Monitoring Grinding Wheel Redress-life Using Support Vector Machines

Condition monitoring is a very important aspect in automated manufacturing processes. Any malfunction of a machining process will deteriorate production quality and efficiency. This paper presents an application of support vector machines in grinding process monitoring. The paper starts with an overview of grinding behaviour. Grinding force is analysed through a Short Time Fourier Transform （STFT） to identify features for condition monitoring. The Support Vector Machine （SVM） methodology is introduced as a powerful tool for the classification of different wheel wear situations. After training with available signal data, the SVM is able to identify the state of a grinding process. The requirement and strategy for using SVM for grinding process monitoring is discussed, while the result of the example illustrates how effective SVMs can be in determining wheel redress-life.

Turbopump Condition Monitoring Using Incremental Clustering and One-class Support Vector Machine

Turbopump condition monitoring is a significant approach to ensure the safety of liquid rocket engine （LRE）.Because of lack of fault samples,a monitoring system cannot be trained on all possible condition patterns.Thus it is important to differentiate abnormal or unknown patterns from normal pattern with novelty detection methods.One-class support vector machine （OCSVM） that has been commonly used for novelty detection cannot deal well with large scale samples.In order to model the normal pattern of the turbopump with OCSVM and so as to monitor the condition of the turbopump,a monitoring method that integrates OCSVM with incremental clustering is presented.In this method,the incremental clustering is used for sample reduction by extracting representative vectors from a large training set.The representative vectors are supposed to distribute uniformly in the object region and fulfill the region.And training OCSVM on these representative vectors yields a novelty detector.By applying this method to the analysis of the turbopump＇s historical test data,it shows that the incremental clustering algorithm can extract 91 representative points from more than 36 000 training vectors,and the OCSVM detector trained on these 91 representative points can recognize spikes in vibration signals caused by different abnormal events such as vane shedding,rub-impact and sensor faults.This monitoring method does not need fault samples during training as classical recognition methods.The method resolves the learning problem of large samples and is an alternative method for condition monitoring of the LRE turbopump.

Monitoring the Heavy Element of Cr in Agricultural Soils Using a Mobile Laser-Induced Breakdown Spectroscopy System with Support Vector Machine

Due to its complicated matrix effects, rapid quantitative analysis of chromium in agricultural soils is difficult without the concentration gradient samples by laser-induced breakdown spectroscopy. To improve the analysis speed and accuracy, two calibration models are built with the support vector machine method： one considering the whole spectra and the other based on the segmental spectra input. Considering the results of the multiple linear regression analysis, three segmental spectra are chosen as the input variables of the support vector regression （SVR） model. Compared with the results of the SVR model with the whole spectra input, the relative standard error of prediction is reduced from 3.18% to 2.61% and the running time is saved due to the decrease in the number of input variables, showing the robustness in rapid soil analysis without the concentration gradient samples.

Combination Method of Principal Component Analysis and Support Vector Machine for On-line Process Monitoring and Fault Diagnosis

On-line monitoring and fault diagnosis of chemical process is extremely important for operation safety and product quality. Principal component analysis (PCA) has been widely used in multivariate statistical process monitoring for its ability to reduce processes dimensions. PCA and other statistical techniques, however, have difficulties in differentiating faults correctly in complex chemical process. Support vector machine (SVM) is a novel approach based on statistical learning theory, which has emerged for feature identification and classification. In this paper, an integrated method is applied for process monitoring and fault diagnosis, which combines PCA for fault feature extraction and multiple SVMs for identification of different fault sources. This approach is verified and illustrated on the Tennessee Eastman benchmark process as a case study. Results show that the proposed PCA-SVMs method has good diagnosis capability and overall diagnosis correctness rate.

Smart Monitoring of Solar Photovoltaic Panels by the Approach of Machine Learning

The exploitation of renewable energy has become a pressing task due to climate change and the recent energy crisis caused by regional conflicts.This has further accelerated the rapid development of the global photovoltaic(PV)market,thereby making the management and maintenance of solar photovoltaic(SPV)panels a new area of business as neglecting it may lead to significant financial losses and failure to combat climate change and the energy crisis.SPV panels face many risks that may degrade their power generation performance,damage their structures,or even cause the complete loss of their power generation capacity during their long service life.It is hoped that these problems can be identified and resolved as soon as possible.However,this is a challenging task as a solar power plant(SPP)may contain hundreds even thousands of SPV panels.To provide a potential solution for this issue,a smart drone-based SPV panel condition monitoring(CM)technique has been studied in this paper.In the study,the U-Net neural network(UNNN),which is ideal for undertaking image segmentation tasks and good at handling small sample size problem,is adopted to automatically create mask images from the collected true color thermal infrared images.The support vector machine(SVM),which performs very well in highdimensional feature spaces and is therefore good at image recognition,is employed to classifying the mask images generated by the UNNN.The research result has shown that with the aid of the UNNN and SVM,the thermal infrared images that are remotely collected by drones from SPPs can be automatically and effectively processed,analyzed,and classified with reasonable accuracy(over 80%).Particularly,the mask images produced by the trained UNNN,which contain less interference items than true color thermal infrared images,significantly benefit the assessing accuracy of the health state of SPV panels.It is anticipated that the technical approach presented in this paper will serve as an inspiration for the exploration of more advanced and dependable smart asset management techniques within the solar power industry.

Comprehensive Overview on Computational Intelligence Techniques for Machinery Condition Monitoring and Fault Diagnosis

Computational intelligence is one of the most powerful data processing tools to solve complex nonlinear problems, and thus plays a significant role in intelligent fault diagnosis and prediction. However, only few com- prehensive reviews have summarized the ongoing efforts of computational intelligence in machinery condition moni- toring and fault diagnosis. The recent research and devel- opment of computational intelligence techniques in fault diagnosis, prediction and optimal sensor placement are reviewed. The advantages and limitations of computational intelligence techniques in practical applications are dis- cussed. The characteristics of different algorithms are compared, and application situations of these methods are summarized. Computational intelligence methods need to be further studied in deep understanding algorithm mech- anism, improving algorithm efficiency and enhancing engineering application. This review may be considered as a useful guidance for researchers in selecting a suit- able method for a specific situation and pointing out potential research directions.

Monitoring models for base flow effect and daily variation of dam seepage elements considering time lag effect

Affected by external environmental factors and evolution of dam performance, dam seepage behavior shows nonlinear time-varying characteristics. In this study, to predict and evaluate the long-term development trend and short-term fluctuation of the dam seepage behavior, two monitoring models were developed, one for the base flow effect and one for daily variation of dam seepage elements. In the first model, to avoid the influence of the time lag effect on the evaluation of seepage variation with the time effect component of seepage elements, the base values of the seepage element and the reservoir water level were extracted using the wavelet multi-resolution analysis method, and the time effect component was separated by the established base flow effect monitoring model. For the development of the daily variation monitoring model for dam seepage elements, all the previous factors, of which the measured time series prior to the dam seepage element monitoring time may have certain influence on the monitored results, were considered. Those factors that were positively correlated with the analyzed seepage element were initially considered to be the support vector machine(SVM) model input factors, and then the SVM kernel function-based sensitivity analysis was performed to optimize the input factor set and establish the optimized daily variation SVM model. The efficiency and rationality of the two models were verified by case studies of the water level of two piezometric tubes buried under the slope of a concrete gravity dam.Sensitivity analysis of the optimized SVM model shows that the influences of the daily variation of the upstream reservoir water level and rainfall on the daily variation of piezometric tube water level are processes subject to normal distribution.

Monitoring Land-Use Change in Nakuru (Kenya) Using Multi-Sensor Satellite Data

Recently land-use change has been the main concern for worldwide environment change and is being used by city and regional planners to design sustainable cities. Nakuru in the central Rift Valley of Kenya has undergone rapid urban growth in last decade. This paper focused on urban growth using multi-sensor satellite imageries and explored the potential benefits of combining data from optical sensors (Landsat, Worldview-2) with Radar sensor data from Advanced Land Observing Satellite (ALOS) Phased Array type L-band Synthetic Aperture Radar (PALSAR) data for urban land-use mapping. Landsat has sufficient spectral bands allowing for better delineation of urban green and impervious surface, Worldview-2 has a higher spatial resolution and facilitates urban growth mapping while PALSAR has higher temporal resolution compared to other operational sensors and has the capability of penetrating clouds irrespective of weather conditions and time of day, a condition prevalent in Nakuru, because it lies in a tropical area. Several classical and modern classifiers namely maximum likelihood (ML) and support vector machine (SVM) were applied for image classification and their performance assessed. The land-use data of the years 1986, 2000 and 2010 were compiled and analyzed using post classification comparison (PCC). The value of combining multi-temporal Landsat imagery and PALSAR was explored and achieved in this research. Our research illustrated that SVM algorithm yielded better results compared to ML. The integration of Landsat and ALOS PALSAR gave good results compared to when ALOS PAL- SAR was classified alone. 19.70 km2 of land changed to urban land-use from non-urban land-use between the years 2000 to 2010 indicating rapid urban growth has taken place. Land-use information is useful for the comprehensive land-use planning and an integrated management of resources to ensure sustainability of land and to achieve social Eq- uity, economic efficiency and environmental sustainability.

Combining flame monitoring techniques and support vector machine for the online identification of coal blends

The combustion behavior of two single coals and three coal blends in a 300 kW coal-fired furnace under variableoperating conditions was monitored by a flame monitoring system based on image processing and spectral analysis. A similaritycoefficient was defined to analyze the similarity of combustion behavior between two different coal types. A total of 20 flamefeatures, extracted by the flame monitoring system, were ranked by weights of their importance estimated using ReliefF, a featureselection algorithm. The mean of the infrared signal was found to have by far the highest importance weight among the flamefeatures. Support vector machine （SVM） was used to identify the coal types. The number of flame features used to build the SVMmodel was reduced from 20 to 12 by combining the methods of ReliefF and SVM, and computational precision was guaranteedsimultaneously. A threshold was found for the relationship between the error rate and similarity coefficient, which were positivelycorrelated. The success rate decreased with increasing similarity coefficient. The results obtained demonstrate that the system canachieve the online＂ identification of coal blends in industry.

Research on Early Fault Self-Recovery Monitoring of Aero-Engine Rotor System

In order to increase robustness of the AERS (Aero-engine Rotor System) and to solve the problem of lacking fault samples in fault diagnosis and the difficulty in identifying early weak fault, we proposed a new method that it not only can identify the early fault of AERS but also it can do self-recovery monitoring of fault. Our method is based on the analysis of the early fault features on AERS, and it combined the SVM (Support Vector Machine) with the stochastic resonance theory and the wavelet packet decomposition and fault self-recovery. First, we zoom the early fault feature signals by using the stochastic resonance theory. Second, we extract the feature vectors of early fault using the multi-resolution analysis of the wavelet packet. Third, we input the feature vectors to a fault classifier, which can be used to identify the early fault of AERS and carry out self-recovery monitoring of fault. In this paper, features of early fault on AERS, the zoom of early fault characteristics, the extraction method of early fault characteristics, the construction of multi-fault classifier and way of fault self-recovery monitoring are studied. Results show that our method can effectively identify the early fault of AERS, especially for identifying of fault with small samples, and it can carry on self-recovery monitoring of fault.

PENETRATION QUALITY EVALUATION IN ROBOTIZED ARC WELDING BASED ON SUPP0RT VECTOR MACHINE

A quality monitoring method by means of support vector machines (SVM) forrobotized gas metal arc welding (GMAW) is introduced. Through the feature extraction of the weldingprocess signal, a SVM classifier is constructed to establish the relationship between the feature ofprocess parameters and the quality of weld penetration. Under the samples obtained from auto partswelding production line, the learning machine with a radial basis function kernel shows goodperformance. And this method can be feasible to identity defect online in welding production.

不平衡数据下基于SVM增量学习的指挥信息系统状态监控方法

针对指挥信息系统历史状态样本有限的特点,基于支持向量机(support vector machines,SVM)设计了一种面向不平衡数据的SVM增量学习方法。针对系统正常/异常状态样本不平衡的情况,首先利用支持向量生成一部分新样本,然后通过分带的思想逐带产生分布更加均匀的新样本以调节原样本集的不平衡比。针对系统监控实时性要求高且在运行过程中会有新样本不断加入的特点,采用增量学习的方式对分类模型进行持续更新,在放松KKT(Karush-Kuhn-Tucker)更新触发条件的基础上,通过定义样本重要度并引入保留率和遗忘率的方式减少了增量学习过程中所需训练的样本数量。为了验证算法的有效性和优越性,实验部分在真实系统中获得的数据集以及UCI数据集中3类6组不平衡数据集中与现有的算法进行了对比。结果表明,所提算法能够有效实现对不平衡数据的增量学习,从而满足指挥信息系统状态监控的需求。

基于PS-InSAR技术的晋城矿区地表形变监测及地质灾害风险预警

地表形变是一种严重的地质灾害现象,不仅严重影响灾害区居民的日常生活,而且会造成巨大的社会经济危害,尤其在采煤区。针对传统地表沉陷监测方法费时费力、无法获取地表沉降面状信息、难以进行地表沉陷灾害评估的不足,基于高分辨率SAR卫星影像,利用永久散射体合成孔径雷达干涉测量(PS-InSAR)技术对山西省晋城市晋城矿区2018年1月至2018年12月期间地表沉陷进行监测,分析获取了该地区地表连续形变情况,并利用该技术获取的海量PS点建立支持向量机(SVM)地质灾害风险评估预警模型,对晋城矿区周边居民点地质灾害风险进行了识别和预测。结果表明:晋城矿区10个煤矿及其周边区域存在较大的地表形变;晋城矿区平均LOS向年平均地表形变速率范围为-37~30.3 mm/a;PS-InSAR技术在晋城矿区地表形变监测中具有可行性,且可以实现矿区地质灾害风险综合识别和预警。

基于混合模型的道岔综合监测系统研究

道岔作为关键的铁路信号设备,也是铁路线路三大薄弱环节之一,其工作质量直接影响列车的运行安全。传统的道岔检测方法过分依赖人工经验,检测效率低下,难以应对现有铁路运行中行车速度快、发车密度高等对道岔维护所带来的严峻挑战,并且现有道岔监测也存在监测项目不全面等问题。为满足工电融合需要,开发了一套基于混合模型的道岔综合监测系统,使用卷积神经网络自动进行特征提取,以获取道岔状态,充分发挥深度学习的自动特征提取优势;采用支持向量机和向量域的混合算法,对正常/故障数据进行分类和异常检测,从而提高故障检测的准确率。测试结果表明:与现有人工巡检方法相比,该系统能够为相关人员提供精准、实时的道岔故障预警,提高维护效率,有效减少人力成本且降低道岔病害的发生概率。

基于PSO-SVM的矿井输送带火灾危险程度预测模型研究

矿井带式输送机输送带火灾严重威胁煤矿安全生产。矿井输送带火灾风险监测预警方面存在指标内在关联性差、火灾不同发展阶段的标志性气体和特征温度不明确、火灾风险预警指标缺失等难题。因此,采用热重红外联用仪和锥型量热仪装置,研究了矿井输送带火灾热解燃烧的特性,热解初始温度段,首先生成较多的CO_(2)、H2O以及少量CO、HCl,温度升至252℃时,HCl生成量开始迅速增加,298℃时HCl的生成量达到峰值1.26%;CO在416℃时生成量开始迅速增加,485℃时CO的生成量达到峰值0.29%;CO_(2)在整个热解过程中产生量最大,总结出CO、CO_(2)、HCl作为输送带火灾监测预警指标。建立了基于粒子群算法优化支持向量机的模型,并对模型进行了最优参数与效果的研究分析。为了验证模型的准确性和可靠性,采用最小二乘误差指标将PSOSVM与SVM预测结果进行对比。

基于EMD分量与小波包能量熵的轧辊磨削颤振在线预测

针对轧辊磨削颤振时的时频域单一处理方法存在部分特征丢失的问题,提出了时频域相结合的方法对信号进行特征处理,并利用智能算法实现轧辊磨削颤振的在线预测。首先,利用经验模态分解(empirical mode decomposition,EMD)方法对振动传感器信号进行分解获得各固有模态函数(intrinsic mode function,IMF),剔除“虚假分量”后计算表征轧辊磨削颤振的时域特征。然后,利用小波包能量熵对声发射传感器信号求解频率段节点能量熵值,获得表征轧辊磨削颤振的频域特征。最后,将上述时频域特征降维后代入智能算法模型实现对轧辊磨削加工的在线预测。结果表明:LV-SVM模型的磨削颤振分类平均准确率达92.75%,模型平均响应时间为0.7765 s;验证了时频域特性的EMD和小波包能量熵方法的LV-SVM在线预测轧辊磨削颤振的有效性。

基于多分类支持向量机的变压器在线监测数据错误模式识别

针对变压器在线油中溶解气体在线监测数据质量问题,统计了200多台监测装置的2020全年数据,总结了3种主要数据错误模式;提出了数据错误模式识别策略和特征参数,构建了多分类支持向量机进行错误数据识别与分类;并利用核主成分分析法和排列组合遍历寻优法对特征向量进行了降维优化。所构建的多分类支持向量机分类器对于H_(2)错误数据识别准确率达到97.5%,对于其他气体达到90%以上。应用所构建的分类器对2020全年数据进行了统计,其中H2的错误数据达到27.14%,C_(2)H_(2)的错误数据达到1.75%。

基于先验统计模型的非侵入负荷辨识算法

针对传统非侵入负荷辨识技术中电热细分能力不足的问题,文中提出了一种基于先验知识与统计学习模型的居民非侵入式负荷辨识算法。文中对洗衣机辅热、电水壶、电饭锅、电热水器等设备进行了电热细分研究,通过设备运行关联算法实现了辅热设备的细分,并在用户有限反馈信息和专家标注的基础上,实现了非辅热设备分类的模型训练。实验结果表明,文中所提技术框架在事件检测负荷辨识算法的基础上实现了电热设备的细分,且在运行状态分解的F1分数指标中取得了0.9以上的优异效果。

基于传感信号采集的电控发动机振动故障监测方法

通过调理振动信号可以更高效地监测振动故障。为此,提出基于传感信号采集的电控发动机振动故障监测方法。首先,搭建电控发电机传感信号采集与处理架构,通过放大传感信号增益、滤波和转换信号模数的方式处理待监测信号,为提高监测准确性奠定可靠的数据基础。通过小波包分解与重构,获取信号的时域参数和小波能谱熵,并构建三维特征量。然后,利用“一对一”分解策略优化孪生支持向量机,构造多元分类器,使其更适用于振动故障监测这一多类别分类问题,再输入待监测信号的特征量,通过确定故障类别实现持续性监测。仿真结果表明:该方法训练耗时的最大值仅为897 ms,对于转子摩擦振动、不平衡振动等5种类型故障的监测准确率始终在97%以上,在缩减训练样本后准确率仍保持在90%以上。

基于机器学习的刀具磨损状态智能预测方法研究

以刀具为研究载体,运用人工智能和智能优化等先进技术,成功实现了刀具磨损状态的智能预测。研究重点在于建立有效的刀具磨损状态预测方法,全面解析刀具磨损机理、形式及磨钝标准等关键信息。同时,构建了自采刀具磨损状态监测平台,以便收集并处理相关数据。在数据处理过程中,采用小波滤噪和EMD-Shannon能量熵进行特征筛选,构建出特征空间数据集,为后续构建预测模型提供坚实的数据基础。结合支持向量机分类算法和智能优化算法,构建出刀具磨损状态的智能预测框架。此框架不仅提高了预测精度,也为维护人员提供了强有力的工具,利于更好地进行刀具磨损状态的预测和维护工作。为增强实际应用价值,将所取得的成果整合至基于MATLAB GUI的刀具磨损状态智能监测原型系统,以图形界面方式呈现预测结果,使用户直观地了解和掌握刀具的磨损状态。结果表明,该方法具有高精度,刀具磨损状态的识别精度可达84%,为相关领域提供了可靠的技术支持。