基于振动信号的轮式机器人地面分类方法研究
|
摘要
为了在行星（如月球和火星）表面探测及在地球表面的危险环境（如沙漠、沼泽、火灾现场、核辐射区域等）中作业，要求自主移动机器人能自主地识别环境，完成使命，避免处于危险境地。地面识别或地面分类是环境识别中重要的一部分。机器人安全有效地穿越不同的地面需要与之相适应的控制策略。当地面发生变化时，自主移动机器人应能适应所穿越的地面。研究地面分类可以解决自主移动机器人在复杂地面的通过性问题，对于提高移动机器人的自主移动性能十分重要。
本文在深入分析综合国内外同类研究的基础上，从地面分类特征提取以及分类器设计这两个基本层次展开对自主轮式机器人地面分类相关理论与技术问题的研究。
本文设计了数据采集的实验，以四轮移动机器人为实验平台，在机器人左前轮轮臂上安装x、 y、 z向加速度计和z向传声器。机器人在沙、碎石、草、土和沥青五种地面上分别以六种速度行驶，提取车轮与地面相互作用的加速度和声压信号作为地面分类的原始数据。
对原始数据进行时域幅值分析，提取地面特征，对于每个传感器数据选用若干个幅值域参数作为地面特征。对于传统的k-近邻（kNN）方法来说，有k值选择问题；用kNN方法解决多类分类问题采用投票法决策，当出现票数相同的情况时，目前尚没有更好的决策策略，尽管随机挑选法是一个实用的策略，但分类精度会降低。针对以上两个问题，本文给出改进的kNN方法，即给出k值的选择方法；对于多种地面分类出现两种以上（包含两种）得票数相同的情况，给出kNN循环寻优的方法。
对于传统的概率神经网络（PNN）方法来说，有平滑因子σ的估计问题，估计得好有利于提高分类精度。以前的学者认为对于所有的样本应选用同一个σ或对于相同维数的样本选用相同的σ，但这种选法不能保证对于所有的测试样本σ都是最优的或较优的，甚至得不到测试结果。针对这一问题，本文给出改进的PNN方法，即给出平滑因子σ的迭代寻优方法。
应用现有的一对一支持向量机（SVM）方法解决多类分类问题采用投票法决策，当出现票数相同的情况时，目前尚没有更好的决策策略。针对这个问题，本文提出改进的一对一SVM方法，即利用LIBSVM中的一对一SVM二值分类程序，对于多种地面分类出现两种以上（包含两种）得票数相同的情况，提出新的算法。从分类准确率和数据处理时间两个方面对改进的kNN、改进的PNN及改进的一对一SVM方法进行比较。
基于时间序列重构的吸引子轨迹矩阵奇异值分解（SVD）的方法原用于故障诊断领域，用于降低原信号中的噪声。本文给出基于奇异值分解（SVD）的特征提取方法，即利用振动信号时间序列重构的吸引子轨迹矩阵奇异值分解的前若干个奇异值作为特征值，取得了好的分类效果。研究基于快速傅里叶变换（FFT）的特征提取方法和基于功率谱密度（PSD）的特征提取方法，阐述二者特征选择的方法。从分类准确率和数据处理时间两个方面对以上三种特征提取方法进行比较。
基于实测数据和相应的分类实验验证了所提方法的效力。
In order to explore in the planets’(such as the moon and Mars) surface and work in thedangerous environment (such as desert, marsh, the scene of the fire, nuclear radiation area, etc)of the earth’s surface, autonomous mobile robots should be able to independently identifyenvironment, complete the mission without a dangerous situation. Terrain identification orterrain classification is an important part of environmental identification. Correspondingcontrol strategy is necessary for robot to travel on different terrain safely and effectively,when the terrain changes, autonomous mobile robot must be able to adapt to the terrain whereit is traversing. Terrain classification can solve the issue of trafficability of autonomousmobile robot in complex terrain. It is very important to improve robot autonomous mobileperformance.
Based on in-depth analysis and synthesis of similar studies home and abroay, the theoryand techniques are researched from the two aspects, i.e. terrain classification featureextraction and classification method.
In this dissertation，experiments for data acquisition are designed. The experimentalplatform is a four-wheeled mobile robot on which arm accelerometers in x, y, zdirections and a microphone in z direction are installed in left front wheel. When the robotis traversing respectively on sand, gravel, grass, soil and asphalt terrain with six differentvelocities, the acceleration and sound pressure signals of wheel-terrain interaction arecollected as the original data.
By time domain amplitude analysis of original data, several parameters of amplitudedomain are selected as the terrain features for each sensor data. To the conventionalk-nearest neighbors (kNN) algorithm, it is necessary to deal with the choice of k, and nowthere is no best decision strategy for the situation when number of votes is the same in theprocess of multi-classification based on voting decisions, though a practical strategy selectedis random method, which is reduces classification accuracy. To the two problems, animproved kNN method was proposed, i.e. the choice method of k was proposed and kNNcycle optimization method was also investigated to deal with the problem that more than twokinds (including two) of terrains have the same number of votes.
To the conventional probabilistic neural network (PNN) method, there is a problem aboutthe estimation of smoothing factor σ which is important to improve the classification accuracy. Previous scholars considered that the same σ was chosen for all samples orsamples of the same dimensions, but it could not make sure that the σ was the best orsub-optimal for all test samples, even there was no result. For the problem, an improved PNNmethod was proposed to deal with the choice of smoothing factor σ by iterativeoptimization method.
The traditional one-against-one support vector machine (SVM) method, now there is nobest decision strategy for the situation when number of votes is the same in the process ofmulti-classification based on voting decisions, an improved one-against-one SVM methodwas rendered to deal with the problem that more than two kinds (including two) of terrainshave the same number of votes based on two-classification program of LIBSVM. Yetimproved kNN，improved PNN and improved one-against-one SVM methods were comparedin terms of classification accuracy and data processing time.
In the field of fault diagnosis, a method based on singular value decomposition of trackmatrix of attractor reconstructed by time series is always used to reduce the noise in originalsignal. Based on singular value decomposition (SVD), a feature extraction method wasproposed using the fore several singular values of track matrix of attractor reconstructed byvibration signals time series as eigenvalues, and better classification effect was achieved.Feature extraction methods based on the fast Fourier transform (FFT) and the power spectraldensity (PSD) were studied, and both feature selections methods were described. Yet the threefeature extraction methods were compared in terms of classification accuracy and dataprocessing time.
Based on measured data, the proposed methods have been validated by correspondingclassification experiments.
本文在深入分析综合国内外同类研究的基础上，从地面分类特征提取以及分类器设计这两个基本层次展开对自主轮式机器人地面分类相关理论与技术问题的研究。
本文设计了数据采集的实验，以四轮移动机器人为实验平台，在机器人左前轮轮臂上安装x、 y、 z向加速度计和z向传声器。机器人在沙、碎石、草、土和沥青五种地面上分别以六种速度行驶，提取车轮与地面相互作用的加速度和声压信号作为地面分类的原始数据。
对原始数据进行时域幅值分析，提取地面特征，对于每个传感器数据选用若干个幅值域参数作为地面特征。对于传统的k-近邻（kNN）方法来说，有k值选择问题；用kNN方法解决多类分类问题采用投票法决策，当出现票数相同的情况时，目前尚没有更好的决策策略，尽管随机挑选法是一个实用的策略，但分类精度会降低。针对以上两个问题，本文给出改进的kNN方法，即给出k值的选择方法；对于多种地面分类出现两种以上（包含两种）得票数相同的情况，给出kNN循环寻优的方法。
对于传统的概率神经网络（PNN）方法来说，有平滑因子σ的估计问题，估计得好有利于提高分类精度。以前的学者认为对于所有的样本应选用同一个σ或对于相同维数的样本选用相同的σ，但这种选法不能保证对于所有的测试样本σ都是最优的或较优的，甚至得不到测试结果。针对这一问题，本文给出改进的PNN方法，即给出平滑因子σ的迭代寻优方法。
应用现有的一对一支持向量机（SVM）方法解决多类分类问题采用投票法决策，当出现票数相同的情况时，目前尚没有更好的决策策略。针对这个问题，本文提出改进的一对一SVM方法，即利用LIBSVM中的一对一SVM二值分类程序，对于多种地面分类出现两种以上（包含两种）得票数相同的情况，提出新的算法。从分类准确率和数据处理时间两个方面对改进的kNN、改进的PNN及改进的一对一SVM方法进行比较。
基于时间序列重构的吸引子轨迹矩阵奇异值分解（SVD）的方法原用于故障诊断领域，用于降低原信号中的噪声。本文给出基于奇异值分解（SVD）的特征提取方法，即利用振动信号时间序列重构的吸引子轨迹矩阵奇异值分解的前若干个奇异值作为特征值，取得了好的分类效果。研究基于快速傅里叶变换（FFT）的特征提取方法和基于功率谱密度（PSD）的特征提取方法，阐述二者特征选择的方法。从分类准确率和数据处理时间两个方面对以上三种特征提取方法进行比较。
基于实测数据和相应的分类实验验证了所提方法的效力。
In order to explore in the planets’(such as the moon and Mars) surface and work in thedangerous environment (such as desert, marsh, the scene of the fire, nuclear radiation area, etc)of the earth’s surface, autonomous mobile robots should be able to independently identifyenvironment, complete the mission without a dangerous situation. Terrain identification orterrain classification is an important part of environmental identification. Correspondingcontrol strategy is necessary for robot to travel on different terrain safely and effectively,when the terrain changes, autonomous mobile robot must be able to adapt to the terrain whereit is traversing. Terrain classification can solve the issue of trafficability of autonomousmobile robot in complex terrain. It is very important to improve robot autonomous mobileperformance.
Based on in-depth analysis and synthesis of similar studies home and abroay, the theoryand techniques are researched from the two aspects, i.e. terrain classification featureextraction and classification method.
In this dissertation，experiments for data acquisition are designed. The experimentalplatform is a four-wheeled mobile robot on which arm accelerometers in x, y, zdirections and a microphone in z direction are installed in left front wheel. When the robotis traversing respectively on sand, gravel, grass, soil and asphalt terrain with six differentvelocities, the acceleration and sound pressure signals of wheel-terrain interaction arecollected as the original data.
By time domain amplitude analysis of original data, several parameters of amplitudedomain are selected as the terrain features for each sensor data. To the conventionalk-nearest neighbors (kNN) algorithm, it is necessary to deal with the choice of k, and nowthere is no best decision strategy for the situation when number of votes is the same in theprocess of multi-classification based on voting decisions, though a practical strategy selectedis random method, which is reduces classification accuracy. To the two problems, animproved kNN method was proposed, i.e. the choice method of k was proposed and kNNcycle optimization method was also investigated to deal with the problem that more than twokinds (including two) of terrains have the same number of votes.
To the conventional probabilistic neural network (PNN) method, there is a problem aboutthe estimation of smoothing factor σ which is important to improve the classification accuracy. Previous scholars considered that the same σ was chosen for all samples orsamples of the same dimensions, but it could not make sure that the σ was the best orsub-optimal for all test samples, even there was no result. For the problem, an improved PNNmethod was proposed to deal with the choice of smoothing factor σ by iterativeoptimization method.
The traditional one-against-one support vector machine (SVM) method, now there is nobest decision strategy for the situation when number of votes is the same in the process ofmulti-classification based on voting decisions, an improved one-against-one SVM methodwas rendered to deal with the problem that more than two kinds (including two) of terrainshave the same number of votes based on two-classification program of LIBSVM. Yetimproved kNN，improved PNN and improved one-against-one SVM methods were comparedin terms of classification accuracy and data processing time.
In the field of fault diagnosis, a method based on singular value decomposition of trackmatrix of attractor reconstructed by time series is always used to reduce the noise in originalsignal. Based on singular value decomposition (SVD), a feature extraction method wasproposed using the fore several singular values of track matrix of attractor reconstructed byvibration signals time series as eigenvalues, and better classification effect was achieved.Feature extraction methods based on the fast Fourier transform (FFT) and the power spectraldensity (PSD) were studied, and both feature selections methods were described. Yet the threefeature extraction methods were compared in terms of classification accuracy and dataprocessing time.
Based on measured data, the proposed methods have been validated by correspondingclassification experiments.
引文
[1] Wilcox B H. Non-geometric hazard detection for a Mars microrover. Proceedings ofthe Conference on Intelligent Robotics in Field, Factory, Service, and Space, vol.2.Houston, TX, USA,1994. Washington, DC, USA: NASA:675-684P
[2]张学工．模式识别．3版．北京：清华大学出版社，2010：10-11，120-130,219,75-78，107页
[3]李晶皎．模式识别．北京：电子工业出版社，2010：2-3页
[4]盛立东．模式识别导论．北京：北京邮电大学出版社，2010：8-9页
[5] Mishkin A, Laubach S. From Prime to Extended Mission: Evolution of the MERTactical Uplink Process. Proceedings of SpaceOps2006Conference, Rome, Italy,2006.AIAA:5689P
[6] Talukder A, Manduchi R, Rankin A, et al. Fast and reliable obstacle detection andsegmentation for cross-country navigation. Proceedings of the IEEE Intelligent VehicleSymposium, vol.2. Versailles, France,2002. Piscataway, NJ, USA: IEEE:610-618P
[7] Cowen R. Opportunity Rolls out of Purgatory. Science News.2005,167(26):413P
[8] Brooks C A. Learning to Visually Predict Terrain Properties for Planetary Rovers:
[PhD dissertation]. Massachusetts Institute of Technology.2009:15-16,39-169,209P
[9] Hebert M, Vandapel N. Terrain classification techniques from ladar data forautonomous navigation. Robotics Institute.2003:411-417P
[10] Vandapel N, Huber D F, Kapuria A, et al. Natural terrain classification using3-D ladardata. Proceedings of the IEEE International Conference on Robotics and Automation.New Orleans, USA,2004. Piscataway, NJ, USA: IEEE:5117-5122P
[11] Manduchi R, Castano A, Talukder A, et al. Obstacle detection and terrain classificationfor autonomous off-road navigation. Robotics and Automation.2005,18:81-102P
[12] Lalonde J F, Vandapel N, Huber D F, et al. Natural terrain classification usingthree-dimensional ladar data for ground robot mobility. Journal of Field Robotics.2006,23(10):839–861P
[13] Iagnemma K, Brooks C A, Dubowsky S. Visual, tactile, and vibration-based terrainanalysis for planetary rovers. IEEE Aerospace Conference Proceedings. Big Sky, MT,USA,2004. Piscataway, NJ, USA: IEEE:841-848P
[14] Brooks C A, Iagnemma K, Dubowsky S. Vibration-based terrain analysis for mobilerobots. Proceedings of the IEEE International Conference on Robotics and Automation.Barcelona, Spain,2005. Piscataway, NJ, USA: IEEE:3415-3420P
[15] Brooks C A, Iagnemma K. Vibration-based terrain classification for planetaryexploration rovers. IEEE Transactions on Robotics.2005,21(6):1185-1191P
[16] Brooks C A, Iagnemma K D. Self-Supervised Classification for Planetary RoverTerrain Sensing. IEEE Aerospace Conference Proceedings. Big Sky, MT, USA,2007.Piscataway, NJ, USA: IEEE:1-9P
[17] Brooks C A, Iagnemma K. Self-supervised terrain classification for planetary surfaceexploration rovers. Journal of Field Robotics.2012,29(3):445-468P
[18] Halatci I, Brooks C A, Iagnemma K. Terrain Classification and Classifier Fusion forPlanetary Exploration Rovers. IEEE Aerospace Conference Proceedings. Big Sky, MT,USA,2007. Piscataway, NJ, USA: IEEE:1-11P
[19] Halatci I, Brooks C A, Iagnemma K. A study of visual and tactile terrain classificationand classifier fusion for planetary exploration rovers. Robotica.2008,26(6):767-779P
[20] Ward C C, Iagnemma K. Speed-independent vibration-based terrain classification forpassenger vehicles. Vehicle System Dynamics.2009,47(9):1095-1113P
[21] Ward C C, Iagnemma K. Classification-based wheel slip detection and detector fusionfor outdoor mobile robots. Proceedings of the IEEE International Conference onRobotics and Automation. Rome, Italy,2007. Piscataway, NJ, USA: IEEE:2730-2735P
[22] Iagnemma K, Ward C C. Classification-based wheel slip detection and detector fusionfor mobile robots on outdoor terrain. Autonomous Robots.2009,26:33-46P
[23] Sadhukhan D and Moore C. Online Terrain Estimation Using Internal Sensors.Proceedings of the Florida Conference on Recent Advances in Robotics. Boca Raton,FL, USA,2003:1-3P
[24] Sadhukhan D. Autonomous Ground Vehicle Terrain Classification Using InternalSensors:[Master’s thesis]. Florida State University,2004:1-74P
[25] DuPont E M, Roberts R G, Selekwa M F, et al. Online terrain classification formobile robots. Proceedings of the ASME international mechanical engineeringcongress and exposition conference. Orlando, FL, USA,2005. New York, USA:ASME:1643-1648P
[26] DuPont E M, Moore C A, Collins E G, et al. Frequency response method for terrainclassification in autonomous ground vehicles. Autonomous Robots.2008,24(4):337-347P
[27] DuPont E M, Roberts R G, Moore C A. Speed Independent TerrainClassifcation. Proceedings of the38th Southeastern Symposium on System Theory.Cookeville, TN, USA,2006. Piscataway, NJ, USA: IEEE:240-244P
[28] DuPont E M, Roberts R G, Moore C A. The Identification of Terrains forMobile Robots Using Eigenspace and Neural Network Methods. Proceedings ofthe Florida Conference on Recent Advances in Robotics. Miami, FL, USA,2006:1-5P
[29] DuPont E M, Moore C A, Roberts R G. Terrain Classification for MobileRobots Traveling at Various Speeds: An Eigenspace Manifold Approach.Proceedings of the IEEE International Conference on Robotics and Automation.Pasadena, CA, USA,2008. Piscataway, NJ, USA: IEEE:3284-3289P
[30] Coyle E, Collins E G, Jr. A Comparison of Classifier Performance forVibration-based terrain Classification. Proceedings of the26thArmy ScienceConference. Orlando, Florida, USA,2008:1-7P
[31] Collins E G, Jr, Coyle E. Vibration-Based Terrain Classification Using SurfaceProfile Input Frequency Responses. Proceedings of the IEEE International Conferenceon Robotics and Automation. Pasadena, CA, USA,2008. Piscataway, NJ, USA: IEEE:3276-3283P
[32] Weiss C, Fr hlich H, Zell A. Vibration-based terrain classification using support vectormachines. Proceedings of the IEEE International Conference on Intelligent Robots andSystems. Beijing, China,2006. Piscataway, NJ, USA: IEEE:4429-4434P
[33] Weiss C, Fechner N, Stark M, et al. Comparison of different approaches tovibration-based terrain classification. Proceedings of the European Conferenace onMobile Robotics. Freiburg, Germany,2007:7-12P
[34] Weiss C, Stark M, Zell A. SVMs for vibration-based terrain classification. Proceedingsof the Autonome Mobile Systeme. Kaiserslautern, Germany,2007. Heidelberg,Germany: Springer:1-7P
[35] Weiss C, Zell A. Novelty detection and online learning for vibration-based terrainclassification. Proceedings of the10th International Conference on IntelligentAutonomous Systems. Baden-Baden, Germany,2008:16–25P
[36] Weiss C, Tamimi H, Zell A. A Combination of Vision-and Vibration-based TerrainClassification. Proceedings of the IEEE International Conference on Intelligent Robotsand Systems. Nice, France. Nice, France,2008. Piscataway, NJ, USA: IEEE:2204-2209P
[37] Komma P, Weiss C, Zell A. Adaptive Bayesian Filtering for Vibration-based TerrainClassification. Proceedings of the IEEE International Conference on Robotics andAutomation. Kobe, Japan,2009. Piscataway, NJ, USA: IEEE:3307-3313P
[38] Ojeda L, Borenstein J, Witus G, et al. Terrain characterization and classification with amobile robot. Journal of Field Robotics.2006,23(2):103-122P
[39] Libby J, Stentz A J. Using sound to classify vehicle-terrain interactions in outdoorenvironments. Proceedings of the IEEE International Conference on Robotics andAutomation. Saint Paul, Minnesota, USA,2012. Piscataway, NJ, USA: IEEE:3559-3566P
[40] Jitpakdee R, Maneewarn T. Neural networks terrain classification using inertialmeasurement unit for an autonomous vehicle. Proceedings of the SICE AnnualConference2008-International Conference on Instrumentation, Control andInformation Technology. Tokyo, Japan,2008. SICE:554-558P
[41] Giguere P, Dudek G. Surface identification using simple contact dynamics for mobilerobots. Proceedings of the IEEE International Conference on Robotics and Automation.Kobe, Japan,2009. Piscataway, NJ, USA: IEEE:3301-3306P
[42] Giguere P, Dudek G. Clustering Sensor Data for Terrain Identification using aWindowless Algorithm. Robotics: Science and Systems IV.2009:25-32P
[43] Mou W, Kleiner A. Online learning terrain classification for adaptive velocity control.Proceedings of the8th IEEE International Workshop on Safety, Security, and RescueRobotics. Bremen, Germany,2010. Piscataway, NJ, USA: IEEE:1-7P
[44] Tick D, Rahman T, Busso C, et al. Indoor robotic terrain classification via angularvelocity based hierarchical classifier selection. Proceedings of the IEEE InternationalConference on Robotics and Automation. Saint Paul, Minnesota, USA,2012.Piscataway, NJ, USA: IEEE:3594-3600P
[45] Park B, Kim J, Lee J. Terrain Feature Extraction and Classification for Mobile RobotsUtilizing Contact Sensors on Rough Terrain. Procedia Engineering.2012,41:846-853P
[46]徐正飞，杨汝清，翁新华．移动机器人四杆地形感知机构的设计．机械工程学报．2003，39(4)：44-48页
[47]乔凤斌，谢霄鹏，杨汝清．六轮移动机器人包容地形研究．机械设计与研究．2004，20(5)：17-19+6页
[48]许宏岩，付宜利，王树国．局部地形变化检测与移动机器人的行为决策．控制与决策．2005，20(8)：951-954页
[49]朱江，王耀南，余洪山，等．未知环境下移动机器人自主感知斜坡地形方法．仪器仪表学报．2010，31(8)：1916-1920页
[50]赵小川，刘培志，张敏，等．一种适用于移动机器人的障碍物快速检测算法及其实现．机器人．2011，33(2)：198-201+214页
[51]郭晏，包加桐，宋爱国，等．基于地形预测与修正的搜救机器人可通过度．机器人．2009，31(5)：445-452页
[52]郭晏，宋爱国，包加桐，等．基于差分进化支持向量机的移动机器人可通过度预测．20011，33(3)：257-264+272页
[53]韩光，赵春霞．融合多可视化特征的可通行性地形分类．华中科技大学学报（自然科学版）．2008，36(增刊Ⅰ)：105-108页
[54]高华，赵春霞，韩光．基于one-class SVM与融合多可视化特征的可通行区域检测．机器人．2011，33(6)：731-735+741页
[55]钱堃，马旭东，戴先中，等．基于层次化SLAM的未知环境级联地图创建方法．机器人．2011，33(6)：736-741页
[56]黄微，张良培，李平湘．基于地形区域分割的复杂地区遥感影像分类．武汉大学学报（信息科学版）．2007，32(9)：791-795页
[57]周访滨，刘学军．基于DTA山体部位分类决策方案的改进与微观地形自动分类研究．西北农业学报．2008，17(3)：343-346页
[58]潘蔚，倪国强，李瀚波．基于遥感图像地形结构-岩性组分分解的岩类多重分形特征研究．地学前缘．2009，16(6)：248-256页
[59]赵连春，刘荣堂，杨予海，等．基于地形因子的草地遥感分类方法的研究．草业科学．2006，23(12)：26-30页
[60]夏伟，刘雁春，黄谟涛，等．基于正交小波变换的海底地形复杂程度分类方法研究．武汉大学学报（信息科学版）．2008，33(6)：631-634页
[61]于吉涛，陈子燊．砂质海滩地形动力分类研究进展．热带地理．2011，31(1)：107-112页
[62] Gora G and Wojna A. A classifier combining rule induction and k-NN method withautomated selection of optimal neighborhood. Proceedings of the Thirteenth EuropeanConference on Machine Learning. Helsinki, Finland,2002. Heidelberg, Germany:Springer:111-123P
[63] Pan J S, Qiao Y L, Sun S H. A fast k nearest neighbors classification algorithm. IEICETrans Fundamentals.2004,87(4):961-963P
[64]乔玉龙，赵春晖，潘正祥．基于Haar小波变换的快速k-近邻分类算法．吉林大学学报（工学版）．2011，41(1)：231-234页
[65] Ferri F, Vidal E. Colour image segmentation and labeling through multiedit-condensing.Pattern Recognition Letters,1992,13(8):561-568P
[66] Segata N, Blanzieri E, Delany S J, et al. Noise reduction for instance-based learningwith a local maximal margin approach. Journal of Intelligent Information Systems.2010,35(2):301-331P
[67] Fayed H A, Atiya A F. A novel template reduction approach for the k-nearest neighbormethod. IEEE Transactions on Neural Networks.2009,20(5):890-896P
[68] Paredes R, Vidal E. Learning prototypes and distances: a prototype reduction techniquebased on nearest neighbor error minimization. Pattern Recognition.2006,39(2):171-179P
[69] Hart P E. The condensed nearest neighbor rule. IEEE Transactions on InformationTheory.1968, IT-14(3):515-516P
[70]周进登，王晓丹．基于最小k近邻错分率编码确定方法及其在多类分类中的应用．控制与决策．2011，26(9)：1295-1302页
[71]蔡曲林，刘普寅．一种新的概率神经网络有监督学习算法．模糊系统与数学．2006,20(6)：83-87页
[72] Specht D F. Enhancement to Probabilistic NeuralNetwork. Proceeding of the IEEEinternational Joint Conference on NeuralNetworks, vol.1. Baltimore, MD, USA,1992.Piscataway, NJ, USA: IEEE:761-768P
[73] Zhong M, Goggeshall D, et al. Gap-Based Estimation: Choosing the SmoothingParameters for Probabilistic and General Regression NeuralNetworks. Proceeding ofthe IEEE Word Congress on Computational Intelligence. Vancouver, BC, Canada,2006.Piscataway, NJ, USA: IEEE:1870-1877P
[74]吴子燕，杨海峰，覃小文，等．基于自适应概率神经网络的损伤模式识别研究．振动与冲击．2008(7)：8-12+182-183页
[75]李春芳，刘连忠，陆震．基于数据场的概率神经网络算法．电子学报．2011，39(8)：1739-1745页
[76]淦文燕,李德毅,王建民．一种基于数据场的层次聚类方法．电子学报．2006，34(2)：258-262页
[77]李德毅，杜鹢．不确定性人工智能．北京：国防工业出版社，2005：207-217页
[78]淦文燕，赫南，李德毅，等．一种基于拓扑势的网络社区发现方法．软件学报．2009，20(8)：2241-2254页
[79] Weston J, Watkins C. Modeling multi-class support vector machines[R]. London:University of London,1998:1-10P
[80] Hsu C W, Lin C J. A comparison of methods for multiclass support vector machine.IEEE Trans on Neural Networks.2002,13(2):415-425P
[81] Platt J C, Cristianini N, Shawe-Taylor J. Large margin DAGs for multiclassclassification. Advances in Neural Information Processing System.2000,12(3):547-553P
[82]连可．基于状态监测的复杂电子系统故障诊断方法研究．电子科技大学博士学位论文．2009：73-75页
[83]唐发明．基于统计学习理论的支持向量机算法研究．华中科技大学博士学位论文．2005：75-77页
[84]谭大伟．五轮可重构移动机器人构型设计及研究．哈尔滨工程大学硕士学位论文．2010
[85]张振宇.崎岖地面移动机器人运动与估计关键技术研究．哈尔滨工程大学硕士学位论文．2011
[86] Xu H, Zhang Z Y, Alipour K, et al. Prototypes selection by multi-objective optimaldesign: application to a reconfigurable robot in sandy terrain. Industrial Robot.2011,38(6):599-613P
[87] Cover T M, Hart P E. Nearest neighbor Pattern Classification. IEEE Transaction onInformation Theory.1967, IT-13(l):21-27P
[88] Nasibov E, Kandemir-Cavas C. Efficiency analysis of KNN and minimumdistance-based classifiers in enzyme family prediction. Computational Biology andChemistry.2009,33(6):461-464P
[89] Zhang R, Jagadish H V, Dai B T, et al. Optimized algorithms for predictive range andKNN queries on moving objects. Information Systems.2010,35(8):911-932P
[90] Yao B, Li F F, Kumar P. K nearest neighbor queries and kNN-joins in large relationaldatabases (almost) for free. Proceeding of the IEEE26th International Conference onData Engineering. Long Beach, CA, USA,2010. Piscataway, NJ, USA: IEEE:4-15P
[91] Toyama J, Kudo M, Imai H. Probably correct k-nearest neighbor search in highdimensions. Pattern Recognition.2010,43(4):1361-1372P
[92]陈凤，杜兰，保铮．一种优化k近邻准则及在雷达HRRP目标识别中的应用．西安电子科技大学学报．2007，34(5)：681-686页
[93]李舜酩，李香莲．振动信号的现代分析技术与应用．北京：国防工业出版社，2008：2-5页
[94]孙亮，禹晶．模式识别原理．北京：北京工业大学出版社，2009：78-80页
[95]杨金福，宋敏，李明爱．一种新的基于距离加权的模板约简K近邻算法．电子与信息学报．2011，33(10)：2378-2383页
[96] Specht D F. Probabilistic Neural Networks for Classification, Mapping or AssociativeMemory. Proceeding of the IEEE International Conference on Neural Networks. SanDiego, USA,1988. Piscataway, NJ, USA: IEEE:525-532P
[97] Specht D F. Probabilistic neural networks. Neural Networks.1990,3(1):109-118P
[98]唐明珠，阳春华，桂卫华，等．代价敏感概率神经网络及其在故障诊断中的应用．控制与决策．2010，25(7)：1074-1078页
[99]李冬辉，刘浩．基于概率神经网络的故障诊断方法及应用．系统工程与电子技术．2004，26(7)：997-999页
[100]高锦红，祝保林，王秋亚，等．概率神经网络及FAAS在植物药分类研究中的应用．光谱实验室,2011，28(1)：128-131页
[101]李庆波，李响，张广军，等．概率神经网络在胃镜样品红外光谱检测中的应用．光谱学与光谱分析．2009，29(6)：1553-1557页
[102]吴婷，颜国正，杨帮华，等．基于有监督学习的概率神经网络的脑电信号分类方法．上海交通大学学报．2008，42(5)：803-806页
[103]黄林，贺鹏，王经民．基于概率神经网络和分形的植物叶片机器识别研究．西北农林科技大学学报（自然科学版）．2008，36(9)：212-218页
[104] Parzen E. On estimation of a probability density function and mode. Annals ofMathematical Statistics.1962,33(3):1065-1076P
[105] Vapnik V N. The nature of statistical learning theory. New York, USA: Springer,1995
[106]丁世飞，齐丙娟，谭红艳．支持向量机理论与算法研究综述．电子科技大学学报．2011，40(1)：2-10页
[107]顾亚祥，丁世飞．支持向量机研究进展．计算机科学．2011，38(2)：14-17页
[108]刘晓亮，丁世飞．SVM用于文本分类的适用性．计算机工程与科学．2010,32(6)：106-108页
[109]谢塞琴，沈福明，邱雪娜．基于支持向量机的人脸识别方法．计算机工程．2009,35(16)：186-188页
[110]林吉良，蒋静坪．基于支持向量机的移动机器人故障诊断．电工技术学报．2008，23(11)：173-177+182页
[111]李颖新，阮晓钢．基于支持向量机的肿瘤分类特征基因选取．计算机研究与发展．2005，42(10)：1796-1801页
[112]温熙森．模式识别与状态监控．北京：科学出版社，2007：327页
[113] Cristianini N, Shawe-Taylor J．支持向量机导论．李国正，王猛，曾华军，译．北京：电子工业出版社，2004：25-26页
[114] Chang C C, Lin C J. LIBSVM: a library for support vector machines[OL].(2001)
[2011-11-14]. http://www.csie.ntu.edu.tw/~cjlin/libsvm.
[115] Boser B E, Guyon I M, Vapnik V N. A training algorithm for optimal margin classifiers.Proceedings of the Fifth Annual Workshop on Computational Learning Theory.Pittsburgh, PA, USA,1992. New York: ACM Press:144-152P
[116] Osuna E, Frenud R, Girosi F. An improved training algorithm for support vectormachines. Proceedings of the IEEE Workshop on Neural Networks for SignalProcessing. New York, USA,1997. Piscataway, NJ, USA: IEEE:276-285P
[117] Syed N, Liu H, Sung K. Incremental learning with support vector machines.Proceedings of the International Joint Conference on Artificial Intelligence. Sweden,1999. Morgan Kaufmann publishers:352-356P
[118] Tang Y C, Jin B, Zhang Y Q, et al. Granular support vector machines for medicalbinary classification problems. Proceedings of the IEEE Symposium on ComputationalIntelligence in Bioinformatics and Computational Biology. La Jolla, CA, USA,2004.Piscataway, NJ, USA: IEEE:73-78P
[119] Lin C F, Wang S D. Fuzzy support vector machines. IEEE Transactions on NeuralNetworks.2002,3(2):464-471P
[120] Jayadcva R, Khemchandani S C. Twin support vector machines for patternclassification. IEEE Trans on Pattern Analysis and Machine Intelligence.2007,29(5):905-910P
[121] Herbrich R, Graepel T, Obermayer K. Large margin rank boundaries for ordinalregression. Advances in Large Margin Classifiers.2000,7:115-132P
[122]连可，陈世杰，周建明，等．基于遗传算法的SVM多分类决策树优化算法研究．控制与决策．2009，24(1)：7-12页
[123] Bottou L, Cortes C, Denker J, et al. Comparison of classifier: a case study inhandwritten digit recognition. Proceedings of the International Conference on PatternRecognition. Los Alamitos, CA,1994. IEEE Computer Society Press:77-82P
[124] Knerr S, Personnaz L, Dreyfus G, et al. Single-layer learning revisited: a stepwiseprocedure for building and training a neural network. Optimization Methods andSoftware.1990,1:23-34P
[125]段江涛，李凌均，张周锁，等．基于支持向量机的机械系统多故障分类方法．农业机械学报．2004，35(4)：144-147页
[126]吕志民，张武军，徐金梧，等．基于奇异谱的降噪方法及其在故障诊断技术中的应用．机械工程学报．1999,35(3)：85-88页
[127]刘献栋，杨绍普，申永军，等．基于奇异值分解的突变信息检测新方法及其应用．机械工程学报．2002,38(6)：102-105页
[128]何田，刘献栋，李其汉．噪声背景下检测突变信息的奇异值分解技术．振动工程学报．2006,19(3)：399-403页
[129]栾方军．在线手写签名认证算法的研究．吉林大学博士学位论文．2007：31-33页
[130]张德丰.MATLAB数字信号处理与应用.北京：清华大学出版社，2010：109-112，264-270页
[131] Cooley J W, Tukey J W. An algorithm for the machine calculation of complex Fourierseries. Mathematics of Computation.1965,19(90):297-301P
[132]沈志远，王黎明，陈方林．基于有限长序列分析的Welch法谱估计研究．计算机仿真．2010，27(12)：391-395页
[133]姚天任，江太辉.数字信号处理.3版.武汉：华中科技大学出版社，2007：369-370页
[134]王世一.数字信号处理.修订版.北京：北京理工大学出版社，2012：106-114页
[135]张思.振动测试与分析技术.北京：清华大学出版社，1992：328-335页
[136] Welch P D. The Use of Fast Fourier Transform for the Estimation of Power Spectra: AMethod Based on Time Averaging Over Short. IEEE Transactions on Audio andElectroacoustics.1967, AU-15(2):70-73P
[137]张峰，石现峰，张学智．Welch功率谱估计算法仿真及分析．西安工业大学学报．2009，29(4)：353-356页
[2]张学工．模式识别．3版．北京：清华大学出版社，2010：10-11，120-130,219,75-78，107页
[3]李晶皎．模式识别．北京：电子工业出版社，2010：2-3页
[4]盛立东．模式识别导论．北京：北京邮电大学出版社，2010：8-9页
[5] Mishkin A, Laubach S. From Prime to Extended Mission: Evolution of the MERTactical Uplink Process. Proceedings of SpaceOps2006Conference, Rome, Italy,2006.AIAA:5689P
[6] Talukder A, Manduchi R, Rankin A, et al. Fast and reliable obstacle detection andsegmentation for cross-country navigation. Proceedings of the IEEE Intelligent VehicleSymposium, vol.2. Versailles, France,2002. Piscataway, NJ, USA: IEEE:610-618P
[7] Cowen R. Opportunity Rolls out of Purgatory. Science News.2005,167(26):413P
[8] Brooks C A. Learning to Visually Predict Terrain Properties for Planetary Rovers:
[PhD dissertation]. Massachusetts Institute of Technology.2009:15-16,39-169,209P
[9] Hebert M, Vandapel N. Terrain classification techniques from ladar data forautonomous navigation. Robotics Institute.2003:411-417P
[10] Vandapel N, Huber D F, Kapuria A, et al. Natural terrain classification using3-D ladardata. Proceedings of the IEEE International Conference on Robotics and Automation.New Orleans, USA,2004. Piscataway, NJ, USA: IEEE:5117-5122P
[11] Manduchi R, Castano A, Talukder A, et al. Obstacle detection and terrain classificationfor autonomous off-road navigation. Robotics and Automation.2005,18:81-102P
[12] Lalonde J F, Vandapel N, Huber D F, et al. Natural terrain classification usingthree-dimensional ladar data for ground robot mobility. Journal of Field Robotics.2006,23(10):839–861P
[13] Iagnemma K, Brooks C A, Dubowsky S. Visual, tactile, and vibration-based terrainanalysis for planetary rovers. IEEE Aerospace Conference Proceedings. Big Sky, MT,USA,2004. Piscataway, NJ, USA: IEEE:841-848P
[14] Brooks C A, Iagnemma K, Dubowsky S. Vibration-based terrain analysis for mobilerobots. Proceedings of the IEEE International Conference on Robotics and Automation.Barcelona, Spain,2005. Piscataway, NJ, USA: IEEE:3415-3420P
[15] Brooks C A, Iagnemma K. Vibration-based terrain classification for planetaryexploration rovers. IEEE Transactions on Robotics.2005,21(6):1185-1191P
[16] Brooks C A, Iagnemma K D. Self-Supervised Classification for Planetary RoverTerrain Sensing. IEEE Aerospace Conference Proceedings. Big Sky, MT, USA,2007.Piscataway, NJ, USA: IEEE:1-9P
[17] Brooks C A, Iagnemma K. Self-supervised terrain classification for planetary surfaceexploration rovers. Journal of Field Robotics.2012,29(3):445-468P
[18] Halatci I, Brooks C A, Iagnemma K. Terrain Classification and Classifier Fusion forPlanetary Exploration Rovers. IEEE Aerospace Conference Proceedings. Big Sky, MT,USA,2007. Piscataway, NJ, USA: IEEE:1-11P
[19] Halatci I, Brooks C A, Iagnemma K. A study of visual and tactile terrain classificationand classifier fusion for planetary exploration rovers. Robotica.2008,26(6):767-779P
[20] Ward C C, Iagnemma K. Speed-independent vibration-based terrain classification forpassenger vehicles. Vehicle System Dynamics.2009,47(9):1095-1113P
[21] Ward C C, Iagnemma K. Classification-based wheel slip detection and detector fusionfor outdoor mobile robots. Proceedings of the IEEE International Conference onRobotics and Automation. Rome, Italy,2007. Piscataway, NJ, USA: IEEE:2730-2735P
[22] Iagnemma K, Ward C C. Classification-based wheel slip detection and detector fusionfor mobile robots on outdoor terrain. Autonomous Robots.2009,26:33-46P
[23] Sadhukhan D and Moore C. Online Terrain Estimation Using Internal Sensors.Proceedings of the Florida Conference on Recent Advances in Robotics. Boca Raton,FL, USA,2003:1-3P
[24] Sadhukhan D. Autonomous Ground Vehicle Terrain Classification Using InternalSensors:[Master’s thesis]. Florida State University,2004:1-74P
[25] DuPont E M, Roberts R G, Selekwa M F, et al. Online terrain classification formobile robots. Proceedings of the ASME international mechanical engineeringcongress and exposition conference. Orlando, FL, USA,2005. New York, USA:ASME:1643-1648P
[26] DuPont E M, Moore C A, Collins E G, et al. Frequency response method for terrainclassification in autonomous ground vehicles. Autonomous Robots.2008,24(4):337-347P
[27] DuPont E M, Roberts R G, Moore C A. Speed Independent TerrainClassifcation. Proceedings of the38th Southeastern Symposium on System Theory.Cookeville, TN, USA,2006. Piscataway, NJ, USA: IEEE:240-244P
[28] DuPont E M, Roberts R G, Moore C A. The Identification of Terrains forMobile Robots Using Eigenspace and Neural Network Methods. Proceedings ofthe Florida Conference on Recent Advances in Robotics. Miami, FL, USA,2006:1-5P
[29] DuPont E M, Moore C A, Roberts R G. Terrain Classification for MobileRobots Traveling at Various Speeds: An Eigenspace Manifold Approach.Proceedings of the IEEE International Conference on Robotics and Automation.Pasadena, CA, USA,2008. Piscataway, NJ, USA: IEEE:3284-3289P
[30] Coyle E, Collins E G, Jr. A Comparison of Classifier Performance forVibration-based terrain Classification. Proceedings of the26thArmy ScienceConference. Orlando, Florida, USA,2008:1-7P
[31] Collins E G, Jr, Coyle E. Vibration-Based Terrain Classification Using SurfaceProfile Input Frequency Responses. Proceedings of the IEEE International Conferenceon Robotics and Automation. Pasadena, CA, USA,2008. Piscataway, NJ, USA: IEEE:3276-3283P
[32] Weiss C, Fr hlich H, Zell A. Vibration-based terrain classification using support vectormachines. Proceedings of the IEEE International Conference on Intelligent Robots andSystems. Beijing, China,2006. Piscataway, NJ, USA: IEEE:4429-4434P
[33] Weiss C, Fechner N, Stark M, et al. Comparison of different approaches tovibration-based terrain classification. Proceedings of the European Conferenace onMobile Robotics. Freiburg, Germany,2007:7-12P
[34] Weiss C, Stark M, Zell A. SVMs for vibration-based terrain classification. Proceedingsof the Autonome Mobile Systeme. Kaiserslautern, Germany,2007. Heidelberg,Germany: Springer:1-7P
[35] Weiss C, Zell A. Novelty detection and online learning for vibration-based terrainclassification. Proceedings of the10th International Conference on IntelligentAutonomous Systems. Baden-Baden, Germany,2008:16–25P
[36] Weiss C, Tamimi H, Zell A. A Combination of Vision-and Vibration-based TerrainClassification. Proceedings of the IEEE International Conference on Intelligent Robotsand Systems. Nice, France. Nice, France,2008. Piscataway, NJ, USA: IEEE:2204-2209P
[37] Komma P, Weiss C, Zell A. Adaptive Bayesian Filtering for Vibration-based TerrainClassification. Proceedings of the IEEE International Conference on Robotics andAutomation. Kobe, Japan,2009. Piscataway, NJ, USA: IEEE:3307-3313P
[38] Ojeda L, Borenstein J, Witus G, et al. Terrain characterization and classification with amobile robot. Journal of Field Robotics.2006,23(2):103-122P
[39] Libby J, Stentz A J. Using sound to classify vehicle-terrain interactions in outdoorenvironments. Proceedings of the IEEE International Conference on Robotics andAutomation. Saint Paul, Minnesota, USA,2012. Piscataway, NJ, USA: IEEE:3559-3566P
[40] Jitpakdee R, Maneewarn T. Neural networks terrain classification using inertialmeasurement unit for an autonomous vehicle. Proceedings of the SICE AnnualConference2008-International Conference on Instrumentation, Control andInformation Technology. Tokyo, Japan,2008. SICE:554-558P
[41] Giguere P, Dudek G. Surface identification using simple contact dynamics for mobilerobots. Proceedings of the IEEE International Conference on Robotics and Automation.Kobe, Japan,2009. Piscataway, NJ, USA: IEEE:3301-3306P
[42] Giguere P, Dudek G. Clustering Sensor Data for Terrain Identification using aWindowless Algorithm. Robotics: Science and Systems IV.2009:25-32P
[43] Mou W, Kleiner A. Online learning terrain classification for adaptive velocity control.Proceedings of the8th IEEE International Workshop on Safety, Security, and RescueRobotics. Bremen, Germany,2010. Piscataway, NJ, USA: IEEE:1-7P
[44] Tick D, Rahman T, Busso C, et al. Indoor robotic terrain classification via angularvelocity based hierarchical classifier selection. Proceedings of the IEEE InternationalConference on Robotics and Automation. Saint Paul, Minnesota, USA,2012.Piscataway, NJ, USA: IEEE:3594-3600P
[45] Park B, Kim J, Lee J. Terrain Feature Extraction and Classification for Mobile RobotsUtilizing Contact Sensors on Rough Terrain. Procedia Engineering.2012,41:846-853P
[46]徐正飞，杨汝清，翁新华．移动机器人四杆地形感知机构的设计．机械工程学报．2003，39(4)：44-48页
[47]乔凤斌，谢霄鹏，杨汝清．六轮移动机器人包容地形研究．机械设计与研究．2004，20(5)：17-19+6页
[48]许宏岩，付宜利，王树国．局部地形变化检测与移动机器人的行为决策．控制与决策．2005，20(8)：951-954页
[49]朱江，王耀南，余洪山，等．未知环境下移动机器人自主感知斜坡地形方法．仪器仪表学报．2010，31(8)：1916-1920页
[50]赵小川，刘培志，张敏，等．一种适用于移动机器人的障碍物快速检测算法及其实现．机器人．2011，33(2)：198-201+214页
[51]郭晏，包加桐，宋爱国，等．基于地形预测与修正的搜救机器人可通过度．机器人．2009，31(5)：445-452页
[52]郭晏，宋爱国，包加桐，等．基于差分进化支持向量机的移动机器人可通过度预测．20011，33(3)：257-264+272页
[53]韩光，赵春霞．融合多可视化特征的可通行性地形分类．华中科技大学学报（自然科学版）．2008，36(增刊Ⅰ)：105-108页
[54]高华，赵春霞，韩光．基于one-class SVM与融合多可视化特征的可通行区域检测．机器人．2011，33(6)：731-735+741页
[55]钱堃，马旭东，戴先中，等．基于层次化SLAM的未知环境级联地图创建方法．机器人．2011，33(6)：736-741页
[56]黄微，张良培，李平湘．基于地形区域分割的复杂地区遥感影像分类．武汉大学学报（信息科学版）．2007，32(9)：791-795页
[57]周访滨，刘学军．基于DTA山体部位分类决策方案的改进与微观地形自动分类研究．西北农业学报．2008，17(3)：343-346页
[58]潘蔚，倪国强，李瀚波．基于遥感图像地形结构-岩性组分分解的岩类多重分形特征研究．地学前缘．2009，16(6)：248-256页
[59]赵连春，刘荣堂，杨予海，等．基于地形因子的草地遥感分类方法的研究．草业科学．2006，23(12)：26-30页
[60]夏伟，刘雁春，黄谟涛，等．基于正交小波变换的海底地形复杂程度分类方法研究．武汉大学学报（信息科学版）．2008，33(6)：631-634页
[61]于吉涛，陈子燊．砂质海滩地形动力分类研究进展．热带地理．2011，31(1)：107-112页
[62] Gora G and Wojna A. A classifier combining rule induction and k-NN method withautomated selection of optimal neighborhood. Proceedings of the Thirteenth EuropeanConference on Machine Learning. Helsinki, Finland,2002. Heidelberg, Germany:Springer:111-123P
[63] Pan J S, Qiao Y L, Sun S H. A fast k nearest neighbors classification algorithm. IEICETrans Fundamentals.2004,87(4):961-963P
[64]乔玉龙，赵春晖，潘正祥．基于Haar小波变换的快速k-近邻分类算法．吉林大学学报（工学版）．2011，41(1)：231-234页
[65] Ferri F, Vidal E. Colour image segmentation and labeling through multiedit-condensing.Pattern Recognition Letters,1992,13(8):561-568P
[66] Segata N, Blanzieri E, Delany S J, et al. Noise reduction for instance-based learningwith a local maximal margin approach. Journal of Intelligent Information Systems.2010,35(2):301-331P
[67] Fayed H A, Atiya A F. A novel template reduction approach for the k-nearest neighbormethod. IEEE Transactions on Neural Networks.2009,20(5):890-896P
[68] Paredes R, Vidal E. Learning prototypes and distances: a prototype reduction techniquebased on nearest neighbor error minimization. Pattern Recognition.2006,39(2):171-179P
[69] Hart P E. The condensed nearest neighbor rule. IEEE Transactions on InformationTheory.1968, IT-14(3):515-516P
[70]周进登，王晓丹．基于最小k近邻错分率编码确定方法及其在多类分类中的应用．控制与决策．2011，26(9)：1295-1302页
[71]蔡曲林，刘普寅．一种新的概率神经网络有监督学习算法．模糊系统与数学．2006,20(6)：83-87页
[72] Specht D F. Enhancement to Probabilistic NeuralNetwork. Proceeding of the IEEEinternational Joint Conference on NeuralNetworks, vol.1. Baltimore, MD, USA,1992.Piscataway, NJ, USA: IEEE:761-768P
[73] Zhong M, Goggeshall D, et al. Gap-Based Estimation: Choosing the SmoothingParameters for Probabilistic and General Regression NeuralNetworks. Proceeding ofthe IEEE Word Congress on Computational Intelligence. Vancouver, BC, Canada,2006.Piscataway, NJ, USA: IEEE:1870-1877P
[74]吴子燕，杨海峰，覃小文，等．基于自适应概率神经网络的损伤模式识别研究．振动与冲击．2008(7)：8-12+182-183页
[75]李春芳，刘连忠，陆震．基于数据场的概率神经网络算法．电子学报．2011，39(8)：1739-1745页
[76]淦文燕,李德毅,王建民．一种基于数据场的层次聚类方法．电子学报．2006，34(2)：258-262页
[77]李德毅，杜鹢．不确定性人工智能．北京：国防工业出版社，2005：207-217页
[78]淦文燕，赫南，李德毅，等．一种基于拓扑势的网络社区发现方法．软件学报．2009，20(8)：2241-2254页
[79] Weston J, Watkins C. Modeling multi-class support vector machines[R]. London:University of London,1998:1-10P
[80] Hsu C W, Lin C J. A comparison of methods for multiclass support vector machine.IEEE Trans on Neural Networks.2002,13(2):415-425P
[81] Platt J C, Cristianini N, Shawe-Taylor J. Large margin DAGs for multiclassclassification. Advances in Neural Information Processing System.2000,12(3):547-553P
[82]连可．基于状态监测的复杂电子系统故障诊断方法研究．电子科技大学博士学位论文．2009：73-75页
[83]唐发明．基于统计学习理论的支持向量机算法研究．华中科技大学博士学位论文．2005：75-77页
[84]谭大伟．五轮可重构移动机器人构型设计及研究．哈尔滨工程大学硕士学位论文．2010
[85]张振宇.崎岖地面移动机器人运动与估计关键技术研究．哈尔滨工程大学硕士学位论文．2011
[86] Xu H, Zhang Z Y, Alipour K, et al. Prototypes selection by multi-objective optimaldesign: application to a reconfigurable robot in sandy terrain. Industrial Robot.2011,38(6):599-613P
[87] Cover T M, Hart P E. Nearest neighbor Pattern Classification. IEEE Transaction onInformation Theory.1967, IT-13(l):21-27P
[88] Nasibov E, Kandemir-Cavas C. Efficiency analysis of KNN and minimumdistance-based classifiers in enzyme family prediction. Computational Biology andChemistry.2009,33(6):461-464P
[89] Zhang R, Jagadish H V, Dai B T, et al. Optimized algorithms for predictive range andKNN queries on moving objects. Information Systems.2010,35(8):911-932P
[90] Yao B, Li F F, Kumar P. K nearest neighbor queries and kNN-joins in large relationaldatabases (almost) for free. Proceeding of the IEEE26th International Conference onData Engineering. Long Beach, CA, USA,2010. Piscataway, NJ, USA: IEEE:4-15P
[91] Toyama J, Kudo M, Imai H. Probably correct k-nearest neighbor search in highdimensions. Pattern Recognition.2010,43(4):1361-1372P
[92]陈凤，杜兰，保铮．一种优化k近邻准则及在雷达HRRP目标识别中的应用．西安电子科技大学学报．2007，34(5)：681-686页
[93]李舜酩，李香莲．振动信号的现代分析技术与应用．北京：国防工业出版社，2008：2-5页
[94]孙亮，禹晶．模式识别原理．北京：北京工业大学出版社，2009：78-80页
[95]杨金福，宋敏，李明爱．一种新的基于距离加权的模板约简K近邻算法．电子与信息学报．2011，33(10)：2378-2383页
[96] Specht D F. Probabilistic Neural Networks for Classification, Mapping or AssociativeMemory. Proceeding of the IEEE International Conference on Neural Networks. SanDiego, USA,1988. Piscataway, NJ, USA: IEEE:525-532P
[97] Specht D F. Probabilistic neural networks. Neural Networks.1990,3(1):109-118P
[98]唐明珠，阳春华，桂卫华，等．代价敏感概率神经网络及其在故障诊断中的应用．控制与决策．2010，25(7)：1074-1078页
[99]李冬辉，刘浩．基于概率神经网络的故障诊断方法及应用．系统工程与电子技术．2004，26(7)：997-999页
[100]高锦红，祝保林，王秋亚，等．概率神经网络及FAAS在植物药分类研究中的应用．光谱实验室,2011，28(1)：128-131页
[101]李庆波，李响，张广军，等．概率神经网络在胃镜样品红外光谱检测中的应用．光谱学与光谱分析．2009，29(6)：1553-1557页
[102]吴婷，颜国正，杨帮华，等．基于有监督学习的概率神经网络的脑电信号分类方法．上海交通大学学报．2008，42(5)：803-806页
[103]黄林，贺鹏，王经民．基于概率神经网络和分形的植物叶片机器识别研究．西北农林科技大学学报（自然科学版）．2008，36(9)：212-218页
[104] Parzen E. On estimation of a probability density function and mode. Annals ofMathematical Statistics.1962,33(3):1065-1076P
[105] Vapnik V N. The nature of statistical learning theory. New York, USA: Springer,1995
[106]丁世飞，齐丙娟，谭红艳．支持向量机理论与算法研究综述．电子科技大学学报．2011，40(1)：2-10页
[107]顾亚祥，丁世飞．支持向量机研究进展．计算机科学．2011，38(2)：14-17页
[108]刘晓亮，丁世飞．SVM用于文本分类的适用性．计算机工程与科学．2010,32(6)：106-108页
[109]谢塞琴，沈福明，邱雪娜．基于支持向量机的人脸识别方法．计算机工程．2009,35(16)：186-188页
[110]林吉良，蒋静坪．基于支持向量机的移动机器人故障诊断．电工技术学报．2008，23(11)：173-177+182页
[111]李颖新，阮晓钢．基于支持向量机的肿瘤分类特征基因选取．计算机研究与发展．2005，42(10)：1796-1801页
[112]温熙森．模式识别与状态监控．北京：科学出版社，2007：327页
[113] Cristianini N, Shawe-Taylor J．支持向量机导论．李国正，王猛，曾华军，译．北京：电子工业出版社，2004：25-26页
[114] Chang C C, Lin C J. LIBSVM: a library for support vector machines[OL].(2001)
[2011-11-14]. http://www.csie.ntu.edu.tw/~cjlin/libsvm.
[115] Boser B E, Guyon I M, Vapnik V N. A training algorithm for optimal margin classifiers.Proceedings of the Fifth Annual Workshop on Computational Learning Theory.Pittsburgh, PA, USA,1992. New York: ACM Press:144-152P
[116] Osuna E, Frenud R, Girosi F. An improved training algorithm for support vectormachines. Proceedings of the IEEE Workshop on Neural Networks for SignalProcessing. New York, USA,1997. Piscataway, NJ, USA: IEEE:276-285P
[117] Syed N, Liu H, Sung K. Incremental learning with support vector machines.Proceedings of the International Joint Conference on Artificial Intelligence. Sweden,1999. Morgan Kaufmann publishers:352-356P
[118] Tang Y C, Jin B, Zhang Y Q, et al. Granular support vector machines for medicalbinary classification problems. Proceedings of the IEEE Symposium on ComputationalIntelligence in Bioinformatics and Computational Biology. La Jolla, CA, USA,2004.Piscataway, NJ, USA: IEEE:73-78P
[119] Lin C F, Wang S D. Fuzzy support vector machines. IEEE Transactions on NeuralNetworks.2002,3(2):464-471P
[120] Jayadcva R, Khemchandani S C. Twin support vector machines for patternclassification. IEEE Trans on Pattern Analysis and Machine Intelligence.2007,29(5):905-910P
[121] Herbrich R, Graepel T, Obermayer K. Large margin rank boundaries for ordinalregression. Advances in Large Margin Classifiers.2000,7:115-132P
[122]连可，陈世杰，周建明，等．基于遗传算法的SVM多分类决策树优化算法研究．控制与决策．2009，24(1)：7-12页
[123] Bottou L, Cortes C, Denker J, et al. Comparison of classifier: a case study inhandwritten digit recognition. Proceedings of the International Conference on PatternRecognition. Los Alamitos, CA,1994. IEEE Computer Society Press:77-82P
[124] Knerr S, Personnaz L, Dreyfus G, et al. Single-layer learning revisited: a stepwiseprocedure for building and training a neural network. Optimization Methods andSoftware.1990,1:23-34P
[125]段江涛，李凌均，张周锁，等．基于支持向量机的机械系统多故障分类方法．农业机械学报．2004，35(4)：144-147页
[126]吕志民，张武军，徐金梧，等．基于奇异谱的降噪方法及其在故障诊断技术中的应用．机械工程学报．1999,35(3)：85-88页
[127]刘献栋，杨绍普，申永军，等．基于奇异值分解的突变信息检测新方法及其应用．机械工程学报．2002,38(6)：102-105页
[128]何田，刘献栋，李其汉．噪声背景下检测突变信息的奇异值分解技术．振动工程学报．2006,19(3)：399-403页
[129]栾方军．在线手写签名认证算法的研究．吉林大学博士学位论文．2007：31-33页
[130]张德丰.MATLAB数字信号处理与应用.北京：清华大学出版社，2010：109-112，264-270页
[131] Cooley J W, Tukey J W. An algorithm for the machine calculation of complex Fourierseries. Mathematics of Computation.1965,19(90):297-301P
[132]沈志远，王黎明，陈方林．基于有限长序列分析的Welch法谱估计研究．计算机仿真．2010，27(12)：391-395页
[133]姚天任，江太辉.数字信号处理.3版.武汉：华中科技大学出版社，2007：369-370页
[134]王世一.数字信号处理.修订版.北京：北京理工大学出版社，2012：106-114页
[135]张思.振动测试与分析技术.北京：清华大学出版社，1992：328-335页
[136] Welch P D. The Use of Fast Fourier Transform for the Estimation of Power Spectra: AMethod Based on Time Averaging Over Short. IEEE Transactions on Audio andElectroacoustics.1967, AU-15(2):70-73P
[137]张峰，石现峰，张学智．Welch功率谱估计算法仿真及分析．西安工业大学学报．2009，29(4)：353-356页