基于工艺参数和监测信号特征的排屑钻削表面粗糙度预测

发布时间：2019-04-02 02:57

【摘要】：排屑钻与普通钻削加工相比,可提高钻削加工质量,显著改善孔加工表面粗糙度质量。但排屑钻加工过程和普通钻削一样都处于半封闭或者封闭环境,孔加工表面粗糙度也难以检测和分析。本文拟结合钻削工艺参数和监测信号特征,开展排屑钻孔加工表面粗糙度预测研究。所开展的主要工作包括监测平台搭建、信号消噪处理、工艺参数和监测信号特征对粗糙度的影响规律、预测模型的建立与验证等方面。(1)排屑钻监控平台搭建与数据采集。搭建排屑钻监控平台,采集排屑钻加工过程中的振动信号、声发射信号,以及所加工孔壁粗糙度值,并采用最小二乘拟合的方法对信号进行趋势项处理。(2)信号消噪处理。针对在钻削加工噪声背景下振动信号特征识别和提取困难的问题,提出了一种小波包分频谱减去噪方法。根据钻削信号在时频域特点,首先将钻削前机床空转信号视为监测信号的“加性噪声”;然后,采用小波包分解将“加性噪声”和监测信号进行分频处理,确定各频带帧数;最后,对各个子频带内“加性噪声”的相应频带进行谱减处理,再重构钻削振动信号。(3)工艺参数和监测信号特征对粗糙度的影响规律。依据所采集的实验数据,首先分析了不同工艺参数对监测信号特征以及孔壁粗糙度的影响规律。然后通过方差分析的方法研究了不同工艺参数对监测信号特征和表面粗糙度影响的显著性。最后,分析了监测信号特征与孔壁表面粗糙度的对应关系。(4)粗糙度预测模型建立与验证。首先确定了神经网络的输入层与输出层节点数,然后针对BP神经网络隐含层节点数无法确定的问题,采用动态调节隐含层节点数的方法,对比不同结构预测值的准确度,确定最优网络结构。最后,通过对试验样本进行仿真分析,对粗糙度预测模型的有效性进行验证。理论分析和实验结果表明:采用本文所建立的粗糙度预测模型,能够有效预测排屑钻表面粗糙度。同时该方法可有效克服传统粗糙度检测采用人工抽检所导致的漏检、检测效率不高等缺点,为实现排屑钻粗糙度预测提供了新的方法和理论基础。
[Abstract]:Compared with common drilling, chip removal drill can improve the quality of drilling and improve the surface roughness of hole machining. However, the process of chip removal drilling is in a semi-closed or closed environment, and the surface roughness of hole machining is also difficult to detect and analyze. In this paper, combining with the parameters of drilling process and the characteristics of monitoring signals, the prediction of surface roughness of chip removal drilling is carried out. The main work includes the construction of monitoring platform, signal de-noising processing, the influence of process parameters and monitoring signal characteristics on roughness, the establishment and verification of prediction model and so on. (1) the construction of monitoring platform for chip removal drill and data acquisition. The monitoring platform of chip removal drill is set up to collect vibration signal, acoustic emission signal and the roughness value of the hole wall in the process of chip removal drilling. The least square fitting method is used to process the trend term of the signal. (2) the signal is de-noised. In order to solve the problem of difficult recognition and extraction of vibration signals in the background of drilling noise, a wavelet packet spectrum division subtract method is proposed. According to the characteristics of drilling signal in time and frequency domain, the machine tool idle signal before drilling is regarded as the "additive noise" of the monitoring signal, and then the "additive noise" and the monitoring signal are processed by using wavelet packet decomposition to determine the frame number of each frequency band. Finally, the corresponding frequency band of "additive noise" in each sub-band is subtracted and then the drilling vibration signal is reconstructed. (3) the influence of technological parameters and monitoring signal characteristics on roughness. Based on the experimental data collected, the influence of different process parameters on the characteristics of the monitoring signal and the roughness of the hole wall was analyzed. Then, the effects of different process parameters on the characteristics of monitoring signals and surface roughness were studied by ANOVA. Finally, the relationship between the characteristics of the monitoring signal and the surface roughness of the hole wall is analyzed. (4) the prediction model of roughness is established and verified. Firstly, the number of nodes in the input layer and the output layer of the neural network is determined. Then, aiming at the problem that the number of hidden layer nodes in the BP neural network cannot be determined, the method of dynamically adjusting the number of nodes in the hidden layer is adopted to compare the accuracy of the predicted values of different structures. The optimal network structure is determined. Finally, the validity of the roughness prediction model is verified by the simulation analysis of the test samples. The theoretical analysis and experimental results show that the roughness prediction model established in this paper can effectively predict the surface roughness of chip removal drills. At the same time, this method can effectively overcome the shortcomings of traditional roughness detection caused by manual sampling, such as low detection efficiency and so on. It provides a new method and theoretical basis for the prediction of chip removal drill roughness.
【学位授予单位】：湘潭大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TG52

【参考文献】