XGBoost机器学习在光电编码器误差补偿中的应用

光电编码器检测系统的误差主要受基准光电编码器测角误差、数据采集误差、检测系统同轴误差影响。其中，基准光电编码器的测角误差可进行补偿。因此设计了一种基于极度梯度提升树（extreme gradient boosting,XGBoost）机器学习的算法用来补偿基准光电编码器的误差。经该算法补偿后，静态精度提高了35倍，标准差由3.62″减小至0.13″，最大误差值由5.53″降低至0.39″。与传统的误差反传（back progagation,BP）神经网络算法以及径向基函数（radial basis function,RBF）神经网络算法补偿效果相比，XGBoost的补偿效果更优。XGBoost机器学习算法有效降低了基准光电编码器的测量误差，提高了光电编码器检测系统的检测精度。     

Thermal error analysis,modeling and compensation of five-axis machine tools

The role of five-axis CNC machine tools (FAMT) in the manufacturing industry is becoming more and more important, but due to the large number of heat sources of FAMT, the thermal error caused by them will be more complicated. To simplify the complicated thermal error model, this paper presents a new modelling method for compensation of the thermal errors on a cradle-type FAMT. This method uses artificial neural network (ANN) and shark smell optimization (SSO) algorithm to evaluate the performance of FAMT, and developing the thermal error compensation system, the compensation model is verified by machining experiments. Generally, the thermal sensitive point screening is performed by a method in which a large number of temperature sensors are arranged randomly, it increases the workload and may cause omission of the heat sensitive point. In this paper, the thermal imager is used to screen out the temperature sensitive points of the machine tool (MT), then the temperature sensor is placed at the position of the heat sensitive point of the FAMT, and the collected thermal characteristic data is used for thermal error modeling. The C-axis heating test, spindle heating test, and the combined movement test are applied in this work, and the results show that the shark smell optimization artificial neural network (SSO-ANN) model was compared to the other two models and verified better performance than back propagation artificial neural network (BP-ANN) model and particle swarm optimization neural network (PSO) model with the same training samples. Finally, a compensation experiment is carried out. The compensation values, which was calculated by the SSO-ANN model are sent to the real-time error compensation controller. The compensation effect of the model is then tested by machining the ‘S’-shaped test piece. Test results show that the 32 % reduction in machining error is achieved after compensation, which means this method improves the accuracy and robustness of the thermal error compensation system.

     

基于神经网络的光纤布拉格光栅触觉信号解耦研究

针对应用于电子皮肤的柔性光纤布拉格光栅触觉传感器的输出信号是施加载荷位置、大小等信息的非线性、多维耦合问题,在对光纤布拉格光栅触觉传感阵列进行力学有限元仿真的基础上,提出了将误差逆传播神经网络和径向基函数神经网络应用于仿真和实验的触觉信号解耦方法。对实验数据的神经网络解耦结果表明,相对于误差逆传播神经网络,径向基函数神经网络具有更强的抗噪声能力,能够更好地逼近含有噪声的触觉多维非线性实验数据之间的映射关系。经径向基函数神经网络解耦后,传感器阵列的空间分辨率为5 mm,对压力位置和大小感知的最小相对误差为3.00%和4.82%。本文的研究成果对开展电子皮肤柔性触觉传感器的研究和推广具有一定的实用价值。     

基于RBF神经网络的湿度传感器建模方法

针对高分子湿度传感器的测量精度易受温度影响的问题，提出了湿度硬件测量电路，建立了高分子湿度传感器的数学模型。此硬件测量电路的湿度传感器选用IH3605，温度传感器选用AD590。此数学模型采用RBF神经网络，利用在线测量的温湿度信号值，应用带遗忘因子的梯度下降算法进行网络参数的调整，同时给出了建模步骤。实际结果表明，该模型的误差小于0．5％RH，具有较高的精度。     

BP神经网络补偿并联机器人定位误差

为减小机构末端定位误差,提高机器人运动精度,分析了所开发的6-DOF精密并联机器人末端位姿的误差来源及以往误差补偿方法的局限性。通过实际测量末端位姿,在精密定位的局部工作空间内,提出了基于BP神经网络的机器人关节空间误差补偿方法。确定了BP神经网络模型,建立了误差补偿的数据样本,并对数据样本进行了标准化,通过实验对比的方法确定了隐层神经元的个数,同时对网络的推广能力进行了验证。经过误差补偿,6-DOF精密并联机器人的平移定位误差下降了80%,转角定位误差下降了60%。该实验结果表明,基于BP神经网络的误差补偿方法对机器人局部工作空间的补偿具有明显的效果,满足精密并联机器人工作的精度要求。     

气体测量中的温湿度补偿新方法

针对气体传感器的温湿度效应,提出了两种新的温湿度补偿方法,即模糊修正因子法和人工神经网络法。实验结果表明,这两种方法软硬件实现简单,补偿精度高,具有一定的工程应用价值。     

基于RBF神经网络集成-模糊加权输出的数字温度传感器误差补偿

数字温度传感器存在非线性误差,在高精度测温系统中需要进行误差补偿。提出了一种基于径向基函数神经网络集成-模糊加权输出(RBFNNE-FWO)的数字温度传感器误差补偿方法:首先根据数字温度传感器的误差特征,提取特征阈值,构造三个相互独立的成员RBFNN;考虑到成员网络之间边界误差补偿问题,构建一种RBFNN集成输出权值模糊调节器,获得RBFNN集成输出权值,从而完成数字温度传感器的全量程误差补偿。与多种方法的比较仿真实验表明,这种RBFNNE-FWO方法的性能最佳、各成员网络边界误差最小,补偿后的数字温度传感器误差减少了两个数量级,大大提高了测温准确度。     

OGSA结合RBF网络数据融合技术在红外气体分析中的应用

针对红外气体分析存在的问题 ,提出光学气体传感器阵列 (OGSA )和径向基函数 (RBF)神经网络技术相结合的方法 ,并将其应用于 SO2 、CO2 浓度检测。该方法利用 RBF网络非线性逼近功能和 OGSA的信息冗余特性 ,消除非目标参量和目标量之间的交叉干扰。结果表明 ,SO2 引用误差由融合处理前的 3.1%降至 2 .0 % ,CO2 由 39%降至 8.6 %。为红外气体分析提供了一种有效的途径 ,具有实用前景。     

基于Elman神经网络的压力传感器温补偿的研究

当受到温度、电磁波等外界因素的影响,压力传感器的测量精度较低,测量值容易有较大的波动.因此,当压力传感器运用于汽车等移动设备时性能下降.以对温度极其敏感的CYJ-101压力传感器为例,建立18组样本数据在Elman模型的神经网络中进行温度补偿训练.仿真结果显示,通过Elman神经网络的温度补偿,传感器的测量误差降低约2...     

Application of ACO-BPN to thermal error modeling of NC machine tool

Thermal errors are the major contributor to the dimensional errors of a workpiece in precision machining. Error compensation technique is a cost-effective way to reduce thermal errors. Accurate modeling of errors is a prerequisite of error compensation. In this paper, four key temperature points of a NC machine tool were obtained based on clustering method. A thermal error model based on the four key temperature points was proposed by using ant colony algorithm-based back propagation neural network (ACO-BPN). The ACO-BPN method improves the prediction accuracy of thermal deformation in the NC machine tool. A thermal error compensation system was developed based on the proposed model, and which has been applied to the NC machine tool in daily production. The results show that the thermal drift in workpiece diameter has been reduced from 33 to 8 μm from its center of tolerance.     

磁力仪温度误差的径向基神经网络补偿模型

磁通门磁力仪参数受温度影响明显,直接影响传感器测量精度,需要研究补偿方法,提高测量精度。采用无磁高低温试验箱测量磁通门传感器温度特性;提出基于径向基神经网络的温度误差补偿方法,分别建立磁通门磁力仪零漂误差补偿模型和刻度因子误差补偿模型。结果表明,径向基神经网络能良好逼近磁通门传感器参数的温度特性;与BP神经网络相比,径向基神经网络在零漂补偿中训练时间更短,精度更高,重复性更好,零漂误差的抑制能力更强。补偿后,磁通门磁力仪零漂误差从7.105 5 nT减少到0.766 1 nT;刻度因子误差从6.3E-3减少到7.2E-5;测量值温度误差由213.6 nT补偿到9.1 nT。提出建立通用的温度补偿模型,在不同磁场环境下经过反复测试,采用训练过的模型补偿后,温度误差均降低一个数量级,提高了磁通门磁力仪温度性能和精度。     

Miniature low-power IR monitor for methane detection

A portable methane gas monitor based on an infrared spectrum absorption principle has been developed using a dual-channel and dual-wavelength pyroelectric infrared detector, active filters around the overtone absorption lines of methane at 3.31 μm, reference filters around the non-absorption lines of methane at 3.93 μm, mid-IR LEDs, a miniature gold-filled cell structure, temperature sensors for gas concentration calibration and compensation, an electrical modulation source, and a highly integrated intelligent controller. A detailed investigation has been carried out to design a low-cost portable IR optical sensor for methane detection that can operate in harsh environments with temperature variations between −10 °C and 40 °C. The infrared detection optics principle used in developing this system is mainly analyzed. A prototype based on this design showed an accuracy of ±0.05%, which meets the technology requirements of lower-power consumption, reduced volume, and wide measurement range.     

基于BP神经网络的矿井一氧化碳检测方法研究

文章采用热催化传感器和电化学式气体传感器的配合使用，为了解决两种传感器对矿井一氧化碳和甲烷气体的“交叉敏感”问题，提出了一种基于BP神经网络技术的传感器系统。研究了采用两种传感器组成的多传感器阵列与BP神经网络相结合来实现一氧化碳气体浓度精确检测的方法。实验证明，利用基于BP神经网络的多传感器阵列模型，能有效的提高对井下一氧化碳气体的测量精度。     

基于神经网络融合的传感器温度误差补偿

海底油气输送管道漏磁检测装置工作于高温高压环境下,其中的InSb霍尔传感器对温度敏感,需要补偿温度误差。该文构建了多传感器融合模型,将多个霍尔传感器和温度传感器的输出用径向基函数（RBF）神经网络进行融合,用遗传算法对网络进行训练。实验室检测数据和反演出的缺陷形状表明,采用神经网络融合方法进行误差补偿,简单方便,霍尔传感器输出的平均温度敏感系数降低了两个数量级。     

Thermal sensor selection for the thermal error modeling of machine tool based on the fuzzy clustering method

Thermal sensor selection is a work of great importance when modeling thermal error. The proper selection of thermal sensors and their locations may greatly improve the prediction accuracy. In this article, the fuzzy C means (FCM) clustering method and the ISODATA method are used to group the data of thermal sensors and a genetic algorithm–back propagation artificial neural network thermal model is established to testify the accuracy. A validity criterion for the FCM method is put forward to guarantee the precision of the model. Both the FCM and the ISODATA methods are effective for thermal sensor selection.     

A pre-compensation method of the systematic contouring error for repetitive command paths

For a repetitive command path, pre-compensating the contouring error by modifying the command path is practical. To obtain the pre-compensation value with better accuracy, this paper proposes the use of a back propagation neural network to extract the function of systematic contouring errors. Furthermore, by using the extracted function, the contouring error can be easily pre-compensated. The experiment results verify that the proposed compensation method can effectively reduce contouring errors.     

神经网络在SnO_2气敏元件浓度测量中的应用

气体传感器属于传感器技术领域,在传感器行业中占有重要的地位.然而气体传感器阵列的交叉敏感性严重影响气体传感器对混合气体的测量.基于Matlab平台的神经网络工具箱,分别构建BP神经网络和RBF(径向基)神经网络,对由涂敷不同敏感材料的声表面波振荡器组成的阵列在4种混合气体灵敏度响应数据进行定量识别研究,结果表明RBF神经网络在气体定量识别方面更具优势.     

基于遗传神经网络算法的发动机性能预测

设计了遗传神经网络预测系统,通过融合分析废气成分的含量、发动机转速、氧传感器等信息预测发动机的工作性能。应用遗传算法的复制、交换、变异过程代替原BP网络的反向传播过程,并针对遗传算法的缺点进行了改进研究。实践证明这种遗传神经网络预测系统收敛速度快、推广性强,具有较高的实际应用价值。     

[成形过程的数据挖掘与深度学习方法]基于机器学习的管材弯曲回弹有效预测与补偿

采用基于优化的误差反向传播(BP)神经网络的机器学习算法建模，提出了考虑材料参数、几何参数等多因素的弯管回弹精确预测和高效控制方法。该方法通过引入非线性惯性权重及遗传算法的杂交算子，改进了粒子群优化(PSO)算法，进而通过改进的PSO算法对BP神经网络进行优化，构建了基于改进的PSO-BP神经网络机器学习回弹预测和补偿模型。以多种规格的铝合金数控弯管构件为对象，将实际生产中不同规格、批次、成形参数下回弹数据作为训练样本，实现了所建机器学习预测模型的应用验证。所建模型获得的预测结果平均相对误差为6.3%，与未优化的BP神经网络等传统模型相比，预测精度最大提高了18.5%，计算时间可从1.5 h缩短至300 s，同时实现了回弹预测与补偿精度以及计算效率的显著提高。     

微双桥结构氢气传感技术研究

基于催化燃烧式气体传感原理,提出了利用MEMS技术将电化学生长的Al_2O_3膜加工成微双桥结构载体,采用平面薄膜工艺制做铂薄膜热敏电阻,涂敷Al_2O_3-ZrO-Th0-Pd和Al_2O_3-ZrO-Th0形成催化敏感桥臂和温湿度补偿桥臂,实现在微双桥上催化桥臂和补偿桥臂的单片集成,制作出氢气传感器.测试结果表明:传感器可实现体积分数0～40 000×10-6氢气检测,具有线性输出特性,温度为5～45 ℃的传感器的最大零点输出优于±1.0%.