Predicting quality characteristics of end-milled parts based on multi-sensor integration using neural networks: individual effects of learning parameters and rules

Artificial neural networks have been shown to have a lot of potential as a means of integrating multi-sensor signals for real-time monitoring of machining processes. However, many questions still remain to be answered on how to optimize the training parameters during the training phase to optimize their subsequent performance, especially in view of the fact that the few published articles have made conflicting recommendations. This paper presents a systematic evaluation of the individual effects of training parameters — learning rate, momentum rate, number of hidden layer nodes, transfer function and learning rule-on the performance of back-propagation networks used for predicting quality characteristics of end-milled parts. Multi-sensor signatures (acoustic emission, spindle vibration, cutting force components and machining time) acquired during circular end-milling of 4140 steel and the corresponding measured quality characteristics (surface roughness and bore tolerance) were used to train the networks. The network is part of a proposed intelligent machining monitoring and diagnostic system for quality assurance of machined parts. The network performances were evaluated using four different criteria: maximum error, rms error, mean error and number of training cycles. One of the results obtained shows that the hyperbolic tangent transfer function gives a better performance than the sigmoid and sine functions respectively. Optimum combinations of training parameters have been observed. The effects of various combinations of training parameters are presented.     

Cutting parameters analysis for the development of a milling process monitoring system based on audible energy sound

Monitoring of machining processes is a critical requirement in the implementation of any unmanned operation in a shop floor and, particularly, in the establishment of Flexible Manufacturing Systems (FMS) and Computer Integrated Manufacturing (CIM) where most of the operations are carried out in an automated way. During the last years, notable efforts have been made to develop reliable and robust monitoring systems based on different types of sensors such as cutting force and torque, motor current and effective power, vibrations, acoustic emission or audible sound energy. This work is focused on this last sensor technology. The basic objective is to characterise the audible sound energy signals generated during different machining operations carried out on a milling machine. In order to achieve this, rotation speed, feed and depth of cut have been analysed separately. The main contributions of this work are, on the one hand, the application of a systematic methodology to set up the cutting tests and, on the other hand, the independent signal analysis of the noise generated by the milling machine used for the cutting tests in order to filter this noise out from the signals obtained during the actual material processing. The classification of audible sound signal features for process monitoring has been obtained by graphical analysis and parallel distributed data processing using a supervised neural network (NN) paradigm.     

Real-time quality monitoring and predicting model based on error propagation networks for multistage machining processes

To ensure the machining processes stability of multistage machining processes (MMPs) and improve the quality of machining processes, a real-time quality monitoring and predicting model based on error propagation networks for MMPs is proposed in this paper. As there are some complicated interactions among different stages in MMPs, a machining error propagation network (MEPN) is proposed and its complexity is discussed to analyze the correlation among different stages in MMPs. Based on these, a real-time quality-monitoring model based on process variation trajectory chart is proposed to monitor the key machining stages extracted by MEPN. Due to the complexity of the correlation in MEPN, it is important and necessary to explore the variation propagation mechanism in MEPN. As for this issue, a machining error propagation model of machining form feature nodes in MEPN is established with the neuron model, which is solved with back-propagation neural network. The mapping relationship among machining errors of quality attributes is described through this node model. Furthermore, a novel equipment synthetic failure probability exponent of machining status nodes in MEPN is established to synthesize equipment’s parameters by using logistic regression to quantitatively analyze the potential-failure and forecast the equipment degradation trend. At last, the machining process of a connecting rod is used to verify the proposed method.     

On-line tool condition monitoring system with wavelet fuzzy neural network

In manufacturing systems such as flexible manufacturing systems (FMS), one of the most important issues is accurate detection of the tool conditions under given cutting conditions. An investigation is presented of a tool condition monitoring system (TCMS), which consists of a wavelet transform preprocessor for generating features from acoustic emission (AE) signals, followed by a high speed neural network with fuzzy inference for associating the preprocessor outputs with the appropriate decisions. A wavelet transform can decompose AE signals into different frequency bands in the time domain. The root mean square (RMS) values extracted from the decomposed signal for each frequency band were used as the monitoring feature. A fuzzy neural network (FNN) is proposed to describe the relationship between the tool conditions and the monitoring features; this requires less computation than a back propagation neural network (BPNN). The experimental results indicate the monitoring features have a low sensitivity to changes of the cutting conditions and FNN has a high monitoring success rate in a wide range of cutting conditions; TCMS with a wavelet fuzzy neural network is feasible.     

Simulation of a complex optical polishing process using a neural network

 Most modern manufacturing processes change their set of parameters during machining in order to work at the optimum state. But in some cases, like computer-controlled polishing, it is not possible to change these parameters during the machining. Then usually a standard set of parameters is chosen which is not adjusted to the specific conditions. To gather the optimum set of parameters anyway simulation of the process prior to manufacturing is a possibility. This research illustrates the successful implementation of a neural network to accomplish such a simulation. The characteristic of this neural network is described along with the decision of the used inputs and outputs. Results are shown and the further usage of the neural network within an automation framework is discussed. The ability to simulate these advanced manufacturing processes is an important contribution to extend automation further and thus increase cost effectiveness.     

In-process tool condition forecasting based on a deep learning method

It is widely acknowledged that machining precision and surface integrity are greatly affected by cutting tool conditions. In order to enable early cutting tool replacement and proactive actions, tool wear conditions should be estimated in advance and updated in real-time. In this work, an approach to in-process tool condition forecasting is proposed based on a deep learning method. A long short-term memory network is designed to forecast multiple flank wear values based on historical data. A residual convolutional neural network is built to enable in-process tool condition monitoring, using raw signals acquired during the machining process. The integration of them enables in-process tool condition forecasting. Median-based correction and mean-based correction are adopted to improve the accuracy. IEEE PHM 2010 challenge data has been used to illustrate and validate this approach. Experimental study and quantitative comparisons showed that future flank wear values could be precisely forecasted during the machining process. The proposed approach contributes to prompt and reliable cutting tool condition forecasting, which will support the decision-making about cutting tool replacement in data-driven smart manufacturing.     

Sensor-Ball system based on IEEE 1451 for monitoring the condition of power transmission lines

The energy industry is developing it safety methods so that it can prevent accidents, diagnose the conditions of power utilities and identify any problems that might arise. In other words, the industry is trying to increase the reliability of power utilities and improve the quality of power utilities by conducting monitoring and diagnosis tasks in real-time by building a sensor network in order to monitor the status of power utilities. This study has applied IEEE 1451, an interface standard of sensor network, for the diagnosis, information exchange and compatibility of the Sensor-Ball, which is a sensor unit that constructs a sensor network for the status monitoring of power transmission lines. In order to accomplish these objectives, we applied IEEE 1451.0 and IEEE 1451.5 to the Sensor-Ball, suggested a reference model, and represented the characteristic information of the Sensor-Ball by use of IEEE 1451 TEDS (Transducer Electronic Data Sheet). By constructing a Sensor-Ball system that has a connected sensor for measuring the atmospheric temperature and line slope, we have identified the operational status and sensor data of TEDS applied through a monitoring program.     

人工神经网络在木材损伤识别中的应用

利用人工神经网络(ANN)对含有不同缺陷程度的木材进行了定性和定量解析.通过选择合理的神经网络结构,建立有效的训练样本集,确定合理的参数及训练方法,对3种不同缺陷类型木材进行了解析,考察了网络的泛化能力.结果表明:该网络能够对不同缺陷程度的木材进行准确地识别,而且本研究为人工神经网络在木材缺陷损伤的定性和定量分析方面提供了一种有效的方法.     

基于异构网络的风光互补电站监控系统设计

针对风光互补电站分布广、距离远、运行时间长、实时监控信息集成度不高的现状,综合运用无线传感网络、现场总线网络、GPRS网络、以太网等工业网络技术和嵌入式系统技术,设计了一种电站监控系统,实现了不同网络之间的数据互联通信。在介绍监控系统总体框架的基础上,重点分析了与异构网络互联相关的节点硬件和软件设计方法。实际测试表明,监控系统自动化程度高,数据传输稳定可靠,能够满足风光互补电站远程自动监控的应用要求。     

基于ARM与ZigBee的矿用通风机监测系统设计

针对传统通风机监测模式的弊端,设计了新型矿用通风机在线监测系统.采用ARM-Linux嵌入式架构与ZigBee无线传感器网络技术实现对通风机运行状态的实时监测.无线传感器网络负责状态数据的采集与传输;采用Qt开发上位机监测软件,实时显示通风机相关部位的温度、振动、风压、风量以及电动机的电量参数,同时对通风机异常状态进行报警;通过手动与触发方式控制通风机传感器节点采集振动加速度数据,并以.txt格式存储于SD卡中以用于通风机故障诊断.整个系统安装简单、使用灵活,实际应用表明该系统能够很好地完成通风机状态监测任务.     

基于无线传感网的核电安全监控系统设计

低碳经济和能源可持续发展的要求下,新能源的开发和使用越来越受到世界各国的青睐。核电在新能源领域发电效率最高,电能质量最稳定,近年来取得长足的进步和丰硕的成果,然而,福岛核辐射事故的发生导致核电发展阻力重重,核电安全问题被提升到前所未有的高度。基于传感网技术的核电安全监控系统采用“检测层-接入层-汇聚层-数据中心”四层传感网结构,实时采集核电厂关键设备的运行状态和参数,并通过传感网传输到远程服务器,在数据中心对采集的数据进行数据挖掘和数据融合处理,对核电厂运行状态进行实时监测和预测,确保核电厂安全运营。核电安全监测网络的生命周期是保证整个网络使用可靠性和稳定性的前提和基础,从节点布局优化的角度出发,研究该监测网络的能量管理方法。     

基于声表面波传感器阵列的气体鉴别算法研究

声表面波(SAW)传感器阵列具有体积小、功耗低、反应灵敏等优点,在食品检测、环境治理、气体鉴别等领域有广泛的应用前景。结合声表面波传感器阵列的原理及特点,建立和优化了声表面波传感器阵列的数学模型,并对数据进行预处理、主成分分析（PCA）以及BP神经网络分析处理,实现了对气体的鉴别分类,取得了好的实验结果。     

In-process regressions and adaptive multicriteria neural networks for monitoring and supervising machining operations

The authors develop a monitoring and supervising system for machining operations using in-process regressions (for monitoring) and adaptive feedforward artificial neural networks (for supervising). The system is designed for: (1) in-process tool life measurement and prediction; (2) supervision of machining operations in terms of the best machining setup; and (3) catastrophic tool failure monitoring. The monitoring system predicts tool life by using different sensors for gathering information based on a regression model that allows for the variations between tools and different machine setups. The regression model makes its prediction by using the history of other tools and combining it with the information obtained about the tool under consideration. The supervision system identifies the best parameters for the machine setup problem within the framework of multiple criteria decision making. The decision maker (operator) considers several criteria, such as cutting quality, production rate and tool life. To make the optimal decision with several criteria, an adaptive feedforward artificial neural network is used to assess the decision maker's preferences. The authors' neural network approach learns from the decision maker's complex behavior and hence, in automatic mode, can make decisions for the decision maker. The approach is not computationally demanding, and experiments demonstrate that its predictions are accurate.     

飞机强度试验中结构健康监测采集控制系统

针对结构健康监测在全机疲劳试验中需要采用多种监测系统的特点,结合飞机强度疲劳试验对数据采集的要求,设计了一种将光栅光纤传感器、压电传感器和声发射传感器等集成为一体的飞机结构健康监测系统.设计了多类型传感器采集控制系统,其中采用了分布式、多任务的结构设计,解决了多台设备采集数据的同步性、有效性和一致性.针对多类传感器的数据源多、数据类型多的特点,设计了通用数据结构解决了多类传感器的数据合并处理问题.通过在某全机疲劳试验中的运用,验证了本系统是一种稳定、高效、易扩展的多传感器健康监测集成系统,能够获取更加可靠、全面的健康监测数据,为健康监测在全机疲劳试验中损伤的准确识别,提供了强有力的支撑.     

基于无线传感器网络的岩体声发射信号监测系统

在分析了现有岩体声发射信号监测系统存在局限性的基础上,利用无线传感器网络和压缩感知技术,设计了一种新型岩体声发射信号监测系统,详细叙述了系统结构和软硬件实现方法,并将其应用于高速公路岩体边坡稳定性监测.实际应用结果表明,系统设计方案合理可行,且由于使用了压缩感知技术,在采样频率为200 kHz的情况下,也可实现声发射信...     

A hybrid neural network model for rule generation and its application to process fault detection and diagnosis

In this paper, a hybrid neural network model, based on the integration of fuzzy ARTMAP (FAM) and the rectangular basis function network (RecBFN), which is capable of learning and revealing fuzzy rules is proposed. The hybrid network is able to classify data samples incrementally and, at the same time, to extract rules directly from the network weights for justifying its predictions. With regards to process systems engineering, the proposed network is applied to a fault detection and diagnosis task in a power generation station. Specifically, the efficiency of the network in monitoring the operating conditions of a circulating water (CW) system is evaluated by using a set of real sensor measurements collected from the power station. The rules extracted are analyzed, discussed, and compared with those from a rule extraction method of FAM. From the comparison results, it is observed that the proposed network is able to extract more meaningful rules with a lower degree of rule redundancy and higher interpretability within the neural network framework. The extracted rules are also in agreement with experts’ opinions for maintaining the CW system in the power generation plant.     

基于声发射和神经网络的复合材料冲击定位

为了提高复合材料结构冲击定位的精度和实时性,基于声发射和神经网络技术,提出了复合材料结构冲击定位两步法,以压电陶瓷（PZT）和自制信号采集系统替代商用声发射仪器,实现了一种能够高精度、实时、在线监测冲击的系统。用小波变换求出原点处冲击源传播到各传感器的波达时间差,用这组时间差修正其他位置上的冲击源到达各传感器的波达时间,利用修正后的波达时间,根据四点圆弧定位算法得到冲击源坐标,实现初步定位;将所求出的位置坐标作为神经网络的输入,训练之后的神经网络可以准确预测冲击位置,实现精确定位。在复合材料板上的试验表明：该方法能快速、准确地识别出冲击位置。     

Wireless Acoustic Sensor Networks and Edge Computing for Rapid Acoustic Monitoring

Passive acoustic monitoring is emerging as a promising solution to the urgent, global need for new biodiversity assessment methods. The ecological relevance of the soundscape is increasingly recognised, and the affordability of robust hardware for remote audio recording is stimulating international interest in the potential for acoustic methods for biodiversity monitoring. The scale of the data involved requires automated methods, however, the development of acoustic sensor networks capable of sampling the soundscape across time and space and relaying the data to an accessible storage location remains a significant technical challenge, with power management at its core. Recording and transmitting large quantities of audio data is power intensive, hampering long-term deployment in remote, off-grid locations of key ecological interest. Rather than transmitting heavy audio data, in this paper, we propose a low-cost and energy efficient wireless acoustic sensor network integrated with edge computing structure for remote acoustic monitoring and in situ analysis. Recording and computation of acoustic indices are carried out directly on edge devices built from low noise primo condenser microphones and Teensy microcontrollers, using internal FFT hardware support. Resultant indices are transmitted over a ZigBee-based wireless mesh network to a destination server. Benchmark tests of audio quality, indices computation and power consumption demonstrate acoustic equivalence and significant power savings over current solutions.      

基于模糊神经网络的WSN无线数据收发单元故障诊断

在一些无线传感器网络(Wireless Sensor Network,WSN)安全监测系统中,节点长时间传输大量数据,导致无线数据收发单元容易出现功率下降和功率放大器(Power Amplifier,PA)被烧毁的现象,而此类故障的诊断方法一般比较复杂且低效。针对上述问题,在分析WSN单元级故障诊断的基础上,利用无线数据收发单元的电流模型,提出了一种基于模糊神经网络的无线数据收发单元故障诊断方法。首先,根据无线数据收发单元中发射消耗的电流与温度和供电电压的关系,建立电流模型;然后,利用聚类算法确定模糊神经网络模型结构,结合混合学习算法优化模糊规则的前件参数和后件参数;最后,提取训练完的模糊神经网络参数,以建立WSN节点故障诊断模型。实验结果表明,提出的无线数据收发单元故障诊断方法的计算量小,诊断准确度高;与高斯过程回归模型相比,其计算量降低了22.4%,诊断准确度提高了17.5%。