With the increase in transport volume, improving rail inspection efficiency becomes more important. Rail inspection is mainly using ultrasonic technology to cover both bulk and subsurface defects. Due to the limitation of data processing and conventional UT technology, it is very hard to deal with the relationship between inspection speed and ultrasound coverage. Now mainstream inspection car can reach maximum speed of less than 80Km/h. In order to achieve higher inspection speed, we are trying to build an inspection system combining phased array (PA) technology and high-speed processing algorithm. Also, fast phased array has been used for improving Probability of Detection (POD) of typical Transverse Defects (TD) in rail head. Meanwhile high-speed processing algorithm gives us potential to process data in parallel and discriminate defect type automatically, such like bolt-hole cracks, transverse defects etc., by setting certain channels’ combination conditions. During the experiment, inspection speed of 108Km/h is simulated. The results shows the system can handle the data flow in such speed.

The monitoring of ship acoustic faults can be carried out from two aspects of noise source and sound transfer path. Beginning with an introduction of the transmissibility function theory, a novel method of binary classification of data recognition is proposed to detect path acoustic fault on-line. The structure output-response data are utilized directly without measuring the excitation force or using the structure model . The functional indicator of principal component assurance criterion(PCAC) measures can be defined by combining transmissibility function with principal component analysis(PCA). Theoretically, the PCAC measures should be equal to “1” when the structures are in good health. Taking the principal component assurance criterion measures under health state as a base value, the value of confidence measures of vibration signals between health state and damage state should less than the base value . And as the damage increases, the value of the principal component assurance criterion should be smaller. Simulations and experiments are designed to validate this method, and the results show the effectiveness of the proposed method.

为了保障高速列车安全运行,采用ARIMA数据挖掘理论,研究了高速列车运动部件车轮的直径和磨耗的变化规律和安全预测.实验结果表明,ARIMA模型适用于车轮尺寸预测,随着输入数据量的增加,预测结果越好,而列车轮径比踏面磨耗的预测精度更高;同时与Least Square Estimation的预测结果进行了比较,结果优于Least Square Estimation.这为进一步将ARIMA应用于铁道领域列车关键部件的寿命和安全预测的研究提供参考.

以高速列车空心车轴超声波探伤为背景,通过声场分析及CIVA软件仿真设计了环形相控阵,评估了有效阵元个数以及触发频率对声场的影响.采用环形相控阵探头的探伤系统能以直线推进的方式对空心轴进行探伤,较现有探伤方式,可大幅提升检测效率、简化机电结构,并可有效避免由于探头旋转所导致的耦合不稳定.

The fault diagnosis of ship structure-borne sound transfer path is mainly based on data binary classification, which requires enough transfer functions or training samples. Actually, the data collection sensors installed on the ship are always limited resulting in quite lack of real training samples. Hence, it is of great significance to solve such a small sample problem for the ship structure acoustic fault identification and location. In this paper, some theoretical studies have been conducted to compare two methods of structure acoustic fault sample expansion including extracting the local Green’s function from cross-correlations of ambient vibration and dynamic monitoring based on the transmissibility function. Then a novel model of fault sample expansion was put forward and verified through a simulation example. The research shows that the proposed two methods have their respective pros and cons. Some suggestions on how to employ these two methods to ship structure health monitoring were also proposed.

The wheel quality is of fundamental significance for the safety of high speed railway. When the train runs at high speed, fatigue crack occurs on the stress concentration area of wheel rim, wheel disk, bore hole and gear seat, wheel seat, unload groove on hollow axle because of the remnants remained from the wheel manufacturing process. The principle of UT (Ultrasonic Testing) and PAUT (Phased Array Ultrasonic Testing) NDT technologies applied in railway wheel-set is introduced and analyzed in this lecture. It is clear that safety of wheel-set cannot be ensured by single NDT method and single maintenance procedure. Based on phased array ultrasonic and other NDT technologies, a comprehensive wheel-set NDT technical safeguard system has been built for high speed train in China, which includes three levels: Dynamic Inspection Systems, Wheel-set Online Inspection System, and Dismounted Wheel-set Inspection System. Now, the safeguard system is widely applied in EMUs maintenance centers and application depots in China, which has a very practical importance for the safety of train and human lives.

The operation safety of locomotive is related to every components on the locomotive. So, understanding the running state of every components tim ely is necessary. And the most difficult in locomotive key components monitorin g is how to define if the components is abnormal due to the judgment is subj ective. Such as, the same shift appear on chain or flexible pipe and on fixed bolt is com pletely different. In this paper, an improved Hierarchical Temporal Memory network was prese nted, and the purpose of the improvement is focused on the construction of the network. What’s more, the adaptability for illumination and translation was considered. Result in this paper show that it is feasible to improve the accuracy of abnormit y warning.

Wheel set is one of the most important components for railway vehicles. Comprehensive NDT wheel set ensuring system and corresponding equipment have been developed in some countries to ensure the safety of wheel set. LU automatic ultrasonic inspection system for in-service wheel set has been widely applied in Chinese high speed train, which uses advanced phased array ultrasonic technology to inspect circumferential, radial and bias fatigue cracks in wheel rim and wheel disk. To automatically, precisely and flexibly posit the probes carrier to the surface of wheel tread, the intelligent robot is used, which makes LU system easy to inspect all types of EMUs manufactured by Bombardier, Siemens, Alston, Kawasaki and China. By the usage of fully automatic control and wireless data transfer technique, LU system can accomplish the inspection of one wheel set within 7 minutes. More than 70 sets of LU system have been applied in CRH EMUs service depots, and hundreds of defects have been detected from 300,000 inspected CRH wheel sets. Now LU system plays a significant role in high speed train wheel set safety ensuring.

The crack detection methods are widely used in wheel-set maintenance, espec ially by ultrasonic and magnetic particle testing. Wheel, gear, and break disk are pressed or shrunk on the axle, and defects at axle press-fit parts can only be used by ultrasonic method by accessing from axle body or end face. This press or shrink processing has significant influence on th e in-service axle defect detection comparing with open defect on the axle. Parts of ultrasonic signal would go into wheel hub, and decrease the total signal power. The condition at press-fit interface makes ultrason ic eflection more strong and as a result makes the signal to noise ratio lower. With long time in service, both sides of press-fit have corrosion fatigue make the ultrasonic back reflection varies fro m one axle to another. In this paper, a series of experiments have been complemented to compare the difference of ultrasonic defect detection between bare axle and axle with press-fit part. By using wav elet denoising algorithm, the defects detecting SNR is obviously increased, and some results are sh own a an example.

In order to realize the automatic judgment flaw of ultrasonic detection through MATLAB and simplify image processing,taking the train wheels for example,the B scanning figure of ultrasonic testing is classified treatment.Based on the image registration method of mutual information,this paper deals with the complex background image,and the result from the image registration that it is operated and got the complement,finally realize the image of defects and the defects are marked through the MATLAB,then display the defects location information.This method completes the defects judgment of automatic ultrasonic testing successfully,has the function on improving the efficiency of the ultrasonic detection on judgment flaw.

Threshold technology which is one of the most important and efficiency technology in image segmentation is based on area image segmentation in many arithmetic. Principle and performance of several image segmentation arithmetic in common use has been involved in the thesis. And the improved Otsu arithmetic has been used for detecting abrasion of locomotive pantograph and catenary in warehouse entry system. So their labor intensity has decreased greatly.