Applied Mechanics and Materials Vols. 333-335

Abstract: Based on EMI and ANN techniques, crack location in coupling beam was studied by the experimental platform which using the WK6500 impedance analyzer. Impedance curves of the different PZT was obtained to diagnosis the situation of coupling beam whether it was perfect or cracked, and then located the crack by neural network technique to if the coupling beam was cracked.

Abstract: The closely combination of Internet of Things and on-site monitoring and control technology is the development tendency of the site monitoring and control technology, especially the special equipment safety monitoring technology. EDGE technology is one way of Internet of Things technology realization. EDGE technology is a digital mobile phone technology which allows improved data transmission rates as a backward-compatible extension of GSM. It is also considered as a pre-3G (2.75G) radio technology and is part of ITU's 3G definition. It is used as one of the aims of the project which allows the operators to keep abreast of the crane operation status. Then a software platform is developed to monitor the real-time stress. A software platform is developed via a lot of preliminary work as mentioned above. The information is collected by FBG sensors and transmitted to the demodulator, and then the software calculates strain by a formula given in. Finally we acquire the real-time stress in different parts of the crane. If the real-time stress exceeds the allowable stress, it issues an alarm signal to the operator.

Abstract: Aim at the non-stationary and time-variation characteristic of the gear-pump fault signal, proposing a condition recognition method for gear pump based on EMD and neural network. Take the vibration acceleration of case as the analysis object. Firstly,deal the signal with EMD, extract the main IMF components, containing the main information components. And then calculate its energy, and use the energy ratio to compose the feature vectors , which is used for BP neural network to identify the gear pump working state. The result reveals the EMD neural network method can recognize the working conditions of mesh-type high-pressure gear pump CB-KP63.

Abstract: Building settlement observation is often measured with leveling. And the elevation difference between the settlement observation points and benchmarks is periodically observed in the building. This paper discusses the problems of how to set the settlement observation points, how to implement the settlement observation and how to arrange the cycle of the settlement observation, this paper gives valuable methods and suggestions in the end.

Abstract: Magnetic flux leakage (MFL) testing, as a modern nondestructive testing technology, has significant application in tank floor corrosion defect inspection. In view of signal characteristics in the MFL testing of tank floor, an improved signal progressing method combining software and hardware was presented. Especially, in order to avoid phase distortion caused by applying common digital filter, zero-phase digital filter which could effectively eliminate the phase distortion was proposed to improve the location accuracy of defect. Experiment results showed that the signal progressing method could improve signal-to-noise ratio, and the peak-peak value of defect signal had a good linear relationship with the depth of defect which showed that the proposed method could meet the requirement of the MFL testing. Comparing results showed that, the maximal location error of zero-phase filter was 2 mm and that of common digital filter was 24 mm, which showed that zero-phase filter had better defect location accuracy.

Abstract: Pattern of number of PD events ("NE" for short) variation detected by UHF online monitoring was found through experiments, demonstrating that in the first half life of oil-paper insulation the NE value declined at the beginning and then kept stable, while it increased sharply in the later half. Least square linear regression was applied to calculate the changing rate ("CR" for short) of the latest 168 measurements of NE, which turned out to be an appropriate indicator of life stage for its monotonicity over time from the initiation of PD till final breakdown. The Grubbs method was also applied to exclude erroneous data and to determine the credible variation of CRs.

Abstract: A machine vision system is developed to detect the cable surface damage of cable-stayed bridge. In the system, four cameras are employed to acquire images around the cable surface. So the same one defect may be split into several images. Image mosaic had to be done to obtain a complete defect image for further process. The feature of cable surface image is simple and its illumination is heterogeneous. So the Scale Invariant Feature Transform (SIFT) feature matching algorithm is suitable for the image mosaic. Firstly, cable surface images should be preprocessed. Secondly, the SIFT algorithm achieves the detection, extraction, description and matching of feature points for defect images. Finally, image fusion is implemented to acquire the integrated image and a complete defect will be showed in the image. Experimental results show that using SIFT for the cable defect image mosaic has good effect on improving the detection accuracy and integrity for cable surface defects.

Abstract: Fault diagnosis is very important to make the system return to normal operation quickly after an accident. This paper diagnoses the specific component failure and failure area when the real-time motion information of inputting protection and switch transferred to a trained artificial neural network model by building an artificial neural network diagnosis model of components such as transmission line, bus bar and transformer, training the artificial neural network through taking the failure rule which is found by the historic fault data as a training sample. This method has obvious advantages in the accuracy and speed of diagnosis compared with the previous artificial neural network and overcomes the shortcomings of the incompletion of training samples and not well dealing with the heuristic knowledge.

Abstract: In view of the differences and errors existed between traditional EMC(electromagnetic compatibility) laboratory and standard laboratory, we introduce the concept of laboratory proficiency testing, and presents the evaluation parameter related to laboratory proficiency testing, such as, correlation coefficient, peak retention rate, connection degree, finally comprehensive similar reached for the laboratory testing capability. We also verified the evaluation parameters proposed in this paper.

Abstract: The main challenges of high-resolution ADC testing are the huge number of samples and the expensive test equipment, especially the requirement of high linearity signal source. In this paper, the scaling and segmentation algorithm which combines SEIR with windows is introduced for high-resolution ADC test. The new approach is validated by simulation with a 24-bit sigma-delta ADC. INL error of the proposed method is ±0.2LSB, which is less than the SEIR method of ±0.5LSB,and less than the histogram method of ±0.3LSB. About 20 million samples are required in the proposed method, which is about 30 times less than the traditional histogram method.