CN114235043A - Cylindrical grinding chatter recognition and tailstock center force online monitoring and measuring device and method - Google Patents

Cylindrical grinding chatter recognition and tailstock center force online monitoring and measuring device and method Download PDF

Info

Publication number
CN114235043A
CN114235043A CN202111525145.XA CN202111525145A CN114235043A CN 114235043 A CN114235043 A CN 114235043A CN 202111525145 A CN202111525145 A CN 202111525145A CN 114235043 A CN114235043 A CN 114235043A
Authority
CN
China
Prior art keywords
grinding
sensor
force
tailstock
signal
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.)
Pending
Application number
CN202111525145.XA
Other languages
Chinese (zh)
Inventor
迟玉伦
戴顺达
徐亮亮
文卓
严妍
陆金雷
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shanghai Machine Tool Factory Co Ltd
University of Shanghai for Science and Technology
Original Assignee
Shanghai Machine Tool Factory Co Ltd
University of Shanghai for Science and Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Shanghai Machine Tool Factory Co Ltd, University of Shanghai for Science and Technology filed Critical Shanghai Machine Tool Factory Co Ltd
Priority to CN202111525145.XA priority Critical patent/CN114235043A/en
Publication of CN114235043A publication Critical patent/CN114235043A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
    • G01D21/00—Measuring or testing not otherwise provided for
    • G01D21/02—Measuring two or more variables by means not covered by a single other subclass
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B24—GRINDING; POLISHING
    • B24B—MACHINES, DEVICES, OR PROCESSES FOR GRINDING OR POLISHING; DRESSING OR CONDITIONING OF ABRADING SURFACES; FEEDING OF GRINDING, POLISHING, OR LAPPING AGENTS
    • B24B49/00—Measuring or gauging equipment for controlling the feed movement of the grinding tool or work; Arrangements of indicating or measuring equipment, e.g. for indicating the start of the grinding operation
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B24—GRINDING; POLISHING
    • B24B—MACHINES, DEVICES, OR PROCESSES FOR GRINDING OR POLISHING; DRESSING OR CONDITIONING OF ABRADING SURFACES; FEEDING OF GRINDING, POLISHING, OR LAPPING AGENTS
    • B24B49/00—Measuring or gauging equipment for controlling the feed movement of the grinding tool or work; Arrangements of indicating or measuring equipment, e.g. for indicating the start of the grinding operation
    • B24B49/003—Measuring or gauging equipment for controlling the feed movement of the grinding tool or work; Arrangements of indicating or measuring equipment, e.g. for indicating the start of the grinding operation involving acoustic means
    • 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
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/243—Classification techniques relating to the number of classes
    • G06F18/2431—Multiple classes

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Mechanical Engineering (AREA)
  • Acoustics & Sound (AREA)
  • Constituent Portions Of Griding Lathes, Driving, Sensing And Control (AREA)

Abstract

The invention relates to a cylindrical grinding chatter recognition and tailstock center force on-line monitoring and measuring device and a method, wherein the device comprises a grinding machine tailstock center subjected to truncation, a three-way force sensor, an acoustic emission sensor, a three-way vibration sensor, a power sensor, a signal amplifier, an AD acquisition card and a computer; according to the method, wavelet packet entropy and empirical mode decomposition are adopted to carry out characteristic analysis on sensor signals, a flutter identification classification model of a support vector machine is utilized, grinding flutter is monitored on line by using the tip force of the tail frame of the grinding machine, online monitoring of the tip force of the tail frame of the grinding machine in the actual machining process is carried out, the relationship between the tip force of the tail frame and the grinding flutter is established, vibration is reduced, the grinding flutter is avoided, and the effect of improving the quality of a product machined by the machine tool is achieved.

Description

Cylindrical grinding chatter recognition and tailstock center force online monitoring and measuring device and method
Technical Field
The invention relates to an external grinding monitoring device and method, in particular to an external grinding chatter recognition and tailstock center force online monitoring and measuring device and method.
Background
The tailstock of the grinding machine is one of important parts of the grinding machine, and the grinding machine tailstock is matched with the grinding machine headstock to support and center a workpiece when the workpiece is ground by the grinding machine, as shown in fig. 1. Grinding is a process with the highest requirement on precision in machine tool machining, however, chattering is a key factor influencing machining quality in the grinding process, and deterioration of indexes such as workpiece form and position errors, dimension errors, surface roughness and waviness can be caused by chattering. In actual grinding, both forced vibration of machine tool machining caused by external factors and self-excited vibration of machining factors inside the machine tool can cause grinding chatter, and the magnitude of the center force of the grinding machine tailstock is one of main factors inducing self-excited vibration of the machine tool to cause chatter, so that the characteristic of improving the chatter resistance of the grinding machine tailstock is one of important measures for improving the high quality, high efficiency and high precision of the current machine tool, and therefore the chatter of the grinding machine tailstock in the actual grinding process needs to be accurately identified on line to analyze the chatter resistance of the grinding machine tailstock, and the proper center force of the tailstock is selected so as to reduce the self-excited vibration of machine tool machining, improve the product quality and prolong the service life of the machine tool. Based on the reasons, the grinding chatter online monitoring testing device and the research method for the tip force of the grinding machine tailstock based on the three-way force sensor, the vibration sensor, the power sensor and the acoustic emission sensor are provided, the tip force of the grinding machine tailstock in the actual machining process is monitored online, vibration is reduced, the chatter phenomenon is avoided, and the product quality processed by a machine tool is improved.
Disclosure of Invention
The invention provides an external circle grinding chatter recognition and tailstock center force online monitoring and measuring device and method based on a force sensor, a three-way vibration sensor, a power sensor and an acoustic emission sensor, which realize online monitoring of grinding machine tailstock center force in the actual machining process so as to analyze the relation between the grinding chatter phenomenon and the tailstock center force, reduce vibration, avoid the grinding chatter phenomenon and achieve the effect of improving the quality of products machined by a machine tool.
In order to achieve the purpose, the technical scheme of the invention is as follows: the utility model provides an online monitoring measuring device of cylindrical grinding shimmy discernment and tailstock center power, includes: the grinding machine tailstock center is subjected to truncation processing, and comprises a grinding machine tailstock center, a three-way force sensor, an acoustic emission sensor, a three-way vibration sensor, a power sensor, a signal amplifier, an AD acquisition card and a computer, wherein the three-way force sensor is connected with the center through a threaded hole and is used for acquiring stress conditions of the center in the X direction and the Y direction in the grinding processing process; the acoustic emission sensor is fixed on the top of the grinding machine by a magnetic gauge stand and is used for acquiring acoustic emission signals under various processing working conditions; the sensitivity of the three-way vibration sensor is fixed on the top of the grinding machine through a magnetic gauge stand at the bottom of the sensor and is used for acquiring vibration signals in the X, Y and Z directions at the top of a tailstock of the grinding machine; the power sensor is connected to a three-way voltage and current position of the machine tool electronic box input grinding wheel spindle motor and used for collecting and monitoring power signals under various working conditions, and the three-way force sensor, the acoustic emission sensor, the three-way vibration sensor and the power sensor respectively transmit collected state signals in the grinding measurement process to a computer through signal lines sequentially passing through a signal amplifier and an AD acquisition card.
Further, the sensitivity of the acoustic emission sensor is 1 dB.
Further, the sensitivity of the three-way vibration sensor is 100 mv/mm/s.
Further, the measuring range of the power sensor is 67.5Kw, and the signal output is 0-10 v.
An online monitoring and measuring method for cylindrical grinding chatter recognition and tailstock center force adopts an online monitoring and measuring device, and the method adopts wavelet packet entropy and empirical mode decomposition to carry out characteristic analysis on sensor signals, utilizes a chatter recognition classification model of a support vector machine to realize online monitoring of grinding chatter of the tailstock center force of a grinding machine, and online monitoring of the tailstock center force of the grinding machine in the actual machining process, establishes the relationship between the tailstock center force and the grinding chatter phenomenon, reduces vibration and avoids the grinding chatter phenomenon.
Furthermore, the selection range of the apex force is set before the wavelet packet entropy and the empirical mode decomposition.
Further, the wavelet packet entropy is used as a method for identifying grinding chatter, and the energy distribution of a decomposition layer is obtained by performing wavelet packet decomposition on an acoustic emission signal:
the wavelet packet coefficient of the jth band of the lth layer is defined as:
Figure BDA0003410041020000021
CJ, I is the wavelet envelope coefficient value of each point after signal decomposition, and the energy of each frequency band is:
Figure BDA0003410041020000022
the total energy of the frequency band is:
Figure BDA0003410041020000031
the L-th layer energy distribution vector T is:
Figure BDA0003410041020000032
combining wavelet packet decomposition coefficient energy distribution with information entropy, and defining the wavelet packet entropy of the L-th layer as follows:
Figure BDA0003410041020000033
meanwhile, the wavelet packet entropy of the J frequency band of the L layer is as follows:
Figure BDA0003410041020000034
for wavelet packet entropy fault information detection, both the wavelet packet entropy value at a single node and the decomposition layer entropy vector can be used as input type feature vectors, so that the entropy vector of the L-th layer of wavelet packet decomposition is as follows:
Figure BDA0003410041020000035
further, the empirical mode decomposition decomposes the acquired vibration signal into a sum of a limited number of intrinsic mode IMF functions, and the "screening" step of obtaining each IMF function is as follows:
(1) all local maximum points of the signal x (t) are determined, then all the local maximum points are connected by using three sample lines to form an upper envelope line, all the local minimum points are connected by using three sample lines to form a lower envelope line, and the two envelope lines envelop all signal data.
(2) The mean of the two envelopes is denoted μ1To find out
y1(t)＝x(t)-μ1
(3) Judgment of y1(t) whether or not IMF, if y1(t) if IMF condition is not satisfied, then y is1(t) repeating steps (1), (2) until y as raw data1(t) satisfies the IMF condition, when y is written1(t)＝c1(t), then c1(t) is the signal x1(t) a first IMF component representing the signal x1(t) the highest frequency component
(4) C is to1(t) from x1(t) separating to obtain a difference signal r with high frequency components removed1(t) is as follows:
r1(t)＝x(t)-c1(t)
will r is1(t) repeating steps (1), (2) and (3) as raw data to obtain a second IMF component c2(t) repeating n times to obtain n IMF components, thus:
Figure BDA0003410041020000041
when c is going ton(t) or rn(t) when a given termination condition is met, the end of the cycle is given by the equation:
Figure BDA0003410041020000042
in the formula, rn(t) is a residual function representing the average trend of the signal, and each IMF component c1(t), c2(t),…,cn(t) contain the components of the frequency bands varying from high to low, respectively.
Further, the support vector machine passes a given (x)i,yi) As training samples, where xi∈RnAs an input vector, yiThe epsilon R is used as an output vector, an input vector is mapped to a high-dimensional space for a multi-classification support vector machine through nonlinear transformation, an optimal decision surface is obtained in the high-dimensional space, accurate classification of samples is achieved, the interval is large, and in order to solve the decision surface in the high-dimensional space, a linear classification function is constructed:
g(x)＝ωTx+b
wherein w is a weight component; b is a classification threshold, and according to the structure risk minimization principle, the classification problem can be converted into the following optimization problem:
Figure BDA0003410041020000043
the constraint conditions are as follows:
Figure BDA0003410041020000044
the penalty factor C in the formula is used for adjusting the penalty degree of the misclassified sample 1; xi is the relaxation coefficient used to represent the number of misclassified samples; further translates into the dualization problem:
Figure BDA0003410041020000045
the constraint conditions are as follows:
Figure BDA0003410041020000051
an optimal decision function can thus be obtained:
Figure BDA0003410041020000052
the regression decision function of the support vector machine can accurately estimate corresponding output for new input outside a sample set according to the decision function, learning and training are carried out by processing acoustic emission sensor signals in grinding processing through wavelet packet entropy, decomposing three-way vibration sensor signals through empirical mode, taking signal values of a power sensor after time domain processing as input vectors of a chatter recognition model of the support vector machine, and dividing the output vectors into three grinding states of 'normal grinding', 'chatter induction' and 'chatter outbreak' to obtain corresponding recognition results of grinding chatter phenomena.
Further, establishing a relationship between tailstock center force and flutter: grinding a workpiece by using the grinding wheel, recording a sensor signal at the tip of the tailstock when the machine tool is in a grinding state by using computer software, and measuring the stress F of the tip by using a force sensor1，F1Is taken from the value of Fx～FySignal value A of acoustic emission sensor at the center of tail stock of grinding machine1Amplitude L of three directions X, Y and Z of the tailstock of the three-way vibration sensorx1,Ly1,Lz1Power sensor signal G1The set of settings 1,2, 3 represents three grinding results of { 'normal grinding', 'flutter inoculation', 'flutter explosion' }, and the grinding results are obtained:
the first set of input vectors:
{F1,Lx1,Ly1,Lz1,A1,G1}
the second set of input vectors:
{F2,Lx2,Ly2,Lz2,A2,G2}
the third set of input vectors:
{F3,Lx3,Ly3,Lz3,A3,G3}
input vector of K-th group:
{FK,Lxk,Lyk,Lzk,Ak,Gk}
Figure BDA0003410041020000061
and importing the input vector set into a flutter recognition model of a support vector machine for flutter recognition and classification to obtain the relationship between tailstock center force and grinding flutter.
The invention has the beneficial effects that:
the invention is based on the comprehensive application of a grinding machine tailstock center, a three-way force sensor, an acoustic emission sensor, a three-way vibration sensor and a power sensor. The grinding flutter online monitoring method for the grinding machine tailstock center force is provided by analyzing the characteristics of sensor signals through wavelet packet entropy and empirical mode decomposition and utilizing a support vector machine flutter identification classification model, so that the grinding flutter online monitoring method for the grinding machine tailstock center force is realized, the online monitoring of the grinding machine tailstock center force in the actual machining process is realized, the relation between the tailstock center force and the grinding flutter phenomenon is established, the vibration is reduced, the grinding flutter phenomenon is avoided, and the effect of improving the quality of products machined by a machine tool is achieved.
Drawings
FIG. 1 is a schematic view of a grinding machine;
FIG. 2 is a schematic view of a grinding machine tailstock center;
FIG. 3 is a schematic view of a grinding machine tailstock center truncation process;
wherein: (A) a truncated tailstock centre front view, (B) a truncated tailstock centre cross-sectional view;
FIG. 4 is a schematic view of a force sensor;
wherein: (A) a force sensor front view, (B) a force sensor top view;
FIG. 5 is a schematic view of a force sensor mated with a tailstock center;
FIG. 6 is a schematic view of the entire measuring apparatus;
FIG. 7 is a graph of tailstock center force magnitude versus chatter.
Detailed Description
The invention is further described with reference to the following figures and examples.
The invention discloses a grinding chatter online monitoring method for the tip force of a grinding machine tailstock based on a force sensor, a three-way vibration sensor, a power sensor and an acoustic emission sensor, wherein a measuring device is arranged on a grinding machine (shown in figures 1 and 6). One end of the upper surface of the grinding machine workbench 3 is provided with a headstock centre 1 through a headstock 2, and the other end of the upper surface of the grinding machine workbench is provided with a tailstock centre 6 through a tailstock shell 4 through a tailstock sleeve 5. The grinding wheel spindle motor 11 drives the grinding wheel 13 to rotate through the grinding wheel spindle 12 to grind the workpiece 14. The measuring device includes: the device comprises a grinding machine tailstock centre 6 subjected to truncation, a force sensor 7, an acoustic emission sensor 8, a three-way vibration sensor 9, a power sensor 10, a signal amplifier, an AD acquisition card and a computer. The dynamic signals of the sensor are transmitted to a computer through signal wires, a signal amplifier and an AD acquisition card in sequence in the process of measuring grinding, and the whole measuring device is shown in figure 6.
(1) Truncated tailstock centre
A grinding machine tailstock centre 6 as shown in figure 2. Length of centre L1Diameter of phi1. In order to accurately measure the stress conditions of the centers in the X direction and the Y direction in the grinding process, the center 6 of the tailstock is required to be arranged at
Figure BDA0003410041020000071
Cut off, see (A) in FIG. 3, and put four each on the two cut-off cross sectionsMKThe center of each threaded hole is d from the center of the cross section of the tip1The distance between two adjacent threaded holes is d2See (B) in fig. 3. Therefore, the three-way force sensor and the tip can be connected through the threaded hole, and the stress conditions of the tip in the X direction and the Y direction in the grinding process can be obtained.
(2) Force sensor and tailstock center matching relation
Force sensor 7, as shown in fig. 4. The diameter of the force sensor 7 is phi2Height is H1See (a) in fig. 4. The upper and lower sides of the force sensor are respectively provided with a square flange, as shown in figure 4 (B), the side length is a, and the height is H2And M is arranged at four corners of the square flangekThe center of each threaded hole is d from the intersection point of the square diagonal1The distance between two adjacent threaded holes is d2. The force sensor can measure the stress in X, Y and Z directions, and the maximum range is 1 KN. In order to research the relation between the tip force and the grinding chatter in the actual grinding process of the tail stock of the grinding machine, the corresponding force application range of the tip force at the tip is controlled to be F through a hydraulic systemx～FyFor this purpose, the force sensor 7 and the truncated tailstock center 6 need to be connected by a threaded hole, as shown in fig. 5.
(3) Acoustic emission sensor
The sensitivity of the acoustic emission sensor 8 is 1dB, and the acoustic emission sensor is used for measuring an acoustic emission signal at the top of the tailstock of the grinding machine. When in use, as shown in fig. 6, the acoustic emission sensor 8 is fixed on the grinding machine tailstock center 6 by a magnetic gauge stand to collect acoustic emission signals under various processing conditions, and the collected acoustic emission signals a are transmitted to a computer through a signal line, a signal amplifier and an AD acquisition card in sequence.
(4) Three-way vibration sensor
The sensitivity of the three-way vibration sensor 9 is 100mv/mm/s, and the three-way vibration sensor is used for measuring vibration signals in X, Y and Z directions at the top of the tailstock of the grinding machine. When the vibration sensor is used, as shown in fig. 6, the magnetic gauge stand at the bottom of the three-way vibration sensor 9 is fixed on the center 6 of the tail frame of the grinding machine to collect and monitor vibration signals under various machining working conditions, and the collected vibration signals in the X, Y and Z directions of the tail frame are transmittedNumber Lx1,Ly1,Lz1And the signals are transmitted to a computer through a signal wire, a signal amplifier and an AD acquisition card in sequence.
(5) Power sensor
The measuring range of the power sensor 10 is 67.5Kw, the signal output is 0-10v, and the power sensor is used for measuring a power signal when a workpiece is machined by the grinding machine. When in use, as shown in fig. 6, the power sensor 10 is connected to a three-way voltage and current input to the grinding wheel spindle motor 11 of the machine tool electronic box to collect and monitor power signals under various working conditions, and the collected grinding power signal G is transmitted to a computer through a signal line, a signal amplifier and an AD acquisition card in sequence.
The online monitoring grinding chatter method comprises the following steps:
setting of selection range of center force
When the grinding machine is used for grinding, the top force of the grinding machine tailstock is controlled to be F through controlling a hydraulic system at the excircle grinding machine tailstockx～FyWithin the range, set conditions per fnAnd (unit: N) obtaining the dynamic force value of the tip of the grinding machine tailstock and the signal parameter value of each sensor of the grinding machine tailstock in a changing way. The tip force is selected according to the following calculation formula:
Figure BDA0003410041020000081
F1＝Fx+fn (2)
F2＝Fx+2fn (3)
F3＝Fx+3fn (4)
FK＝Fx+Kfn (5)
therefore, the signal values of K groups of tailstock center forces and the values of other sensors are taken as input vectors of a grinding flutter model of the support vector machine in the grinding process.
Second, flutter on-line identification method
1) Wavelet packet entropy
Wavelet packet transformation is an effective method for identifying unstable phenomena such as flutter, and the like, can provide more fine decomposition for flutter monitoring signals, can decompose the signals into matched frequency bands in a self-adaptive manner according to a binary tree principle, and improves time-frequency resolution. The information entropy is a commonly used measurement method for uncertainty of complex signals, the larger the entropy is, the more unstable the information is, otherwise, the smaller the entropy is, the more stable the information is, and meanwhile, the flutter phenomenon is monitored online to obtain the complex uncertain signals, the information entropy theory and the wavelet packet decomposition theory are combined to form the wavelet packet entropy, and the unstable signals monitored under the fault working condition of grinding flutter can be effectively identified by combining the two methods, so that the wavelet packet entropy is used as an identification method of grinding flutter, and the wavelet packet decomposition is firstly carried out on acoustic emission signals to obtain the energy distribution of a decomposition layer:
the wavelet packet coefficient of the J-th frequency band of the L-th layer is defined as
Figure BDA0003410041020000091
CJ,IIs the wavelet packet coefficient value of each point after signal decomposition, and the energy of each frequency band is
Figure BDA0003410041020000092
Total energy of frequency band of
Figure BDA0003410041020000093
The L-layer energy distribution vector T is:
Figure BDA0003410041020000094
wavelet packet entropy provides a measure of the amount of fault information, a measure of the degree of sequence unknowns, and can be used to estimate the complexity of various signals, thereby combining wavelet packet decomposition coefficient energy distribution with information entropy to define the wavelet packet entropy of the L-th layer as
Figure BDA0003410041020000095
Meanwhile, the wavelet packet entropy of the J frequency band of the L layer is rho
Figure BDA0003410041020000096
For wavelet packet entropy fault information detection, both the wavelet packet entropy value at a single node and the decomposition layer entropy vector can be used as input type feature vectors.
The entropy vector h of the L-th layer of wavelet packet decomposition is:
Figure BDA0003410041020000101
2) empirical mode decomposition
The method comprises the steps of performing Empirical Mode Decomposition (EMD) on vibration signals acquired by a vibration sensor, wherein the decomposition principle is that the vibration signals are assumed to be composed of different Intrinsic Mode Functions (IMFs), the Intrinsic Mode Functions (IMFs) obtained by vibration signal decomposition can be linear or nonlinear, and IMF components must meet two conditions: the number of extreme points and the number of zero-crossing points are the same or have one difference at most, and the upper envelope line and the lower envelope line of the two-dimensional space phase-locked loop are locally symmetrical about a time axis. The vibration signal thus acquired can be decomposed into the sum of a finite number of Intrinsic Mode Functions (IMFs), and the "screening" steps to obtain each IMF function are as follows:
(1) all local maximum points of the signal x (t) are determined, then all the local maximum points are connected by using three sample lines to form an upper envelope line, all the local minimum points are connected by using three sample lines to form a lower envelope line, and the two envelope lines envelop all signal data.
(2) The mean of the two envelopes is denoted μ1To find out
y1(t)＝x(t)-μ1 (13)
(3) Judgment of y1(t) whether or not IMF, if y1(t) if IMF condition is not satisfied, then y is1(t) repeating steps (1), (2) until y as raw data1(t) satisfies the IMF condition, when y is written1(t)＝c1(t), then c1(t) is the signal x1(t) a first IMF component representing the signal x1(t) the highest frequency component
(4) C is to1(t) from x1(t) separating to obtain a difference signal r with high frequency components removed1(t) immediately have
r1(t)＝x(t)-c1(t) (14)
Will r is1(t) repeating steps (1), (2) and (3) as raw data to obtain a second IMF component c2And (t), repeating the steps for n times to obtain n IMF components. Thus, there is
Figure BDA0003410041020000111
When c is going ton(t) or rn(t) when a given termination condition is satisfied, the end of the cycle is obtained from the above equation
Figure BDA0003410041020000112
In the formula, rn(t) is a residual function representing the average trend of the signal. And each IMF component c1(t),c2(t),…,cn(t) contain the components of the frequency bands varying from high to low, respectively.
3) Support vector machine
By giving (x)i,yi) As training samples, where xi∈RnAs an input vector, yiAnd E.R is an output vector, the input vector can be mapped to a high-dimensional space through nonlinear transformation for a multi-classification support vector machine, and an optimal decision surface is obtained in the high-dimensional space, so that accurate classification of samples is realized and the interval is large. To solve a decision surface located in a high dimensional space, a linear classification function is constructed:
g(x)＝ωTx+b (17)
wherein w is a weight component; b is a classification threshold. According to the principle of minimizing the structural risk, the classification problem can be converted into the following optimization problem:
Figure BDA0003410041020000113
the constraint conditions are as follows:
Figure BDA0003410041020000114
the penalty factor C in the formula is used for adjusting the penalty degree of the misclassified sample 1; ξ is the relaxation coefficient used to represent the number of misclassified samples.
Further translates into the dualization problem:
Figure BDA0003410041020000115
the constraint conditions are as follows:
Figure BDA0003410041020000121
from this, an optimal decision function f (x):
Figure BDA0003410041020000122
the decision function of the regression of the support vector machine can accurately estimate corresponding output for new input outside a sample set according to the decision function, the signal value of an acoustic emission sensor after grinding is processed through wavelet entropy processing, the signal of a three-way vibration sensor is decomposed through empirical mode, the signal value of a power sensor after time domain processing is taken as the input vector of the support vector machine chatter recognition model for learning and training, and the output vector is divided into three grinding states of 'normal grinding', 'chatter induction' and 'chatter outbreak' to obtain the corresponding recognition result of the grinding chatter phenomenon.
Thirdly, establishing the relation between the tip force and the flutter of the tailstock
(1) The force sensor is connected with the center of the grinding machine tailstock subjected to the truncation processing through a threaded hole, and the force sensor, a signal amplifier, an AD acquisition card and a computer are sequentially connected through a signal line. The three-way vibration and acoustic emission sensor is respectively fixed near the tip of the tail frame of the grinding machine through a magnetic gauge stand, so that the vibration amplitude and acoustic emission signals at the tip of the tail frame are respectively measured when the grinding machine grinds, and the power sensor collects power signals during grinding. The signal line is connected with the corresponding sensor, the signal amplifier, the AD acquisition card and the computer. And debugging the whole acquisition test system to work normally.
(2) Grinding a workpiece by using the grinding wheel, recording a sensor signal at the tip of the tailstock when the machine tool is in a grinding state by using computer software, and measuring the stress F of the tip by using a force sensor1，F1Is taken from the value of Fx～FySignal value A of acoustic emission sensor at the center of tail stock of grinding machine1Amplitude L of three directions X, Y and Z of the tailstock of the three-way vibration sensorx1,Ly1,Lz1Power sensor signal G1The set {1, 2, 3} represents three grinding results of { 'normal grinding', 'flutter inoculation', 'flutter explosion' }. Can obtain
The first set of input vectors:
{F1,Lx1,Ly1,Lz1,A1,G1} (23)
the second set of input vectors:
{F2,Lx2,Ly2,Lz2,A2,G2} (24)
the third set of input vectors:
{F3,Lx3,Ly3,Lz3,A3,G3} (25)
input vector of K-th group:
{FK,Lxk,Lyk,Lzk,Ak,Gk} (26)
Figure BDA0003410041020000131
and importing the input vector set into a support vector machine flutter identification model for flutter identification and classification to obtain the relationship between tailstock center force and grinding flutter.
(3) As can be seen from FIG. 7, if the magnitude of the center force is Fc2～Fc3Within the range, when the result of chattering recognition is 1 for normal grinding, the center force of the grinding is within the normal acceptable range, if the magnitude of the center force is Fc1～Fc2Within the range, the result of the vibration identification is 2 at the vibration inoculation stage, a large amount of vibration is inoculated when the center force is selected to carry out grinding processing according to the range, the quality of the processed surface of a workpiece is reduced when the center force is adjusted to F when the center force is continuously processed, and the center force is required to be adjusted to Fc2～Fc3In the normal grinding tip force range, if the tip force is Fc3～Fc4Within the range, the result of the vibration identification is 3 at the stage of a large number of explosions of grinding vibration, and the problem of scrapping of the machined part and the like can be caused by selecting the region range according to the tip force (F)c1,Fc2,Fc3,Fc4∈Fx～Fy)。

Claims (10)

1. The utility model provides an online monitoring measuring device of cylindrical grinding shimmy discernment and tailstock center power which characterized in that includes: the grinding machine tailstock center is subjected to truncation processing, and comprises a grinding machine tailstock center, a three-way force sensor, an acoustic emission sensor, a three-way vibration sensor, a power sensor, a signal amplifier, an AD acquisition card and a computer, wherein the three-way force sensor is connected with the center through a threaded hole and is used for acquiring stress conditions of the center in the X direction and the Y direction in the grinding processing process; the acoustic emission sensor is fixed on the top of the grinding machine by a magnetic gauge stand and is used for acquiring acoustic emission signals under various processing working conditions; the sensitivity of the three-way vibration sensor is fixed on the top of the grinding machine through a magnetic gauge stand at the bottom of the sensor and is used for acquiring vibration signals in the X, Y and Z directions at the top of a tailstock of the grinding machine; the power sensor is connected to a three-way voltage and current input part of the grinding wheel spindle motor of the machine tool electronic box and used for collecting and monitoring power signals under various working conditions, and the three-way force sensor, the acoustic emission sensor, the three-way vibration sensor and the power sensor respectively transmit collected state signals in the grinding measurement process to a computer through signal lines sequentially passing through a signal amplifier and an AD acquisition card.
2. The cylindrical grinding chatter recognition and tailstock center force online monitoring and measuring device according to claim 1, characterized in that: the sensitivity of the acoustic emission sensor is 1 dB.
3. The cylindrical grinding chatter recognition and tailstock center force online monitoring and measuring device according to claim 1, characterized in that: the sensitivity of the three-way vibration sensor is 100 mv/mm/s.
4. The cylindrical grinding chatter recognition and tailstock center force online monitoring and measuring device according to claim 1, characterized in that: the measuring range of the power sensor is 67.5Kw, and the signal output is 0-10 v.
5. An external grinding chatter recognition and tailstock center force online monitoring method, which adopts the external grinding chatter recognition and tailstock center force online monitoring and measuring device of any one of claims 1-4, and is characterized in that: the method adopts wavelet packet entropy and empirical mode decomposition to carry out characteristic analysis on sensor signals, utilizes a flutter identification classification model of a support vector machine to realize online monitoring of grinding flutter of the tip force of the grinding machine tailstock, online monitoring of the tip force of the grinding machine tailstock in the actual processing process, establishes the relationship between the tip force of the tailstock and the grinding flutter phenomenon, reduces vibration and avoids the grinding flutter phenomenon.
6. The cylindrical grinding chatter recognition and tailstock center force online monitoring method according to claim 5, characterized in that: and setting a tip force selection range before wavelet packet entropy and empirical mode decomposition.
7. The cylindrical grinding chatter recognition and tailstock center force online monitoring method according to claim 5, characterized in that: the wavelet packet entropy is used as a method for identifying grinding chatter, and firstly, wavelet packet decomposition is carried out on an acoustic emission signal to obtain energy distribution of a decomposition layer:
the wavelet packet coefficient of the jth band of the lth layer is defined as:
Figure FDA0003410041010000021
CJ,Ithe wavelet packet coefficient value of each point after signal decomposition, and the energy of each frequency band is:
Figure FDA0003410041010000022
the total energy of the frequency band is:
Figure FDA0003410041010000023
the L-th layer energy distribution vector T is:
Figure FDA0003410041010000024
combining wavelet packet decomposition coefficient energy distribution with information entropy, and defining the wavelet packet entropy rho of the L-th layer as:
Figure FDA0003410041010000025
meanwhile, the wavelet packet entropy of the J frequency band of the L layer is as follows:
Figure FDA0003410041010000026
for wavelet packet entropy fault information detection, both the wavelet packet entropy value at a single node and the decomposition layer entropy vector can be used as input type feature vectors, so that the entropy vector h of the L-th layer of wavelet packet decomposition is as follows:
Figure FDA0003410041010000027
。
8. the cylindrical grinding chatter recognition and tailstock center force online monitoring method according to claim 5, characterized in that: the empirical mode decomposition decomposes the acquired vibration signal into the sum of a limited number of intrinsic mode IMF functions, and the steps of screening to obtain each IMF function are as follows:
(1) all local maximum points of the signal x (t) are determined, then all the local maximum points are connected by using three sample lines to form an upper envelope line, all the local minimum points are connected by using three sample lines to form a lower envelope line, and the two envelope lines envelop all signal data.
(2) The mean of the two envelopes is denoted μ1To find out
y1(t)＝x(t)-μ1
(3) Judgment of y1(t) whether or not IMF, if y1(t) if IMF condition is not satisfied, then y is1(t) repeating steps (1), (2) until y as raw data1(t) satisfies the IMF condition, when y is written1(t)＝c1(t), then c1(t) is the signal x1(t) a first IMF component representing the signal x1(t) highest frequencyComponent(s) of
(4) C is to1(t) from x1(t) separating to obtain a difference signal r with high frequency components removed1(t) is as follows:
r1(t)＝x(t)-c1(t)
will r is1(t) repeating steps (1), (2) and (3) as raw data to obtain a second IMF component c2(t) repeating n times to obtain n IMF components, thus:
Figure FDA0003410041010000031
when c is going ton(t) or rn(t) when a given termination condition is met, the end of the cycle is given by the equation:
Figure FDA0003410041010000032
in the formula, rn(t) is a residual function representing the average trend of the signal, and each IMF component c1(t),c2(t),…,cn(t) contain the components of the frequency bands varying from high to low, respectively.
9. The cylindrical grinding chatter recognition and tailstock center force online monitoring method according to claim 5, characterized in that: the support vector machine passes a given (x)i,yi) As training samples, where xi∈RnAs an input vector, yiThe epsilon R is used as an output vector, an input vector is mapped to a high-dimensional space for a multi-classification support vector machine through nonlinear transformation, an optimal decision surface is obtained in the high-dimensional space, accurate classification of samples is achieved, the interval is large, and in order to solve the decision surface in the high-dimensional space, a linear classification function is constructed:
g(x)＝ωTx+b
wherein w is a weight component; b is a classification threshold, and according to the structure risk minimization principle, the classification problem can be converted into the following optimization problem:
Figure FDA0003410041010000041
the constraint conditions are as follows:
Figure FDA0003410041010000042
the penalty factor C in the formula is used for adjusting the penalty degree of the misclassification sample; xi is the relaxation coefficient used to represent the number of misclassified samples;
further translates into the dualization problem:
Figure FDA0003410041010000043
the constraint conditions are as follows:
Figure FDA0003410041010000044
from this, an optimal decision function f (x):
Figure FDA0003410041010000045
the regression decision function of the support vector machine can accurately estimate corresponding output for new input outside a sample set according to the decision function, learning and training are carried out by processing acoustic emission sensor signals in grinding processing through wavelet packet entropy, decomposing three-way vibration sensor signals through empirical mode, taking signal values of a power sensor after time domain processing as input vectors of a chatter recognition model of the support vector machine, and dividing the output vectors into three grinding states of 'normal grinding', 'chatter induction' and 'chatter outbreak' to obtain corresponding recognition results of grinding chatter phenomena.
10. The cylindrical grinding chatter recognition and tailstock center force online monitoring method according to claim 5, characterized in that: establishing the relation between tailstock center force and flutter: grinding a workpiece by using the grinding wheel, recording a sensor signal at the tip of the tailstock when the machine tool is in a grinding state by using computer software, and measuring the stress F of the tip by using a force sensor1，F1Is taken from the value of Fx～FySignal value A of acoustic emission sensor at the center of tail stock of grinding machine1Amplitude L of three directions X, Y and Z of the tailstock of the three-way vibration sensorx1,Ly1,Lz1Power sensor signal G1And setting a set (1, 2, 3) to represent three grinding results of { 'normal grinding', 'flutter inoculation', 'flutter explosion' }, and obtaining:
the first set of input vectors:
{F1,Lx1,Ly1,Lz1,A1,G1}
the second set of input vectors:
{F2,Lx2,Ly2,Lz2,A2,G2}
the third set of input vectors:
{F3,Lx3,Ly3,Lz3,A3,G3}
input vector of K-th group:
{FK,Lxk,Lyk,Lzk,Ak,Gk}
Figure FDA0003410041010000051
and importing the input vector set into a flutter recognition model of a support vector machine for flutter recognition and classification to obtain the relationship between tailstock center force and grinding flutter.
CN202111525145.XA 2021-12-14 2021-12-14 Cylindrical grinding chatter recognition and tailstock center force online monitoring and measuring device and method Pending CN114235043A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111525145.XA CN114235043A (en) 2021-12-14 2021-12-14 Cylindrical grinding chatter recognition and tailstock center force online monitoring and measuring device and method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111525145.XA CN114235043A (en) 2021-12-14 2021-12-14 Cylindrical grinding chatter recognition and tailstock center force online monitoring and measuring device and method

Publications (1)

Publication Number Publication Date
CN114235043A true CN114235043A (en) 2022-03-25

Family

ID=80755866

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111525145.XA Pending CN114235043A (en) 2021-12-14 2021-12-14 Cylindrical grinding chatter recognition and tailstock center force online monitoring and measuring device and method

Country Status (1)

Country Link
CN (1) CN114235043A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115555920A (en) * 2022-10-12 2023-01-03 山东大学 Online flutter detection method and system based on adaptive variational modal decomposition

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103786069A (en) * 2014-01-24 2014-05-14 华中科技大学 Flutter online monitoring method for machining equipment
CN105930602A (en) * 2016-04-28 2016-09-07 上海交通大学 Optimal weighted wavelet package entropy-based chattering detection method
CN107330375A (en) * 2017-06-05 2017-11-07 天津大学 Lathe flutter on-line monitoring method based on empirical mode decomposition and Sample Entropy feature
CN110411719A (en) * 2019-07-05 2019-11-05 上海理工大学 Grinding machine tailstock dynamic stiffness measurement device and evaluation method
CN110605663A (en) * 2019-09-17 2019-12-24 上海理工大学 Cut-in grinding wheel passivation prediction method based on time constant
CN111694320A (en) * 2020-06-10 2020-09-22 大连理工大学 Machine tool self-adaptive control method considering flutter suppression
CN112733603A (en) * 2020-12-11 2021-04-30 江苏大学 Frequency conversion scroll compressor fault diagnosis method based on improved VMD and SVM

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103786069A (en) * 2014-01-24 2014-05-14 华中科技大学 Flutter online monitoring method for machining equipment
CN105930602A (en) * 2016-04-28 2016-09-07 上海交通大学 Optimal weighted wavelet package entropy-based chattering detection method
CN107330375A (en) * 2017-06-05 2017-11-07 天津大学 Lathe flutter on-line monitoring method based on empirical mode decomposition and Sample Entropy feature
CN110411719A (en) * 2019-07-05 2019-11-05 上海理工大学 Grinding machine tailstock dynamic stiffness measurement device and evaluation method
CN110605663A (en) * 2019-09-17 2019-12-24 上海理工大学 Cut-in grinding wheel passivation prediction method based on time constant
CN111694320A (en) * 2020-06-10 2020-09-22 大连理工大学 Machine tool self-adaptive control method considering flutter suppression
CN112733603A (en) * 2020-12-11 2021-04-30 江苏大学 Frequency conversion scroll compressor fault diagnosis method based on improved VMD and SVM

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
迟玉伦: "基于功率信号的切入式磨削工艺优化关键技术研究", 中国博士学位论文全文数据库 工程科技Ⅰ辑, no. 04, pages 022 - 7 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115555920A (en) * 2022-10-12 2023-01-03 山东大学 Online flutter detection method and system based on adaptive variational modal decomposition
CN115555920B (en) * 2022-10-12 2024-05-10 山东大学 Online chatter detection method and system based on adaptive variation modal decomposition

Similar Documents

Publication Publication Date Title
US11761930B2 (en) Prediction method of part surface roughness and tool wear based on multi-task learning
CN110355608B (en) Cutter wear amount prediction method based on self-attention mechanism and deep learning
CN111366123B (en) Part surface roughness and cutter wear prediction method based on multi-task learning
CN110647943B (en) Cutting tool wear monitoring method based on evolution data cluster analysis
CN111687689A (en) Cutter wear state prediction method and device based on LSTM and CNN
CN111716150B (en) Evolution learning method for intelligently monitoring cutter state
CN112247674B (en) Cutter wear prediction method
CN111037365A (en) Cutter state monitoring data set enhancing method based on generative countermeasure network
CN109158954A (en) A kind of ultrasonic cutting-tool wear state recognition methods and system based on acoustical signal processing
CN113378725A (en) Cutter fault diagnosis method, equipment and storage medium based on multi-scale-channel attention network
CN112528955B (en) High-frequency element machining size precision prediction method and system
CN112207631B (en) Method for generating tool detection model, method, system, device and medium for detecting tool detection model
CN112372371B (en) Method for evaluating abrasion state of numerical control machine tool cutter
CN109605128B (en) Milling chatter online detection method based on power spectrum entropy difference
CN110434676A (en) A kind of boring monitoring chatter method of multisensor time-frequency characteristics fusion
CN107350900A (en) A kind of tool condition monitoring method based on the extraction of chip breaking time
CN112247673B (en) Woodworker cutter wear state diagnosis method based on genetic BP neural network
CN116787225A (en) Rotary cutter abrasion monitoring method based on sound vibration signal characteristic fusion graph
CN114235043A (en) Cylindrical grinding chatter recognition and tailstock center force online monitoring and measuring device and method
Du et al. Intelligent turning tool monitoring with neural network adaptive learning
CN114354184A (en) Deep learning-based method and device for establishing health early warning model of main shaft of large-scale rotating equipment
CN114227382B (en) Cutter damage monitoring system and method based on novel capsule network
CN109434562A (en) Milling cutter state of wear recognition methods based on partition clustering
CN114378653B (en) Cylindrical grinding chatter online identification and monitoring method based on BP neural network
CN114888635B (en) Cutter state monitoring method

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