【作者】 宋阳；

【导师】 刘富；

【作者基本信息】 吉林大学 ， 系统工程， 2021， 硕士

【摘要】 刀具磨损状态实时监测技术是先进制造系统的关键技术,是机械加工中的一个重要环节,有效的刀具状态监测对于提高生产效率、降低生产成本、改善产品质量等具有十分重要的意义。但是在实际应用过程中,现有的刀具磨损状态监测设备往往存在检测精度不足、设备庞大、造价昂贵、前期数据标定工作量大的问题,因此本文在分析刀具状态监测技术研究现状的基础上,从振动信号入手,对刀具磨损状态监测进行了研究,主要的工作内容如下:首先,搭建一种基于新式仿生柔性振动传感元件的刀具振动信号检测系统。新式仿生柔性振动传感元件是一种仿蝎子缝感受器感知机理的感知结构,具有检测精度高、制造成本低的优势。考虑工业检测中数控机床内部加工环境的影响,根据仿生柔性振动传感元件结构设计封装装置,将其封装为刚性的仿生传感器,使之能够在机床加工过程中精准、实时地检测刀具振动信号。搭建信号采集系统,将刀具振动信号上传至上位机进行存储,在实时检测过程中,记录下实时刀具状态标签。其次,采用支持向量机算法对采集到的刀具振动信号进行刀具正常工作和后刀面故障两种状态的诊断。首先采用小波包去噪去除刀具振动信号中的环境噪声,然后将信号归一化,提取信号的时域特征和频域特征,主要包括最大值、均值、方差、裕度系数、脉冲因子、重心频率等,之后采用支持向量机算法对刀具振动信号样本特征进行训练,建立刀具故障诊断模型,实验结果验证了本文所提的基于仿生传感器的刀具振动信号检测系统的有效性。最后,通过改进的无参数K-means聚类算法实现刀具的磨损状态分析。传统K-means算法需要人为确定参数,而且随机选取初始聚类中心会导致局部最优的问题,结合基于密度峰值的聚类算法,提出一种改进的可以自动确定聚类个数和聚类中心的无参数K-means算法。首先计算每个样本点的密度和离散度并建立决策图,然后根据决策图的位置信息,计算样本点与其邻近点的距离均值,根据数据点的距离分布规律,确定聚类中心和聚类个数,将筛选的聚类中心和聚类个数作为K-means聚类算法的聚类中心和聚类个数,实现无参数K-means算法改进。并在高斯数据集和UCI数据集上,验证了算法的有效性和可行性,最后将本文提出的改进算法用于对机床刀具振动信号数据进行聚类,实现刀具磨损状态识别。综上,本文进行了一种基于蝎子振动感知机理的刀具磨损状态监测研究,封装新式仿生柔性振动传感元件检测刀具振动信号,提高了检测精度,降低了检测成本;采用支持向量机算法对采集到的刀具振动信号进行刀具正常工作和后刀面故障两种状态诊断;提出了一种改进的无参数K-means算法应用于刀具磨损状态监测,能够根据刀具状态变化特性识别刀具运行状态,在获得较优刀具状态识别结果的同时,减少人工数据标定,对于准确实时提取数控机床刀具运行状态具有较高的实际应用价值。更多还原

【Abstract】 The real-time monitoring of tool wear status is a key technology of advanced manufacturing systems and an important part of machining.Effective tool status monitoring is of great significance for improving production efficiency,reducing production costs,and improving product quality.However,in the actual application process,the existing tool wear condition monitoring equipment often has the problems of insufficient detection accuracy,huge equipment,expensive cost,and large workload of preliminary data calibration.Therefore,this article analyzes the current research status of tool condition monitoring technology.Starting from the vibration signal,the monitoring of the running state of the tool is researched,and the main work content is as follows:Firstly,a tool vibration signal detection system based on a new bionic flexible vibration sensor element was built.The new bionic flexible vibration sensitive element is a sensing structure that mimics the sensing mechanism of the scorpion seam receptor,and has the advantages of high detection accuracy and low manufacturing cost.Considering the influence of the internal processing environment of CNC machine tools in industrial testing,the packaging device is designed according to the structure of the bionic flexible vibration sensitive element,and the rigid bionic sensor is packaged to enable it to detect tool vibration signals accurately and in real time during machine tool processing.Set up a signal acquisition system,upload the tool vibration signal to the host computer for storage,and record the real-time tool status label during the real-time detection process.Secondly,this thesis uses the support vector machine algorithm to diagnose the tool’s normal working and flank faults on the collected tool vibration signals.First,use wavelet packet denoising to remove the environmental noise in the tool vibration signal,and then normalize the signal to extract the time-domain and frequency-domain features of the signal,mainly including maximum value,mean value,variance,margin coefficient,pulse factor,center of gravity Frequency etc.,After that,the support vector machine algorithm is used to train the tool vibration signal sample characteristics,and the tool fault diagnosis model is established.The experimental results verify the effectiveness of the tool vibration signal detection system based on the bionic sensor proposed in this thesis.Finally,this thesis realizes the analysis of tool wear status through an improved parameter-free K-means clustering algorithm.The traditional K-means algorithm needs to manually determine the parameters,and the random selection of the initial clustering center will lead to the local optimal problem.Combined with the clustering algorithm based on the density peak,an improved algorithm that can automatically determine the number of clusters and the clustering center is proposed.K-means algorithm without parameters.First calculate the density and dispersion of each sample point and establish a decision diagram.Then,according to the location information of the decision diagram,calculate the average distance between the sample point and its neighbors,and determine the cluster center and cluster number according to the distance distribution law of the data points.The selected cluster center and the number of clusters are used as the cluster center and cluster number of the K-means clustering algorithm to realize the improvement of the parameter-free K-means algorithm.And on the Gaussian data set and UCI data set,the effectiveness and feasibility of the algorithm are verified.Finally,the improved algorithm proposed in this thesis is used to cluster machine tool vibration signal data to realize tool wear status recognition.In summary,this thesis has carried out a research on tool wear status monitoring based on the scorpion vibration sensing mechanism,and encapsulated a new type of bionic flexible vibration sensitive element to detect tool vibration signals,which improved the detection accuracy and reduced the detection cost;support vector machine algorithm is used to diagnose the tool vibration signals in two states: normal working and flank faults;an improved parameter-free K-means algorithm is proposed for tool wear status monitoring,which can identify the running state of the tool according to the change characteristics of the tool state,and while obtaining a better tool state recognition result.It has high practical application value for accurately and real-time extracting the running state of CNC machine tools.更多还原