基于M估计器的支持向量机算法及其应用

训练样本的准确性对回归分析模型有很大的影响,然而训练样本中难免会出现一些造成分析模型失效的奇异点。 为克服奇异点对回归模型的影响,本文提出了一种基于M估计器的支持向量机（M-SVM）。它采用M估计器的目标函数代替最小二乘支持向量机（LS-SVM）目标函数中的残差平方和,同时提出了M-SVM的迭代求解算法,并将该算法应用于含有奇异点的低维仿真数据回归和汽油近红外光谱定量分析中。实验结果证明,相比于其他的支持向量机,M-SVM具有更好的稳健性和分析精度。     

Outlier Detection And Estimation In NonLinear Time Series

Abstract. The problem of identifying the time location and estimating the amplitude of outliers in nonlinear time series is addressed. A model‐based method is proposed for detecting the presence of additive or innovational outliers when the series is generated by a general nonlinear model. We use this method for identifying and estimating outliers in bilinear, self‐exciting threshold autoregressive and exponential autoregressive models. A simulation study is performed to test the proposed procedures and comparing them with the methods based on linear models and linear interpolators. Finally, our results are applied for detecting outliers in the Canadian lynx trappings and in the sunspot numbers data.     

一种新型融合离群点的稳态检测方法

针对带有离群点的数据稳态检测,采用分布图法对离群点进行剔除;为了保持数据的完整性,提出用灰色预测值替代离群点值;最后用3δ法则进行稳态检验。如此,数据的稳态与非稳态便会区分开来。与现有稳态检测方法相比,分布图法快速有效地克服了离群点对稳态检测结果不准确的影响,降低了过程中个别异常数据带来的误诊率;灰色预测方法使离群点的替代值更贴近真实值,从而得到的过程数据比现有方法得到的数据更可靠。仿真结果证实了该方法的有效性和优越性。     

Autoregressive modeling of near-IR spectra and MLR to predict RON values of gasolines

A new calibration method that accurately predicts the Research Octane Number (RON) values of gasoline fractions, based on their infrared spectra, is presented. This model combines Linear Predictive Coding (LPC) and multiple linear regression (MLR) as an integrated estimation technique. Spectral information from the 4800-3520 cm⁻¹ range was initially encoded into Linear Predictive (LP) coefficients, which were used as predictor variables in the MLR model against RON values. The model was trained and tested on an extensive data set (384 gasoline samples) and found to ensure prediction accuracy of 0.3 RON Root Mean Squared Error (RMSE). The LPC technique was found to be efficient in capturing spectral features of the entire range, related to the RON characteristics of the gasoline samples, without the need of any pretreatment on the experimental raw data. The small number of input variables in the regression model ensures a robust, easy-to-use and high accuracy prediction model.     

一种改进的动态数据校正离群值检测法

Data reconciliation technology can decrease the level of corruption of process data due to measurement noise, but the presence of outliers caused by process peaks or unmeasured disturbances will smear the reconciled results. Based on the analysis of limitation of conventional outlier detection algorithms, a modified outlier detection method in dynamic data reconciliation （DDR） is proposed in this paper. In the modified method, the outliers of each variable are distinguished individually and the weight is modified accordingly. Therefore, the modified method can use more information of normal data, and can efficiently decrease the effect of outliers. Simulation of a continuous stirred tank reactor （CSTR） process verifies the effectiveness of the proposed algorithm.     

傅立叶近红外光谱法测定汽油中苯含量

介绍了傅立叶红外光谱测定汽油中苯含量的方法.以标准分析方法为参照,利用傅立叶红外光谱仪测定汽油的红外光谱,采用一阶微分和偏最小二乘法对汽油中的苯含量建立校正模型.该模型测定未知样品汽油苯含量的结果与标准分析方法的偏差符合标准方法的要求.     

A Bayesian approach to robust process identification with ARX models

In the context of process industries, outlying observations mostly represent a large random error resulting from irregular process disturbances, instrument failures, or transmission problems. Statistical analysis of process data contaminated with outliers may lead to biased parameter estimation and plant‐model mismatch. The problem of process identification in the presence of outliers has received great attention and a wide variety of outlier identification approaches have been proposed. However, there is a great need to seek for more general solutions and a robust framework to deal with different types of outliers. The main objective of this work is to formulate and solve the robust process identification problem under a Bayesian framework. The proposed solution strategy not only yields maximum a posteriori estimates of model parameters but also provides hyperparameters that determine data quality as well as prior distribution of model parameters. Identification of a simulated continuous fermentation reactor is considered to show the effectiveness and robustness of the proposed Bayesian framework. The advantages of the method are further illustrated through an experimental case study of a pilot‐scale continuous stirred tank heater. © 2012 American Institute of Chemical Engineers AIChE J, 59: 845–859, 2013     

分数阶微分在红外光谱数据预处理中的应用

比较了Grumwald-Letnikov分数阶微分算法和常用的Savitzky-Golay算法对汽油样品近红外光谱数据1阶微分和2阶微分结果。对25个汽油样品和41个煤炭样品的近红外光谱数据通过Savitzky-Golay算法进行平滑,平滑后1阶微分和平滑后2阶微分处理;通过Grumwald-Letnikov算法进行平滑后的0.2～2.2阶的21个阶次的微分处理。汽油样品的处理数据结合汽油的辛烷值、初馏点指标数据和煤炭样品的处理数据结合煤炭挥发分、氢含量和氮含量指标数据分别通过PLS建立数据模型,利用留一法全交互验证选取最优主成分。通过预测残差平方和(PRESS)和相关系数(R)对数据,处理方法进行评估。结果表明:分数阶微分可以应用于近红外光谱的数据预处理,对于相同数据的不同指标取得最优值的分数阶微分的阶次是不同的。     

Using principal component analysis technique in the instrumental shade sorting of textile fabrics

In the present study, the CCC shade sorting method was employed with CMC(2:1) color difference formula on the colorimetric data (CIEL*a* b*) of 37 fabric color sets. The k‐means non‐hierarchical clustering technique was also combined with the CCC shade sorting method to increase its efficiency. The results of this combined method showed a slightly better performance, as compared with the CCC method. Also, a new proposed shade sorting method by the application of principal components analysis (PCA) technique was used to identify and remove the outliers in each of the color sets. The results of separating the outliers showed that although the diameter of group criterion was improved significantly, the number of groups, the number of singleton groups, and the number of groups with low samples were increased considerably. Finally, in a second new proposed shade sorting method, PCA was used as a data reduction tool on the colorimetric data of the 37 color sets. Then, the two first principal components in combination with a k‐means clustering technique were used for the clustering of the samples in each color set. The results of this second new proposed method were found to be similar to the CCC method considering number of group and fabric consumption criteria. The second new proposed method revealed a moderately worse result, with regard to the diameter of group criterion, than the CCC method.     

基于动态单类随机森林的非线性过程监控方法

针对高维化工过程中存在的非线性和动态特性，提出了一种基于动态单类随机森林（dynamic one-class random forest，DOCRF（的过程监控方法。对正常运行状态下的过程数据进行稀疏性分析，根据其反分布产生离群点数据。利用典型变量分析对正常数据进行相关性分析，分别将正常数据和离群点数据投影到典型变量空间，利用典型变量空间数据训练单类随机森林。基于单类随机森林模型根据待检测样本与正常数据的相似度构造监控统计量进行故障检测。在Tennessee Eastman过程的仿真结果表明，所提DOCRF方法总体优于单类支持向量机方法。     

Adulteration detection of Brazilian gasoline samples by statistical analysis

Unfortunately, addition of organic solvents (heavy aliphatic, light aliphatic and aromatic hydrocarbons) in Brazilian gasoline is very frequent, and this illegal practice does not guarantee gasoline quality. Organic solvent adulterations of gasoline samples have been investigated. For characterization and comparison of these samples, physico-chemical parameters together with gas chromatographic analyses data were proposed as the factors for multivariate analysis. Hierarchical clusters analysis was used to improve the detection of the type of solvent and their relative proportion used for this practice. More detailed information of their compositions was revealed. It was found that using physico-chemical properties of gasoline samples together with statistical analysis are a useful method to adulteration detection.     

一种新颖的鲁棒动态数据校正方法

Outlier in one variable will smear the estimation of other measurements in data reconciliation （DR）. In this article, a novel robust method is proposed for nonlinear dynamic data reconciliation, to reduce the influence of outliers on the result of DR. This method introduces a penalty function matrix in a conventional least-square objective function, to assign small weights for outliers and large weights for normal measurements. To avoid the loss of data information, element-wise Mahalanobis distance is proposed, as an improvement on vector-wise distance, to construct a penalty function matrix. The correlation of measurement error is also considered in this article. The method introduces the robust statistical theory into conventional least square estimator by constructing the penalty weight matrix and gets not only good robustness but also simple calculation. Simulation of a continuous stirred tank reactor, verifies the effectiveness of the proposed algorithm.     

Eliminating material dependency in spectra measurement via non‐neighbouring band regression

A multispectral imaging system, after necessary calibration, can measure the spectral reflectances of colour samples accurately at a high spatial resolution. A limitation is that agreement of its measurements with those of a reference spectrophotometer is affected by the reflective characteristics of sample materials. The state‐of‐the‐art methods aim to improve interinstrument agreement using the spectral values of neighbouring bands. However, it is observed that non‐neighbouring bands are more effective in modelling interinstrument agreement. Inspired by this observation, the present paper proposes a method for eliminating material dependency by least‐squares regression among non‐neighbouring spectral bands. The fundamental issue of band selection is solved using a binary differential evolution algorithm. Experimental results confirm that the proposed method is effective in reflectance correction in terms of both spectral and colorimetric accuracy. The method is of practical application to multispectral imaging systems when measuring the spectral reflectances of colour samples with different materials.     

A novel robust dynamic method for NOx emissions prediction in a thermal power plant

NOx is a harmful by-product of coal-fired boilers, and accurate prediction of NOx emissions in the outlet of a boiler is essential for environmental protection. In recent years, data-driven models have been widely studied and applied in this area. However, dynamic characteristics are ignored by many existing models, leading to sub-optimal performance. Besides, outliers that occur in the operation data have adverse effects on the efficacy of these prediction models. To address these issues, this paper presents a novel method for predicting NOx concentration via integrating a robust dynamic probabilistic approach and the long short-term memory (LSTM). First, mutual information (MI) is applied to determine the input variables. Subsequently, a robust probabilistic method is proposed to extract dynamic latent features considering outliers. On this basis, the generated latent variables are further utilized to train the LSTM-based model, with which the intrinsic relation between inputs and NOx values are obtained. Based on the application to a 660 MW thermal power plant, the superiority of the proposed method is demonstrated in terms of high prediction accuracy.     

SEARCHING FOR ADDITIVE OUTLIERS IN NONSTATIONARY TIME SERIES*

Abstract. Recently, Vogelsang (1999) proposed a method to detect outliers which explicitly imposes the null hypothesis of a unit root. It works in an iterative fashion to select multiple outlier in a given series. We show, via simulations, that, under the null hypothesis of no outliers, it has the right size in finite samples to detect a single outlier but, when applied in an iterative fashion to select multiple outliers, it exhibits severe size distortions towards finding an excessive number of outliers. We show that his iterative method is incorrect and derive the appropriate limiting distribution of the test at each step of the search. Whether corrected or not, we also show that the outliers need to be very large for the method to have any decent power. We propose an alternative method based on first‐differenced data that has considerably more power. We also show that our method to identify outliers leads to unit root tests with more accurate finite sample size and robustness to departures from a unit root. The issues are illustrated using two US/Finland real‐exchange rate series.     

Blackbody-based calibration for temperature calculations in the visible and near-IR spectral ranges using a spectrometer

This paper presents, the results of a method used to create a blackbody-based referenced calibration curve for a spectrometer in the visible and near-IR range. This method would allow the use of optical temperature measurements in high temperature furnaces when distance, environment, and emissivity effects are not accurately known. A probe containing a lens connected to a fiber-optic cable is inserted into a furnace and aimed toward a hot wall source. Spectral intensity data is fed back to a spectrometer and then to a monitoring computer. Initial data is taken along with another method to measure the source temperature, usually a thermocouple or IR-gun. The spectral data is compared to the blackbody intensities generated from the source temperature to create the calibration curve. This calibration curve is then used to correct intensities for temperature calculations using a spectrometer where furnace conditions are similar to those of the calibration data. This calibration method provides much more accurate temperature measurement results than the common practice of using a halogen reference. The results in the visible range compare favorably with those taken in the near-IR range under the same conditions.     

Robust identification of piecewise/switching autoregressive exogenous process

A robust identification approach for a class of switching processes named PWARX (piecewise autoregressive exogenous) processes is developed in this article. It is proposed that the identification problem can be formulated and solved within the EM (expectation‐maximization) algorithm framework. However, unlike the regular EM algorithm in which the objective function of the maximization step is built upon the assumption that the noise comes from a single distribution, contaminated Gaussian distribution is utilized in the process of constructing the objective function, which effectively makes the revised EM algorithm robust to the latent outliers. Issues associated with the EM algorithm in the PWARX system identification such as sensitivity to its starting point as well as inability to accurately classify “un‐decidable” data points are examined and a solution strategy is proposed. Data sets with/without outliers are both considered and the performance is compared between the robust EM algorithm and regular EM algorithm in terms of their parameter estimation performance. Finally, a modified version of MRLP (multi‐category robust linear programming) region partition method is proposed by assigning different weights to different data points. In this way, negative influence caused by outliers could be minimized in region partitioning of PWARX systems. Simulation as well as application on a pilot‐scale switched process control system are used to verify the efficiency of the proposed identification algorithm. © 2009 American Institute of Chemical Engineers AIChE J, 2010     

钢铁企业高炉煤气发生量异常数据检测

从高炉煤气生产的实际工况出发,对异常数据产生的原因和特点进行分析。针对现有异常检测方法运算效率低下的问题,提出一种改进的局部异常因子检测算法。该算法首先利用五数总括法剔除掉大量的正常数据,然后再用一种相对k距离的比值表示剩余离群点的异常程度,进而判断异常数据。仿真实验表明:改进方法检测所需的时间比传统的局部异常因子方法检测所需的时间更少,且检测效果更加准确、直观。     

Determination of octane numbers for clean gasoline using dielectric spectroscopy

In the present work, dielectric spectroscopy (DES) in association with partial least squares (PLS) multivariate calibration method was employed to determine octane numbers (research octane number or RON and motor octane number or MON) of clean gasoline samples. The factor number included in PLS model was obtained according to the lowest sum of squares of predicted residual error (PRESS) in calibration set. The performance of the final model was evaluated according to PRESS and correlation coefficient (R). The optimal factor numbers are 9 in both RON and MON PLS calibration models, which were achieved with PRESS = 2.74 and R = 0.9598 in RON calibration set and the lowest PRESS = 2.72 and R = 0.8983 in MON calibration set. In validation set, PRESS = 1.00 and R = 0.9552 for RON and PRESS = 0.47 and R = 0.9105 for MON were obtained. Results indicated that PLS multivariate calibration models based on DES data were proven suitable as a practical analytical method for predicting octane numbers of clean gasoline.