CN109590805A - A kind of determination method and system of turning cutting tool working condition - Google Patents
A kind of determination method and system of turning cutting tool working condition Download PDFInfo
- Publication number
- CN109590805A CN109590805A CN201811541220.XA CN201811541220A CN109590805A CN 109590805 A CN109590805 A CN 109590805A CN 201811541220 A CN201811541220 A CN 201811541220A CN 109590805 A CN109590805 A CN 109590805A
- Authority
- CN
- China
- Prior art keywords
- data
- frequency
- amplitude
- determining
- neural network
- 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
- 238000007514 turning Methods 0.000 title claims abstract description 122
- 238000000034 method Methods 0.000 title claims abstract description 48
- 238000005520 cutting process Methods 0.000 title abstract description 17
- 238000012545 processing Methods 0.000 claims abstract description 68
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 65
- 238000000513 principal component analysis Methods 0.000 claims abstract description 21
- 238000010606 normalization Methods 0.000 claims description 73
- 238000012549 training Methods 0.000 claims description 47
- 239000011159 matrix material Substances 0.000 claims description 17
- 238000012847 principal component analysis method Methods 0.000 claims description 9
- 238000013528 artificial neural network Methods 0.000 claims description 8
- 238000012216 screening Methods 0.000 claims description 8
- 238000003062 neural network model Methods 0.000 claims description 5
- 238000012886 linear function Methods 0.000 claims description 4
- 230000008569 process Effects 0.000 abstract description 7
- 230000009466 transformation Effects 0.000 abstract description 2
- 238000012544 monitoring process Methods 0.000 abstract 2
- 230000005856 abnormality Effects 0.000 abstract 1
- 239000012141 concentrate Substances 0.000 abstract 1
- 238000005070 sampling Methods 0.000 description 14
- 238000010586 diagram Methods 0.000 description 7
- 230000002159 abnormal effect Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 238000003754 machining Methods 0.000 description 5
- 239000000463 material Substances 0.000 description 5
- 230000006870 function Effects 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 210000002569 neuron Anatomy 0.000 description 3
- 238000011176 pooling Methods 0.000 description 3
- 238000006073 displacement reaction Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000005284 excitation Effects 0.000 description 2
- 238000001228 spectrum Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 150000001875 compounds Chemical class 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23Q—DETAILS, COMPONENTS, OR ACCESSORIES FOR MACHINE TOOLS, e.g. ARRANGEMENTS FOR COPYING OR CONTROLLING; MACHINE TOOLS IN GENERAL CHARACTERISED BY THE CONSTRUCTION OF PARTICULAR DETAILS OR COMPONENTS; COMBINATIONS OR ASSOCIATIONS OF METAL-WORKING MACHINES, NOT DIRECTED TO A PARTICULAR RESULT
- B23Q17/00—Arrangements for observing, indicating or measuring on machine tools
- B23Q17/09—Arrangements for observing, indicating or measuring on machine tools for indicating or measuring cutting pressure or for determining cutting-tool condition, e.g. cutting ability, load on tool
- B23Q17/0952—Arrangements for observing, indicating or measuring on machine tools for indicating or measuring cutting pressure or for determining cutting-tool condition, e.g. cutting ability, load on tool during machining
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23Q—DETAILS, COMPONENTS, OR ACCESSORIES FOR MACHINE TOOLS, e.g. ARRANGEMENTS FOR COPYING OR CONTROLLING; MACHINE TOOLS IN GENERAL CHARACTERISED BY THE CONSTRUCTION OF PARTICULAR DETAILS OR COMPONENTS; COMBINATIONS OR ASSOCIATIONS OF METAL-WORKING MACHINES, NOT DIRECTED TO A PARTICULAR RESULT
- B23Q17/00—Arrangements for observing, indicating or measuring on machine tools
- B23Q17/09—Arrangements for observing, indicating or measuring on machine tools for indicating or measuring cutting pressure or for determining cutting-tool condition, e.g. cutting ability, load on tool
- B23Q17/0952—Arrangements for observing, indicating or measuring on machine tools for indicating or measuring cutting pressure or for determining cutting-tool condition, e.g. cutting ability, load on tool during machining
- B23Q17/0957—Detection of tool breakage
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B23—MACHINE TOOLS; METAL-WORKING NOT OTHERWISE PROVIDED FOR
- B23Q—DETAILS, COMPONENTS, OR ACCESSORIES FOR MACHINE TOOLS, e.g. ARRANGEMENTS FOR COPYING OR CONTROLLING; MACHINE TOOLS IN GENERAL CHARACTERISED BY THE CONSTRUCTION OF PARTICULAR DETAILS OR COMPONENTS; COMBINATIONS OR ASSOCIATIONS OF METAL-WORKING MACHINES, NOT DIRECTED TO A PARTICULAR RESULT
- B23Q17/00—Arrangements for observing, indicating or measuring on machine tools
- B23Q17/09—Arrangements for observing, indicating or measuring on machine tools for indicating or measuring cutting pressure or for determining cutting-tool condition, e.g. cutting ability, load on tool
- B23Q17/0952—Arrangements for observing, indicating or measuring on machine tools for indicating or measuring cutting pressure or for determining cutting-tool condition, e.g. cutting ability, load on tool during machining
- B23Q17/0985—Arrangements for observing, indicating or measuring on machine tools for indicating or measuring cutting pressure or for determining cutting-tool condition, e.g. cutting ability, load on tool during machining by measuring temperature
Landscapes
- Engineering & Computer Science (AREA)
- Mechanical Engineering (AREA)
- Image Analysis (AREA)
Abstract
The present invention discloses a kind of determination method and system of turning cutting tool working condition.The determination method and system of turning cutting tool working condition provided by the invention after the parameter of turning cutting tool working condition is carried out type mark, concentrate the data of each type to carry out Fourier transformation each parameter signal, obtain the frequency and amplitude of each data.Frequency and amplitude are normalized, its natural number less than or equal to 255 is made, corresponding pixel is then generated according to frequency, amplitude and data type.It is analyzed using the convolutional neural networks method based on principal component analysis, the working condition of real-time monitoring cutter, real-time monitoring can be carried out to Tutrning Process, finds abnormality when truning fixture work in time, provided safeguard to improve quality and the efficiency of ultrasonic wave turnery processing.
Description
Technical Field
The invention relates to the field of ultrasonic waves, in particular to a method and a system for determining the working state of a turning tool.
Background
With the development of scientific technology, precision and ultra-precision machining technology occupies an increasingly important position. The requirements of the mechanical manufacturing industries such as automobiles, ships, aerospace and the like on the processing quality and the processing precision of various top products and precision mechanical parts are higher and higher, and for some workpieces made of special materials and in complex shapes, the traditional processing method is difficult to adopt, so that the ultrasonic processing technology is developed unprecedentedly.
Ultrasonic vibration cutting is an important component of ultrasonic machining technology, and has superior machining performance compared with common cutting. The ultrasonic vibration turning ensures normal cutting processing of parts through a high-frequency and micro-amplitude processing mode, and deformation of workpiece materials cannot occur due to extremely short contact time. In the whole cutting process, the cutter and the workpiece material are in discontinuous contact, and the cutter and the workpiece are periodically separated through the mechanical and ultrasonic compound action of the cutter on the workpiece, so that the traditional cutting process is fundamentally changed. Due to the introduction of ultrasonic vibration, the material removal mechanism of common turning is changed, the material is removed mainly through the mechanical cutting action, the high-frequency micro-impact action and the like, and the processing effect which is difficult to achieve by common turning can be achieved. However, the existing detection technology cannot monitor the working state of the cutting tool in real time, and only can identify the cutting tool after the cutting tool is damaged or broken, so that the application range of the turning device is greatly limited.
Disclosure of Invention
The invention aims to provide a method and a system for determining the working state of a turning tool, which can monitor the working state of the tool in real time, can predict in advance and find the abnormal state existing during the working of a turning device in time, and thus effectively improve the quality and efficiency of ultrasonic turning.
In order to achieve the purpose, the invention provides the following scheme:
a method of determining an operating condition of a turning tool, the method comprising:
acquiring parameter data to be identified of a target turning tool, wherein the parameter data to be identified comprises temperature data at a tool edge, pressure data at the tool edge, tool edge position data and tool edge speed data;
inputting the parameter data to be identified into a convolutional neural network model to obtain the working state of the target turning tool, wherein the input of the convolutional neural network model is the parameter data to be identified of the turning device, and the output of the convolutional neural network model is the working state of the turning device; the method for establishing the convolutional neural network model specifically comprises the following steps:
acquiring a training sample data set; the training sample data set comprises: the parameter signal set of the turning device in the normal state and the calibrated normal state; the parameter signal set of the turning device fault state and the calibrated fault state; the method comprises the steps that a parameter signal set of a turning device critical fault state and a calibrated critical fault state are set, wherein the critical fault state is a state that the turning device is about to break down, and the parameter signal set comprises temperature data at a tool edge of the turning device, pressure data at the tool edge, tool edge position data and tool edge speed data in corresponding states;
marking a data type of temperature data in the parameter signal set as a natural number n less than or equal to 2551Marking the data type of the pressure data as a natural number n less than or equal to 2552Marking the data type of the knife edge position data as a natural number n less than or equal to 2553Marking the data type of the knife edge speed data as a natural number n less than or equal to 2554Wherein n is1≠n2≠n3≠n4；
Carrying out Fourier transform on each type of data in the parameter signal set to obtain the frequency and amplitude of each data;
normalizing the frequency and the amplitude of each datum to obtain the frequency and the amplitude after the normalization, wherein the frequency and the amplitude after the normalization are natural numbers less than or equal to 255;
determining a pixel point corresponding to each data according to the frequency after the normalization processing, the amplitude after the normalization processing and the data type;
selecting at least 256 pixel points from the pixel points corresponding to each data type in each state to generate an image matrix;
obtaining length k of convolution filter of initial convolutional neural network1And width k2；
Screening out all sizes k from the image matrix1×k2Forming a training data block;
performing principal component analysis on the training data block by using a principal component analysis method to obtain principal components of the training data block;
determining a filter bank of an initial convolutional neural network according to the principal component;
and generating a convolution neural network model for determining the working state of the turning tool according to the filter bank.
Optionally, the data type of the temperature data is marked as 0, the data type of the pressure data is marked as 1, the data type of the knife edge position data is marked as 2, and the data type of the knife edge speed data is marked as 3.
Optionally, the normalizing the frequency and the amplitude of each data to obtain the frequency and the amplitude after the normalizing, specifically includes:
according to the formula: f ═ Pmax-Pmin)×(fw-fmin)/(fmax-fmin)+PminNormalizing the frequency of each datum to obtain a normalized frequency, wherein f represents the normalized frequency, and P represents the normalized frequencymax＝255，Pmin＝0，fmaxRepresenting the maximum frequency value, fminRepresenting the minimum frequency value, fwRepresenting the frequency before normalization processing;
according to the formula: a ═ Pmax-Pmin)×(Ax-Amin)/(Axmax-Axmin)+PminNormalizing the amplitude of each datum to obtain a normalized amplitude, wherein A represents the normalized amplitude, and A ismaxDenotes the maximum amplitude, AminDenotes the minimum amplitude, AxTo representNormalizing the amplitude before processing.
Optionally, the determining a pixel point corresponding to each data according to the frequency after the normalization processing, the amplitude after the normalization processing, and the data type specifically includes:
determining the R component of a pixel point corresponding to each datum according to the frequency after normalization processing;
determining the G component of the pixel point according to the amplitude after the normalization processing;
and generating the B component of the pixel point corresponding to the data according to the data type corresponding to each data.
Optionally, after the input data block of the convolutional neural network model is convolved with the convolutional filter, the convolutional output of the convolutional layer is obtained by exciting through a nonlinear function.
A system for determining the working condition of a turning tool, said system comprising:
the data acquisition module is used for acquiring parameter data to be identified of a target turning tool, wherein the parameter data to be identified comprises temperature data at a tool edge, pressure data at the tool edge, tool edge position data and tool edge speed data;
the working state determining module is used for inputting the parameter data to be identified into a convolutional neural network model to obtain the working state of the target turning tool, the input of the convolutional neural network model is the parameter data to be identified of the turning device, the output of the convolutional neural network model is the working state of the turning device, and the convolutional neural network model establishing subsystem comprises:
the training sample acquisition module is used for acquiring a training sample data set; the training sample data set comprises: the parameter signal set of the turning device in the normal state and the calibrated normal state; the parameter signal set of the turning device fault state and the calibrated fault state; the method comprises the steps that a parameter signal set of a turning device critical fault state and a calibrated critical fault state are set, wherein the critical fault state is a state that the turning device is about to break down, and the parameter signal set comprises temperature data at a tool edge of the turning device, pressure data at the tool edge, tool edge position data and tool edge speed data in corresponding states;
a data type marking module for marking the data type of the temperature data in the parameter signal set as a natural number n less than or equal to 2551Marking the data type of the pressure data as a natural number n less than or equal to 2552Marking the data type of the knife edge position data as a natural number n less than or equal to 2553Marking the data type of the knife edge speed data as a natural number n less than or equal to 2554Wherein n is1≠n2≠n3≠n4；
The Fourier transform module is used for carrying out Fourier transform on each type of data in the parameter signal set to obtain the frequency and amplitude of each data;
the normalization processing module is used for performing normalization processing on the frequency and the amplitude of each datum to obtain the frequency and the amplitude after the normalization processing, and the frequency and the amplitude after the normalization processing are natural numbers less than or equal to 255;
the pixel point determining module is used for determining a pixel point corresponding to each datum according to the frequency after the normalization processing, the amplitude after the normalization processing and the data type;
the image matrix generation module is used for selecting at least 256 pixels from the pixel points corresponding to each data type in each state to generate an image matrix;
a convolution filter size obtaining module for obtaining the length k of the convolution filter of the initial convolution neural network1And width k2；
A screening module for screening out all the sizes k from the image matrix1×k2Image block ofForming a training data block;
the principal component analysis module is used for carrying out principal component analysis on the training data block by using a principal component analysis method to obtain principal components of the training data block;
the filter bank determining module is used for determining a filter bank of the initial convolutional neural network according to the principal component;
and the convolutional neural network model determining module is used for generating a convolutional neural network model for determining the working state of the turning tool according to the filter bank.
Optionally, the data type marking module includes:
the temperature data marking unit is used for marking the data type of the temperature data as 0;
a pressure data marking unit, wherein the data type of the pressure data is marked as 1;
the position data marking unit is used for marking the data type of the knife edge position data as 2;
and the speed data marking unit is used for marking the data type of the knife edge speed data as 3.
Optionally, the normalization processing module includes:
a frequency normalization processing unit for, according to the formula: f ═ Pmax-Pmin)×(fw-fmin)/(fmax-fmin)+PminNormalizing the frequency of each datum to obtain a normalized frequency, wherein f represents the normalized frequency, and P represents the normalized frequencymax＝255，Pmin＝0，fmaxRepresenting the maximum frequency value, fminRepresenting the minimum frequency value, fwRepresenting the frequency before normalization processing;
an amplitude normalization processing unit for, according to the formula: a ═ Pmax-Pmin)×(Ax-Amin)/(Axmax-Axmin)+PminNormalizing the amplitude of each datum to obtain a normalized amplitude, wherein A represents the normalized amplitude, and A ismaxDenotes the maximum amplitude, AminDenotes the minimum amplitude, AxRepresenting the amplitude value before normalization processing.
Optionally, the pixel point determining module includes:
the R component determining unit is used for determining the R component of the pixel point corresponding to each datum according to the frequency after the normalization processing;
the G component determining unit is used for determining the G component of the pixel point according to the amplitude after the normalization processing;
and the B component determining unit is used for generating the B component of the pixel point corresponding to the data according to the data type corresponding to each data.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
the invention provides a method and a system for determining the working state of a turning tool, which mark the data type of temperature data for identifying the working state of the turning tool as a natural number n less than or equal to 2551Labeling the data type of the pressure data as a natural number n less than or equal to 2552The data type of the knife edge position data is marked as a natural number n less than or equal to 2553Marking the data type of the knife edge speed data as a natural number n less than or equal to 2554Wherein n is1≠n2≠n3≠n4. And carrying out Fourier transform on each type of data in each parameter signal set to obtain the frequency and amplitude of each data. Normalizing the frequency and the amplitude of each datum to obtain the frequency and the amplitude after the normalization, wherein the frequency and the amplitude after the normalization are natural numbers less than or equal to 255; according to the frequency after the normalization processing and the amplitude after the normalization processingAnd determining the pixel point corresponding to each data according to the data type. The principal component of the training data block is obtained by using a principal component analysis method, a filter bank of an initial convolutional neural network is determined according to the principal component, a convolutional neural network model for determining the working state of the turning tool is generated according to the filter bank, the working state of the tool can be monitored in real time according to the working parameters of the cutting tool, the abnormal state existing during the working of the turning device can be predicted in advance and found in time, and therefore the quality and the efficiency of ultrasonic turning are effectively improved.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.
Fig. 1 is a flowchart of a method for determining a working state of a turning tool according to embodiment 1 of the present invention;
fig. 2 is a flowchart of a method for establishing the convolutional neural network model according to an embodiment of the present invention;
FIG. 3 is a block diagram of a system for determining the operating condition of a turning tool according to an embodiment of the present invention;
fig. 4 is a block diagram of a convolutional neural network model building subsystem according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention aims to provide a method and a system for determining the working state of a turning tool, which can monitor the working state of the tool in real time, can predict in advance and find the abnormal state existing during the working of a turning device in time, and thus effectively improve the quality and efficiency of ultrasonic turning.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
Fig. 1 is a flowchart of a method for determining the working condition of a turning tool according to embodiment 1 of the present invention. As shown in fig. 1, a method for determining the working condition of a turning tool, the method comprising:
step 101: acquiring parameter data to be identified of a target turning tool, wherein the parameter data to be identified comprises temperature data at a tool edge, pressure data at the tool edge, tool edge position data and tool edge speed data;
step 102: inputting the parameter data to be identified into a convolutional neural network model to obtain the working state of the target turning tool, wherein the input of the convolutional neural network model is the parameter data to be identified of the turning device, the output of the convolutional neural network model is the working state of the turning device, and the convolutional neural network model is established by adopting a principal component analysis method and a convolutional neural network.
Fig. 2 is a flowchart of a method for establishing the convolutional neural network model according to an embodiment of the present invention. As shown in fig. 2, the method for establishing the convolutional neural network model specifically includes:
step 201: acquiring a training sample data set; the training sample data set comprises: the parameter signal set of the turning device in the normal state and the calibrated normal state; the parameter signal set of the turning device fault state and the calibrated fault state; the method comprises the steps of calibrating a parameter signal set of a turning device critical fault state, wherein the critical fault state is a state that the turning device is about to fail, namely the turning device fails within a set time threshold value in the current working state, and the parameter signal set comprises temperature data of a tool edge of the turning device, pressure data of the tool edge, tool edge position data and tool edge speed data in corresponding states.
Step 202: the type of the marked data: marking a data type of temperature data in the parameter signal set as a natural number n less than or equal to 2551Marking the data type of the pressure data as a natural number n less than or equal to 2552Marking the data type of the knife edge position data as a natural number n less than or equal to 2553Marking the data type of the knife edge speed data as a natural number n less than or equal to 2554Wherein n is1≠n2≠n3≠n4. In this embodiment, the data type of the temperature data is marked as 0, the data type of the pressure data is marked as 1, the data type of the knife edge position data is marked as 2, and the data type of the knife edge velocity data is marked as 3.
Step 203: and carrying out Fourier transform on each type of data in the parameter signal set to obtain the frequency and amplitude of each data.
Step 204: and normalizing the frequency and the amplitude of each datum to obtain the frequency and the amplitude after the normalization, wherein the frequency and the amplitude after the normalization are both natural numbers less than or equal to 255.
The normalizing the frequency and the amplitude of each datum to obtain the frequency and the amplitude after the normalizing comprises:
according to the formula: f ═ Pmax-Pmin)×(fw-fmin)/(fmax-fmin)+PminNormalizing the frequency of each datum to obtain a normalized frequency, wherein f represents the normalized frequency, and P represents the normalized frequencymax＝255，Pmin＝0，fmaxRepresenting the maximum frequency value, fminRepresenting the minimum frequency value, fwRepresenting the frequency before normalization processing;
according to the formula: a ═ Pmax-Pmin)×(Ax-Amin)/(Axmax-Axmin)+PminNormalizing the amplitude of each datum to obtain a normalized amplitude, wherein A represents the normalized amplitude, and A ismaxDenotes the maximum amplitude, AminDenotes the minimum amplitude, AxRepresenting the amplitude value before normalization processing.
Step 205: and determining a pixel point corresponding to each datum according to the frequency after the normalization processing, the amplitude after the normalization processing and the data type.
The determining a pixel point corresponding to each data according to the frequency after the normalization processing, the amplitude after the normalization processing and the data type specifically includes:
determining the R component of a pixel point corresponding to each datum according to the frequency after normalization processing;
determining the G component of the pixel point according to the amplitude after the normalization processing;
and generating the B component of the pixel point corresponding to the data according to the data type corresponding to each data.
Step 206: at least 256 pixel points are selected from the pixel points corresponding to each data type in each state, and an image matrix is generated.
Step 207: obtaining length k of convolution filter of initial convolutional neural network1And width k2。
Step 208: fromScreening out all the sizes k in the image matrix1×k2Forming a training data block.
Step 209: and carrying out principal component analysis on the training data block by using a principal component analysis method to obtain a principal component of the training data block.
Step 210: and determining a filter bank of the initial convolutional neural network according to the principal component.
Step 211: and generating a convolution neural network model for determining the working state of the turning tool according to the filter bank. After the input data block of the convolutional neural network model is convolved with the convolutional filter, excitation is carried out through a nonlinear function, and convolutional output of the convolutional layer is obtained.
The implementation process of the method for determining the working state of the turning tool provided by the invention is as follows:
(1) and collecting various parameter indexes in the turning process through a sensor.
And acquiring various parameters to be identified through a sensor on the ultrasonic device. The temperature of the current state of the cutter edge is collected by a temperature sensor, the pressure of the current state of the air at the cutter edge in the machining process is collected by a pressure sensor, the real-time position data of the cutter edge is collected by a position sensor, and the real-time speed data of the cutter edge is collected by a speed sensor. And analyzing the working state of the cutting device in the current state according to the information returned by the sensor.
(2) And processing the acquired data:
the temperature, pressure, position and speed of different signals are numbered, and the number types are respectively 0,1, 2 and 3. Then, Fourier transform is carried out on each signal to obtain different frequency spectrums. Two parameters, the frequency and the corresponding amplitude at this frequency, are available in the spectrogram. Finally, the three parameters of frequency, amplitude and sampling type are mapped to R, G, B three parameters of a pixel. Thus, the working state of the ultrasonic device can be analyzed by utilizing the neural network algorithm.
The method comprises the following specific steps:
STEP1, and Fourier transform is carried out on the collected signals.
Assuming the collected temperature is t (t), the temperature signal is fourier transformed:
a corresponding spectrum is obtained.
Thus, it is possible to obtain a different frequency ω when the sampling type is constant at 0iAmplitude A ofi. By analogy, Fourier transformation of pressure, position and speed can be obtained.
STEP2, normalizing the sampled frequency and value.
Since the pixel parameters RGB all have values between 0 and 255, the frequency and amplitude are normalized to make better use of the neural network algorithm.
Ri＝(Rmax-Rmin)(fw-fmin)/(fmax-fmin)+Rmin
Wherein R ismin＝0，Rmax＝255，fmaxRepresenting the input maximum frequency value, fminRepresenting the minimum frequency value, fwFor the frequency of the input, RiThe value of the corresponding parameter R after normalization for a specific frequency.
Gi＝((Gmax-Gmin)(Ax-Amin)/(Amax-Amin)+Gmin)
Wherein G ismin＝0，Gmax＝255，AmaxRepresenting the input maximum amplitude, AminDenotes the minimum amplitude, AxTo be transportedAmplitude of incident, GiThe value of the corresponding parameter G after normalization for a particular amplitude.
STEP3, obtains data that can be input to the neural network.
The parameter B represents a different sample type. The current state is temperature sampling, and B is 0. Thus, a pixel point (R) is obtainedi,Gi0), the number of pixel points depends on the value range of the subscript i. Similarly, when the sampling types are pressure, displacement and speed, the pixel points (R) can be obtained respectivelyj,Gj,1)，(Rk,Gk,3)，(Rz,Gz,2)。
3. Establishing a convolution neural network model for determining the working state of the turning tool based on Principal Component Analysis (PCA) and a convolution neural network:
and initializing a convolutional neural network by utilizing principal component analysis unsupervised pre-training, obtaining hidden layer representation of the turning device to be identified by minimizing reconstruction errors, and further learning to obtain a filter set containing training data statistical characteristics. In order to improve the training speed, avoid trapping overfitting while ensuring the characteristic sparsity, the linear function of the convolutional neural network model is improved into a nonlinear correction function. In order to reduce the loss of the characteristic caused by down-sampling and enhance the robustness, a probability maximization sampling rule is introduced, and the characteristic is subjected to local contrast standardization after the convolution layer.
The method comprises the following specific steps:
STEP1, principal component analysis unsupervised training
A specific turning device corresponding to four different sampling types and respectively taking 256 pixel points (R)i,Gi,0),i＝0,1,...,255，(Rj,Gj,0),j＝0,1,...,255，(Rk,Gk0), k ═ 0,1,. multidot.255, and (R)z,Gz0), z 0, 1.., 255 make up a 32 × 32 block of data, assuming that the input training data of the convolutional neural network has N blocks of size mxn (m × N × 32)Wherein N is the number of turning devices, the data block is the parameter index collected by the sensor, and the size of the convolution filter is k1×k2For the ith turning device I of training dataiIn IiTake out all k1×k2Is represented asWherein xi,jFor turning devices IiVector representation of the jth data block in (1), then xi,jRemoving the mean value to obtain a turning device IiData block data ofSimilarly, the data block data of the training data is:
and (3) minimizing the reconstruction error by utilizing a principal component analysis method to solve a characteristic vector:
wherein F represents norm, ILIs an L × L identity matrix, V is a covariance matrix XXTThe first L eigenvectors of size (k)1k2) X1, then V represents the principal component of the data block input into the turning device, and the principal component analysis is used to perform unsupervised training on the data block X to obtain the weight of the initialized filter of the convolutional neural network and initialize the filter bank of the convolutional neural networkCan be expressed as:
wherein,represents the vectorMapping to a matrixql(XXT) Is shown by XXTL principal components of (1).
Principal components of the training data local data block are obtained through principal component analysis unsupervised training, the principal components can represent local features of the data to the maximum extent, and the principal information of the local features and the change and difference between the features can be well obtained as a filter.
STEP2, establishing convolution neural network model
The conventional convolutional neural network is composed of convolutional layers, downsampling layers and full-connection layers. And carrying out convolution on input data and the neurons in the convolution layer to obtain a plurality of characteristics, then carrying out blurring and generalization in the lower sampling layer, and finally outputting the characteristics for identifying the working state of the turning device through the full connection layer. The neural network is applied to the working state analysis of the turning process. After the input data block is convoluted with the filter, the input data block is excited by a nonlinear excitation function to obtain the output characteristic x of the l layer(1)。
Wherein ". sup." is a convolution operation,is the output of the jth neuron after the ith layer of convolution,the output of the ith neuron of the l-1 th layer, i.e. the input data of the l layer,in order to be a filter, the filter is,is the offset, f (●) is a non-linear function, as used herein, the non-linear function f (x) max (x).
In order to improve the invariance of the characteristics, increase the sparsity and the robustness of the model and improve the recognition rate, the local contrast standardization treatment is added after the convolution layer, and a characteristic diagram x is output(2)The formula of the local contrast normalization process is as follows:
wherein,features representing corresponding positions (i, j) of layer l-1The output value of the contrast normalization is calculated,mean and variance of the local domain N (i, j) are respectively represented.
In order to keep the characteristic invariance, the main characteristics of the image are kept, meanwhile, the parameters and the calculated amount are reduced, overfitting is prevented, the generalization capability of the model is improved, and a down-sampling layer which takes the maximum value as a sampling rule performs down-sampling on x(2)Performing a modulo and generalization to obtain x(3). In this embodiment, the pooling layer of the convolutional neural network adopts a maximum pooling mode shown in formula (6):
wherein imax，jmaxCorresponding respectively to the index of the element of the largest pooling kernel area,the corresponding is the maximum value of the pooled nuclear region.
After downsampling, the convolutional neural network model inputs all feature maps into a full connection layer (FC), and obtains features finally used for classification after passing through a hidden layer, and the features are directly input into a softmax classifier, so that classification and identification of targets can be performed. The operating state of the turning device can be analyzed in this way.
Fig. 3 is a block diagram of a system for determining an operating state of a turning tool according to an embodiment of the present invention. As shown in fig. 3, a system for determining the working condition of a turning tool, said system comprising:
the data acquisition module 301 is configured to acquire parameter data to be identified of a target turning tool, where the parameter data to be identified includes temperature data at a tool edge, pressure data at the tool edge, tool edge position data, and tool edge speed data.
And the working state determining module 302 is configured to input the parameter data to be identified into a convolutional neural network model to obtain a working state of the target turning tool, where the input of the convolutional neural network model is the parameter data to be identified of the turning device, and the output of the convolutional neural network model is the working state of the turning device.
Fig. 4 is a block diagram of a convolutional neural network model building subsystem according to an embodiment of the present invention. As shown in fig. 4, the convolutional neural network model building subsystem includes:
a training sample obtaining module 401, configured to obtain a training sample data set; the training sample data set comprises: the parameter signal set of the turning device in the normal state and the calibrated normal state; the parameter signal set of the turning device fault state and the calibrated fault state; the method comprises the steps of calibrating a critical fault state of the turning device, and obtaining a parameter signal set of the critical fault state of the turning device and a calibrated critical fault state, wherein the critical fault state is a state in which the turning device is about to fail, and the parameter signal set comprises temperature data at a tool edge of the turning device, pressure data at the tool edge, tool edge position data and tool edge speed data in corresponding states.
A data type marking module 402 for marking a data type of the temperature data in the parameter signal set as a natural number n less than or equal to 2551Marking the data type of the pressure data as a natural number n less than or equal to 2552Marking the data type of the knife edge position data as a natural number n less than or equal to 2553Marking the data type of the knife edge speed data as a natural number n less than or equal to 2554Wherein n is1≠n2≠n3≠n4。
The data type marking module 402 comprises:
the temperature data marking unit is used for marking the data type of the temperature data as 0;
a pressure data marking unit, wherein the data type of the pressure data is marked as 1;
the position data marking unit is used for marking the data type of the knife edge position data as 2;
and the speed data marking unit is used for marking the data type of the knife edge speed data as 3.
A fourier transform module 403, configured to perform fourier transform on each type of data in the parameter signal set, so as to obtain a frequency and an amplitude of each data.
A normalization processing module 404, configured to perform normalization processing on the frequency and the amplitude of each data to obtain a normalized frequency and a normalized amplitude, where the normalized frequency and the normalized amplitude are both natural numbers less than or equal to 255.
The normalization processing module 404 includes:
a frequency normalization processing unit for, according to the formula: f ═ Pmax-Pmin)×(fw-fmin)/(fmax-fmin)+PminNormalizing the frequency of each datum to obtain a normalized frequency, wherein f represents the normalized frequency, and P represents the normalized frequencymax＝255，Pmin＝0，fmaxRepresenting the maximum frequency value, fminRepresenting the minimum frequency value, fwRepresenting the frequency before normalization processing;
an amplitude normalization processing unit for, according to the formula: a ═ Pmax-Pmin)×(Ax-Amin)/(Axmax-Axmin)+PminNormalizing the amplitude of each datum to obtain a normalized amplitude, wherein A represents the normalized amplitude, and A ismaxDenotes the maximum amplitude, AminDenotes the minimum amplitude, AxRepresenting the amplitude value before normalization processing.
And a pixel point determining module 405, configured to determine a pixel point corresponding to each data according to the normalized frequency, the normalized amplitude, and the data type.
The pixel point determining module 405 includes:
the R component determining unit is used for determining the R component of the pixel point corresponding to each datum according to the frequency after the normalization processing;
the G component determining unit is used for determining the G component of the pixel point according to the amplitude after the normalization processing;
and the B component determining unit is used for generating the B component of the pixel point corresponding to the data according to the data type corresponding to each data.
The image matrix generating module 406 is configured to select at least 256 pixels from the pixel points corresponding to each data type in each state, and generate an image matrix.
A convolution filter size obtaining module 407 for obtaining the length k of the convolution filter of the initial convolutional neural network1And width k2。
A screening module 408 for screening out all sizes k from the image matrix1×k2Forming a training data block.
And a principal component analysis module 409, configured to perform principal component analysis on the training data block by using a principal component analysis method to obtain a principal component of the training data block.
A filter bank determining module 410, configured to determine a filter bank of the initial convolutional neural network according to the principal component.
And the convolutional neural network model determining module 411 is used for generating a convolutional neural network model for determining the working state of the turning tool according to the filter bank.
The working process of the system for determining the working state of the turning tool provided by the invention is as follows:
step1, collecting various parameter indexes in turning through a sensor.
Various parameter indexes are collected through a sensor on the ultrasonic device. The temperature of the current state near the knife edge is collected by a temperature sensor, the air pressure of the current state of the air near the knife edge in the machining process is collected by a pressure sensor, the real-time position of the knife edge is collected by a position sensor, and the real-time speed of the knife edge is collected by a speed sensor. And analyzing the working state of the cutting device in the current state according to the information returned by the sensor.
Step2, processing the collected data
The following procedure is exemplified for temperature acquisition:
STEP1, Fourier transform the collected temperature signal
STEP2, normalizing the sampled frequency and value.
Ri＝((Rmax-Rmin)(fw-fmin)/(fmax-fmin)+Rmin)
RiThe value of the corresponding parameter R after normalization for a specific frequency.
Gi＝((Gmax-Gmin)(Ax-Amin)/(Amax-Amin)+Gmin)
GiThe value of the corresponding parameter G after normalization for a particular amplitude.
STEP3, obtains data that can be input to the neural network.
The parameter B represents a different sample type. The current state is temperature sampling, and B is 0. Thus, a pixel point (R) is obtainedi,Gi0), the number of pixel points depends on the value range of the subscript i. Similarly, when the sampling types are pressure, displacement and speed, the pixel points (R) can be obtained respectivelyj,Gj,1)，(Rk,Gk,3)，(Rz,Gz,2)。
Step3, establishing a convolutional neural network model based on a Principal Component Analysis (PCA) and convolutional neural network method:
STEP1, cutting the data collected from the turning device in the training set to the same size k as the filter1×k2The data block of (1) constitutes data X according to the principle of equation (1);
STEP2, and obtaining the initial filter weight of the convolutional neural network from data X through unsupervised training by using principal component analysis according to the principles of formula (2) and formula (3).
STEP3, calculating the feature x of the convolution not linearized by equation (4)(1)。
STEP4, by equation (5) for x(1)Carrying out local contrast standardization and outputting a characteristic diagram x(2)。
STEP5 downsampling layer pair x downsampled for maximum value by sampling rule(2)Performing a modulo and generalization to obtain x(3)。
STEP6, map x(3)Repeating the steps 1-5 as input data, and outputting the characteristic diagram x(4)。
STEP7, map x(4)And combining the two into a column vector which is used as the input of a full connection layer, and outputting the identification result by using a softmax classifier.
STEP8, calculating the difference between the recognition result and the mark, and adjusting the update parameters from top to bottom through a back propagation algorithm
And STEP9, inputting data of the test turning device, and analyzing and identifying the working state of the test turning device by using the filter set obtained by training and the full-connection network weight parameters.
Because when the turning device is in abnormal working condition, can make the turning device local high temperature, the turning precision descends, and tool bit life reduces, reduces turning efficiency. Therefore, the maintenance of the normal working state is an important guarantee for reducing the turning force and improving the turning efficiency. The method comprises the steps of firstly collecting real-time data during ultrasonic turning, then processing the collected data, analyzing the collected data by using a Convolutional Neural Network (CNN) algorithm based on Principal Component Analysis (PCA), and judging whether the ultrasonic turning device is in a normal working state in real time, so that the turning process is monitored in real time, abnormal states of the turning device during working are found in time, and guarantee is provided for improving the quality and efficiency of ultrasonic turning.
The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.
The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.
Claims (9)
1. A method for determining the working condition of a turning tool, characterized in that the method for determining comprises:
acquiring parameter data to be identified of a target turning tool, wherein the parameter data to be identified comprises temperature data at a tool edge, pressure data at the tool edge, tool edge position data and tool edge speed data;
inputting the parameter data to be identified into a convolutional neural network model to obtain the working state of the target turning tool, wherein the input of the convolutional neural network model is the parameter data to be identified of the turning device, and the output of the convolutional neural network model is the working state of the turning device; the method for establishing the convolutional neural network model specifically comprises the following steps:
acquiring a training sample data set; the training sample data set comprises: the parameter signal set of the turning device in the normal state and the calibrated normal state; the parameter signal set of the turning device fault state and the calibrated fault state; the method comprises the steps that a parameter signal set of a turning device critical fault state and a calibrated critical fault state are set, wherein the critical fault state is a state that the turning device is about to break down, and the parameter signal set comprises temperature data at a tool edge of the turning device, pressure data at the tool edge, tool edge position data and tool edge speed data in corresponding states;
marking a data type of temperature data in the parameter signal set as a natural number n less than or equal to 2551Marking the data type of the pressure data as a natural number n less than or equal to 2552Marking the data type of the knife edge position data as a natural number n less than or equal to 2553Marking the data type of the knife edge speed data as a natural number n less than or equal to 2554Wherein n is1≠n2≠n3≠n4；
Carrying out Fourier transform on each type of data in the parameter signal set to obtain the frequency and amplitude of each data;
normalizing the frequency and the amplitude of each datum to obtain the frequency and the amplitude after the normalization, wherein the frequency and the amplitude after the normalization are natural numbers less than or equal to 255;
determining a pixel point corresponding to each data according to the frequency after the normalization processing, the amplitude after the normalization processing and the data type;
selecting at least 256 pixel points from the pixel points corresponding to each data type in each state to generate an image matrix;
obtaining length k of convolution filter of initial convolutional neural network1And width k2；
Screening out from the image matrixAll sizes are k1×k2Forming a training data block;
performing principal component analysis on the training data block by using a principal component analysis method to obtain principal components of the training data block;
determining a filter bank of an initial convolutional neural network according to the principal component;
and generating a convolution neural network model for determining the working state of the turning tool according to the filter bank.
2. The determination method according to claim 1, characterized in that the data type of the temperature data is marked as 0, the data type of the pressure data is marked as 1, the data type of the knife-edge position data is marked as 2, and the data type of the knife-edge velocity data is marked as 3.
3. The method according to claim 2, wherein the step of normalizing the frequency and the amplitude of each datum to obtain the frequency and the amplitude after normalization comprises:
according to the formula: f ═ Pmax-Pmin)×(fw-fmin)/(fmax-fmin)+PminNormalizing the frequency of each datum to obtain a normalized frequency, wherein f represents the normalized frequency, and P represents the normalized frequencymax＝255，Pmin＝0，fmaxRepresenting the maximum frequency value, fminRepresenting the minimum frequency value, fwRepresenting the frequency before normalization processing;
according to the formula: a ═ Pmax-Pmin)×(Ax-Amin)/(Axmax-Axmin)+PminNormalizing the amplitude of each datum to obtain a normalized amplitude, wherein A represents the normalized amplitude, and A ismaxDenotes the maximum amplitude, AminDenotes the minimum amplitude, AxRepresenting the amplitude value before normalization processing.
4. The method according to claim 1, wherein the determining a pixel point corresponding to each data according to the normalized frequency, the normalized amplitude, and the data type specifically includes:
determining the R component of a pixel point corresponding to each datum according to the frequency after normalization processing;
determining the G component of the pixel point according to the amplitude after the normalization processing;
and generating the B component of the pixel point corresponding to the data according to the data type corresponding to each data.
5. The method of claim 1, wherein the convolutional layer convolution output is obtained by convolving the input data block of the convolutional neural network model with a convolution filter and then exciting with a non-linear function.
6. A system for determining the working condition of a turning tool, said system comprising:
the data acquisition module is used for acquiring parameter data to be identified of a target turning tool, wherein the parameter data to be identified comprises temperature data at a tool edge, pressure data at the tool edge, tool edge position data and tool edge speed data;
the working state determining module is used for inputting the parameter data to be identified into a convolutional neural network model to obtain the working state of the target turning tool, the input of the convolutional neural network model is the parameter data to be identified of the turning device, and the output of the convolutional neural network model is the working state of the turning device; the convolutional neural network model building subsystem comprises:
the training sample acquisition module is used for acquiring a training sample data set; the training sample data set comprises: the parameter signal set of the turning device in the normal state and the calibrated normal state; the parameter signal set of the turning device fault state and the calibrated fault state; the method comprises the steps that a parameter signal set of a turning device critical fault state and a calibrated critical fault state are set, wherein the critical fault state is a state that the turning device is about to break down, and the parameter signal set comprises temperature data at a tool edge of the turning device, pressure data at the tool edge, tool edge position data and tool edge speed data in corresponding states;
a data type marking module for marking the data type of the temperature data in the parameter signal set as a natural number n less than or equal to 2551Marking the data type of the pressure data as a natural number n less than or equal to 2552Marking the data type of the knife edge position data as a natural number n less than or equal to 2553Marking the data type of the knife edge speed data as a natural number n less than or equal to 2554Wherein n is1≠n2≠n3≠n4；
The Fourier transform module is used for carrying out Fourier transform on each type of data in the parameter signal set to obtain the frequency and amplitude of each data;
the normalization processing module is used for performing normalization processing on the frequency and the amplitude of each datum to obtain the frequency and the amplitude after the normalization processing, and the frequency and the amplitude after the normalization processing are natural numbers less than or equal to 255;
the pixel point determining module is used for determining a pixel point corresponding to each datum according to the frequency after the normalization processing, the amplitude after the normalization processing and the data type;
the image matrix generation module is used for selecting at least 256 pixels from the pixel points corresponding to each data type in each state to generate an image matrix;
a convolution filter size obtaining module for obtaining the length k of the convolution filter of the initial convolution neural network1And width k2；
A screening module for screening out all the sizes k from the image matrix1×k2Forming a training data block;
the principal component analysis module is used for carrying out principal component analysis on the training data block by using a principal component analysis method to obtain principal components of the training data block;
the filter bank determining module is used for determining a filter bank of the initial convolutional neural network according to the principal component;
and the convolutional neural network model determining module is used for generating a convolutional neural network model for determining the working state of the turning tool according to the filter bank.
7. The determination system of claim 6, wherein the data type tagging module comprises:
the temperature data marking unit is used for marking the data type of the temperature data as 0;
a pressure data marking unit, wherein the data type of the pressure data is marked as 1;
the position data marking unit is used for marking the data type of the knife edge position data as 2;
and the speed data marking unit is used for marking the data type of the knife edge speed data as 3.
8. The determination system according to claim 7, wherein the normalization processing module comprises:
a frequency normalization processing unit for, according to the formula: f ═ Pmax-Pmin)×(fw-fmin)/(fmax-fmin)+PminNormalizing the frequency of each datum to obtain a normalized frequency, wherein f represents the normalized frequency, and P represents the normalized frequencymax＝255，Pmin＝0，fmaxRepresenting the maximum frequency value, fminRepresenting the minimum frequency value, fwRepresenting the frequency before normalization processing;
an amplitude normalization processing unit for, according to the formula: a ═ Pmax-Pmin)×(Ax-Amin)/(Axmax-Axmin)+PminNormalizing the amplitude of each data to obtainObtaining the normalized amplitude value, wherein A represents the normalized amplitude value, AmaxDenotes the maximum amplitude, AminDenotes the minimum amplitude, AxRepresenting the amplitude value before normalization processing.
9. The determination system of claim 6, wherein the pixel point determination module comprises:
the R component determining unit is used for determining the R component of the pixel point corresponding to each datum according to the frequency after the normalization processing;
the G component determining unit is used for determining the G component of the pixel point according to the amplitude after the normalization processing;
and the B component determining unit is used for generating the B component of the pixel point corresponding to the data according to the data type corresponding to each data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811541220.XA CN109590805B (en) | 2018-12-17 | 2018-12-17 | A kind of determination method and system of turning cutting tool working condition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811541220.XA CN109590805B (en) | 2018-12-17 | 2018-12-17 | A kind of determination method and system of turning cutting tool working condition |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109590805A true CN109590805A (en) | 2019-04-09 |
| CN109590805B CN109590805B (en) | 2019-11-29 |
Family
ID=65962650
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811541220.XA Active CN109590805B (en) | 2018-12-17 | 2018-12-17 | A kind of determination method and system of turning cutting tool working condition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109590805B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112556819A (en) * | 2020-09-18 | 2021-03-26 | 集美大学 | Detection apparatus for ultrasonic machining cutter |
| CN113601265A (en) * | 2021-09-01 | 2021-11-05 | 重庆科技学院 | Method for estimating energy consumption ratio of front tool face and rear tool face of cutter in machining |
| CN114091544A (en) * | 2021-11-29 | 2022-02-25 | 西南交通大学 | TBM disc cutter abrasion identification system based on vibration signal and neural network |
| CN114144280A (en) * | 2019-06-13 | 2022-03-04 | 德克尔马霍泽巴赫有限公司 | Control device for use on a numerically controlled machine tool, and machine tool comprising a control device |
| CN115837472A (en) * | 2023-02-20 | 2023-03-24 | 成都大金航太科技股份有限公司 | Annular end face deep and narrow groove chip breaking turning method |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1927830A2 (en) * | 2006-11-30 | 2008-06-04 | Matsushita Electric Works, Ltd. | Device for overall machine tool monitoring |
| CN102091972A (en) * | 2010-12-28 | 2011-06-15 | 华中科技大学 | Numerical control machine tool wear monitoring method |
| CN105196114A (en) * | 2015-11-05 | 2015-12-30 | 西安科技大学 | Real-time online tool wear monitoring method based on wavelet analysis and neural network |
| CN106002483A (en) * | 2016-05-04 | 2016-10-12 | 北京信息科技大学 | Intelligent tool fault diagnosis method |
| CN107194404A (en) * | 2017-04-13 | 2017-09-22 | 哈尔滨工程大学 | Submarine target feature extracting method based on convolutional neural networks |
| CN107584334A (en) * | 2017-08-25 | 2018-01-16 | 南京航空航天大学 | A kind of complex structural member numerical control machining cutter status real time monitor method based on deep learning |
| CN108319962A (en) * | 2018-01-29 | 2018-07-24 | 安徽大学 | A kind of Tool Wear Monitoring method based on convolutional neural networks |
| CN108942409A (en) * | 2018-08-26 | 2018-12-07 | 西北工业大学 | The modeling and monitoring method of tool abrasion based on residual error convolutional neural networks |
-
2018
- 2018-12-17 CN CN201811541220.XA patent/CN109590805B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1927830A2 (en) * | 2006-11-30 | 2008-06-04 | Matsushita Electric Works, Ltd. | Device for overall machine tool monitoring |
| CN102091972A (en) * | 2010-12-28 | 2011-06-15 | 华中科技大学 | Numerical control machine tool wear monitoring method |
| CN105196114A (en) * | 2015-11-05 | 2015-12-30 | 西安科技大学 | Real-time online tool wear monitoring method based on wavelet analysis and neural network |
| CN106002483A (en) * | 2016-05-04 | 2016-10-12 | 北京信息科技大学 | Intelligent tool fault diagnosis method |
| CN107194404A (en) * | 2017-04-13 | 2017-09-22 | 哈尔滨工程大学 | Submarine target feature extracting method based on convolutional neural networks |
| CN107584334A (en) * | 2017-08-25 | 2018-01-16 | 南京航空航天大学 | A kind of complex structural member numerical control machining cutter status real time monitor method based on deep learning |
| CN108319962A (en) * | 2018-01-29 | 2018-07-24 | 安徽大学 | A kind of Tool Wear Monitoring method based on convolutional neural networks |
| CN108942409A (en) * | 2018-08-26 | 2018-12-07 | 西北工业大学 | The modeling and monitoring method of tool abrasion based on residual error convolutional neural networks |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114144280A (en) * | 2019-06-13 | 2022-03-04 | 德克尔马霍泽巴赫有限公司 | Control device for use on a numerically controlled machine tool, and machine tool comprising a control device |
| CN112556819A (en) * | 2020-09-18 | 2021-03-26 | 集美大学 | Detection apparatus for ultrasonic machining cutter |
| CN112556819B (en) * | 2020-09-18 | 2024-03-01 | 集美大学 | Detection device for ultrasonic processing cutter |
| CN113601265A (en) * | 2021-09-01 | 2021-11-05 | 重庆科技学院 | Method for estimating energy consumption ratio of front tool face and rear tool face of cutter in machining |
| CN113601265B (en) * | 2021-09-01 | 2022-05-27 | 重庆科技学院 | Method for estimating energy consumption ratio of front tool face and rear tool face of cutter in machining |
| CN114091544A (en) * | 2021-11-29 | 2022-02-25 | 西南交通大学 | TBM disc cutter abrasion identification system based on vibration signal and neural network |
| CN115837472A (en) * | 2023-02-20 | 2023-03-24 | 成都大金航太科技股份有限公司 | Annular end face deep and narrow groove chip breaking turning method |
| CN115837472B (en) * | 2023-02-20 | 2023-06-09 | 成都大金航太科技股份有限公司 | Chip breaking turning method for deep and narrow groove of annular end face |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109590805B (en) | 2019-11-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109590805B (en) | A kind of determination method and system of turning cutting tool working condition | |
| Wang et al. | Data-driven prognostic method based on self-supervised learning approaches for fault detection | |
| EP3693823A1 (en) | Apparatus and method of detecting fault | |
| CN108178037A (en) | A kind of elevator faults recognition methods based on convolutional neural networks | |
| CN100573100C (en) | Method for Discriminating Gas-liquid Two Phase Flow based on digital image processing techniques | |
| CN111383227A (en) | Neural network-based tool wear type identification method and wear value determination method | |
| US11275975B2 (en) | Fault detection classification | |
| CN112733885A (en) | Point cloud identification model determining method and point cloud identification method and device | |
| You et al. | An efficient lightweight neural network using BiLSTM-SCN-CBAM with PCA-ICEEMDAN for diagnosing rolling bearing faults | |
| CN111860446A (en) | Detection system and method for unknown mode of satellite remote measurement time sequence data | |
| CN114714145A (en) | Method for enhancing, comparing, learning and monitoring tool wear state by using Gelam angular field | |
| CN113554621A (en) | Tool wear state identification system and method based on wavelet scale map and depth migration learning | |
| CN112149612A (en) | Marine organism recognition system and recognition method based on deep neural network | |
| Patil et al. | Health Monitoring of Milling Tool Inserts Using CNN Architectures Trained by Vibration Spectrograms. | |
| Zhang et al. | An image processing approach to machine fault diagnosis based on visual words representation | |
| Kong et al. | Semi-supervised classification of wafer map based on ladder network | |
| CN114707551A (en) | Multi-mode data fusion method for industrial internet equipment fault diagnosis | |
| CN115829942A (en) | Electronic circuit defect detection method based on non-negative constraint sparse self-encoder | |
| Fu et al. | Unsupervised rolling bearing fault diagnosis method across working conditions based on multiscale convolutional neural network | |
| CN117851813A (en) | Sensor abnormality detection method based on deep learning and principal component analysis | |
| CN111754497B (en) | Industrial defect detection method and system based on geometric algebra | |
| Amar Bouzid et al. | CNC milling cutters condition monitoring based on empirical wavelet packet decomposition | |
| CN116579985A (en) | Wafer defect detection method based on refindet model | |
| CN111578956A (en) | Visual SLAM positioning method based on deep learning | |
| CN115511063A (en) | Small sample structure fault diagnosis method based on Transformer neural network |
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 |