 238000005259 measurement Methods 0.000 claims description 7
- 230000008569 process Effects 0.000 claims description 6
- 239000011159 matrix material Substances 0.000 claims description 4
- 238000005070 sampling Methods 0.000 claims description 4
- 238000012544 monitoring process Methods 0.000 claims description 3
- 230000006978 adaptation Effects 0.000 claims description 2
- 238000013016 damping Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000003754 machining Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 244000145845 chattering Species 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000002184 metal Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000087 stabilizing effect Effects 0.000 description 1
- 238000010998 test method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B19/00—Programme-control systems
- G05B19/02—Programme-control systems electric
- G05B19/18—Numerical control [NC], i.e. automatically operating machines, in particular machine tools, e.g. in a manufacturing environment, so as to execute positioning, movement or co-ordinated operations by means of programme data in numerical form
- G05B19/408—Numerical control [NC], i.e. automatically operating machines, in particular machine tools, e.g. in a manufacturing environment, so as to execute positioning, movement or co-ordinated operations by means of programme data in numerical form characterised by data handling or data format, e.g. reading, buffering or conversion of data
- G05B19/4086—Coordinate conversions; Other special calculations
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B2219/00—Program-control systems
- G05B2219/30—Nc systems
- G05B2219/35—Nc in input of data, input till input file format
- G05B2219/35356—Data handling
Landscapes
- Engineering & Computer Science (AREA)
- Human Computer Interaction (AREA)
- Manufacturing & Machinery (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Numerical Control (AREA)
Abstract
The invention provides a method and a system for predicting tool tip modal parameters of different numerical control machine tools based on transfer learning, wherein the method comprises the following steps: selecting a plurality of milling centers of the same model, respectively carrying out modal hammering experiments at the same position on a machine tool of each milling center, measuring the FRFs of the tool nose, and comparing the FRFs difference of the tool nose of the same tool of different numerical control machines of the same type; selecting any machine tool as a source machine tool, performing a spindle rotation speed gradual lifting experiment within the range of the spindle rotation speed and the operation space of the machine tool until flutter occurs, recording the flutter frequency and the corresponding limit axial cutting depth, and identifying tool tip modal parameters at different positions and speeds; establishing a tool nose modal parameter regression prediction model related to the position-speed of the source machine tool by adopting a Kriging method, and accurately predicting tool nose modal parameters of the machine tool at different positions and rotating speeds; identifying the corresponding tool nose modal parameters to obtain an optimal prediction model of the tool nose modal parameters of the target machine tool to predict the tool nose modal parameters.
Description
Technical Field
The invention relates to the field of modal parameters of numerical control machines, in particular to a method and a system for predicting tool nose modal parameters of different numerical control machines based on transfer learning.
Background
Self-excited vibration, also known as chatter vibration, is one of the most serious contributing factors in metal cutting of numerically controlled machine tools. It not only reduces the processing performance, but also shortens the service life of the cutter. To prevent chatter, the most common method is to select chatter-free cutting parameters by stabilizing the lobe diagram (SLD). Generally, the SLD method requires a tool tip FRF, which can be easily obtained through a test mode analysis (EMA), but numerous scholars have proved that the tool tip FRF is also closely related to the position of a moving part of a numerical control machine and the rotation speed of a spindle, and the tool tip modal parameters in an operating state cannot be obtained through the EMA.
In addition, for the large-scale processing of enterprises, the use of different individual machine tools of the same type is necessarily involved, and due to the factors of random deviation in the manufacturing and assembling processes, uncertain degradation degree in the service process, complex interface effect of a machine tool joint part, different field working conditions and the like of different individual machine tools, even if the different machine tool individuals of the same type have great difference, the difference causes that the tool tip modal parameters of the same tool-tool shank assembly are different under the same condition, so that the prediction of the milling process stability is finally influenced.
Through years of development, remarkable progress is made in the depth and the breadth of a method for predicting tool nose modal parameters of a numerical control machine tool related to position and speed, but the tool nose modal parameters in the current numerical control machine tool running state are only independently considered to influence the position of a running part and the rotating speed of a main shaft, the position and the speed of the numerical control machine tool in actual machining have time-varying performance, the two factors have coupling performance on the influence of the tool nose modal parameters, particularly, for the use of different individual machine tools of the same type in batch machining, no method is found to be researched and discussed at present aiming at the problem, and the existing other tool nose modal parameter prediction methods are limited in application range, prediction precision and efficiency and still have certain limitation in the aspect of actual application.
Disclosure of Invention
The invention provides a method and a system for predicting tool tip modal parameters of different numerical control machines based on transfer learning, and aims to solve the problems that the existing method for predicting the tool tip modal parameters of other numerical control machines is limited in application range, prediction accuracy and efficiency.
In order to achieve the purpose, the invention provides a method for predicting tool nose modal parameters of different numerical control machine tools based on transfer learning, which comprises the following steps:
and 4, performing a milling experiment on the target machine tool at a preset position and at a preset spindle rotation speed by using the plurality of milling centers of the same model except the source machine tool, identifying corresponding tool nose modal parameters, fusing the data of the source machine tool with the data of the target machine tool, reducing the difference between the edge probability distribution and the conditional probability distribution of the modal parameter data of the source machine tool and the target machine tool by using a weight self-adaptive joint distribution adaptive migration learning method, and finishing iterative training of the migration learning model after a migration learning loss curve is stable to obtain an optimal prediction model of the tool nose modal parameters of the target machine tool for predicting the tool nose modal parameters.
Wherein, the step 1 specifically comprises:
a low-quality broadband one-way accelerometer is arranged at the tool tip of each numerical control machine tool of the same type and is used for monitoring the vibration of the tool tip X, Y in the direction;
the accelerometer is connected with an acquisition system, the system sampling rate of the acquisition system is 8192Hz, and the average value is obtained every five times in the measurement process.
Wherein, the step 2 specifically comprises:
the preset maximum stroke of the source machine tool is X Y Z850X 500X 560mm, and the total number of machine tool experiment position points determined after preliminary experiments is 18;
the rotating speed range of the main shaft of the source machine tool is 50-8000rpm, the rotating speed range of the selected experimental main shaft is 4300 plus 7200rpm, the interval of the rotating speed ranges is 200rpm, and 15 groups of main shaft rotating speeds are provided;
and (3) according to the experimental position point of the machine tool and the rotating speed of the main shaft, carrying out milling experiments under 270 groups of position-speed combinations, gradually changing the axial cutting depth until chatter occurs, and identifying corresponding tool nose modal parameters through inverse stability solution.
Wherein, the step 3 specifically comprises:
decomposing the Kriging approximate model into a cutter nose modal parameter related to the predicted position-speed and a random function from normal distribution;
expressing the random function part in a covariance matrix form, and further expressing a function of 2 sample points and position-speed correlation;
fitting a function related to position-speed by adopting a Gaussian function and an exponential function, and identifying unknown parameters in the function;
and calculating the error between the approximate value and the true value of the Kriging model, adjusting the related parameters and minimizing the prediction error.
Wherein, the step 4 specifically comprises:
determining 20 groups of key position-speed combinations to be tested of a target machine tool according to the test result of a source machine tool;
identifying tool nose modal parameters under the key position-speed combination of the target machine tool according to the inverse stability solution;
reducing edge probability distribution and conditional probability distribution of tool nose modal parameter data of a source machine tool and a target machine tool by adopting a weight self-adaptive joint distribution adaptive migration learning method;
and substituting the fused new data into the Kriging model in the step 3, training to obtain a tool nose modal parameter prediction model of the target machine tool, and predicting the tool nose modal parameter.
The invention also provides a tool tip modal parameter prediction system of different numerical control machine tools based on transfer learning, which comprises the following steps:
the data measurement and comparison module is used for selecting a plurality of milling centers of the same type, respectively carrying out modal hammering experiments at the same position on a machine tool of each milling center, measuring the tool nose FRFs of the machine tool, and comparing the tool nose FRFs difference of the same tool of different numerical control machines of the same type;
the parameter identification module is used for selecting any machine tool as a source machine tool, performing a main shaft rotating speed gradual increasing experiment within the rotating speed and running space range of a main shaft of the machine tool until flutter occurs, recording the flutter frequency and the corresponding limit axial cutting depth, and identifying tool tip modal parameters at different positions and speeds;
the model establishing module is used for establishing a tool nose modal parameter regression prediction model related to the position-speed of the source machine tool by adopting a Kriging method, and accurately predicting tool nose modal parameters of the machine tool at different positions and rotating speeds;
and the parameter prediction module is used for the plurality of milling centers with the same model, the rest of the milling centers except the source machine tool are target machine tools, milling experiments are carried out on the target machine tools at preset positions and preset spindle rotation speeds, corresponding tool nose modal parameters are identified, the source machine tool data and the target machine tool data are fused, the difference between the edge probability distribution and the conditional probability distribution of the modal parameter data of the source machine tool and the target machine tool is reduced by a weight self-adaptive joint distribution adaptive transfer learning method, and after a transfer learning loss curve is stable, iterative training of the transfer learning model is finished, so that the optimal tool nose modal parameter prediction model of the target machine tool is obtained for tool nose modal parameter prediction.
The scheme of the invention has the following beneficial effects:
the invention provides a method and a system for predicting tool nose modal parameters of different numerical control machine tools based on transfer learning. Furthermore, the tool tip modal parameter prediction of the same type of different individual numerical control machine tools is realized through a transfer learning method, the generalization of a tool tip modal parameter prediction model and the prediction efficiency of modal parameters are improved, and the manual test and marking cost is greatly saved.
Other advantages of the present invention will be described in detail in the detailed description that follows.
Drawings
FIG. 1 is a schematic flow chart illustrating a method for predicting modal parameters according to the present invention;
FIG. 2 is a comparison result of FRFs differences of tool tips of the same type of different numerical control machine tools;
FIG. 3 is the comparison result of the FRFs difference of the tool tips of different numerically-controlled machine tools of the same type;
FIG. 4 (a) shows the result of Kriging prediction of the cutting edge natural frequency for a source machine tool diameter of 8mm, and FIG. 4 (b) shows the result of Kriging prediction of the cutting edge damping ratio; fig. 4 (c) and 4 (d) show actual recognition results of the tip natural frequency and the damping ratio, respectively;
FIG. 5 is a basic flow of tool nose modal parameter prediction of a target machine tool based on weight adaptive joint distribution adaptive transfer learning according to the present invention;
fig. 6 (a) shows a diagram of the modal parameter calculation stability lobe of one of the position-speed combination predictions of the target machine tool, which is the limit cut depth prediction result, and fig. 6 (b) shows a diagram of the chattering frequency prediction result;
Detailed Description
In order to make the technical problems, technical solutions and advantages of the present invention more apparent, the following detailed description is given with reference to the accompanying drawings and specific embodiments. It is to be understood that the embodiments described are only a few embodiments of the present invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, and are only for convenience of description and simplicity of description, but do not indicate or imply that the device or element being referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus, should not be construed as limiting the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
In the description of the present invention, it is to be noted that, unless otherwise explicitly specified or limited, the terms "mounted", "connected" and "connected" are to be understood broadly, for example, as being either a locked connection, a detachable connection, or an integral connection; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood in specific cases to those skilled in the art.
In addition, the technical features involved in the different embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
As shown in fig. 1, an embodiment of the present invention provides a method for predicting tool nose modal parameters of different numerically-controlled machine tools based on migration learning, including:
and 4, performing a milling experiment on the target machine tool at a preset position and at a preset spindle rotation speed by using the plurality of milling centers of the same model except the source machine tool, identifying corresponding tool nose modal parameters, fusing the data of the source machine tool with the data of the target machine tool, reducing the difference between the edge probability distribution and the conditional probability distribution of the modal parameter data of the source machine tool and the target machine tool by using a weight self-adaptive joint distribution adaptive migration learning method, and finishing iterative training of the migration learning model after a migration learning loss curve is stable to obtain an optimal prediction model of the tool nose modal parameters of the target machine tool for predicting the tool nose modal parameters.
In the embodiment of the invention, 5 milling centers (VMC850E) with the same model are selected as experimental objects, modal hammering experiments are respectively carried out at the same position of 5 machine tools, the tool nose FRFs of the machine tools are measured, and the difference of the tool nose FRFs of the same tool of different numerical control machines with the same model is compared; selecting any machine tool as a source machine tool, performing a main shaft rotating speed gradual lifting experiment within the rotating speed and operating space range of a main shaft of the machine tool until flutter occurs, recording the flutter frequency and the corresponding limit axial cutting depth, and identifying tool tip modal parameters at different positions and speeds by an inverse stability solution method; establishing a tool nose modal parameter regression prediction model related to the position-speed of the source machine tool by adopting a Kriging method, and accurately predicting tool nose modal parameters of the machine tool at different positions and rotating speeds; and (3) taking other numerical control machines as target machine tools, performing milling experiments on the target machine tools at a few positions and at the rotating speed of a main shaft, identifying corresponding tool nose modal parameters of the target machine tools, fusing data of the source machine tools and data of the target machine tools, and reducing the difference between the edge probability distribution and the conditional probability distribution of the modal parameter data of the source machine tools and the target machine tools by a weight self-adaptive joint distribution adaptive migration learning method so as to finish iterative training of the migration learning model after a migration learning loss curve is stable to obtain an optimal prediction model of the tool nose modal parameters of the target machine tools. Considering the influence of the time-varying property of the position of a moving part of the machine tool and the speed of the main shaft on the tool nose modal parameter in the machining process of the numerical control machine tool, a tool nose modal parameter prediction model related to the position-speed is established, the model considers the coupling influence of the position and the speed, and the prediction result is more accurate. Furthermore, the tool tip modal parameter prediction of the same type of different individual numerical control machine tools is realized through a transfer learning method, the generalization of a tool tip modal parameter prediction model and the prediction efficiency of modal parameters are improved, and the manual test and marking cost is greatly saved.
Wherein, the step 1 specifically comprises: a low-quality broadband one-way accelerometer is arranged at the tool tip of each numerical control machine tool of the same type and is used for monitoring the vibration of the tool tip X, Y in the direction; the accelerometer is connected with an acquisition system, the system sampling rate of the acquisition system is 8192Hz, and the average value is obtained every five times in the measurement process.
Wherein, the step 2 specifically comprises:
the preset maximum stroke of the source machine tool is X Y Z850X 500X 560mm, and the total number of machine tool experiment position points determined after preliminary experiments is 18;
the rotating speed range of the main shaft of the source machine tool is 50-8000rpm, the rotating speed range of the selected experimental main shaft is 4300 plus 7200rpm, the interval of the rotating speed ranges is 200rpm, and 15 groups of main shaft rotating speeds are provided;
and (3) according to the experimental position point of the machine tool and the rotating speed of the main shaft, carrying out milling experiments under 270 groups of position-speed combinations, gradually changing the axial cutting depth until chatter occurs, and identifying corresponding tool nose modal parameters through inverse stability solution.
Wherein, the step 3 specifically comprises:
decomposing a Kriging approximate model into a random function and a random function, wherein the random function is used for predicting the cutter nose modal parameter related to position-speed and the random function is normally distributed;
expressing the random function part in a covariance matrix form, and further expressing a function of 2 sample points and position-speed correlation;
fitting a function related to position-speed by adopting a Gaussian function and an exponential function, and identifying unknown parameters in the function;
and calculating the error between the approximate value and the true value of the Kriging model, adjusting the related parameters and minimizing the prediction error.
Wherein, the step 4 specifically comprises:
determining 20 groups of key position-speed combinations to be tested of a target machine tool according to the test result of a source machine tool;
identifying tool nose modal parameters under the key position-speed combination of the target machine tool according to the inverse stability solution;
reducing edge probability distribution and conditional probability distribution of tool nose modal parameter data of a source machine tool and a target machine tool by adopting a weight self-adaptive joint distribution adaptive migration learning method;
and substituting the fused new data into the Kriging model in the step 3, training to obtain a tool nose modal parameter prediction model of the target machine tool, and predicting the tool nose modal parameter.
Further, a low-mass broadband one-way accelerometer is arranged at the tool tip of 5 numerically-controlled machine tools of the same type so as to monitor the vibration of the tool tip X, Y in the direction. The accelerometer is connected to an acquisition system (LMS-SCADAS-Mobile-SCM05) with a sampling rate of 8192Hz, and the measurement process is averaged every fifth time to reduce random errors. According to the experimental test method, the FRFs of the tool tips are measured at the same position (X-300, Y-0, Z-200) of 5 machine tools respectively, and the differences are compared. Taking the machine tool # 1 as a source machine tool, wherein the maximum stroke X Y Z of the tested numerical control machine tool is 850X 500X 560mm, and the total number of the machine tool test position points determined after preliminary pre-tests is 18; the rotating speed range of the machine tool spindle is 50-8000rpm, the rotating speed range of the experimental spindle is 4300-7200rpm, the interval of the rotating speed ranges is 200rpm, and 15 groups of spindle rotating speeds are total; milling experiments were performed in a total of 270 sets of position-speed combinations, by gradually changing the axial cut-depth until chatter occurred, and then identifying the corresponding tool tip modal parameters by "inverse stability solution". Decomposing a Kriging approximate model into a random function and a random function, wherein the random function is used for predicting the cutter nose modal parameter related to position-speed and the random function is normally distributed; expressing the random function part in a covariance matrix form, and further expressing a function of 2 sample points and position-speed correlation; fitting a function related to position-speed by adopting a Gaussian function and an exponential function, and identifying unknown parameters in the function; and calculating the error between the approximate value and the true value of the Kriging model, adjusting the related parameters and minimizing the prediction error. Determining 20 groups of key position-speed combinations to be tested of a target machine tool according to the test result of a source machine tool; identifying tool nose modal parameters under the key position-speed combination of the target machine tool according to the inverse stability solution; training a tool nose modal parameter prediction model of a target machine tool by using source machine tool data, and reducing edge probability distribution and conditional probability distribution of tool nose modal parameter data of the source machine tool and the target machine tool by adopting a weight self-adaptive joint distribution adaptive migration learning method; and substituting the fused new data into a Kriging model in S3, and training to obtain a tool nose modal parameter prediction model of the target machine tool for predicting the tool nose modal parameters.
The invention also provides a tool tip modal parameter prediction system of different numerical control machine tools based on transfer learning, which comprises the following steps:
the data measurement and comparison module is used for selecting a plurality of milling centers of the same type, respectively carrying out modal hammering experiments at the same position on a machine tool of each milling center, measuring the tool nose FRFs of the machine tool, and comparing the tool nose FRFs difference of the same tool of different numerical control machines of the same type;
the parameter identification module is used for selecting any machine tool as a source machine tool, performing a main shaft rotating speed gradual lifting experiment within the rotating speed and running space range of a main shaft of the machine tool until flutter occurs, recording the flutter frequency and the corresponding limit axial cutting depth, and identifying tool tip modal parameters at different positions and speeds;
the model establishing module is used for establishing a tool nose modal parameter regression prediction model related to the position-speed of the source machine tool by adopting a Kriging method, and accurately predicting tool nose modal parameters of the machine tool at different positions and rotating speeds;
and the parameter prediction module is used for carrying out milling experiments on the target machine tool at a preset position and a preset spindle rotating speed, identifying corresponding tool nose modal parameters, fusing source machine tool data and target machine tool data, reducing the difference between the edge probability distribution and the condition probability distribution of the modal parameter data of the source machine tool and the target machine tool by a weight self-adaptive joint distribution adaptation transfer learning method, and ending iterative training of the transfer learning model after a transfer learning loss curve is stable to obtain an optimal tool nose modal parameter prediction model of the target machine tool for tool nose modal parameter prediction.
The invention provides a method and a system for predicting tool tip modal parameters of different numerical control machines based on transfer learning, as shown in figure 2, the method firstly carries out modal hammering experiments under the same condition of the same type of different numerical control machines, and determines that the tool tip modal parameters of the same type of different numerical control machines are different even under the same condition. In order to build an accurate prediction model of the position-velocity related tool tip modal parameters of the source machine tool, milling experiments must be performed for a sufficient number of times to identify the tool tip modal parameters, the machine tool # 1 is used as the source machine tool, the maximum stroke X Y Z of the tested numerically-controlled machine tool is 850X 500X 560mm, and a total of 18 machine tool experimental position points determined after preliminary pre-experiments are performed, as shown in fig. 3. The rotating speed range of the machine tool spindle is 50-8000rpm, and for the high-precision milling center, the rotating speed range of the experimental spindle is 4300-7200rpm, the interval of the rotating speed ranges is 200rpm, and 15 groups of spindle rotating speeds are provided. Therefore, a total of 270 tests were performed, with the workpiece being made of Al6061 and having dimensions of 160X 300 mm. The tool nose natural frequency and the damping ratio (assuming that the cutting force coefficient is unchanged) under the position-speed combination are identified through an 'inverse stability solution', the tool nose natural frequency and the damping ratio at other positions are respectively obtained through the method, and 270 sets of tool nose modal parameters related to the position-speed in the milling process of the source machine tool are obtained together. Regression modeling is carried out on tool nose modal parameters related to the position-speed of the source machine tool by adopting a Kriging method, and a model prediction result and an actual identification result are shown in FIG. 4. Since the position is represented by three-axis coordinates (X, Y, Z) of the machine tool, if the spindle rotation speed factor is considered at the same time, 5 coordinate axes are required for visualization, which cannot be realized in a 3-dimensional space. Therefore, in order to visualize the influence of position-speed on the natural frequency of the tool nose, 18 selected machine tool space positions are represented by numbers 1-18, the position coordinate of the position P1 in this chapter is (100,100,100), the coordinate of the position P2 is (400,100,100), when Kriging modeling is carried out, the space between P1 and P2 is equally divided into 10 equal parts, the corresponding machine tool position coordinate is (P2-P1) × 0.1+ P1, and modeling and prediction of the tool nose modal parameters are carried out between the other two adjacent positions by a similar method.
In order to migrate the tool nose modal parameter prediction model of the source machine tool to the target machine tool, the basic flow is as shown in fig. 5, and 20 sets of tool nose modal parameters of the key position-speed combination, which need to be measured by the tool of the target machine tool, are determined according to the tool nose modal parameter prediction model of the Kriging source machine tool, as shown in table 1. Using a machine tool # 1 as a source machine tool and machine tools # 2, #3, #4 and #5 as target machine tools, obtaining new Y by using a tool nose modal parameter of a key position-speed combination of a target machine tool cutter and carrying out adaptive migration learning transformation on the tool nose modal parameter of the source machine tool after adaptive weight joint distribution adaptationnewAnd then jointly training a Kriging target machine tool modal parameter regression model by 270+ 20-290 groups of data of the source machine tool data and the target machine tool data.
TABLE 1 selected key position-velocity combinations for target machine tool
Finally, in order to verify the reliability of the tool nose modal parameter of the target machine tool predicted by adopting the migration learning, taking the milling parameter P-V (600-8N/m2And 6.26X 108N/m2. According to the synthesized tool nose FRFs and the cutting force coefficient, the milling process SLD obtained by the calculation of the two-degree-of-freedom milling stability model is shown in figure 6, and in order to verify the accuracy of the SLD prediction result, a spindle rotating speed step-by-step lifting method is adopted in the spindle rotating speed range of 5100 and 7100rpmAnd gradually changing the axial cutting depth in the periphery to carry out milling vibration experiments, wherein the axial cutting depth ranges from 0.2 mm to 1.2mm, and the axial cutting depth is increased by 0.1mm each time until vibration occurs.
While the foregoing is directed to the preferred embodiment of the present invention, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention as defined in the appended claims.
Claims (6)
1. A tool nose modal parameter prediction method of different numerical control machine tools based on transfer learning is characterized by comprising the following steps:
step 1, selecting a plurality of milling centers of the same type, respectively carrying out modal hammering experiments at the same position on a machine tool of each milling center, measuring the tool nose FRFs of the machine tool, and comparing the difference of the tool nose FRFs of the same tool of different numerical control machines of the same type;
step 2, selecting any machine tool as a source machine tool, performing a main shaft rotating speed gradual lifting experiment within the main shaft rotating speed and operating space range of the machine tool until flutter occurs, recording flutter frequency and corresponding limit axial cutting depth, and identifying tool tip modal parameters at different positions and speeds;
step 3, establishing a tool nose modal parameter regression prediction model related to the position-speed of the source machine tool by adopting a Kriging method, and accurately predicting tool nose modal parameters of the machine tool at different positions and rotating speeds;
and 4, performing a milling experiment on the target machine tool at a preset position and at a preset spindle rotation speed by using the plurality of milling centers of the same model except the source machine tool, identifying corresponding tool nose modal parameters, fusing the data of the source machine tool with the data of the target machine tool, reducing the difference between the edge probability distribution and the conditional probability distribution of the modal parameter data of the source machine tool and the target machine tool by using a weight self-adaptive joint distribution adaptive migration learning method, and finishing iterative training of the migration learning model after a migration learning loss curve is stable to obtain an optimal prediction model of the tool nose modal parameters of the target machine tool for predicting the tool nose modal parameters.
2. The method for predicting tool nose modal parameters of different numerically-controlled machine tools based on transfer learning according to claim 1, wherein the step 1 specifically comprises:
a low-quality broadband one-way accelerometer is arranged at the tool tip of each numerical control machine tool of the same type and is used for monitoring the vibration of the tool tip X, Y in the direction;
the accelerometer is connected with an acquisition system, the system sampling rate of the acquisition system is 8192Hz, and the average value is obtained every five times in the measurement process.
3. The method for predicting tool nose modal parameters of different numerically-controlled machine tools based on transfer learning according to claim 1, wherein the step 2 specifically comprises:
the preset maximum stroke of the source machine tool is X Y Z850X 500X 560mm, and the total number of machine tool experiment position points determined after preliminary experiments is 18;
the rotating speed range of the main shaft of the source machine tool is 50-8000rpm, the rotating speed range of the selected experimental main shaft is 4300 plus 7200rpm, the interval of the rotating speed ranges is 200rpm, and 15 groups of main shaft rotating speeds are provided;
and (3) according to the experimental position point of the machine tool and the rotating speed of the main shaft, carrying out milling experiments under 270 groups of position-speed combinations, gradually changing the axial cutting depth until chatter occurs, and identifying corresponding tool nose modal parameters through inverse stability solution.
4. The method for predicting tool nose modal parameters of different numerically-controlled machine tools based on transfer learning according to claim 1, wherein the step 3 specifically comprises:
decomposing a Kriging approximate model into a random function and a random function, wherein the random function is used for predicting the cutter nose modal parameter related to position-speed and the random function is normally distributed;
expressing the random function part in a covariance matrix form, and further expressing a function of 2 sample points and position-speed correlation;
fitting a function related to position-speed by adopting a Gaussian function and an exponential function, and identifying unknown parameters in the function;
and calculating the error between the approximate value and the true value of the Kriging model, adjusting the related parameters and minimizing the prediction error.
5. The method for predicting tool nose modal parameters of different numerically-controlled machine tools based on the transfer learning of claim 4, wherein the step 4 specifically comprises:
determining 20 groups of key position-speed combinations to be tested of a target machine tool according to the test result of a source machine tool;
identifying tool nose modal parameters under the key position-speed combination of the target machine tool according to the inverse stability solution;
reducing edge probability distribution and conditional probability distribution of tool nose modal parameter data of a source machine tool and a target machine tool by adopting a weight self-adaptive joint distribution adaptive migration learning method;
and substituting the fused new data into the Kriging model in the step 3, training to obtain a tool nose modal parameter prediction model of the target machine tool, and predicting the tool nose modal parameter.
6. The utility model provides a different digit control machine tool knife tip modal parameter prediction system based on migration learning which characterized in that includes:
the data measurement and comparison module is used for selecting a plurality of milling centers of the same type, respectively carrying out modal hammering experiments at the same position on a machine tool of each milling center, measuring the tool nose FRFs of the machine tool, and comparing the tool nose FRFs difference of the same tool of different numerical control machines of the same type;
the parameter identification module is used for selecting any machine tool as a source machine tool, performing a main shaft rotating speed gradual lifting experiment within the rotating speed and running space range of a main shaft of the machine tool until flutter occurs, recording the flutter frequency and the corresponding limit axial cutting depth, and identifying tool tip modal parameters at different positions and speeds;
the model establishing module is used for establishing a tool nose modal parameter regression prediction model related to the position-speed of the source machine tool by adopting a Kriging method, and accurately predicting tool nose modal parameters of the machine tool at different positions and rotating speeds;
and the parameter prediction module is used for carrying out milling experiments on the target machine tool at a preset position and a preset spindle rotating speed, identifying corresponding tool nose modal parameters, fusing source machine tool data and target machine tool data, reducing the difference between the edge probability distribution and the condition probability distribution of the modal parameter data of the source machine tool and the target machine tool by a weight self-adaptive joint distribution adaptation transfer learning method, and ending iterative training of the transfer learning model after a transfer learning loss curve is stable to obtain an optimal tool nose modal parameter prediction model of the target machine tool for tool nose modal parameter prediction.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210319572.0A CN114675598B (en) | 2022-03-29 | 2022-03-29 | Method and system for predicting tool nose modal parameters of different numerical control machine tools based on transfer learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210319572.0A CN114675598B (en) | 2022-03-29 | 2022-03-29 | Method and system for predicting tool nose modal parameters of different numerical control machine tools based on transfer learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114675598A true CN114675598A (en) | 2022-06-28 |
| CN114675598B CN114675598B (en) | 2024-06-25 |
Family
ID=82076111
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210319572.0A Active CN114675598B (en) | 2022-03-29 | 2022-03-29 | Method and system for predicting tool nose modal parameters of different numerical control machine tools based on transfer learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114675598B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116679614A (en) * | 2023-07-08 | 2023-09-01 | 四川大学 | Multi-feature cutter comprehensive adaptation method based on evolution game |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112488397A (en) * | 2020-12-01 | 2021-03-12 | 合肥工业大学 | Load prediction method under extreme scene based on modal decomposition and transfer learning |
| CN113077043A (en) * | 2021-03-17 | 2021-07-06 | 华中科技大学 | Machine tool nose dynamic characteristic prediction method based on improved graph convolution network |
| WO2021163618A1 (en) * | 2020-02-14 | 2021-08-19 | Novartis Ag | Method of predicting response to chimeric antigen receptor therapy |
| CN114072807A (en) * | 2020-09-10 | 2022-02-18 | 南京航空航天大学 | Milling robot multi-mode frequency response prediction method based on small sample transfer learning |
-
2022
- 2022-03-29 CN CN202210319572.0A patent/CN114675598B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2021163618A1 (en) * | 2020-02-14 | 2021-08-19 | Novartis Ag | Method of predicting response to chimeric antigen receptor therapy |
| CN114072807A (en) * | 2020-09-10 | 2022-02-18 | 南京航空航天大学 | Milling robot multi-mode frequency response prediction method based on small sample transfer learning |
| WO2022051973A1 (en) * | 2020-09-10 | 2022-03-17 | 南京航空航天大学 | Milling robot multi-modal frequency response prediction method based on small-sample transfer learning |
| CN112488397A (en) * | 2020-12-01 | 2021-03-12 | 合肥工业大学 | Load prediction method under extreme scene based on modal decomposition and transfer learning |
| CN113077043A (en) * | 2021-03-17 | 2021-07-06 | 华中科技大学 | Machine tool nose dynamic characteristic prediction method based on improved graph convolution network |
Non-Patent Citations (1)
| Title |
|---|
| KAI LI, CHAOCHAO QIU: "A weighted adaptive transfer learning for tool tip dynamics prediction of different machine tools", 《COMPUTERS & INDUSTRIAL ENGINEERING》, 27 May 2022 (2022-05-27) * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116679614A (en) * | 2023-07-08 | 2023-09-01 | 四川大学 | Multi-feature cutter comprehensive adaptation method based on evolution game |
| CN116679614B (en) * | 2023-07-08 | 2024-02-02 | 四川大学 | Multi-feature cutter comprehensive adaptation method based on evolution game |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114675598B (en) | 2024-06-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109822399B (en) | Numerical control machine tool cutter wear state prediction method based on parallel deep neural network | |
| CN114237155B (en) | Error prediction and compensation method, system and medium for multi-axis numerical control machining | |
| US6400998B1 (en) | Generation of measurement program in NC machining and machining management based on the measurement program | |
| CN107589723B (en) | A kind of static-dynamic state optimization method of numerically-controlled machine tool Milling Process stability | |
| CN107627152B (en) | Numerical control machining chip control method based on BP neural network | |
| CN115562161B (en) | On-line monitoring-based cutter cutting path machining error compensation method | |
| US11914339B2 (en) | Datum selection methods and systems for minimizing hole position errors in group hole machining of large components | |
| CN111487923A (en) | Swing position error detection and identification method for CA double-swing five-axis numerical control machine tool | |
| CN114675598A (en) | Method and system for predicting tool tip modal parameters of different numerical control machines based on transfer learning | |
| CN113158371A (en) | Dynamic cutting force prediction system for high-speed milling and parameter optimization method | |
| CN110703686A (en) | On-line measuring path planning method for blade section of blisk | |
| Yu et al. | Integration of tool error identification and machining accuracy prediction into machining compensation in flank milling | |
| CN114840932A (en) | Method for improving TC4 titanium alloy surface roughness prediction precision through multi-factor coupling | |
| CN118466395A (en) | Dynamic sensitivity analysis method for numerical control machine tool error term | |
| CN109202539A (en) | A kind of composite material weak separation polymorphic structure online test method | |
| CN117348546A (en) | Flexible production line process quality evaluation method | |
| Nielsen | Specifications, operators and uncertainties | |
| CN110362038B (en) | Test piece for identifying online detection capability of five-axis linkage numerical control machine tool and detection method | |
| CN113427320A (en) | Multi-axis on-machine measurement planning method for reducing measurement uncertainty | |
| CN110370076A (en) | Free form surface in-situ measuring method based on machining tool | |
| CN110465831A (en) | A kind of processing On-line Measuring Method of turbine blade | |
| CN115741234B (en) | Milling cutter face abrasion loss measuring method | |
| CN114547888B (en) | Process design and verification method for milling surface precision distribution consistency | |
| CN116451156B (en) | Hole feature recognition method and sequencing method for blade air film holes | |
| CN115793572B (en) | Self-adaptive machining method for welding boss of aviation casing part |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |