WO2021046737A1 - 基于ssae-lstm模型的深孔加工刀具磨损量监测方法 - Google Patents

基于ssae-lstm模型的深孔加工刀具磨损量监测方法 Download PDF

Info

Publication number
WO2021046737A1
WO2021046737A1 PCT/CN2019/105282 CN2019105282W WO2021046737A1 WO 2021046737 A1 WO2021046737 A1 WO 2021046737A1 CN 2019105282 W CN2019105282 W CN 2019105282W WO 2021046737 A1 WO2021046737 A1 WO 2021046737A1
Authority
WO
WIPO (PCT)
Prior art keywords
layer
network
ssae
deep hole
data
Prior art date
Application number
PCT/CN2019/105282
Other languages
English (en)
French (fr)
Inventor
刘阔
厉大维
沈明瑞
刘庆君
任慧民
王永青
Original Assignee
大连理工大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 大连理工大学 filed Critical 大连理工大学
Priority to PCT/CN2019/105282 priority Critical patent/WO2021046737A1/zh
Publication of WO2021046737A1 publication Critical patent/WO2021046737A1/zh

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
    • 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
    • B23Q17/12—Arrangements for observing, indicating or measuring on machine tools for indicating or measuring vibration
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology

Definitions

  • the invention belongs to the technical field of tool wear state monitoring, and is specifically a method for monitoring tool wear in deep hole machining based on an SSAE-LSTM model.
  • the cutting area is generally inside the deep hole, which is filled with a large amount of cutting fluid and chips, and the position of the tool has always been Movement, so the machine operator cannot determine the wear status of the deep hole machining tool with the naked eye.
  • the machine tool operator can only rely on personal experience to determine the cutting state of the tool inside the deep hole by observing the outflow of chips and touching the tool bar to perceive the vibration, and cannot accurately determine the wear state of the tool.
  • the tool wear exceeds the usable range, if the judgment cannot be made in time and corresponding measures are taken, the parts may be scrapped.
  • the residual convolution network is trained to establish tool wear by using cutting force, vibration and acoustic emission signals.
  • the model takes tool wear as output, and establishes a model from signal to wear based on the method of supervised learning, which solves the problem of tool wear prediction.
  • the image data acquisition module is used to shoot the video of the tool cutting workpiece during the cutting process, and the image data preprocessing module is used to extract the video The extracted images are positioned, cropped and normalized.
  • the edge computing module integrated with the knife discriminator is used to receive the processed images, and the pre-trained convolutional neural network forward inference is used to obtain the broken Knife discrimination result.
  • the cutting area is located inside the deep hole, so the cutting force signal and tool image data are difficult to obtain, and it is impossible to monitor the amount of deep hole tool wear.
  • the present invention proposes a method for monitoring the wear of deep hole machining tools.
  • the purpose of the present invention is to provide an effective method for monitoring tool wear in deep hole machining, and solve the problem that it is difficult to monitor tool wear in deep hole machining.
  • the technical scheme of the present invention is as follows: firstly, two three-directional acceleration sensors are respectively installed outside the two tool holder bearing bushes of the deep hole processing machine tool, and a microphone is installed at the processing entrance of the deep hole workpiece to collect and process Toolbar vibration and cutting sound data in the process; then use the collected data to greedily train the stacked sparse autoencoder (SSAE) network layer by layer, and use the trained SSAE network to perform feature selection on the data to obtain the condensed data ; Then use the simplified data to train the long-short-term memory (LSTM) network.
  • SSAE stacked sparse autoencoder
  • the LSTM network can be used for tool wear prediction; in real-time monitoring, the trained The SSAE network and the LSTM network are combined into an SSAE-LSTM model. Real-time vibration and sound data are input into the SSAE-LSTM model, and the model outputs the tool wear.
  • a method for monitoring tool wear in deep hole machining based on the SSAE-LSTM model characterized in that the steps are as follows:
  • the first step is to collect vibration and sound information during deep hole processing
  • the second step is the construction and training of the SSAE model
  • f( ⁇ ) represents the activation function of the neuron, Represents the weight of the input layer and the hidden layer, Indicates the weight of the hidden layer and the output layer, Shows the bias of each neuron in the hidden layer, Represents the bias of each neuron in the output layer;
  • the loss function of SAE is defined as the mean square error L between the reconstructed data and the original data:
  • SAEs are formed into a stacked autoencoder network SSAE, and the SSAE network is trained by a greedy layer-by-layer training method.
  • the process is: first use sample data to train the first layer of the network, and this layer converts the sample data to the output of the hidden layer The feature vector A composed of values; then A is used as the input of the second layer to train to obtain the feature vector B of the second layer; and so on, the training of each layer is completed; after the training is completed, all the sample data is input into the SSAE network to obtain Feature vector;
  • the third step is the construction and training of the deep LSTM network
  • the memory unit of the long-short-term memory LSTM network is updated once every time step t, then the value of its input gate i t and the candidate state value of the memory unit They are:
  • the current value C t of the memory unit at time t is calculated from the values of the input gate, the forget gate and the candidate state of the memory unit:
  • x t is the input of the memory unit at time t;
  • W is the weight parameter of the model;
  • b is the deviation vector of the model;
  • ⁇ and tanh are the Sigmoid activation function and Tanh activation function; stacked multi-layer LSTM networks to form a deep LSTM The internet;
  • the fourth step real-time monitoring of tool wear in deep hole machining
  • the beneficial effects of the present invention through the method, the dependence on the experience of the machine tool operator is reduced, the processing efficiency is improved, and the rejection rate is reduced. Compared with cutting force monitoring method and image monitoring method, the hardware cost is relatively low.
  • the SSAE network performs adaptive feature selection on the original data, removes redundant data, improves the generalization ability of the prediction model, and accelerates the prediction speed and accuracy.
  • Figure 1 is a flowchart of tool wear monitoring for deep hole machining.
  • Figure 2 is a schematic diagram of the sensor layout of a deep hole processing machine tool.
  • Figure 3 shows the monitoring model of tool wear in deep hole machining
  • Figure 4 is a graph of tool wear prediction.
  • the first step is to collect vibration and sound information during deep hole processing
  • the #1 three-way acceleration sensor 5 and #2 three-way acceleration sensor 7 are attached to the side of the tool holder bearing bush with a magnetic seat, and the installation method is shown in Figure 2. Set the sampling frequency to 5000Hz to collect vibration and sound signals during processing.
  • the second step is the construction and training of the SSAE network
  • the collected data is divided into 5000 training samples and 1000 test samples, each sample contains 7000 data.
  • Stack 3 SAEs to build an SSAE network.
  • the number of neurons in the input layer and output layer of the first SAE is 7000, and the number of neurons in the hidden layer is 4000.
  • the number of neurons in the input and output layer of the second SAE is 4000, and the number of neurons in the hidden layer is 2000.
  • the number of neurons in the input and output layer of the third SAE is 2000, and the number of neurons in the hidden layer is 700.
  • the greedy training method is used to train the SSAE network layer by layer with training samples. Use test samples to test the SSAE network. After passing the test, all samples are input into the SSAE network, and after feature extraction, 5000 feature vectors of the training set and 1000 feature vectors of the test set are obtained.
  • the third step is the construction and training of the deep LSTM network
  • Use the test set feature vector to test the trained deep LSTM network, and get the average test error of 0.0121mm, which is lower than the set maximum average error ⁇ 0.025mm, and the model test is qualified.
  • the fourth step real-time monitoring of tool wear in deep hole machining

Abstract

一种基于SSAE-LSTM模型的深孔加工刀具磨损量的监测方法，先在深孔加工机床的两个刀杆（6）保持架轴瓦外部分别安装两个三向加速度传感器（5，7），在深孔工件（1）加工入口处安装传声器（3），采集加工过程中的刀杆振动和切削声音数据；然后利用采集到的数据对堆叠自编码器进行贪婪逐层训练，用训练好的堆叠自编码器对数据进行特征选择，得到精简后的数据；再用精简后的数据对长短时记忆网络进行训练。如果训练预测误差低于设定的δ值，则模型可用于刀具磨损量预测；在实时监测时，将实时的振动和声音数据输入训练好的堆叠自编码器和长短时记忆网络中，网络输出刀具的磨损量。该方法可实现深孔加工过程中刀具磨损量的监测。

Description

基于SSAE-LSTM模型的深孔加工刀具磨损量监测方法 技术领域
本发明属于刀具磨损状态监测技术领域，具体为一种基于SSAE-LSTM模型的深孔加工刀具磨损量监测方法。
背景技术
在加工大深径比的孔类零件时，如钻削、镗削、扩孔等，切削区域一般出于深孔内部，深孔内部充斥着大量的切削液和切屑，且刀具的位置一直在运动，所以机床操作者，无法用肉眼去确定深孔加工刀具的磨损状态。机床操作者只能依靠个人经验，通过观察流出的切屑、触摸刀杆感知其振动，来确定深孔内部刀具的切削状态，无法准确的判断刀具的磨损状态。当刀具磨损量超出可用范围时，如果不能及时做出判断并采取相应措施，可能会导致零件的报废。
在专利“基于残差卷积神经网络的刀具磨损量的建模和监测方法”(申请号：CN201810977274.4)中利用切削力、振动以及声发射信号训练残差卷积网络建立刀具磨损量的模型，以刀具磨损量作为输出，根据监督式学习的方法建立从信号到磨损量的模型，解决了刀具磨损量预测的问题。在专利“一种基于深度学习的数控机床断刀检测系统及方法”，申请号：CN201910228970.X中利用图像数据采集模块拍摄切削加工过程中刀具切削工件的视频，利用图像数据预处理模块提取视频中的图像，并对提取的图像进行定位、裁剪和归一化处理，利用集成有断刀判别器的边缘计算模块接收处理后的图像，并利用预先训练的卷积神经网络前向推理得到断刀判别结果。但是由于深孔加工，切削区域位于深孔内部，所以切削力信号和刀具图像数据难以获取，无法实现深孔刀具磨损量的监测。
通过分析上述专利可知，上述刀具监测方法难以在深孔加工中使用。本发 明针对深孔加工中刀具磨损量难以监测的问题，提出了一种对深孔加工刀具进行磨损量监测的方法。
发明内容
本发明的目的为提供一种有效的深孔加工刀具磨损量监测方法，解决深孔加工刀具磨损量难以监测的问题。
为解决上述问题，本发明的技术方案为：首先，在深孔加工机床的两个刀杆保持架轴瓦外部分别安装两个三向加速度传感器，在深孔工件加工入口处安装一个传声器，采集加工过程中的刀杆振动和切削声音数据；然后利用采集到的数据对堆叠稀疏自编码器(SSAE)网络进行贪婪逐层训练，用训练好的SSAE网络对数据进行特征选择，得到精简后的数据；再用精简后的数据对长短时记忆(LSTM)网络进行训练，如果训练预测误差低于设定的δ值，则该LSTM网络可用于刀具磨损量预测；在实时监测时，将训练好的SSAE网络和LSTM网络组合成SSAE-LSTM模型，将实时的振动和声音数据输入SSAE-LSTM模型中，模型输出刀具的磨损量。
本方法的具体步骤如下：
1.一种基于SSAE-LSTM模型的深孔加工刀具磨损量监测方法，其特征在于，步骤如下：
第一步，深孔加工过程中的振动和声音信息采集
将加速度传感器通过磁座吸附在刀杆保持架的轴瓦外部，将传声器放置于深孔工件加工入口处，对加工过程中的刀杆振动以及切削噪声进行采集；
第二步，SSAE模型的构建和训练
将采集到的振动和声音数据拆分成样本数据；设拆分后的某一个训练样本数据为x＝[x 1,x 2,…,x i]，则稀疏自动编码器SAE的输入输出关系表示为：
Figure PCTCN2019105282-appb-000001
Figure PCTCN2019105282-appb-000002
式中，f(·)表示神经元的激活函数，
Figure PCTCN2019105282-appb-000003
表示输入层与隐含层的权重，
Figure PCTCN2019105282-appb-000004
表示隐含层与输出层的权重，
Figure PCTCN2019105282-appb-000005
示隐含层各神经元的偏置，
Figure PCTCN2019105282-appb-000006
表示输出层各神经元的偏置；
将SAE的损失函数定义为重构后的数据与原始数据的均方误差L：
Figure PCTCN2019105282-appb-000007
式中，
Figure PCTCN2019105282-appb-000008
是SAE器的输出数据；β是稀疏惩罚项参数，作用是稀疏惩罚项在损失函数当中所占的比重；ρ是稀疏性参数；
Figure PCTCN2019105282-appb-000009
是隐含层第g个神经元的激活度；通过误差反向传播和梯度下降法调整SAE各层参数，使损失函数值达到最小；
将多个SAE组成堆叠自编码器网络SSAE，采用贪婪逐层训练法训练SSAE网络，其过程为：首先利用样本数据来训练网络的第1层，该层将样本数据转换为由隐含层输出值组成的特征向量A；然后把A作为第2层的的输入，训练得到第2层的特征向量B；以此类推至各层训练完毕；训练完成后，将所有样本数据输入SSAE网络，得到特征向量；
第三步，深度LSTM网络的构建和训练
设长短时记忆LSTM网络的记忆单元在每一个时间步长t更新一次，则其输入门的值i t和记忆单元候选状态值
Figure PCTCN2019105282-appb-000010
分别为：
i t＝σ(W i·[h t-1,x t]+b i)
Figure PCTCN2019105282-appb-000011
则其t时刻遗忘门的值f t为：
f t＝σ(W f·[h t-1,x t]+b f)
由输入门、遗忘门和记忆单元候选状态的值计算出记忆单元t时刻当前值C t：
Figure PCTCN2019105282-appb-000012
最后可得记忆单元输出值h t为：
h t＝σ(W o[h t-1,x t]+b o)*tanh(C t)
式中，x t为t时刻记忆单元的输入；W为模型的权重参数；b为模型的偏差向量；σ和tanh为Sigmoid激活函数和Tanh激活函数；将多层LSTM网络堆叠起来，构成深度LSTM网络；
将特征向量分成训练集特征向量和测试集特征向量；以训练集特征向量作为深度LSTM网络的输入数据，对深度LSTM网络进行训练，输出深孔加工刀具的磨损量；然后再用测试集特征向量对深度LSTM网络进行测试。如果预测平均误差低于设定的δ值，则测试合格，模型用于磨损量监测；反之，则测试不合格，回到第二步重新对网络进行训练；
第四步，深孔加工刀具磨损量的实时监测
将训练好的SSAE网络和深度LSTM网络进行连接生成SSAE-LSTM模型；在实际加工过程中，将实时数据输入至SSAE-LSTM模型中，输出刀具磨损量。
本发明的有益效果：通过该方法，降低对机床操作工经验的依赖，提高加工效率，降低废品率。相较于切削力监测法和图像监测法，硬件成本相对较低。通过SSAE网络对原始数据进行自适应的特征选择，去除冗余数据，提高预测模型的泛化能力，加快预测速度和准确度。
附图说明
图1为深孔加工刀具磨损量监测流程图。
图2为深孔加工机床传感器布置示意图。
图3为深孔加工刀具磨损量监测模型
图4为刀具磨损量预测图。
图中：1-工件；2-机床齿轮箱；3-传声器；4-床身；5-#1三向加速度传感器；6-刀杆；7-#2三向加速度传感器。
具体实施方式
为了使本发明的技术方案和有益效果更加清晰明了，下面结合深孔加工刀具磨损量监测的具体实施方式并参照附图，对本发明作详细说明。本实施例是以本发明的技术方案为前提进行的，给出了详细的实施方式和具体的操作过程，但本发明的保护范围不限于下述的实施例。
以一台卧式深孔镗床加工深孔为例，详细说明本发明的实施方式。
第一步，深孔加工过程中的振动和声音信息采集
将传声器3固定在工件右端，对准工件内孔。#1三向加速度传感器5和#2三向加速度传感器7用磁座吸附到刀杆保持架轴瓦侧面，安装方式如图2所示。设置采样频率5000Hz，采集加工过程中的振动和声信号。
第二步，SSAE网络的构建和训练
将采集到的数据拆分成5000个训练样本和1000个测试样本，每个样本包含7000个数据。
堆叠3个SAE构建SSAE网络。第一个SAE输入层和输出层神经元个数均为7000，隐层神经元个数为4000。第二个SAE输入输出层神经元个数均为4000，隐层神经元个数为2000。第三个SAE输入输出层神经元个数为2000，隐层神经元个数为700。采用贪婪训练法，用训练样本对SSAE网络逐层进行训练。使用测试样本对SSAE网络进行测试。测试合格后，将所有样本输入SSAE网络，经过特征提取后，得到精简后的训练集特征向量5000个和测试集特征向量1000个。
第三步，深度LSTM网络的构建与训练
构建深度LSTM网络，其输入层神经元个数为700，两个隐层神经元个数分别为200和100，输出层神经元个数为1。取训练集特征向量作为深度LSTM网络的输入数据，对深度LSTM网络进行训练，输出刀具的磨损量。采用测试集特征向量对训练好的深度LSTM网络进行测试，得到测试平均误差为0.0121mm，低于设定最大平均误差δ＝0.025mm，模型测试合格。
第四步，深孔加工刀具磨损量的实时监测
将训练好的SSAE网络和深度LSTM网络进行连接生成SSAE-LSTM模型，如图3所示。把实时振动和声音数据输入至模型中，对深孔加工过程中的刀具磨损量进行监测。监测结果如图3所示，预测平均误差为0.0142mm，预测效果良好。

Claims (1)

  1. 一种基于SSAE-LSTM模型的深孔加工刀具磨损量监测方法，其特征在于，步骤如下：
    第一步，深孔加工过程中的振动和声音信息采集
    将加速度传感器通过磁座吸附在刀杆保持架的轴瓦外部，将传声器放置于深孔工件加工入口处，对加工过程中的刀杆振动以及切削噪声进行采集；
    第二步，SSAE模型的构建和训练
    将采集到的振动和声音数据拆分成样本数据；设拆分后的某一个训练样本数据为x＝[x 1,x 2,…,x i]，则稀疏自动编码器SAE的输入输出关系表示为：
    Figure PCTCN2019105282-appb-100001
    Figure PCTCN2019105282-appb-100002
    式中，f(·)表示神经元的激活函数，
    Figure PCTCN2019105282-appb-100003
    表示输入层与隐含层的权重，
    Figure PCTCN2019105282-appb-100004
    表示隐含层与输出层的权重，
    Figure PCTCN2019105282-appb-100005
    示隐含层各神经元的偏置，
    Figure PCTCN2019105282-appb-100006
    表示输出层各神经元的偏置；
    将SAE的损失函数定义为重构后的数据与原始数据的均方误差L：
    Figure PCTCN2019105282-appb-100007
    式中，
    Figure PCTCN2019105282-appb-100008
    是SAE器的输出数据；β是稀疏惩罚项参数，作用是稀疏惩罚项在损失函数当中所占的比重；ρ是稀疏性参数；
    Figure PCTCN2019105282-appb-100009
    是隐含层第g个神经元的激活度；通过误差反向传播和梯度下降法调整SAE各层参数，使损失函数值达到最小；
    将多个SAE组成堆叠自编码器网络SSAE，采用贪婪逐层训练法训练SSAE网络，其过程为：首先利用样本数据来训练网络的第1层，该层将样本数据转换为由隐含层输出值组成的特征向量A；然后把A作为第2层的的输入，训练 得到第2层的特征向量B；以此类推至各层训练完毕；训练完成后，将所有样本数据输入SSAE网络，得到特征向量；
    第三步，深度LSTM网络的构建和训练
    设长短时记忆LSTM网络的记忆单元在每一个时间步长t更新一次，则其输入门的值i t和记忆单元候选状态值
    Figure PCTCN2019105282-appb-100010
    分别为：
    i t＝σ(W i·[h t-1,x t]+b i)
    Figure PCTCN2019105282-appb-100011
    则其t时刻遗忘门的值f t为：
    f t＝σ(W f·[h t-1,x t]+b f)
    由输入门、遗忘门和记忆单元候选状态的值计算出记忆单元t时刻当前值C t：
    Figure PCTCN2019105282-appb-100012
    最后可得记忆单元输出值h t为：
    h t＝σ(W o[h t-1,x t]+b o)*tanh(C t)
    式中，x t为t时刻记忆单元的输入；W为模型的权重参数；b为模型的偏差向量；σ和tanh为Sigmoid激活函数和Tanh激活函数；将多层LSTM网络堆叠起来，构成深度LSTM网络；
    将特征向量分成训练集特征向量和测试集特征向量；以训练集特征向量作为深度LSTM网络的输入数据，对深度LSTM网络进行训练，输出深孔加工刀具的磨损量；然后再用测试集特征向量对深度LSTM网络进行测试；如果预测平均误差低于设定的δ值，则测试合格，模型用于磨损量监测；反之，则测试不合格，回到第二步重新对网络进行训练；
    第四步，深孔加工刀具磨损量的实时监测
    将训练好的SSAE网络和深度LSTM网络进行连接生成SSAE-LSTM模型； 在实际加工过程中，将实时数据输入至SSAE-LSTM模型中，输出刀具磨损量。
PCT/CN2019/105282 2019-09-11 2019-09-11 基于ssae-lstm模型的深孔加工刀具磨损量监测方法 WO2021046737A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/CN2019/105282 WO2021046737A1 (zh) 2019-09-11 2019-09-11 基于ssae-lstm模型的深孔加工刀具磨损量监测方法

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2019/105282 WO2021046737A1 (zh) 2019-09-11 2019-09-11 基于ssae-lstm模型的深孔加工刀具磨损量监测方法

Publications (1)

Publication Number Publication Date
WO2021046737A1 true WO2021046737A1 (zh) 2021-03-18

Family

ID=74866775

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/105282 WO2021046737A1 (zh) 2019-09-11 2019-09-11 基于ssae-lstm模型的深孔加工刀具磨损量监测方法

Country Status (1)

Country Link
WO (1) WO2021046737A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113560955A (zh) * 2021-06-09 2021-10-29 西安电子科技大学 一种数控机床刀具剩余使用寿命预测方法、系统及应用
CN117236381A (zh) * 2023-11-08 2023-12-15 智能制造龙城实验室 一种基于三元组长短时记忆神经网络的刀具磨损监测方法

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS59211856A (ja) * 1983-05-13 1984-11-30 シ−メンス、アクチエンゲゼルシヤフト 電気変換信号の評価方法
DE3627796C1 (en) * 1986-08-16 1987-10-22 Klaus Dipl-Ing Nordmann Device for monitoring the state and breakage of rotating tools by means of measurements of structure-borne sound
JP2000233369A (ja) * 1999-02-15 2000-08-29 Noritake Co Ltd 研削状態監視装置およびドレッシング状態監視装置
CN108319962A (zh) * 2018-01-29 2018-07-24 安徽大学 一种基于卷积神经网络的刀具磨损监测方法
CN109048492A (zh) * 2018-07-30 2018-12-21 北京航空航天大学 基于卷积神经网络的刀具磨损状态检测方法、装置及设备
CN110153802A (zh) * 2019-07-04 2019-08-23 西南交通大学 一种基于卷积神经网络和长短时记忆神经网络联合模型的刀具磨损状态辨识方法

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS59211856A (ja) * 1983-05-13 1984-11-30 シ−メンス、アクチエンゲゼルシヤフト 電気変換信号の評価方法
DE3627796C1 (en) * 1986-08-16 1987-10-22 Klaus Dipl-Ing Nordmann Device for monitoring the state and breakage of rotating tools by means of measurements of structure-borne sound
JP2000233369A (ja) * 1999-02-15 2000-08-29 Noritake Co Ltd 研削状態監視装置およびドレッシング状態監視装置
CN108319962A (zh) * 2018-01-29 2018-07-24 安徽大学 一种基于卷积神经网络的刀具磨损监测方法
CN109048492A (zh) * 2018-07-30 2018-12-21 北京航空航天大学 基于卷积神经网络的刀具磨损状态检测方法、装置及设备
CN110153802A (zh) * 2019-07-04 2019-08-23 西南交通大学 一种基于卷积神经网络和长短时记忆神经网络联合模型的刀具磨损状态辨识方法

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113560955A (zh) * 2021-06-09 2021-10-29 西安电子科技大学 一种数控机床刀具剩余使用寿命预测方法、系统及应用
CN113560955B (zh) * 2021-06-09 2022-09-09 西安电子科技大学 一种数控机床刀具剩余使用寿命预测方法、系统及应用
CN117236381A (zh) * 2023-11-08 2023-12-15 智能制造龙城实验室 一种基于三元组长短时记忆神经网络的刀具磨损监测方法
CN117236381B (zh) * 2023-11-08 2024-02-02 智能制造龙城实验室 一种基于三元组长短时记忆神经网络的刀具磨损监测方法

Similar Documents

Publication Publication Date Title
CN110576336A (zh) 基于ssae-lstm模型的深孔加工刀具磨损量监测方法
US20210197335A1 (en) Data Augmentation Method Based On Generative Adversarial Networks In Tool Condition Monitoring
Wang et al. Multiscale convolutional attention network for predicting remaining useful life of machinery
WO2021128576A1 (zh) 基于生成式对抗网络的刀具状态监测数据集增强方法
WO2021046737A1 (zh) 基于ssae-lstm模型的深孔加工刀具磨损量监测方法
CN113375941B (zh) 高速动车组轴承的开集故障诊断方法
WO2017191249A1 (en) Speech enhancement and audio event detection for an environment with non-stationary noise
CN106271881B (zh) 一种基于SAEs和K-means的刀具破损监测方法
CN109872415B (zh) 一种基于神经网络的车速估计方法及系统
Li et al. Chatter identification of thin-walled parts for intelligent manufacturing based on multi-signal processing
CN111832432A (zh) 一种基于小波包分解和深度学习的刀具磨损实时预测方法
Kumar et al. Fault diagnosis of bearings through vibration signal using Bayes classifiers
EP3945472A3 (en) Method of and system for online machine learning with dynamic model evaluation and selection
Zheng et al. TCM in milling processes based on attention mechanism-combined long short-term memory using a sound sensor under different working conditions
CN113707175B (zh) 基于特征分解分类器与自适应后处理的声学事件检测系统
Yoon et al. Full ceramic bearing fault diagnosis using LAMSTAR neural network
CN115221973A (zh) 一种基于增强加权异质集成学习的航空轴承故障诊断方法
Wan et al. Chatter detection for micro milling considering environment noises without the requirement of dominant frequency
Li et al. Deep learning based end-to-end rolling bearing fault diagnosis
Xiao et al. Research on gearbox fault diagnosis system based on BP neural network optimized by particle swarm optimization
Zafar et al. Tool health monitoring using airborne acoustic emission and a PSO-optimized neural network
Zhang et al. CNN4GCDD: A one-dimensional convolutional neural network-based model for gear crack depth diagnosis
WO2021148392A1 (de) Verfahren und vorrichtung zur objektidentifikation basierend auf sensordaten
Medina et al. A LSTM neural network approach using vibration signals for classifying faults in a gearbox
CN111753876A (zh) 一种基于深度神经网络的产品质量检测方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19944718

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19944718

Country of ref document: EP

Kind code of ref document: A1