CN114571286A - Tool wear state monitoring method and system based on friction electrification principle - Google Patents
Tool wear state monitoring method and system based on friction electrification principle Download PDFInfo
- Publication number
- CN114571286A CN114571286A CN202210263842.0A CN202210263842A CN114571286A CN 114571286 A CN114571286 A CN 114571286A CN 202210263842 A CN202210263842 A CN 202210263842A CN 114571286 A CN114571286 A CN 114571286A
- Authority
- CN
- China
- Prior art keywords
- sensor
- cutter
- tool
- monitored
- wear
- 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
Images
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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Abstract
The invention relates to a method and a system for monitoring the wear state of a cutter based on the principle of triboelectrification, which relate to the field of monitoring the wear state of the cutter, and comprise the steps of determining a neural network model based on the acquired voltage signal of the monitored cutter and the wear state of the cutter, and determining the wear degree of the monitored cutter; the voltage signal of the monitored cutter comprises a voltage waveform diagram of the monitored cutter and a voltage waveform diagram of a machined workpiece, which are acquired by sensing equipment; the sensing equipment comprises a waveform generator, a first sensor and a second sensor, wherein the first sensor and the second sensor are connected with the input end of the waveform generator; the first sensor is arranged on a monitored cutter, the second sensor is arranged on a machined workpiece, and the first sensor and the second sensor are both manufactured based on a triboelectrification principle; the workpiece to be machined is a workpiece machined by the monitored tool. The invention has the characteristics of easy installation, low cost, strong applicability and high accuracy.
Description
Technical Field
The invention relates to the field of monitoring of a tool wear state, in particular to a tool wear state monitoring method and system based on a friction electrification principle.
Background
According to statistics, the faults caused by the tool problems account for 22.4% of the total faults, and workers in actual production judge the wear state of the tool by own experience, such as the number and the processing time of the processed workpieces, the surface quality and the processing noise condition of the processed workpieces, and the formation of chips. This presents some problems: the cutter is changed before the service life of the cutter is reached, so that the cost of the cutter is increased; the tool is worn seriously, so that the tool changing influences the processing precision of the processed workpiece, and even damages the processed workpiece and a machine tool. It is necessary to apply an intelligent tool wear state monitoring technology.
The current cutter wear state monitoring technology mainly comprises the following steps: tool wear state monitoring based on cutting force signals, tool wear state monitoring based on sound field signals, tool wear state monitoring based on vibration signals, and tool wear state monitoring based on machine vision. However, the machining process is an unstable nonlinear process, and the tool wear is greatly related to tool materials, processed workpiece materials, processing conditions and the like. The traditional sensor signal acquisition device is difficult to install on a processed workpiece or a workbench and has higher relative cost; and the machine vision method is difficult to be suitable for machining under complex cutting conditions. Therefore, a tool wear state monitoring method which is easy to install, low in cost, high in applicability and high in accuracy is urgently needed to be researched.
Disclosure of Invention
The invention aims to provide a tool wear state monitoring method and system based on a triboelectrification principle, and the method and system have the characteristics of easiness in installation, low cost, strong applicability and high accuracy.
In order to achieve the purpose, the invention provides the following scheme:
a tool wear state monitoring method based on a friction electrification principle comprises the following steps:
acquiring a voltage signal of a monitored cutter; the voltage signal of the monitored cutter comprises a voltage waveform diagram of the monitored cutter and a voltage waveform diagram of a processed workpiece, wherein the voltage waveform diagram of the monitored cutter is acquired by sensing equipment; the sensing equipment comprises a waveform generator, a first sensor and a second sensor, wherein the first sensor and the second sensor are connected with the input end of the waveform generator; the first sensor is arranged on the monitored cutter, the second sensor is arranged on the machined workpiece, and the first sensor and the second sensor are both manufactured on the basis of the principle of triboelectrification; the workpiece to be processed is a workpiece processed by the monitored cutter;
and determining a neural network model based on the voltage signal of the monitored tool and the tool wear state, and determining the wear degree of the monitored tool.
Optionally, the first sensor and the second sensor are both nano friction generator sensor modules; wherein the nano friction generator sensor module comprises at least one nano friction generator sensor.
Optionally, when the nano friction generator sensor module comprises a plurality of nano friction generator sensors with the same structure, the plurality of nano friction generator sensors are arranged in a stacked manner;
the nanometer friction generator sensor comprises a fixed end and an opening and closing end, and the opening and closing ends of the nanometer friction generator sensor face oppositely.
Optionally, the nano friction generator sensor includes a first substrate, a first electrode layer, a first friction layer, a second electrode layer, and a second substrate, which are sequentially stacked, and the first friction layer and the second friction layer can perform opening and closing operations under a micro vibration.
Optionally, the determining a neural network model based on the voltage signal of the monitored tool and the tool wear state, and determining the wear degree of the monitored tool specifically include:
carrying out noise reduction pretreatment on the voltage signal of the monitored cutter;
and inputting the voltage signal of the monitoring cutter subjected to noise reduction pretreatment into a cutter wear state determination neural network model so as to determine the wear degree of the monitored cutter.
Optionally, the tool wear state determining neural network model is determined based on a sample data set and a neural network; the sample data set comprises a plurality of sample tool voltage signals and the corresponding wear degree of each sample tool voltage signal; the sample cutter voltage signal comprises a voltage waveform diagram of a sample cutter and a voltage waveform diagram of a sample processing workpiece, wherein the voltage waveform diagram of the sample cutter is acquired by sensing equipment and subjected to noise reduction pretreatment; the sample processing workpiece is a workpiece processed by the sample cutter.
Optionally, the degree of wear is light wear, moderate wear, or extreme wear; the neural network comprises a feature extraction network layer, a feature fusion layer, a measurement function and a classification layer which are connected in sequence;
the feature extraction network layer is configured to:
respectively extracting the characteristic vector of the voltage waveform diagram of the monitored cutter subjected to noise reduction pretreatment and the characteristic vector of the voltage waveform diagram of the processed workpiece subjected to noise reduction pretreatment;
or respectively extracting the characteristic vector of the voltage waveform diagram of the sample cutter subjected to noise reduction pretreatment and the characteristic vector of the voltage waveform diagram of the sample processing workpiece subjected to noise reduction pretreatment;
the feature fusion layer is used for fusing the feature vectors output by the feature extraction network layer to obtain fusion vectors;
the measurement function layer is used for calculating the similarity between the fusion vector and different wear degree vectors respectively and determining a similarity value corresponding to each wear degree;
the classification layer is used for respectively calculating a first probability, a second probability and a third probability based on the similarity value corresponding to each wear degree, determining a maximum probability based on the first probability, the second probability and the third probability, and then determining the wear degree corresponding to the maximum probability as the wear degree of the monitored tool or the sample tool;
the first probability is the probability that the monitored tool or the sample tool is slightly worn; the second probability is the probability that the monitored tool or sample tool is of moderate wear; the third probability is the probability that the monitored tool or sample tool is of extreme wear.
Optionally, the feature extraction network layer includes a first convolution layer, a first pooling layer, a second convolution layer, a second pooling layer, a flat layer, and a full connection layer, which are connected in sequence.
A tool wear state monitoring system based on a triboelectric principle, comprising:
the data acquisition module is used for acquiring a voltage signal of the monitored cutter; the voltage signal of the monitored cutter comprises a voltage waveform diagram of the monitored cutter and a voltage waveform diagram of a processed workpiece, wherein the voltage waveform diagram of the monitored cutter is acquired by sensing equipment; the sensing equipment comprises a waveform generator, a first sensor and a second sensor, wherein the first sensor and the second sensor are connected with the input end of the waveform generator; the first sensor is arranged on the monitored cutter, the second sensor is arranged on the machined workpiece, and the first sensor and the second sensor are both manufactured on the basis of the principle of triboelectrification; the workpiece to be processed is a workpiece processed by the monitored cutter;
and the monitoring module is used for determining a neural network model based on the voltage signal of the monitored cutter and the cutter wear state, and determining the wear degree of the monitored cutter.
Optionally, the first sensor and the second sensor are both nano friction generator sensor modules; wherein the nano friction generator sensor module comprises at least one nano friction generator sensor;
when the nano friction generator sensor module comprises a plurality of nano friction generator sensors with the same structure, the plurality of nano friction generator sensors are arranged in a stacked mode;
the nanometer friction generator sensor comprises a fixed end and an opening and closing end, and the opening and closing ends of the nanometer friction generator sensor face oppositely.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
in consideration of the characteristics of high cost, difficult installation and the like of the traditional sensor, the invention uses the characteristic of the sensor manufactured based on the triboelectrification principle as a special sensor to acquire a wear signal, namely a voltage waveform diagram, of the monitored cutter; when the image features are automatically extracted, compared with the traditional feature extraction method, the convolutional neural network acquired by the method can play a good role in reducing data and accelerating the operation time of the algorithm, can completely express the features, and simultaneously neglects the influence of insignificant small features on the identification of the wear state of the cutter, and has the advantages of simple algorithm implementation, high identification rate of the wear state of the cutter, high operation efficiency and the like.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used 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 inventive exercise.
Fig. 1 is a schematic structural diagram of a nano friction generator sensor module according to an embodiment of the present invention;
FIG. 2 is a general schematic diagram of a tool wear state monitoring device based on a triboelectrification principle according to an embodiment of the present invention;
FIG. 3 is a flowchart of a tool wear status classification method according to an embodiment of the present invention;
FIG. 4 is a graph of tool wear categories provided by an embodiment of the present invention;
FIG. 5 is a block diagram of a convolutional neural network model according to an embodiment of the present invention;
FIG. 6 is a three-dimensional structure diagram of a convolutional neural network model according to an embodiment of the present invention;
FIG. 7 is a schematic flow chart of a tool wear state monitoring method based on a triboelectrification principle according to an embodiment of the present invention;
fig. 8 is a schematic structural diagram of a tool wear state monitoring method based on a triboelectric principle 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.
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.
Example one
The embodiment of the invention relates to the field of nano friction generators (TENG for short) and cutter wear state classification, in particular to a method which is based on a friction electrification principle and can monitor the current cutter wear state in real time, namely a cutter wear state monitoring method based on a nano friction generator, which mainly comprises the following steps: 1) a nano friction generator sensor module; 2) and a tool wear state classification module.
The embodiment of the invention provides a method for monitoring the wear state of a cutter by using a nano friction generator as a special sensor, which can perform noise reduction treatment on a voltage waveform diagram by acquiring voltage signals output by TENG on a workpiece to be processed and the cutter to be monitored in real time, and input the voltage waveform diagram into a parameterized neural network designed by the embodiment of the invention as data, wherein the network structure comprises the following steps: and finally, outputting the current corresponding tool wear state.
Fig. 1 is a schematic structural diagram of a sensor module of a nano friction generator according to an embodiment of the present invention. As shown in fig. 1, the nano-friction generator sensor module employed in the embodiment of the present invention is a simple contact separation mode "Z" type structure. The nano-friction generator sensor module includes a first group TENG1, a second group TENG2, a third group TENG 3; taking the first TENG1 as an example, the first TENG1 includes a first friction unit 11 and a second friction unit 12, one end of the first friction unit 11 is connected with one end of the second friction unit 12, and the other end of the first friction unit 11 is not connected with one end of the second friction unit 12, so that the first friction unit 11 and the second friction unit 12 are separated and contacted under the slight vibration; the first friction unit 11 includes, in order from top to bottom, a first substrate 111, a first electrode layer 112, and a first friction layer 113; the first friction unit 12 sequentially includes a second friction layer 123, a second electrode layer 122 and a second substrate 121 from top to bottom, that is, the second friction layer 123 and the first friction layer 113 realize separation and contact under micro-vibration; a second group TENG2, a third group TENG3 identical to the first group TENG 1.
As shown in fig. 2, in the embodiment of the present invention, the nano friction generator sensor modules are respectively disposed on the workpiece to be processed and the monitored tool, and are separated from each other by TENG contact caused by minute vibration during processing to generate a voltage, and then a voltage waveform diagram is acquired by an oscilloscope to serve as an input of the neural network, and at the same time, a wear value of the current monitored tool is measured offline and a current wear state is determined to serve as an output of the neural network. A waveform map corresponds to a wear state, thus forming a set of data samples, and 300 sets of data samples are collected and transmitted to a computer as a data set.
Step S1: for a complete tool life cycle, the main categories are slight wear, moderate wear and sharp wear. The method comprises the steps of collecting 300 groups of voltage signal oscillograms output by TENG in different abrasion stages, manually marking the voltage signal oscillograms to form a label, namely, creating a new document, storing an image name and the corresponding label, and vectorizing the label by adopting one-hot coding: slight wear is marked a1(1,0, 0); moderate wear is marked by mark a2(0,1, 0); acute wear mark a3The tool wear category is shown in fig. 3 for (0,0, 1). After the setting is finished, pressing 8: the scale of 2 is randomly divided into a training set and a test set.
Step S2: and performing denoising pretreatment on all acquired voltage waveform diagrams, wherein a wavelet threshold denoising method is adopted for denoising. First, the dimensions need to be reset, i.e. the width and height dimensions of the original image are set to 28px and 28px, which are indicated as (28, 28); then, the wavelet threshold denoising processing is carried out on the image after the size resetting, wherein the selection of the wavelet basis, the threshold and the threshold function can be carried out according to the following processes:
(1) the wavelet basis function adopts 'sym 8', and the number of layers is 5;
(2) the threshold value is selected by the following formula:
in the formula, cD1Representing detail coefficients of the first-layer decomposition, N representing data length, and M representing calculation median;
(3) the threshold function adopts a soft and hard threshold compromise method, and is expressed by a formula as follows:
wherein x represents a signal data value;
step S3: taking the preprocessed voltage waveform as input and the corresponding tool wear state as output, a parameterized neural network model P (y | x, θ) is established, as shown in fig. 4, which includes: the device comprises a feature extraction layer, a feature fusion layer, a measurement function layer and a classification layer;
step S4: initializing feature extraction network model parameters, learning rate, iteration step number and the like on a training set;
(1)w0＝[0(1),0(2),...,0(n)]Twhere n is the dimension of the image and w0 represents the weights in the network model parameters;
(2) b is 0, where b represents the bias in the network model parameters;
(3) α is 0.001, where α is a learning rate. The learning rate determines the speed of updating the parameters, and if the learning rate is too high, the global optimal value can be exceeded;
(4) δ 1000, where δ is the number of iteration steps.
Step S5: and respectively extracting the voltage waveform characteristics of the workpiece TENG and the cutter TENG through the characteristic extraction layer. As shown in fig. 5, a layer structure of the feature extraction layer extracts features related to tool wear, such as time domain, frequency domain, and time-frequency domain, in a waveform diagram by using multiple layers of convolution and pooling, and outputs a feature vector with a size of 3 through a full connection layer, which specifically includes the following steps:
the first layer is a convolutional layer and the network inputs a compressed image of the original image, with dimensions 28 x 1. The convolution kernel size is 3 x 3, depth is 32, and the activation function is RELU.
The second layer was a pooled layer, with maximum pooling, a characteristic map size of 28 x 32 for the network input, and a filter size of 2 x 2.
The third layer is a convolutional layer, the size of the characteristic graph of the network input is 14 × 32, the size of the convolutional kernel is 3 × 3, the depth is 64, and the activation function is RELU.
The fourth layer is the pooling layer, using the maximum pooling operator, the net input signature size is 7 × 64, and the filter size is 2 × 2.
The fifth layer is a flat layer, with the output being a feature vector of 3136 parameters.
The sixth layer is a fully connected layer, and the output is a feature vector of 3 parameters.
The depth of each layer of the network is shown in fig. 6, and the present invention uses RELU as an activation function, whose function expression is:
step S6: and performing feature fusion through the feature fusion layer, and obtaining the optimal feature vector by adopting a weighted average mode.
Step S7: the measurement function layer measures the similarity between two vectors a and b mainly through a cosine function, and is represented by a similarity c:
for example: get the characteristic vector v ═ 10,1,2, slight abrasion a1(1,0,0), moderate abrasion a2(0,1,0), rapid wear a3＝(0,0,1)，
Step S8: the classification layer calculates the probability that the current cutter wear belongs to three states respectively through similarity, outputs a feature vector with the size of 3, represents the probability value that the current cutter belongs to each wear state, and obtains the current wear state, and can be expressed by a formula as follows:
for example:
for the same reason p2＝0.0769，p30.1539, the vector obtained by the classification layer is the result of weighted summation of the three states and their corresponding probability values, i.e. p1a1+p2a2+p3a3That is, the output category is light wear, which is more likely to belong to light wear (1,0,0) (0.7692,0.0769, 0.1539).
Step S9: an error exists between the obtained probability vector and the corresponding one-hot label, the current cross entropy loss of forward propagation is calculated, and the parameter theta of the optimized parameterized neural network model is updated in a random gradient descent mode in backward propagation, and the specific process can be expressed as follows:
wherein θ represents a parameter before update, θ'Indicates the updated parameter, alpha indicates the learning rate,
it indicates derivation, L (θ) indicates cross entropy loss, y (θ) indicates an actual value, y' (θ) indicates a predicted value, and n indicates the number of data.
Step S10: and inputting the test set into a parameterized neural network model, and outputting the prediction probability and classification result of the wear state of the test tool.
Step S11: and predicting the wear state of the monitored tool according to the parameterized neural network model.
The technical characteristic points of the embodiment of the invention are as follows:
firstly, the TENG is prepared into a special sensor by utilizing the characteristics of different output magnitudes of the TENG such as voltage, current, power and the like under different contact degrees. TENG can be used as a stable sensor for various processing, and power supply is not needed, so that the complex steps of the traditional sensor are omitted, and the TENG is more environment-friendly;
secondly, the TENG is arranged on a workpiece and a cutter by utilizing the characteristics of easy preparation, convenient installation and the like, so that signal data acquisition is facilitated;
thirdly, a Z-shaped structure is adopted, so that the voltage output efficiency of TENG is improved;
fourthly, acquiring a voltage waveform diagram output by TENG through an oscilloscope, and judging the current tool wear state by using the waveform diagram as the input of a neural network;
fifthly, denoising the oscillogram by wavelet threshold denoising to make the obtained characteristics more effective;
sixthly, a parameterized neural network model is established, and the method comprises the following steps: the system comprises a feature extraction network, a feature fusion layer, a measurement function and a classification layer;
seventhly, a TENG output voltage waveform diagram corresponding to the workpiece and the cutter is fused, so that the prediction accuracy is improved;
eighthly, the measuring function adopts a cosine similarity function for judging the similarity between vectors;
the embodiment of the invention has the advantages that: the TENG is adopted as a sensor, so that the cost is low, the manufacturing is easy, the installation is convenient and fast, and the like; the sensor can be used as a stable sensor for various processing, does not need power supply, avoids the complicated steps of the traditional sensor, and is more environment-friendly; the low-frequency energy can be well collected by adopting a frictional electrification principle; the Z-shaped structure is adopted, so that the voltage output efficiency can be improved; the wavelet threshold denoising processing oscillogram is adopted, so that the characteristics are more effective; the prediction accuracy is improved by fusing the characteristics of the TENG oscillograms of the workpiece and the cutter; the similarity between vectors is judged by adopting a measurement function, so that the prediction efficiency and accuracy are improved; in conclusion, the method provided by the invention utilizes the characteristics of TENG, such as convenience, high efficiency and stability, and combines a deep learning method to accurately predict the current tool wear state in real time.
Example two
As shown in fig. 7, the method for monitoring the wear state of a tool based on the principle of triboelectrification according to this embodiment includes:
step 701: acquiring a voltage signal of a monitored cutter; the voltage signal of the monitored cutter comprises a voltage waveform diagram of the monitored cutter and a voltage waveform diagram of a machined workpiece, which are acquired by sensing equipment; the sensing equipment comprises a waveform generator, a first sensor and a second sensor, wherein the first sensor and the second sensor are connected with the input end of the waveform generator; the first sensor is arranged on the monitored cutter, the second sensor is arranged on the machined workpiece, and the first sensor and the second sensor are both manufactured on the basis of the principle of triboelectrification; the workpiece to be machined is a workpiece machined by the monitored tool.
Step 702: and determining a neural network model based on the voltage signal of the monitored tool and the tool wear state, and determining the wear degree of the monitored tool.
On the basis of the embodiment shown in fig. 7, the first sensor and the second sensor are both nano friction generator sensor modules; the nano friction generator sensor module at least comprises one nano friction generator sensor, and when the nano friction generator sensor module comprises a plurality of nano friction generator sensors with the same structure, the plurality of nano friction generator sensors are arranged in a stacking mode; the nanometer friction generator sensor comprises a fixed end and an opening and closing end, and the opening and closing ends of the nanometer friction generator sensor face oppositely.
Further, the nano friction generator sensor comprises a first substrate, a first electrode layer, a first friction layer, a second electrode layer and a second substrate which are sequentially stacked, and the first friction layer and the second friction layer can be opened and closed under the action of micro vibration.
On the basis of the embodiment described in fig. 7, step 702 specifically includes:
carrying out noise reduction pretreatment on the voltage signal of the monitored cutter; and inputting the voltage signal of the monitoring cutter subjected to noise reduction pretreatment into a cutter wear state determination neural network model so as to determine the wear degree of the monitored cutter.
On the basis of the embodiment illustrated in fig. 7, the tool wear state determination neural network model is determined based on a sample data set and a neural network; the sample data set comprises a plurality of sample tool voltage signals and the corresponding wear degree of each sample tool voltage signal; the sample cutter voltage signal comprises a voltage waveform diagram of a sample cutter and a voltage waveform diagram of a sample processing workpiece, wherein the voltage waveform diagram of the sample cutter is acquired by sensing equipment and subjected to noise reduction pretreatment; the sample processing workpiece is a workpiece processed by the sample cutter.
Further, the degree of wear is slight wear, moderate wear, or extreme wear; the neural network comprises a feature extraction network layer, a feature fusion layer, a measurement function and a classification layer which are connected in sequence;
the feature extraction network layer is configured to:
respectively extracting a characteristic vector of a voltage waveform diagram of the monitored cutter subjected to noise reduction pretreatment and a characteristic vector of a voltage waveform diagram of the processed workpiece subjected to noise reduction pretreatment;
or extracting the characteristic vector of the voltage waveform diagram of the sample cutter after the noise reduction pretreatment and the characteristic vector of the voltage waveform diagram of the sample processing workpiece after the noise reduction pretreatment respectively.
And the feature fusion layer is used for fusing the feature vectors output by the feature extraction network layer to obtain fusion vectors.
And the measurement function layer is used for calculating the similarity between the fusion vector and different wear degree vectors respectively and determining the similarity value corresponding to each wear degree.
The classification layer is used for respectively calculating a first probability, a second probability and a third probability based on the similarity value corresponding to each wear degree, determining a maximum probability based on the first probability, the second probability and the third probability, and then determining the wear degree corresponding to the maximum probability as the wear degree of the monitored tool or the sample tool.
The first probability is the probability that the monitored tool or the sample tool is slightly worn; the second probability is the probability that the monitored tool or sample tool is of moderate wear; the third probability is the probability that the monitored tool or sample tool is of extreme wear.
The feature extraction network layer comprises a first convolution layer, a first pooling layer, a second convolution layer, a second pooling layer, a flat layer and a full-connection layer which are sequentially connected.
EXAMPLE III
As shown in fig. 8, a tool wear state monitoring system based on the principle of triboelectrification according to an embodiment of the present invention includes:
the data acquisition module 801 is used for acquiring a voltage signal of the monitored cutter; the voltage signal of the monitored cutter comprises a voltage waveform diagram of the monitored cutter and a voltage waveform diagram of a processed workpiece, wherein the voltage waveform diagram of the monitored cutter is acquired by sensing equipment; the sensing equipment comprises a waveform generator, a first sensor and a second sensor, wherein the first sensor and the second sensor are connected with the input end of the waveform generator; the first sensor is arranged on the monitored cutter, the second sensor is arranged on the machined workpiece, and the first sensor and the second sensor are both manufactured on the basis of the principle of triboelectrification; the workpiece to be machined is a workpiece machined by the monitored tool.
And the monitoring module 802 is configured to determine a neural network model based on the voltage signal of the monitored tool and the tool wear state, and determine the wear degree of the monitored tool.
Further, the first sensor and the second sensor are both nano friction generator sensor modules; wherein the nano friction generator sensor module comprises at least one nano friction generator sensor; when the nano friction generator sensor module comprises a plurality of nano friction generator sensors with the same structure, the plurality of nano friction generator sensors are arranged in a stacked mode; the nanometer friction generator sensor comprises a fixed end and an opening and closing end, and the opening and closing ends of the nanometer friction generator sensor face oppositely.
In order to solve the defect problem of monitoring the wear state of the existing cutter, the real-time performance and accuracy of identification are met, and meanwhile the cost is reduced. The embodiment of the invention provides a tool wear state monitoring method and system based on a friction electrification principle, and tool wear state monitoring is realized on the basis of taking TENG (tool orientation control) with low cost and high efficiency as a sensor. The method specifically comprises the following steps:
the invention takes the characteristics of high cost, difficult installation and the like of the traditional sensor into consideration, and takes the characteristics of TENG as a special sensor; when the collected image features are considered, the convolutional neural network is used for automatically extracting the features, compared with the traditional feature extraction method, the method has the functions of reducing data and accelerating the operation time of the algorithm, and the influence of insignificant small features on the identification of the wear state of the cutter is ignored while the features are expressed completely. The method has the advantages of simple algorithm realization, high cutter wear state recognition rate and high operation efficiency, and the wavelet threshold denoising processing is required to be carried out on the image after the acquisition due to the consideration of the noise influence of the acquired image, so that the calculation data volume is obviously reduced, the recognition rate is improved, and the final discrimination result is obtained by a high-efficiency and accurate method. The method extracts tool wear features in an image, calculates similarity among sample feature vectors through a measurement function, carries out three classifications on tool wear states by utilizing the similarity, integrates the factors, realizes acquisition of tool wear related signals by utilizing a TENG-based method, and carries out state classification and identification by extracting image features.
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 (10)
1. A tool wear state monitoring method based on a friction electrification principle is characterized by comprising the following steps:
acquiring a voltage signal of a monitored cutter; the voltage signal of the monitored cutter comprises a voltage waveform diagram of the monitored cutter and a voltage waveform diagram of a processed workpiece, wherein the voltage waveform diagram of the monitored cutter is acquired by sensing equipment; the sensing equipment comprises a waveform generator, a first sensor and a second sensor, wherein the first sensor and the second sensor are connected with the input end of the waveform generator; the first sensor is arranged on the monitored cutter, the second sensor is arranged on the machined workpiece, and the first sensor and the second sensor are both manufactured on the basis of the principle of triboelectrification; the workpiece to be processed is a workpiece processed by the monitored cutter;
and determining a neural network model based on the voltage signal of the monitored tool and the tool wear state, and determining the wear degree of the monitored tool.
2. The tool wear state monitoring method based on the triboelectric principle according to claim 1, wherein the first sensor and the second sensor are both nano friction generator sensor modules; wherein the nano friction generator sensor module comprises at least one nano friction generator sensor.
3. The tool wear state monitoring method based on the triboelectric principle according to claim 2, wherein when the nano friction generator sensor module comprises a plurality of nano friction generator sensors with the same structure, the plurality of nano friction generator sensors are arranged in a stacked manner;
the nanometer friction generator sensor comprises a fixed end and an opening and closing end, and the opening and closing ends of the nanometer friction generator sensor face oppositely.
4. The method for monitoring the wear state of the tool based on the triboelectric charging principle according to claim 2 or 3, wherein the nano-friction generator sensor comprises a first substrate, a first electrode layer, a first friction layer, a second electrode layer and a second substrate which are sequentially stacked, and the first friction layer and the second friction layer can perform opening and closing operations under the condition of micro-vibration.
5. The method for monitoring the wear state of the tool based on the triboelectric principle according to claim 1, wherein the determining the neural network model based on the voltage signal of the monitored tool and the wear state of the tool to determine the wear degree of the monitored tool specifically comprises:
carrying out noise reduction pretreatment on the voltage signal of the monitored cutter;
and inputting the voltage signal of the monitoring cutter subjected to noise reduction pretreatment into a cutter wear state determination neural network model so as to determine the wear degree of the monitored cutter.
6. The tool wear state monitoring method based on the triboelectric charging principle according to claim 1, wherein the tool wear state determination neural network model is determined based on a sample data set and a neural network; the sample data set comprises a plurality of sample tool voltage signals and the corresponding wear degree of each sample tool voltage signal; the sample cutter voltage signal comprises a voltage waveform diagram of a sample cutter and a voltage waveform diagram of a sample processing workpiece, wherein the voltage waveform diagram of the sample cutter is acquired by sensing equipment and subjected to noise reduction pretreatment; the sample processing workpiece is a workpiece processed by the sample cutter.
7. The tool wear state monitoring method based on the triboelectric charging principle according to claim 6, wherein the wear degree is slight wear, moderate wear or sharp wear; the neural network comprises a feature extraction network layer, a feature fusion layer, a measurement function and a classification layer which are connected in sequence;
the feature extraction network layer is configured to:
respectively extracting the characteristic vector of the voltage waveform diagram of the monitored cutter subjected to noise reduction pretreatment and the characteristic vector of the voltage waveform diagram of the processed workpiece subjected to noise reduction pretreatment;
or respectively extracting the characteristic vector of the voltage waveform diagram of the sample cutter subjected to noise reduction pretreatment and the characteristic vector of the voltage waveform diagram of the sample processing workpiece subjected to noise reduction pretreatment;
the feature fusion layer is used for fusing the feature vectors output by the feature extraction network layer to obtain fusion vectors;
the measurement function layer is used for calculating the similarity between the fusion vector and different wear degree vectors respectively and determining a similarity value corresponding to each wear degree;
the classification layer is used for respectively calculating a first probability, a second probability and a third probability based on the similarity value corresponding to each wear degree, determining a maximum probability based on the first probability, the second probability and the third probability, and then determining the wear degree corresponding to the maximum probability as the wear degree of the monitored tool or the sample tool;
the first probability is the probability that the monitored tool or the sample tool is slightly worn; the second probability is the probability that the monitored tool or sample tool is of moderate wear; the third probability is the probability that the monitored tool or sample tool is of extreme wear.
8. The method for monitoring the wear state of the tool based on the triboelectric charging principle according to claim 7, wherein the feature extraction network layer comprises a first convolution layer, a first pooling layer, a second convolution layer, a second pooling layer, a flat layer and a full-connection layer which are connected in sequence.
9. A tool wear state monitoring system based on a triboelectric principle, comprising:
the data acquisition module is used for acquiring a voltage signal of the monitored cutter; the voltage signal of the monitored cutter comprises a voltage waveform diagram of the monitored cutter and a voltage waveform diagram of a processed workpiece, wherein the voltage waveform diagram of the monitored cutter is acquired by sensing equipment; the sensing equipment comprises a waveform generator, a first sensor and a second sensor, wherein the first sensor and the second sensor are connected with the input end of the waveform generator; the first sensor is arranged on the monitored cutter, the second sensor is arranged on the machined workpiece, and the first sensor and the second sensor are both manufactured on the basis of the principle of triboelectrification; the workpiece to be processed is a workpiece processed by the monitored cutter;
and the monitoring module is used for determining a neural network model based on the voltage signal of the monitored cutter and the cutter wear state, and determining the wear degree of the monitored cutter.
10. The tool wear state monitoring system based on the principle of triboelectrification according to claim 9, wherein the first sensor and the second sensor are both nano friction generator sensor modules; wherein the nano friction generator sensor module comprises at least one nano friction generator sensor;
when the nano friction generator sensor module comprises a plurality of nano friction generator sensors with the same structure, the plurality of nano friction generator sensors are arranged in a stacked mode;
the nanometer friction generator sensor comprises a fixed end and an opening and closing end, and the opening and closing ends of the nanometer friction generator sensor face oppositely.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210263842.0A CN114571286B (en) | 2022-03-17 | 2022-03-17 | Tool wear state monitoring method and system based on friction electrification principle |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210263842.0A CN114571286B (en) | 2022-03-17 | 2022-03-17 | Tool wear state monitoring method and system based on friction electrification principle |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114571286A true CN114571286A (en) | 2022-06-03 |
| CN114571286B CN114571286B (en) | 2023-03-31 |
Family
ID=81775411
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210263842.0A Active CN114571286B (en) | 2022-03-17 | 2022-03-17 | Tool wear state monitoring method and system based on friction electrification principle |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114571286B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0760318A (en) * | 1993-08-26 | 1995-03-07 | Sumitomo Metal Ind Ltd | Plasticification processing method of metallic material |
| US5651648A (en) * | 1996-02-22 | 1997-07-29 | Virginia Tech Intellectual Properties, Inc. | Method for reducing ceramic tool wear and friction in machining/cutting applications |
| US7198043B1 (en) * | 2005-08-26 | 2007-04-03 | Gang Zhang | Method for extending diamond tool life in diamond machining of materials that chemically react with diamond |
| CN201960178U (en) * | 2011-01-24 | 2011-09-07 | 海克斯康测量技术(青岛)有限公司 | Milling robot with hollow milling spindle |
| CN108972152A (en) * | 2018-10-12 | 2018-12-11 | 哈尔滨理工大学 | A kind of sound-power detection method monitoring abrasion of cutting tool state |
| CN109623494A (en) * | 2019-01-18 | 2019-04-16 | 四川大学 | A kind of three-in-one sensor fixture and multi-modal cutting-tool wear state monitor system |
| CN113600896A (en) * | 2021-08-05 | 2021-11-05 | 豪丰茂五金制品(太仓)有限公司 | Method for monitoring wear state signal of milling cutter of numerical control machine tool |
-
2022
- 2022-03-17 CN CN202210263842.0A patent/CN114571286B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0760318A (en) * | 1993-08-26 | 1995-03-07 | Sumitomo Metal Ind Ltd | Plasticification processing method of metallic material |
| US5651648A (en) * | 1996-02-22 | 1997-07-29 | Virginia Tech Intellectual Properties, Inc. | Method for reducing ceramic tool wear and friction in machining/cutting applications |
| US7198043B1 (en) * | 2005-08-26 | 2007-04-03 | Gang Zhang | Method for extending diamond tool life in diamond machining of materials that chemically react with diamond |
| CN201960178U (en) * | 2011-01-24 | 2011-09-07 | 海克斯康测量技术(青岛)有限公司 | Milling robot with hollow milling spindle |
| CN108972152A (en) * | 2018-10-12 | 2018-12-11 | 哈尔滨理工大学 | A kind of sound-power detection method monitoring abrasion of cutting tool state |
| CN109623494A (en) * | 2019-01-18 | 2019-04-16 | 四川大学 | A kind of three-in-one sensor fixture and multi-modal cutting-tool wear state monitor system |
| CN113600896A (en) * | 2021-08-05 | 2021-11-05 | 豪丰茂五金制品(太仓)有限公司 | Method for monitoring wear state signal of milling cutter of numerical control machine tool |
Non-Patent Citations (1)
| Title |
|---|
| 吴凤和等: "基于卷积门控循环神经网络的刀具磨损状态监测", 《计量学报》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114571286B (en) | 2023-03-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111721535B (en) | Bearing fault detection method based on convolution multi-head self-attention mechanism | |
| CN107941537B (en) | A kind of mechanical equipment health state evaluation method | |
| CN111242202B (en) | Method for monitoring wear state of turning tool based on metric learning | |
| CN111716150B (en) | Evolution learning method for intelligently monitoring cutter state | |
| CN103034170B (en) | Numerical control machine tool machining performance prediction method based on intervals | |
| CN110647943A (en) | Cutting tool wear monitoring method based on evolutionary data cluster analysis | |
| CN111037365A (en) | Cutter state monitoring data set enhancing method based on generative countermeasure network | |
| Liu et al. | Online monitoring and measurements of tool wear for precision turning of stainless steel parts | |
| CN110596506A (en) | Converter fault diagnosis method based on time convolution network | |
| CN111633467A (en) | Cutter wear state monitoring method based on one-dimensional depth convolution automatic encoder | |
| CN106271881B (en) | A kind of Condition Monitoring of Tool Breakage method based on SAEs and K-means | |
| CN108388915A (en) | A method of automatically extracting modal parameters using cluster | |
| CN103941645B (en) | Thin-wall part complex working condition machining state monitoring method | |
| CN113378725A (en) | Cutter fault diagnosis method, equipment and storage medium based on multi-scale-channel attention network | |
| CN108171119B (en) | SAR image change detection method based on residual error network | |
| CN113554621B (en) | Cutter wear state identification system and method based on wavelet scale map and deep migration learning | |
| CN109238447A (en) | A kind of blind source separation method of tether vibration signal | |
| CN113158814A (en) | Bearing health state monitoring method based on convolution self-encoder | |
| CN112207631B (en) | Method for generating tool detection model, method, system, device and medium for detecting tool detection model | |
| CN111783544B (en) | Method for building diamond milling grinding head state monitoring system for processing ceramic mobile phone backboard | |
| CN116593157A (en) | Complex working condition gear fault diagnosis method based on matching element learning under small sample | |
| CN114346761A (en) | Cutter wear condition detection method for generating countermeasure network based on improved conditions | |
| CN110458071B (en) | DWT-DFPA-GBDT-based optical fiber vibration signal feature extraction and classification method | |
| CN111855202A (en) | Gear box fault diagnosis method and system | |
| Wang et al. | Bidirectional gated recurrent deep learning neural networks for smart acoustic emission sensing of natural fiber–reinforced polymer composite machining process |
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 |