CN112395809A - Method for detecting surface vibration line defects of machined part - Google Patents
Method for detecting surface vibration line defects of machined part Download PDFInfo
- Publication number
- CN112395809A CN112395809A CN202011310382.XA CN202011310382A CN112395809A CN 112395809 A CN112395809 A CN 112395809A CN 202011310382 A CN202011310382 A CN 202011310382A CN 112395809 A CN112395809 A CN 112395809A
- Authority
- CN
- China
- Prior art keywords
- vibration
- flutter
- detection model
- chatter
- acceleration response
- 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
- 230000007547 defect Effects 0.000 title claims abstract description 56
- 238000000034 method Methods 0.000 title claims abstract description 52
- 238000001514 detection method Methods 0.000 claims abstract description 94
- 238000003754 machining Methods 0.000 claims abstract description 47
- 230000004044 response Effects 0.000 claims abstract description 40
- 230000001133 acceleration Effects 0.000 claims abstract description 39
- 238000012545 processing Methods 0.000 claims abstract description 32
- 238000001914 filtration Methods 0.000 claims abstract description 10
- 238000005520 cutting process Methods 0.000 claims description 37
- 238000012549 training Methods 0.000 claims description 22
- 238000012706 support-vector machine Methods 0.000 claims description 18
- 230000008569 process Effects 0.000 claims description 14
- 238000004422 calculation algorithm Methods 0.000 claims description 11
- 238000003801 milling Methods 0.000 claims description 10
- 238000004590 computer program Methods 0.000 claims description 8
- 238000012360 testing method Methods 0.000 claims description 8
- 238000003860 storage Methods 0.000 claims description 7
- 238000010801 machine learning Methods 0.000 claims description 6
- 238000007689 inspection Methods 0.000 claims description 4
- 238000002372 labelling Methods 0.000 claims description 4
- 230000002411 adverse Effects 0.000 abstract description 3
- 230000000694 effects Effects 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 6
- 239000000463 material Substances 0.000 description 4
- 239000011159 matrix material Substances 0.000 description 4
- 239000013598 vector Substances 0.000 description 4
- 238000005299 abrasion Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000004088 simulation Methods 0.000 description 3
- 230000015556 catabolic process Effects 0.000 description 2
- 238000006731 degradation reaction Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000005284 excitation Effects 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000013016 damping Methods 0.000 description 1
- 230000006866 deterioration Effects 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000001172 regenerating effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 230000003746 surface roughness Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/04—Constraint-based CAD
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/14—Force analysis or force optimisation, e.g. static or dynamic forces
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Automatic Control Of Machine Tools (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The invention discloses a method for detecting surface vibration line defects of machined parts, which comprises the following steps: measuring acceleration response signals of a cutter along the horizontal and vertical directions of a machine tool in the current machining process of a part, filtering cutter-pass harmonic components in the acceleration response signals, and extracting wavelet entropy characteristics reflecting machining instability strength in the acceleration response signals and recording the wavelet entropy characteristics as vibration characteristics; inputting the vibration characteristics into a pre-trained flutter detection model, and judging the flutter state of a weak rigidity processing system for processing parts at present; if the workpiece is in a flutter state, the current machining surface of the part has a chatter mark defect. After each part is machined, according to the accuracy of the current flutter detection model, an incremental learning mode is adopted on the basis of the existing flutter detection model, and information which can have adverse effects on the judgment precision is gradually eliminated by utilizing continuously accumulated actually-measured vibration information, so that the accuracy of the flutter detection model is improved, and the detection precision of the machined surface chatter mark defects is high.
Description
Technical Field
The invention belongs to the field of machining defect detection, and particularly relates to a method for detecting a surface vibration line defect of a machined part.
Background
The surface vibration line defect of the part determines the service performance and the fatigue life of the part. The surface vibration lines are marks repeatedly engraved on the surface of the part by the vibration displacement of the cutter teeth in the machining process. Chatter defects are often caused by chatter in the machining vibrations, and chatter characteristics and chatter defect characteristics show high similarity in frequency distribution. Therefore, the chatter information similar to the vibration mark defect characteristics can be extracted from the machining vibration signal to be used as a judgment basis for judging whether the vibration mark defect exists.
In the machining process, accurate judgment of machining chatter provides conditions for timely avoiding deterioration of the machined surface and reducing the surface roughness of the workpiece. However, the existing judgment indexes of the flutter state usually come from learning and mining of the existing machining process data, how to effectively extract flutter components from original cutting force and vibration signals is clarified by adopting different signal analysis technologies, and the state of a machining system is detected based on an extracted flutter component information training model, and the technologies need to master flutter frequency information in advance without exception; however, in the machining process, due to the removal of workpiece materials and the aggravation of tool abrasion, the flutter frequency and the working condition of the flutter frequency can be changed along with the degradation of the system, so that the flutter detection model has parameter drift, the detection precision of the model is gradually reduced, the flutter state of the machining system cannot be accurately judged, and the chatter mark defect on the machined surface cannot be accurately detected.
Disclosure of Invention
Aiming at the defects or improvement requirements of the prior art, the invention provides a method for detecting the chatter mark defects of the machined surface, and aims to solve the technical problem that the accuracy of detecting the chatter mark defects of the surface is low due to parameter drift of a chatter detection model caused by removal of workpiece materials and abrasion of a cutter in the machining process in the prior art.
In order to achieve the above object, in a first aspect, the present invention provides a method for detecting a chatter mark defect on a surface of a machined part, including the steps of:
s1, measuring acceleration response signals of the tool in the horizontal and vertical directions of the machine tool in the current machining process of the part, filtering off tool-pass harmonic components irrelevant to surface vibration mark defects in the acceleration response signals, and extracting wavelet entropy characteristics reflecting machining instability strength in the acceleration response signals and recording the wavelet entropy characteristics as vibration characteristics;
s2, inputting the vibration characteristics into a pre-trained flutter detection model, and judging the flutter state of the weak rigidity processing system for processing the part at present; if the workpiece is in a flutter state, the current machining surface of the part has a chatter mark defect;
wherein the flutter detection model is a machine learning model; counting the vibration line detection results of all processing surfaces obtained in the whole processing process of a part to obtain the vibration line detection result of the surface of the part every time the part is processed, and comparing the vibration line detection result of the surface of the part with a standard vibration line quality detection result to obtain vibration line defect detection precision, namely the accuracy of a vibration detection model; if the accuracy of the chatter detection model is smaller than a preset accuracy threshold, marking the chatter state of the machining system corresponding to the vibration characteristics of the part in the whole machining process according to the standard chatter texture quality inspection result of the part, then training the chatter detection model in an incremental mode by taking the vibration characteristics of the part in the whole machining process as input and the corresponding marks as output, updating parameters of the chatter detection model, and correcting the change of chatter detection accuracy caused by workpiece material removal and cutter abrasion.
Further preferably, the training method of the chatter detection model includes:
s01, establishing a dynamic equation of the weak rigidity processing system, extracting vibration characteristics under different cutting parameters from acceleration response signals of the tool along the horizontal and vertical directions of the machine tool under different cutting parameters in the processing system according to the dynamic equation, and respectively marking the vibration states corresponding to the vibration characteristics;
s02, taking vibration characteristics under different cutting parameters as input, and correspondingly marking as output training machine learning models to obtain flutter detection models;
s03, acquiring acceleration response signals of the tool along the horizontal and vertical directions of the machine tool in the actual machining process under different cutting parameters, extracting actual values of the vibration characteristics under different cutting parameters, and respectively labeling the flutter states corresponding to the actual values of the vibration characteristics;
and S04, taking actual values of the vibration characteristics under different cutting parameters as input and corresponding labels as output, training a flutter detection model in an incremental mode, and updating the flutter detection model.
Further preferably, step S01 includes the steps of:
s011, establishing a dynamic equation of the weak rigidity machining system, and identifying a cutting force coefficient and a modal parameter in the dynamic equation by combining full-tooth milling and modal test;
and S012, solving the kinetic equation by adopting a Runge-Kutta method based on the cutting force coefficient and the modal parameter to obtain acceleration response signals of the cutter along the horizontal and vertical directions of the machine tool under different cutting parameters, filtering off the cutter pass harmonic component irrelevant to the surface vibration pattern defect in each acceleration response signal, and extracting wavelet entropy characteristics reflecting the processing instability strength in each acceleration response signal, namely vibration characteristics under different cutting parameters.
Further preferably, a notch filtering method is adopted to filter out knife-pass harmonic components which are not related to the surface vibration mark defects in each acceleration response signal.
Further preferably, the flutter detection model is a kernel support vector machine model.
Further preferably, in the training process, an SMO algorithm is used to solve an optimal decision boundary corresponding to the flutter detection model for distinguishing whether the model is in the flutter state.
In a second aspect, the present invention further provides a computer-readable storage medium, where the computer-readable storage medium includes a stored computer program, where when the computer program is executed by a processor, the computer program controls an apparatus in which the storage medium is located to execute the method for detecting the chatter mark defects on the surface of the machined part, provided by the first aspect of the present invention.
In general, compared with the prior art, the above technical solution contemplated by the present invention can achieve the following beneficial effects:
1. the invention provides a method for detecting the surface vibration line defects of a machined part, which is characterized in that vibration characteristics are input into a pre-trained vibration detection model, and the vibration state of a weak rigidity machining system for machining the part at present is judged; if the workpiece is in a flutter state, the current machining surface of the part has a chatter mark defect; after each part is machined, according to the accuracy of the current flutter detection model, an incremental learning mode is continuously adopted on the basis of the existing flutter detection model, and information which can have adverse effects on the judgment precision is gradually eliminated by utilizing continuously accumulated actually-measured vibration information so as to improve the accuracy of the flutter detection model. The problem of among the prior art because the model precision is not enough to lead to the skew actual value of simulation vibration result is solved, the detection precision of processing surface chatter marks defect is higher.
2. The method for detecting the vibration line defects on the surface of the machined part, provided by the invention, comprises the steps of establishing a motion equation of a weak rigidity machining system to obtain an acceleration response signal when a vibration detection model is pre-trained, extracting a vibration signal and then carrying out corresponding vibration marking; establishing a flutter detection model, and solving an optimal decision boundary corresponding to the flutter detection model and used for distinguishing whether the flutter detection model is in a flutter state by adopting an SMO algorithm; aiming at the problem that the simulation vibration result deviates from the actual value due to insufficient model precision, collecting vibration signals in the actual machining process to carry out corresponding flutter labeling in order to avoid the degradation of the judgment precision of unstable machining working conditions; and gradually adding the measured vibration signals into the learning process of the kernel support vector machine for judging the boundary by adopting an incremental learning mode, and updating the decision boundary on the basis of the original decision boundary. The method adopts an incremental learning mode on the basis of the existing flutter detection model, and avoids the requirement of zero-basis learning on large sample size.
3. According to the method for detecting the vibration line defects of the machined surface, provided by the invention, in the machining process of the part, the vibration signals of the machining system are collected, and the vibration characteristics, which are highly similar to the vibration line defects, in the vibration signals are extracted, so that the vibration signals can be used as a judgment basis for judging whether the vibration line defects exist. The method can detect the defects of the machined vibration lines in real time, and does not need to take out the parts for offline detection after the parts are machined, so that the detection efficiency is greatly improved.
Drawings
FIG. 1 is a flowchart of a method for detecting chatter marks on a surface of a machined part according to embodiment 1 of the present invention;
fig. 2 is a schematic diagram of a dynamic milling process provided in embodiment 1 of the present invention.
Fig. 3 is a schematic diagram of a decision boundary corresponding to the incremental learning pre-flutter detection model provided in embodiment 1 of the present invention;
fig. 4 is a schematic diagram of a decision boundary corresponding to the incremental learning post-flutter detection model provided in embodiment 1 of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
Examples 1,
A method for detecting the chatter mark defects on the surface of a machined part, as shown in figure 1, comprises the following steps:
s1, measuring acceleration response signals of the tool in the horizontal and vertical directions of the machine tool in the current machining process of the part, filtering off tool-pass harmonic components irrelevant to surface vibration mark defects in the acceleration response signals, and extracting wavelet entropy characteristics reflecting machining instability strength in the acceleration response signals and recording the wavelet entropy characteristics as vibration characteristics;
s2, inputting the vibration characteristics into a pre-trained flutter detection model, and judging the flutter state of the weak rigidity processing system for processing the part at present; if the workpiece is in a flutter state, the current machining surface of the part has a chatter mark defect;
wherein the flutter detection model is a machine learning model; in this embodiment, the kernel support vector machine model.
Specifically, the training method of the flutter detection model includes:
s01, establishing a dynamic equation of the weak rigidity processing system, extracting vibration characteristics under different cutting parameters from acceleration response signals of the tool along the horizontal and vertical directions of the machine tool under different cutting parameters in the processing system according to the dynamic equation, and respectively marking the vibration states corresponding to the vibration characteristics;
specifically, step S01 includes the following steps:
s011, establishing a dynamic equation of a weak rigidity machining system (specifically, the dynamic equation of a machine tool-cutter system), and identifying a cutting force coefficient and a modal parameter in the dynamic equation by combining full-tooth milling and modal test;
in the embodiment, a weak rigidity processing system is taken as an example of a milling processing system for detailed description, and fig. 2 is a schematic diagram of a dynamic milling process. In this embodiment, the workpiece vibration is much weaker than the vibration of the machine tool-tool system. Therefore, the vibration of the machine tool-cutter system is a key factor causing chatter defects on the machined surface. Machine tool-tool systems can be generally simplified to spring-damped systems (shown in fig. 2), with alternating interaction forces, i.e., cutting forces, between the tool workpieces. The machine tool-cutter system can vibrate under the condition of alternating cutting load, and when machining vibration exists, the surface of a workpiece can be damaged by the violent vibration of the cutter, so that vibration line defects are formed. In this embodiment, only considering the single modes in the two directions (respectively, the X direction and the Y direction) of the horizontal direction and the vertical direction of the machine tool, the motion equation of the milling system can be expressed as follows:
m, C, K is a modal mass matrix, a modal damping matrix and a modal stiffness matrix respectively; kc(t)[q(t)-q(t-T)]Is an excitation with a regenerative effect; f0Is a steady state force excitation relative to surface position error; kc(T) is a cutting force coefficient matrix, [ q (T) -q (T-T)]The dynamic cutting thickness is generated by the vibration at the time T and the vibration before one tooth passing period T.
And S012, solving a kinetic equation of the machining system by adopting a Runge-Kutta method based on the cutting force coefficient and the modal parameter to obtain acceleration response signals of the cutter along the horizontal and vertical directions of the machine tool under different cutting parameters, filtering off cutter pass harmonic components irrelevant to surface vibration pattern defects in each acceleration response signal, and extracting wavelet entropy characteristics reflecting machining instability strength in each acceleration response signal, namely vibration characteristics under different cutting parameters.
Specifically, the motion equation of the milling system is converted into a first order differential equation applicable to a 4-order Runge-Kutta method:
wherein the content of the first and second substances,
after the acceleration response signals are obtained by solving through a 4-order Runge Kutta method, notch filtering is carried out on the acceleration response, harmonic components of a tooth-through period in the acceleration response are filtered, the frequency response value of the signal point to be filtered in each acceleration response signal is 0, the frequency response value of other signal points is 1, and knife-through harmonic components irrelevant to surface vibration mark defects in each acceleration response signal are filtered. And then extracting wavelet entropy characteristics reflecting processing instability strength in each acceleration response signal, namely vibration characteristics under different cutting parameters.
S02, taking vibration characteristics under different cutting parameters as input, and correspondingly marking as output training machine learning models to obtain flutter detection models;
specifically, the flutter detection model is a kernel support vector machine model. In the training process, an SMO algorithm is adopted to solve an optimal decision boundary which is corresponding to the flutter detection model and used for distinguishing whether the flutter detection model is in a flutter state or not. The optimal decision boundary is a decision boundary for distinguishing whether the vibration characteristic sample is in a flutter state or not and is an optimal decision boundary searched by the kernel support vector machine model. When the support vector machine trains the classifier, the optimal classifier is searched according to the classification interval of the classifier, the classification interval is the vertical distance between the extreme positions of the sample points at the two sides of the decision boundary, which are closest to the decision boundary, different classification intervals exist on the decision boundaries in different directions, and the decision boundary with the maximum interval is the searched optimal decision boundary. The model structure of the kernel support vector machine model is simple, and the incremental learning speed is higher. In the training process, an SMO algorithm is adopted to solve an optimal decision boundary which is corresponding to the flutter detection model and used for distinguishing whether the flutter detection model is in a flutter state or not. The optimal decision boundary is a decision boundary for distinguishing whether the vibration characteristic sample is in a flutter state or not and is an optimal decision boundary searched by the kernel support vector machine model. Compared with the common solution dual problem method, the SMO algorithm of the kernel support vector machine obtains a better solution at a higher speed, and the algorithm is simpler in structure.
Specifically, the basic description of the support vector machine optimization problem is as follows:
s.t.yi(ωTxi+b)≥1i＝1,2,...,n
where | ω | | is the reciprocal of the classification interval, n is the number of training samples (i.e., the number of vibration features used for training), and xiIs the ith vibration characteristic, yiIs the label corresponding to the ith vibration feature, f (x)i)＝ωTxi+ b is the decision boundary equation.
Because data can not be completely linearly separable, a relaxation variable xi and a penalty variable C are introduced, and in addition, for the optimization problem with inequality constraint, the constraint and the optimization function are written into a Lagrange function by utilizing a Lagrange multiplier method:
then, according to the requirement KKT condition of the optimal value, the form of the kernel support vector machine model suitable for the SMO algorithm can be written as follows:
s.t.C≥αi≥0i＝1,2,...,n
αithe terms and b terms are continuously updated in the SMO algorithm until an optimal decision boundary is found, that is, the decision boundary of the kernel support vector machine can be represented as:
s03, acquiring acceleration response signals of the tool along the horizontal and vertical directions of the machine tool in the actual machining process under different cutting parameters, extracting actual values of the vibration characteristics under different cutting parameters, and respectively labeling the flutter states corresponding to the actual values of the vibration characteristics; specifically, a machining test platform is set up, and an accelerometer is adopted to measure and acquire acceleration response signals of the cutter along the horizontal and vertical directions of the machine tool in the actual machining process under different cutting parameters.
And S04, taking actual values of the vibration characteristics under different cutting parameters as input and corresponding labels as output, training a flutter detection model in an incremental mode, and updating the flutter detection model. Specifically, the training process is as above.
It should be noted that, in the invention, every time a part is machined, the vibration line detection results of each machined surface obtained in the whole machining process of the part are counted to obtain the vibration line detection result of the part surface, and the vibration line detection result of the part surface is compared with the standard vibration line quality detection result to obtain the vibration line defect detection precision, namely the accuracy of the vibration detection model; if the accuracy of the chatter detection model is smaller than a preset accuracy threshold (the preset accuracy threshold is 97% in this embodiment), marking the chatter state of the processing system corresponding to the vibration characteristics of the part in the whole processing process according to the standard chatter texture quality inspection result of the part, then training the chatter detection model in an incremental manner by taking the vibration characteristics of the part in the whole processing process as input and corresponding marking as output, updating parameters of the chatter detection model, and correcting the change of the chatter detection accuracy caused by workpiece material removal and tool wear. Otherwise, incremental training and updating are not performed on the flutter detection model.
In particular, incremental learning refers to a learning system that can continuously learn new knowledge from new samples and can preserve a large portion of previously learned knowledge. When the measured samples are gradually added into the learning of the decision boundary of the kernel support vector machine, the boundary is not required to be completely reconstructed, but only the change caused by newly added data is updated on the basis of the original boundary.
Specifically, based on the lagrangian function of the kernel support vector machine model, the first order condition is simplified to the KKT condition:
the initial data set is divided into three categories:
②0＜αic, S set, boundary support vector set;
③αiset C, E, set error support vectors.
Coefficient alpha of edge support vector at each step incrementiE, synchronously changing with the threshold b, keeping balance of all elements in the current data set and synchronously updating the decision boundary under the condition that the KKT condition is always met. After calculation by the SMO algorithm, the decision boundary is updated, and the error rate is improved from 3.33% to 2.33%. Specifically, fig. 3 is a schematic diagram of a decision boundary corresponding to an incremental learning pre-flutter detection model, and fig. 4 is a schematic diagram of a decision boundary corresponding to an incremental learning post-flutter detection model. The abscissa and ordinate of fig. 3 and 4 represent wavelet entropy features extracted from two different acceleration sensor signals, respectively. "+" represents a sample with surface vibration pattern defect in the test set, and "o" represents a sample without surface vibration pattern defect in the test set, wherein the test set is a wavelet entropy characteristic sample set for testing the precision of the detection model; the curve line in the graph is a surface vibration line defect judgment decision boundary obtained by learning according to a support vector machine model. It can be known from fig. 3 and 4 that the updated boundary can be used to make a more accurate judgment on whether the defect exists or not by using the incremental learning.
In conclusion, the method for detecting the surface chatter mark defects of the machined part in the embodiment of the invention develops research around chatter stability in the milling process, establishes a motion equation of a milling system to obtain an acceleration response signal when determining a chatter detection model, extracts a vibration signal and then performs corresponding chatter marking; establishing a flutter detection model (a kernel support vector machine model), and solving an optimal decision boundary corresponding to the flutter detection model and used for distinguishing whether the flutter detection model is in a flutter state by adopting an SMO algorithm; collecting vibration signals in the actual processing process to carry out corresponding flutter marking; and gradually adding the measured vibration signals into the learning process of the kernel support vector machine for judging the boundary by adopting an incremental learning mode, and updating the boundary on the basis of the original boundary. And after each part is machined, judging whether incremental learning is carried out or not according to the accuracy of the current chatter detection model, if the accuracy of the current chatter detection model is smaller than a preset precision threshold, marking the chatter state of the machining system corresponding to the vibration characteristics of the part in the whole machining process according to the standard chatter texture quality inspection result of the part, training the chatter detection model in an incremental mode, and updating the chatter detection model. The method continuously adopts an incremental learning mode on the basis of the existing flutter detection model, and gradually eliminates some information which can generate adverse effect on the judgment precision by utilizing continuously accumulated actually-measured vibration information so as to improve the accuracy of the flutter detection model. The method solves the problem that the simulation vibration result deviates from an actual value due to insufficient model precision in the prior art, and obtains the method for improving the detection precision of the vibration line defect of the machined surface by correcting the judgment condition of the flutter state by using the actually measured vibration information in an incremental learning mode.
Examples 2,
A computer-readable storage medium, which includes a stored computer program, wherein when the computer program is executed by a processor, the apparatus where the storage medium is located is controlled to execute the method for detecting the chatter mark defects on the surface of the machined part provided in embodiment 1 of the present invention. The related technical scheme is the same as embodiment 1, and is not described herein.
It will be understood by those skilled in the art that the foregoing is only a preferred embodiment of the present invention, and is not intended to limit the invention, and that any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (7)
1. A method for detecting the vibration line defects on the surface of a machined part is characterized by comprising the following steps:
s1, measuring acceleration response signals of the tool along the horizontal and vertical directions of the machine tool in the current machining process of the part, filtering off tool-pass harmonic components in the acceleration response signals, and extracting wavelet entropy characteristics reflecting machining instability strength in the acceleration response signals and recording the wavelet entropy characteristics as vibration characteristics;
s2, inputting the vibration characteristics into a pre-trained flutter detection model, and judging the flutter state of the weak rigidity processing system for processing the part at present; if the workpiece is in a flutter state, the current machining surface of the part has a chatter mark defect;
wherein the flutter detection model is a machine learning model; counting the vibration line detection results of all processing surfaces obtained in the whole processing process of a part to obtain the vibration line detection result of the surface of the part every time the part is processed, and comparing the vibration line detection result of the surface of the part with a standard vibration line quality detection result to obtain vibration line defect detection precision, namely the accuracy of a vibration detection model; if the accuracy of the flutter detection model is smaller than a preset precision threshold value, marking the flutter state of the processing system corresponding to the vibration characteristic of the part in the whole processing process according to the standard chatter texture quality inspection result of the part, and training the flutter detection model in an incremental mode by taking the vibration characteristic of the part in the whole processing process as input and the corresponding mark as output, and updating the flutter detection model.
2. The method for detecting the chatter mark defects on the surface of the machined part as claimed in claim 1, wherein the training method of the chatter detection model comprises:
s01, establishing a kinetic equation of the weak rigidity processing system, extracting vibration characteristics under different cutting parameters from acceleration response signals of the tool along the horizontal and vertical directions of the machine tool under different cutting parameters in the processing system according to the kinetic equation, and respectively marking the vibration states corresponding to the vibration characteristics;
s02, taking the vibration characteristics under the different cutting parameters as input, and correspondingly marking as output training machine learning models to obtain flutter detection models;
s03, acquiring acceleration response signals of the tool along the horizontal and vertical directions of the machine tool in the actual machining process under different cutting parameters, extracting actual values of the vibration characteristics under different cutting parameters, and respectively labeling the flutter states corresponding to the actual values of the vibration characteristics;
and S04, training the flutter detection model in an incremental mode by taking the actual values of the vibration characteristics under different cutting parameters as input and taking the corresponding labels as output, and updating the flutter detection model.
3. The method for detecting the chatter mark defects on the surface of a machined part as claimed in claim 2, wherein said step S01 comprises the steps of:
s011, establishing a dynamic equation of the weak rigidity processing system, and identifying a cutting force coefficient and modal parameters in the dynamic equation by combining full-tooth milling and modal test;
and S012, solving the kinetic equation by adopting a Runge-Kutta method based on the cutting force coefficient and the modal parameter to obtain acceleration response signals of the cutter along the horizontal and vertical directions of the machine tool under different cutting parameters, filtering cutter-pass harmonic components in each acceleration response signal, and extracting wavelet entropy characteristics reflecting processing instability strength in each acceleration response signal, namely vibration characteristics under different cutting parameters.
4. The method for detecting the surface moire defect of the machined part as claimed in claim 3, wherein a notch filtering method is adopted to filter out the knife-pass harmonic component which is irrelevant to the surface moire defect in each acceleration response signal.
5. The method for detecting the surface chatter mark defects of a machined part according to any one of claims 1-4, wherein the chatter detection model is a kernel support vector machine model.
6. The method for detecting the chatter mark defects on the surface of the machined part as claimed in claim 5, wherein in the training process, an SMO algorithm is adopted to solve an optimal decision boundary corresponding to the chatter detection model and used for distinguishing whether the chatter state exists or not.
7. A computer-readable storage medium, comprising a stored computer program, wherein when the computer program is executed by a processor, the computer program controls an apparatus on which the storage medium is located to execute the method for detecting the chatter mark defects on the surface of a machined part according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011310382.XA CN112395809B (en) | 2020-11-20 | 2020-11-20 | Method for detecting vibration pattern defect on surface of machined part |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011310382.XA CN112395809B (en) | 2020-11-20 | 2020-11-20 | Method for detecting vibration pattern defect on surface of machined part |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112395809A true CN112395809A (en) | 2021-02-23 |
| CN112395809B CN112395809B (en) | 2023-12-19 |
Family
ID=74606838
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011310382.XA Active CN112395809B (en) | 2020-11-20 | 2020-11-20 | Method for detecting vibration pattern defect on surface of machined part |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112395809B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113065277A (en) * | 2021-03-11 | 2021-07-02 | 自然资源部国土卫星遥感应用中心 | Method for high-resolution remote sensing satellite flutter detection and modeling in cooperation with multi-load data |
| CN113094925A (en) * | 2021-04-23 | 2021-07-09 | 湖南工业大学 | Milling chatter stability prediction method |
| CN114918679A (en) * | 2022-06-24 | 2022-08-19 | 沈阳建筑大学 | Continuous machining and heat treatment production line and data acquisition and processing system |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104786101A (en) * | 2015-04-29 | 2015-07-22 | 常州信息职业技术学院 | Monitoring method for vertical milling cutting vibration |
| CN105108584A (en) * | 2015-07-21 | 2015-12-02 | 上海交通大学 | Turning chatter detection method |
| WO2018185993A1 (en) * | 2017-04-04 | 2018-10-11 | Dmg森精機株式会社 | Main-shaft speed control device |
| CN110434676A (en) * | 2019-07-29 | 2019-11-12 | 北京理工大学 | A kind of boring monitoring chatter method of multisensor time-frequency characteristics fusion |
| JP2020023039A (en) * | 2018-07-25 | 2020-02-13 | 株式会社ジェイテクト | Grinding quality estimation model generation device, grinding quality estimation device, operation command data adjustment model generation device of grinding machine and operation command data updating device of grinding machine |
-
2020
- 2020-11-20 CN CN202011310382.XA patent/CN112395809B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104786101A (en) * | 2015-04-29 | 2015-07-22 | 常州信息职业技术学院 | Monitoring method for vertical milling cutting vibration |
| CN105108584A (en) * | 2015-07-21 | 2015-12-02 | 上海交通大学 | Turning chatter detection method |
| WO2018185993A1 (en) * | 2017-04-04 | 2018-10-11 | Dmg森精機株式会社 | Main-shaft speed control device |
| JP2020023039A (en) * | 2018-07-25 | 2020-02-13 | 株式会社ジェイテクト | Grinding quality estimation model generation device, grinding quality estimation device, operation command data adjustment model generation device of grinding machine and operation command data updating device of grinding machine |
| CN110434676A (en) * | 2019-07-29 | 2019-11-12 | 北京理工大学 | A kind of boring monitoring chatter method of multisensor time-frequency characteristics fusion |
Non-Patent Citations (1)
| Title |
|---|
| 代艳霞;: "铝合金薄壁件的铣削加工颤振的建模与识别", 机械设计与制造, no. 05, pages 109 - 112 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113065277A (en) * | 2021-03-11 | 2021-07-02 | 自然资源部国土卫星遥感应用中心 | Method for high-resolution remote sensing satellite flutter detection and modeling in cooperation with multi-load data |
| CN113065277B (en) * | 2021-03-11 | 2021-10-08 | 自然资源部国土卫星遥感应用中心 | High-resolution remote sensing satellite flutter detection and modeling method in cooperation with multi-load data |
| CN113094925A (en) * | 2021-04-23 | 2021-07-09 | 湖南工业大学 | Milling chatter stability prediction method |
| CN114918679A (en) * | 2022-06-24 | 2022-08-19 | 沈阳建筑大学 | Continuous machining and heat treatment production line and data acquisition and processing system |
| CN114918679B (en) * | 2022-06-24 | 2024-04-12 | 沈阳建筑大学 | Continuous machining and heat treatment production line and data acquisition and processing system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112395809B (en) | 2023-12-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112395809B (en) | Method for detecting vibration pattern defect on surface of machined part | |
| US9508042B2 (en) | Method for predicting machining quality of machine tool | |
| Shi et al. | A reinforced k-nearest neighbors method with application to chatter identification in high-speed milling | |
| US20180307203A1 (en) | Machining defect factor estimation device | |
| CN112765778B (en) | Bogie lateral stability identification method and device and computer equipment | |
| CN104898565A (en) | Improved database for chatter predictions | |
| CN110928237B (en) | Vibration signal-based numerical control machining center flutter online identification method | |
| CN110889091B (en) | Machine tool thermal error prediction method and system based on temperature sensitive interval segmentation | |
| CN112405072B (en) | On-line monitoring method and device for cutting chatter of machine tool | |
| CN113487533B (en) | Part assembly quality digital detection system and method based on machine learning | |
| CN115311629B (en) | Abnormal bending precision monitoring system of bending machine | |
| CN110434676A (en) | A kind of boring monitoring chatter method of multisensor time-frequency characteristics fusion | |
| CN104657526A (en) | Method for predicting machining quality of machine tool | |
| CN115019158A (en) | Image recognition-based marine pollution area recognition method and system | |
| CN113554621A (en) | Tool wear state identification system and method based on wavelet scale map and depth migration learning | |
| CN116186499A (en) | Numerical control machining surface roughness prediction method | |
| CN111062093B (en) | Automobile tire service life prediction method based on image processing and machine learning | |
| CN113314142B (en) | Engine fault detection method, system, storage medium and equipment | |
| CN117036358B (en) | Method and system for detecting tool wear of numerical control machine tool | |
| CN116307938B (en) | Health state assessment method for feeding system of machining center | |
| CN112946072A (en) | Abrasive belt wear state monitoring method based on machine learning | |
| CN116167983A (en) | Rail web alignment method, system and terminal based on minimum DTW distance | |
| CN112419304B (en) | Multi-stage target detection method and device for one-dimensional data | |
| CN112529099A (en) | Robot milling chatter identification method | |
| CN110015319B (en) | Track corrugation identification method, device, equipment and storage medium |
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 |