A number of induction machine is widely used in people's daily life and agriculture manufacture, and the chance to meet all kinds of breakdown is large. Electrical machine's breakdown not only damage electrical machine itself, but also make electrical machine suddenly break down and product line breakdown, result in great economy's loss and fatal result. So, research on electrical machine's fault diagnosis' technology, Possess significant theory's meaning and society economic benefit. Texts according to induction drive motor's common breakdown's feature, Directing to the past research settle electrical machine's breakdown's means's fault, Bring up base on improve the smallest two ride stand by vector machine (LS- SVM) de induction drive motor's fault diagnosis' means. Adopt muddleheaded optimization algorithm, Seek come out possess compare strong popularize ability de parameter. Combine induction machine's structural parameter with performance export model de build mold example, with cross validation techniques to compare confirm research means be used for motor fault diagnosis de validity. A number of induction machine is widely used in people's daily life and agriculture manufacture, and the chance to meet all kinds of breakdown is large. Electrical machine's breakdown not only can damage electrical machine in itself Seriously breakdown can make electrical machine suddenly suspend air, product line breakdown, result in great economy's loss and fatal result. So, Researching on electrical machine's fault diagnosis' technology, Enhance motor's credibility, find the fault in time, Diagnose and obviate breakdown possess great society and economic benefits. Proceed motor's online fault diagnose mechanism research, give back can promote electrical machine systematic debug and monitor and control technical development, Much good field control electrical machine's normally execution ([1]). At present, not a few domestic and abroad scholars already proceed plenty of research on electromotor breakdown Diagnose means chief have under several kind of: Base on parse model means is in need of establishing general work condition quantity under normal situation (stator voltage, current, energizing voltage, current, deserving, have not achievements' power, empty carry characteristic curve) with relational model between energizing current. Means validity consist in right establish electrical machine de mechanism mathematic model, because affecting factor complex, Difficult compare big ([2,3]). Base on oscillatory means, this' a kind of means is take advantage of vibration transducer come to bearing of motor or machine de oscillation proceeding monitor. Vibration transducer's outcoming signal by signal disposal (many for spectral analysis) behind, With already know de breakdown's characteristic frequency compare with compare, Thus look for breakdown's nature and faulty part. Base on oscillatory fault diagnosis' means de principle is according to different breakdown or abnormal can bring forth different frequence de oscillate come judge de ([4]). But, oscillate checkout de limitation compare big, First of all, oscillate monitor need supply sensor de install part, This to some's electromotor compare difficult; Next, oscillate signal incur surrounding environmental vibration noise affect compare big, Disturb under at complex background, hard to right judgement breakdown. In electrical machine de fault diagnosis, normal executing state de data's sample is very easy acquire de, But breakdown's sample but not easy acquire. Breakdown's data de lack already become restrict fault diagnosis' developing a important reason. Vapnik wait for people in bring up de a kind of base on stat study theoretical pattern recognition's new concept-stand by vector machine in 1992 (SVM) ([5]). SVM is a kind of many entity stat machine's learning method, adapt to in a kind of sample easy get but another kind of sample very difficult get de occasion, SVM possess very good extensive melt popularize ability, can popularize go many classify problem ([6],[7]). In induction drive motor's fault diagnosis, Only depend on great mass of data's sample and little breakdown sample can establish rise categorizer under normal executing state. SVM is use linear function suppose spatial learning algorithm at high dimension feature space, overcome nonlinearity classify problematic" dimension's number calamities ", Not easy appear part pole little, pass study problem, Already at image recognition ([8]) ,speech recognition ([9]) ,face recognition ([10]) and biochemics and medical science's domain get success application. Will SVM this a kind of good machine's learning method should be used for electromotor de fault diagnosis, have no doubt can for electromotor de fault diagnosis bring new energy ([11]).
