Real-Time Prediction of Workpiece Errors for a CNC Turning Centre, Part 1. Measurement and Identification

This paper analyses the error sources of the workpiece in bar turning, which mainly derive from the geometric error of machine tools, i.e. the thermally induced error, the error arising from machine–workpiece–tool system deflection induced by the cutting forces. A simple and low-cost compact measuring system combining a fine touch sensor and Q-setter of machine tools (FTS FQ) is developed, and applied to measure the workpiece dimensions. An identification method for workpiece errors is also presented. The workpiece errors which are composed of the geometric error, thermal error, and cutting force error can be identified according to the measurement results of each step. The model of the geometric error of a two-axis CNC turning centre is established rapidly based on the measurement results by using an FTSFQ setter and coordinate measuring machine (CMM). Experimental results show that the geometric error can be compensated by modified NC commands in bar turning.     

Real-Time Prediction of Workpiece Errors for a CNC Turning Centre, Part 2. Modelling and Estimation of Thermally Induced Errors

Real-time modelling and estimation of thermally induced error have very important contributions for precision machining. In this paper, a novel radial basis function (RBF) neural network, which combines a regression tree and a radial basis function, has been considered and selected to model the thermally induced errors of CNC turning centres. Then, a simple and low-cost compact measurement system is applied to measure the time variant thermal errors of CNC turning centres. The thermally induced errors are estimated in real-time using the trained RBF neural network. The application of the measurement system and the novel RBF neural network are described in detail in this paper. The proposed approach is verified through some tests under different cutting conditions.     

Methodology of Parametric Programming for Error Compensation on CNC Centres

Although work has been carried out on parametric programming on CNC centres, there have been few papers which focus on error compensation. Parametric programming for error compensation is presented in this paper on the basis of a simple model of machining system deflections induced by the radial cutting force in CNC turning operations. The resulting errors are introduced as compensation values to the conventional tool movements along the programmed tool path. This can result in a complex tool path. Parametric programming is applied to handle this complexity for error compensation.     

Characterisation of the workpiece dilatation phenomenon during machining using the neural network method. Application to NC turning

Dilatation of workpieces during machining is a major source of defects. With the current trend for re-treatment of cutting and cleaning fluids becoming compulsory, lubrication by a stream of oil and dry machining are becoming more widely used in aluminium alloy machining. Indeed, this makes it easier to recycle chippings and greatly simplifies the cleaning and grease removal phases for workpieces that are compulsory before any finishing surface treatment. However, the workpiece’s deformation during machining must be taken into account. This is especially true for NC turning of machining diameters with very tight tolerances. Here we propose a method based on the use of a neural network intended to model changes in the workpiece’s dimensions to correct tool paths. This study covered machining of workpieces made of 2017 T4 aluminium alloy during the turning phase. We first conducted preliminary tests on a workpiece to highlight workpiece dilatation. We then implemented a neural network to predict this deformation to be able to compensate for it. The results of a first test campaign gave us knowledge of the network then a second test campaign was used to validate that network. To finish off, we machined a test workpiece in order to test and analyse network performance.     

A Study of High-Precision CNC Lathe Thermal Errors and Compensation

This study is addressed at the thermal deformation errors resulting from temperature rise that contribute to 40%–70% of the precision errors in machining at a turning centre, and proposes an economic, accurate, and quick measurement method. It also investigates the thermal error differentials between static idle turning and in the actual cutting environment. The temperature measurement units are intelligent IC temperature sensors with correction circuits. The A/D card extracts and transforms data and saves data in the computer files, and the displacement sensor measures the displacement deviation online during cutting. The temperatures and the deviation of thermal drifts so obtained are used to establish the relationship function using multivariable linear regression and nonlinear exponential regression models, respectively. Finally, this paper compares software compensation methods for the thermal-drift relationship. As proven by experiments, the software compensation method can limit the thermal error of a turning centre to within 5 μm. Moreover, the software compensation for the thermal error relationship using a single variable nonlinear exponent regression model can reduce the error by 40% to 60%.     

基于神经网络的细长轴车削加工尺寸误差预测研究

为优化细长轴车削加工,应用人工神经网络方法建立使用跟刀架车削细长轴时的加工尺寸误差预测模型,并基于获得的预测模型研究切削用量对尺寸误差的影响。试验结果表明,该模型具有良好的预测精度,为细长轴车削加工切削用量的选择提供了依据。     

ARX Modelling and Compensation of Roundness Errors in Taper Turning

The forecasting compensatory control (FCC) strategy is successfully applied to the taper turning process to improve the roundness accuracy of the workpiece during cutting. This strategy, which is based on active error sensing, stochastic modelling and forecasting control, is capable of compensating both repeatable and non-repeatable errors. In the stochastic modelling, the effect of cutting force is considered, thus yielding an autoregressive model with exogeneous input (ARX). The resultant force is obtained from the outputs of the piezoelectric force sensor while the relative motion error between the tool and the workpiece is determined by means of a master taper and a capacitive sensor. The forecast error is sent to a piezoelectric actuator that moves the cutter to compensate for the error motion. Practical tests were performed on an experimental lathe with the FCC system using models with different orders and forgetting factors. A maximum improvement of 33% can be achieved by this FCC strategy using ARX (3,3) and forgetting factor of 0.997.     

基于电流测量的数控机床切削力误差建模与实时补偿

从数控机床主轴驱动系统的传动机理出发，系统地研究主轴伺服电机电流信号与切削力之间的关系。利用模糊神经网络理论完成切削力误差建模，研制了数控机床的切削力误差实时补偿系统，并通过实例进行了验证。该研究方案避免了传统用测力仪进行切削力监测花费大、调试难、可靠性低等问题，所建切削力误差模型鲁棒性强，具有重要的工程实用价值。     

Thermal Error Measurement and Real Time Compensation System for the CNC Machine Tools Incorporating the Spindle Thermal Error and the Feed Axis Thermal Error

Thermally induced errors have been significant factors affecting machine tool accuracy. In this paper, the thermal spindle error and thermal feed axis error have been considered, and a measurement/compensation system for thermal error is introduced. Several modelling techniques for thermal errors are also implemented for the thermal error prediction; i.e. multiple linear regression, neural network, and the system identification methods, etc. The performances of the thermal error modelling techniques are evaluated and compared, showing that the system identification method is the optimum model having the least deviation. The thermal error model for the feed axis is composed of geometric terms and thermal terms. The volumetric errors are calculated, combining the spindle thermal error and feed axis thermal error. In order to compensate for the thermal error in real-time, the coordinates of the CNC controller are modified in the PMC program. After real-time compensation, the machine tool accuracy improved about 4–5 times. ID="A1" Correspondence and offprint requests to: Dr H. J. Pahk, School of Mechanical and Aerospace Engineering, Seoul National University, San 56–1, Shinlim-Dong, Kwanak-Ku, Seoul 151–742, Korea. E-mail: hjpahk@plaza.snu.ac.kr     

基于RBF神经网络的数字闭环光纤陀螺温度误差补偿

为了消除数字闭环光纤陀螺温度误差,设计了基于径向基函数（RBF）神经网络的温度误差补偿方案,对该方案所采用的标度因数误差模型和偏置误差模型进行了研究。首先,根据光纤陀螺的温度误差分布情况设计了标度因数误差和偏置误差联合补偿的方案。接着,将基于多尺度分析的噪声和趋势项分离算法应用于建模数据预处理,以提高建模数据准确性。然后,建立了RBF神经网络模型,并改进模型的学习方法以防止网络的过拟合。最后,讨论了模型输入向量对神经网络规模的影响。温度补偿的结果表明：标度因数误差模型的残差均方（RMS）达到0.73 ,偏置误差模型的RMS达到0.051 。该建模方法可以满足中、高精度光纤陀螺实时温度补偿的要求。     

Automated measurement and compensation of thermally induced error maps in machine tools

In this paper, a direct method of machine tool calibration is adopted to model and predict thermally induced errors in machine tools. This method uses a laser ball bar (LBB) as the calibration instrument and is implemented on a two-axis computerized numerical control turning center (CNC). Rather than individually measuring the parametric errors to build the error model of the machine, the total positioning errors at the cutting tool and spindle thermal drifts are rapidly measured using the LBB within the same experimental setup. Unlike conventional approaches, the spindle thermal drifts are derived from the true spindle position and orientation measured by the LBB. A neural network is used to build a machine model in an incremental fashion by correlating the measured errors with temperature gradients of the various heat sources during a regular thermal duty cycle. The machine model developed by the neural network is further tested using random thermal duty cycles. The performance of the system is also evaluated through cutting tests under various thermal conditions. A substantial improvement in the overall accuracy was obtained.     

In-Process Modelling and Estimation of Thermally Induced Errors of a Machine Tool During Cutting

A compact measurement system was developed to measure the time variant machine tool errors during cutting. The system is composed of a gauge, a sensor holder, 5 gap-sensors and a PC. The gauge is made of invar which has a very low thermal expansion coefficient, and is often used to measure the thermally induced errors of a machine tool. A new neural network model was considered to estimate the time variant machine tool errors during cutting using a new concept of input values. The detail of the model proposed is described in the paper together with experimental methodologies using a compact measurement system to examine the validity of this approach. These schemes were implemented on a small vertical-machining centre.     

An accuracy design approach for a multi-axis NC machine tool based on reliability theory

Accuracy design constitutes an important role in machine tool designing. It is used to determine the permissible level of each error parameter of a machine tool, so that any criterion can be optimized. Geometric, thermal-induced, and cutting force-induced errors are responsible for a large number of comprehensive errors of a machine tool. These errors not only influence the machining accuracy but are also of great importance for accuracy design to be performed. The aim of this paper is the proposal of a general approach that simultaneously considered geometric, thermal-induced, and cutting force-induced errors, in order for machine tool errors to be allocated. By homogeneous transformation matrix (HTM) application, a comprehensive error model was developed for the machining accuracy of a machine tool to be acquired. In addition, a generalized radial basis function (RBF) neural network modeling method was used in order for a thermal and cutting force-induced error model to be established. Based on the comprehensive error model, the importance sampling method was applied for the reliability and sensitivity analysis of the machine tool to be conducted, and two mathematical models were presented. The first model predicted the reliability of the machine tool, whereas the second was used to identify and optimize the error parameters with larger effect on the reliability. The permissible level of each geometric error parameter can therefore be determined, whereas the reliability met the design requirement and the cost of this machining was optimized. An experiment was conducted on a five-axis machine tool, and the results confirmed the proposed approach being able to display the accuracy design of the machine tool.     

基于神经模糊控制理论的数控机床热误差建模

将基于神经模糊控制理论的建模方法--模糊神经网络建模法应用到数控机床热误差建模当中,讨论了热误差模糊神经网络的结构及建模原理;对大型数控龙门导轨磨床主轴箱系统进行建模试验,采用非接触式红外温度测量仪和千分表分别测量主轴箱系统温度值与主轴热误差,得到两组独立的试验数据,一组用来建立主轴箱系统热误差模糊神经网络预报模型,另一组用来对模型进行验证。试验结果表明,模糊神经网络模型预测精度高,泛化能力强;将模糊神经网络建模方法与径向基函数神经网络建模方法进行综合对比,分析结果表明,模糊神经网络建模方法具有更好的建模效率、建模鲁棒性及预测性能。     

车削过程切削力的计算机数值仿真

切削力是表征切削过程最重要特征的物理量,其动态变化将直接影响加工过程中刀具与工件的相对位移、刀具磨损和表面加工质量等,所以对切削力建模是进行加工过程物理仿真研究的基础。因此在基于实时工况的切削实验研究基础上,考虑切削参数的因素,利用BP（back pmpagation）神经网络建立车削过程中的切削力的仿真模型。通过大量的样本训练,使神经网络能够对切削力进行较准确地数值仿真。     

柔性和摩擦力不确定条件下RBF神经网络自适应轨迹跟踪方法

提出一种基于径向基神经网络(Radial basis function, RBF)的力/位置混合自适应控制方法并用于机器人轨迹跟踪控制,解决机器人柔性末端执行器轨迹跟踪过程中柔性和摩擦力模型难以精确描述的问题。RBF神经网络是一种高效的前馈式神经网络,具有其他前向网络所不具有的非线性逼近性能和全局最优特性,并且网络结构简单,训练速度快。设计一种基于RBF神经网络非线性逼近能力来估计模型中的不确定参数的自适应控制器,给出控制器中神经网络权值更新规则,并证明所设计控制器输出力和位置误差的最终一致有界性。将该控制器应用于风管清扫机器人仿真试验,结果表明该自适应控制器能很好地用于柔性和摩擦力不确定条件下轨迹跟踪控制,与传统自适应控制方法相比具有更精确的跟踪特性和更强的鲁棒性。     

Real-Time Diagnostics of Metal-Cutting Machines by Means of Recurrent LSTM Neural Networks

Real-time diagnostics of modules in metal-cutting machines may be based on neural-network algorithms for simulation of the standard process, identification of defects, and the introduction of corrections in the cutting machine’s control system. The machining conditions in normal operation of the machine are recorded by means of a trained neural network with long short-term memory (LSTM network). In real-time operation, the difference between the standard neural-network model and the actual process characteristics is used to determine the type of defect and the module of the machine where it occurs on the basis of a second neural network, the classification unit.     

The monitoring of flank wear on the CBN tool in the hard turning process

In precision hard turning, tool flank wear is one of the major factors contributing to the geometric error and thermal damage in a machined workpiece. Tool wear not only directly reduces the part geometry accuracy but also increases the cutting forces drastically. The change in cutting forces causes instability in the tool motion, and in turn, more inaccuracy. There are demands for reliably monitoring the progress of tool wear during a machining process to provide information for both correction of geometric errors and to guarantee the surface integrity of the workpiece. A new method for tool wear monitoring in precision hard turning is presented in this paper. The flank wear of a CBN tool is monitored by feature parameters extracted from the measured passive force, by the use of a force dynamometer. The feature parameters include the passive force level, the frequency energy and the accumulated cutting time. An ANN model was used to integrate these feature parameters in order to obtain more reliable and robust flank wear monitoring. Finally, the results from validation tests indicate that the developed monitoring system is robust and consistent for tool wear monitoring in precision hard turning.     

基于径向基函数神经网络的压电式六维力传感器解耦算法

针对四点支撑结构的压电式六维力传感器线性度差,维间耦合严重的问题,提出了基于径向基函数(RBF)神经网络的解耦算法。分析了耦合产生的主要原因,建立了RBF神经网络模型。通过对六维力传感器进行标定实验获取解耦所需的实验数据,并对实验数据进行处理。然后采用RBF神经网络优化传感器输出系统的多维非线性解耦算法,解耦出传感器的输入输出映射关系,得到解耦后的传感器输出数据。对传感器解耦后的数据分析表明:采用RBF神经网络的解耦算法得到的最大Ⅰ类误差和Ⅱ类误差分别为1.29%、1.56%。结果显示:采用RBF神经网络的解耦算法,能够更加有效地减小传感器的Ⅰ类误差和Ⅱ类误差,满足了传感器两类误差指标均低于2%的要求。该算法有效地提高了传感器的测量精度,基本解决了传感器解耦困难的难题,     

The effect of high frequency and duty cycle in electrochemical microdrilling

A useful method for increasing the metal removal rate and the tool life of turning systems is by controlling a constant cutting force. This study developed a hybrid self-organizing fuzzy and radial basis-function neural-network controller (HSFRBNC) for turning systems to maintain constant cutting force operation. The HSFRBNC uses a radial basis function neural-network to regulate both the learning rate and the weighting distribution of a self-organizing fuzzy controller (SOFC) in real time to appropriate values, instead of obtaining these values by trial and error. It not only eliminates the difficulties of finding appropriate membership functions and fuzzy rules in the design of the fuzzy logic controller but also solves the problem of determining suitable parameters of the SOFC. The HSFRBNC has better control performance than the SOFC for manipulating constant cutting force in the turning system, as shown in the simulation results.