Applied Mechanics and Materials Vols. 20-23

Abstract: The mechanism of effect of the acceleration on acceleration sensor is preliminary elucidated. Using the three-dimensional finite element method, the temperature field in sensing element of the sensor under the effect of the different acceleration have been obtained by a series of procedure , such as model building ,meshing ,loads applying and equation solving. In the process ANSYS2FLOTRAN CFD program is applied. By the numerical results and experiments, the effect of input acceleration on accleration sensor is analogous to the effect of input tilt angle on the sensor, which also will change the temperature of gas flow at two thermal resistors ,so lead to change of the current difference of both thermistor wires, finally the bridge outputs a voltage signal corresponding to the acceleration.

Abstract: The motion estimation algorithm based on the Region of Interest has been proposed for the contradictions between accuracy and computational complexity. According to different levels of physical activity, the image is divided into regions of interest (ROI) and background (BG). For ROI, motion estimation based on redundant wavelet domain (RDWT-FS-ARPS) is presented in order to ensure the accuracy of motion estimation and the speed. And for BG, adaptive rood pattern search based on wavelet domain (DWT-ARPS) is proposed so that the computational complexity is of a dominant position while maximizing the accuracy of motion estimation. Experiment shows that the method is practical and effective to meet human visual system, and it can solve the contradictions between accuracy and computational complexity of the motion estimation to a certain extent. Besides, it provides a foundation for the compression of the region of interest and real-time transmission, especially suitable for monitoring, video phone system.

Abstract: The embedding parameters of electroencephalogram (EEG) time series, i.e., the embedding dimension and delay time, are used together as the input features of artificial neural network for distinguishing between normal and epileptic EEG time series. Cao’s method and mutual information method are applied for computing the embedding dimension and delay time of normal and epileptic EEG time series, respectively. The probabilistic neural network (PNN) is used in this paper for distinguishing between normal and epileptic EEG time series. The results of the simulation show that the overall accuracy as high as 100% can be achieved by using the method proposed in this paper, and that the accuracy obtained based on the both parameters is better than that obtained based on each of the two parameters respectively.

Abstract: Constrained with structures of the meter panels, the automatic measurements are required to take different ways. In this paper, a new approach to the automatic measurement of voltmeter with a cylindrical panel and nonlinear scale is presented. Firstly, one real-time image captured by a camera is processed to locate the region of the pointer and the scales. Secondly, according to the shape features of the dial, a 5×1 filter is adopted to reduce the noises, then a thinning algorithm is applied to extract the framework of the pointer and the scales, and a thinning correction algorithm is used to compensate for the slight deflection of pixels. Thirdly, a method for cylindrical rectification is used to compute the precise position of the pointer and the scales. Finally, error sources are discussed. Experimental results show that this system is of speediness and errorless and can meet the engineering requirements.

Abstract: Reliability of CTCS-3 train operation and control system is necessary to ensure train running safely and efficiently. Thus, train control system simulation should be used for lots of tests and validations. This paper is focus on the theory and period of fault injection systematically. According to characters of simulation of train control system and advantages of fault injection, application of fault injection in the field of train control system simulation is proposed for reliability test and to obtain system information after fault injection. Main structure and detailed functions, structures of fault injection system is designed and realized. This system could simulate system fault to cause rapid failure in train control system during the experiment. By analyzing the simulation result, the conclusion could be obtained that application of fault injection could help to improve reliability and fault tolerance of C3 simulation.

Abstract: In a noninvasive brain-computer interface (BCI), EEG feature extraction is a key part for improving classification accuracy and resulting information transfer rate, and it has a crucial and decisive role. In this paper, three different methods were proposed that combine spatial filtering with autoregressive model for EEG feature extraction. Six subjects participated in the BCI experiment during which they were asked to imagine movements of left hand and right hand. Each subject carried out four sessions and each session contained 120 trials. EEG data recordings were used for off-line analysis and the 10 leads around C3 and C4 were chosen for feature extraction. Autoregressive model coefficients and the parameters derived from other three methods were proposed as classification features. Fisher discriminant analysis (FDA) was used as linear classifier. The results show that classification accuracy rates obtained from the three proposed methods are far higher than those acquired from autoregressive model coefficients. At the same time the classification results of each subject are very stable, proving the effectiveness of these novel feature methods.

Abstract: The paper presents a novel quantum neural network (QNN) model with variable selection for short term load forecasting. In the proposed QNN model, first, the combiniation of maximum conditonal entropy theory and principal component analysis method is used to select main influential factors with maximum correlation degree to power load index, thus getting effective input variables set. Then the quantum neural network forecating model is constructed. The proposed QNN forecastig model is tested for certain province load data. The experiments and the performance with QNN neural network model are given, and the results showed the method could provide a satisfactory improvement of the forecasting accuracy compared with traditional BP network model.

Abstract: An unavoidable disadvantage of most traditional sampling methods is the large amount of samples data preservation during sampling process. Equivalent Time Sampling is a popular sampling method which usually used in oscilloscope technology. When input signals have the similar features as the jump signal, whose information focus in a very short period of time, the Equivalent Time Sampling’s disadvantage will become more obvious. This paper proposed a new sampling method named Vertical Sampling based on testing time via ADC. High-speed voltage comparator, ADC, and mathematical model between phase and time are used in this method. The whole system obtains unknown input signal’s tendency information and the time difference information to accomplish the sampling process. Experimental results proved this new method’s effectiveness on reducing the samples for three kinds of signals especially for the jump signals.

Abstract: The paper discusses the irregular parts packing problem based on an improved immune genetic algorithm, and a NIGA based on crowing mechanism is proposed. For improving the packing efficiency, the graphics-matching algorithms and clustering algorithm in the pretreatment of the part graphics are introduced. Matching algorithm of surplus rectangle as decoding algorithm for local optimization is proposed for automatic layout. In solving the large-scale packing problem, the application of immunity operator and niche genetic algorithm based on crowing mechanism improves the global optimization performance and velocity of convergence. The algorithms are effective and feasibility for solving the packing problem in the hull construction automatic packing system.

Abstract: 31P MRS(31Phosphorus Magnetic Resonance Spectroscopy) is a non invasive protocol for analyzing the energetic metabolism and biomedical changes in cellular level. Evaluation of 31P MRS is important in diagnosis and treatment of many hepatic diseases. In this paper, we apply back-propagation neural network (BP) and self-organizing map (SOM) neural network to analyze 31P MRS data to distinguish three diagnostic classes of cancer, normal and cirrhosis tissue. 66 samples of 31P MRS data are selected including cancer, normal and cirrhosis tissue. Four experiments are carried out. Good performance is achieved with limited samples. Experimental results prove that neural network models based on 31P MRS data offer an alternative and promising technique for diagnostic prediction of liver cancer in vivo.