采用神经网络和支持向量机预测啤酒中乙酸含量

啤酒中的乙酸大多是酵母在发酵过程中产生的。乙酸含量对啤酒风味的影响显著，尤其是含量高于闽值时。因此，控制乙酸的含量对保证啤酒风味一致性非常重要。在本项研究中，采用人工神经网络和支持向量机（SVM）来预测啤酒发酵结束时的乙酸含量。啤酒发酵过程参数和啤酒中乙酸含量之间的关系采用偏最小二乘（PLS）回归法、反向传播神经网络（BP—NN）、径向基函数神经网络（RBF—NN）和最小二乘支持向量机（LS-SVM）进行建模。本研究中所使用的数据来自同一品牌啤酒的146个生产批次。LS—SVM和RBF预测乙酸含量要优于RBP—NN和PLS。对比RBF—NN和LS—SVM，RBF—NN构建的模型可靠性更好，但预测的准确性要低一些。SVM有较好的泛化性，但是模型的可靠性较低。总之，在这项研究中，预测大生产啤酒发酵中的乙酸含量时，LS-SVM模型要优于RBF。

基于形态学小波理论和SVM神经网络的人脸识别

主要研究了快速识别人脸的基本算法,它包括人脸检测和人脸识别两部分.人脸检测部分利用肤色电平的聚类特性和形态学处理检测出准人脸图像,再利用小波特征提取出特征进行人脸认证.人脸识别部分采用支持向量机(SVM)神经网络进行人脸识别.支持向量机神经网络对二类判别具有很强的识别能力.对于N类判别需连续使用N次.该方法识别速度快,且不受发型、头饰、眼镜等的影响.仿真证明了该方法的有效性.

一种基于支持向量机的齿轮箱故障诊断方法

提出了一种基于多分类支持向量机(简称MSVM)的齿轮箱故障诊断方法。先根据齿轮箱故障机理和振动特点,探讨了齿轮箱故障诊断试验方案。再测取齿轮箱振动信号,并提取了能反映齿轮箱运转信息的时频域特征参数。通过结合投票法和决策树的基本思想,有针对性地构造了多分类支持向量机决策结构并将其应用于齿轮箱故障诊断。实际齿轮箱故障诊断试验结果表明,该决策结构较好地解决了小样本学习问题,避免了人工神经网络进行诊断时出现的过学习、收敛速度慢、泛化能力弱等缺点,能有效应用于齿轮箱故障诊断。

基于暂态时-频特征差异的配电网高阻接地故障识别方法

高阻接地故障发生时,故障特征微弱,传统故障识别方法存在特征提取困难、阈值选取灵活性较差的技术瓶颈,导致极端故障场景下出现漏判。为此,提出基于暂态时-频特征差异的配电网高阻接地故障识别方法。首先,结合小波包香农熵量化分析高阻接地故障与正常扰动工况暂态信号的时频分布,发现二者存在显著差异:频域上,扰动工况信号的能量集中于低频,而高阻故障信号能量分布相对均匀;时域上,扰动工况信号能量集中于时间窗的前半段,高阻故障信号能量在整个时间窗内均匀分布。在此基础上,以暂态信号时-频域波形作为输入样本,将传统卷积神经网络(convolutional neural networks,CNN)模型中的softmax分类器改进为支持向量机(support vector machine,SVM)分类器,构建适应配电网高阻接地故障识别小样本场景下的CNN-SVM复合分类模型,以卷积层作为特征提取器,以SVM作为分类器,实现高阻接地故障识别。最后,为论证所提方法具有强适应性的内在原因,利用LIME可解释性分析算法可视化展现模型训练过程中的高关注度区域,从模型分类原理层面证明所提方法不受各种故障条件的影响,克服了传统故障识别方法在极端故障场景下出现漏判的缺陷,能准确识别配电线路末端10 kΩ高阻接地故障。

发酵过程中生物量浓度的在线估计

在发酵过程中,像生物量浓度等变量都是进行实验室的离线采样分析,这往往由于存在较大的时间延迟而不能及时地进行过程控制,达不到指导生产的目的.而软测量技术为该问题提出了一个很好的解决办法.基于神经网络与最小二乘支持向量机分别建立了生物量浓度的在线检测软测量模型.模型分为两类:黑箱模型与混合模型.模型的训练与验证数据都是取自真实的实验过程诺西肽发酵.结果表明软测量方法对生物量浓度具有很好的预估性能,而且加入先验知识的混合模型精度更高.

基于CNN-SVM性别组合分类的单通道语音分离

实际语音分离时,混合语音的说话人性别组合相关信息往往是未知的。若直接在普适的模型上进行分离,语音分离效果欠佳。为了更好地进行语音分离,本文提出一种基于卷积神经网络-支持向量机(CNN-SVM)的性别组合判别模型,来确定混合语音的两个说话人是男-男、男-女还是女-女组合,以便选用相应性别组合的分离模型进行语音分离。为了弥补传统单一特征表征性别组合信息不足的问题,本文提出一种挖掘深度融合特征的策略,使分类特征包含更多性别组合类别的信息。本文的基于CNN-SVM性别组合分类的单通道语音分离方法,首先使用卷积神经网络挖掘梅尔频率倒谱系数和滤波器组特征的深度特征,融合这两种深度特征作为性别组合的分类特征,然后利用支持向量机对混合语音性别组合进行识别,最后选择对应性别组合的深度神经网络/卷积神经网络(DNN/CNN)模型进行语音分离。实验结果表明,与传统的单一特征相比,本文所提的深度融合特征可以有效提高混合语音性别组合的识别率;本文所提的语音分离方法在主观语音质量评估(PESQ)、短时客观可懂度(STOI)、信号失真比(SDR)指标上均优于普适的语音分离模型。

Fault detection in flotation processes based on deep learning and support vector machine

Effective fault detection techniques can help flotation plant reduce reagents consumption,increase mineral recovery,and reduce labor intensity.Traditional,online fault detection methods during flotation processes have concentrated on extracting a specific froth feature for segmentation,like color,shape,size and texture,always leading to undesirable accuracy and efficiency since the same segmentation algorithm could not be applied to every case.In this work,a new integrated method based on convolution neural network(CNN)combined with transfer learning approach and support vector machine(SVM)is proposed to automatically recognize the flotation condition.To be more specific,CNN function as a trainable feature extractor to process the froth images and SVM is used as a recognizer to implement fault detection.As compared with the existed recognition methods,it turns out that the CNN-SVM model can automatically retrieve features from the raw froth images and perform fault detection with high accuracy.Hence,a CNN-SVM based,real-time flotation monitoring system is proposed for application in an antimony flotation plant in China.

Flame image recognition of alumina rotary kiln by artificial neural network and support vector machine methods

Based on the Fourier transform, a new shape descriptor was proposed to represent the flame image. By employing the shape descriptor as the input, the flame image recognition was studied by the methods of the artificial neural network(ANN) and the support vector machine(SVM) respectively. And the recognition experiments were carried out by using flame image data sampled from an alumina rotary kiln to evaluate their effectiveness. The results show that the two recognition methods can achieve good results, which verify the effectiveness of the shape descriptor. The highest recognition rate is 88.83% for SVM and 87.38% for ANN, which means that the performance of the SVM is better than that of the ANN.

特高压线路工程的工程量组合预测研究

在分析特高压(ultra high voltage,UHV)线路工程的工程量影响因素基础上,根据已有特高压线路工程相关数据特点,提出支持向量机、BP神经网络以及工程相似度三种工程量预测方法,针对单一预测方法的局限性,为进一步提高预测精度,构建基于偏差平方和最小的组合预测模型,组合预测模型可以多角度搜集数据信息,实现各种预测模型之间的取长补短。通过样本测试表明该组合预测模型明显降低了最大误差,缩小了误差波动范围。同时考虑到不可量化因素对特高压线路工程量的影响,利用数理统计中置信区间的估计得到工程量的区间预测值,为特高压线路工程量管控提供技术支撑。

Estimation of wear performance of AZ91 alloy under dry sliding conditions using machine learning methods

The wear behavior of AZ91 alloy was investigated by considering different parameters,such as load(10−50 N),sliding speed(160−220 mm/s)and sliding distance(250−1000 m).It was found that wear volume loss increased as load increased for all sliding distances and some sliding speeds.For sliding speed of 220 mm/s and sliding distance of 1000 m,the wear volume losses under loads of 10,20,30,40 and 50 N were calculated to be 15.0,19.0,24.3,33.9 and 37.4 mm3,respectively.Worn surfaces show that abrasion and oxidation were present at a load of 10 N,which changes into delamination at a load of 50 N.ANOVA results show that the contributions of load,sliding distance and sliding speed were 12.99%,83.04%and 3.97%,respectively.The artificial neural networks(ANN),support vector regressor(SVR)and random forest(RF)methods were applied for the prediction of wear volume loss of AZ91 alloy.The correlation coefficient(R2)values of SVR,RF and ANN for the test were 0.9245,0.9800 and 0.9845,respectively.Thus,the ANN model has promising results for the prediction of wear performance of AZ91 alloy.

Analyses and predictions of rock cuttabilities under different confining stresses and rock properties based on rock indentation tests by conical pick

The rock indentation tests by a conical pick were conducted to investigate the rock cuttability correlated to confining stress conditions and rock strength.Based on the test results,the regression analyses,support vector machine(SVM)and generalized regression neural network(GRNN)were used to find the relationship among rock cuttability,uniaxial confining stress applied to rock,uniaxial compressive strength(UCS)and tensile strength of rock material.It was found that the regression and SVM-based models can accurately reflect the variation law of rock cuttability,which presented decreases followed by increases with the increase in uniaxial confining stress and the negative correlation to UCS and tensile strength of rock material.Based on prediction models for revealing the optimal stress condition and determining the cutting parameters,the axial boom roadheader with many conical picks mounted was satisfactorily utilized to perform rock cutting in hard phosphate rock around pillar.

Support vector regression-based internal model control

This paper proposes a design of internal model control systems for process with delay by using support vector regression(SVR).The proposed system fully uses the excellent nonlinear estimation performance of SVR with the structural risk minimization principle.Closed-system stability and steady error are analyzed for the existence of modeling errors.The simulations show that the proposed control systems have the better control performance than that by neural networks in the cases of the training samples with small size and noises.

Joint application of feature extraction based on EMD-AR strategy and multi-class classifier based on LS-SVM in EMG motion classification

This paper presents an effective and efficient combination of feature extraction and multi-class classifier for motion classification by analyzing the surface electromyografic(sEMG) signals. In contrast to the existing methods,considering the non-stationary and nonlinear characteristics of EMG signals,to get the more separable feature set,we introduce the empirical mode decomposition(EMD) to decompose the original EMG signals into several intrinsic mode functions(IMFs) and then compute the coefficients of autoregressive models of each IMF to form the feature set. Based on the least squares support vector machines(LS-SVMs) ,the multi-class classifier is designed and constructed to classify various motions. The results of contrastive experiments showed that the accuracy of motion recognition is improved with the described classification scheme. Furthermore,compared with other classifiers using different features,the excellent performance indicated the potential of the SVM techniques embedding the EMD-AR kernel in motion classification.

SVM model for estimating the parameters of the probability-integral method of predicting mining subsidence

A new mathematical model to estimate the parameters of the probability-integral method for mining subsidence prediction is proposed.Based on least squares support vector machine(LS-SVM) theory, it is capable of improving the precision and reliability of mining subsidence prediction.Many of the geological and mining factors involved are related in a nonlinear way.The new model is based on statistical theory(SLT) and empirical risk minimization(ERM) principles.Typical data collected from observation stations were used for the learning and training samples.The calculated results from the LS-SVM model were compared with the prediction results of a back propagation neural network(BPNN) model.The results show that the parameters were more precisely predicted by the LS-SVM model than by the BPNN model.The LS-SVM model was faster in computation and had better generalized performance.It provides a highly effective method for calculating the predicting parameters of the probability-integral method.