Abstract:Spherical joint is a commonly multidegreeoffreedom mechanical hinge which has many advantages such as compact structure, good flexibility, and high carrying capacity. Realization of its multidimensional angular displacement measurement is of great significance in the prediction, feedback, and control of the system motion error. Firstly, the application of spherical joint and its structural characteristics were presented in the paper. Then, the motion description of the spherical joint and needed angles for measurement were analyzed. A review of multidimensional angular displacement measurement method, including structural decoupling detection method, opticalbased detection method and magneticfieldbased detection method, at home and abroad was provided, Finally, the development of research on multidimensional angular displacement measurement method for spherical joint was summarized. The focus and the difficulty of the research were pointed out, and the challenges and the breakthroughs in the key technologies were also stated.

Abstract:The complex and various defects of the steel surface bring great difficulty to the feature extraction and selection. Therefore, this paper proposes a new RAdaBoost future selection method with a fusion of feature selection and sample weights updated. The proposed algorithm selects features and reduces the dimension of features via Relief feature selection according to updated samples in each cyle of AdaBoost algorithm, and uses reduced features to remove noise samples by intra class difference among samples, and then update sample library according to dynamic weight of AdaBoost. The weak classifiers are trained by the resulting optimal features, and combined to generate the final AdaBoost strong classifier, and detect and locate strip surface defects by AdaBoost two classifiers. Aiming at a variety of defects such as scratch, wrinkle, mountain, stain, etc. in the actual strip production line, the experimental results show that the proposed RAdaBoost algorithm can effectively extract features with high distinction and independence and reduce the feature dimension, and simultaneously improve the accuracy of defect detection.

Abstract:Aiming at the random error of MEMS gyroscope is the main factor that restricts its precision and application range, the Kalman filter estimation method based on regression moving average (ARMA) model is proposed in this paper. Firstly, based on the results of Allan variance analysis, the quantization noise, angle random walk and zero bias instability are the main parts of the MEMS gyroscope random noise. Then, the stability of MEMS gyroscope random noise is tested by using time series analysis. Finally, based on the random drift of the autoregressive moving average (ARMA) model, a discrete Kalman filter equation is built to actualize its error estimation and compensation. The results of static vehicle and dynamic environment of digital noise reduction and Kalman filtering compensation experiments show that the Kalman filter estimation method based on the ARMA model has more obvious advantages in MEMS Gyroscope random error compensation.

Abstract:Empirical mode decomposition（EMD）method attenuates the signals’ energy and generates false signals in decomposing signal noise, which leads to incorrect detection results. In order to solve this problem, a stochastic resonance method under Levy noise after denoised by EMD decomposition is presented in this paper. After decomposed by EMD, the noisy signals are handled by overlaying, averaging and resampling to meet the condition of stochastic resonance. An adaptive algorithm is used to optimize system parameters, and then the processed signal can generate stochastic resonance in bistable system to achieve precise detection. The theoretical analysis and experimental results prove that the method can detect singlefrequency signal and multifrequency signal under the same characteristic exponent with the Levy noise. The experimental results demonstrate that the SNR of singlefrequency signal can increase 14 dB in the case of SNR of -28 dB. The spectral amplitude of the 5 Hz spectrum is increased from 311.8 to 724 and 10 Hz spectrum amplitude is increased from 138.9 to 143.2. This method that reduces the residual noise energy and false signal can improve the signal energy in a complex noisy condition. Compared to EMD decomposition which cannot determine the signal components, this method can achieve the detection effect better.

Abstract:To help nursing staff in senile apartment find the elderly fall and other actions timely, an action recognition method based on video surveillance is proposed. Firstly, the foreground images are extracted by the GMM background modeling method in HS color space. Feature extraction is performed by combining the motion features and morphological features. And action recognition can be achieved by HMM with Gaussian output. The method proposed in this paper can adapt to the changes of illumination. The method also has good robustness to the change of motion direction and motion range, and the recognition accuracy rate reaches 90%. The result shows that the method can meet the basic requirements of action recognition and the method has certain practical value.

Abstract:To solve the timeconsuming problem of image stitching algorithm based on KAZE, a simple and effective image stitching algorithm based on AKAZE is proposed. Firstly, AKAZE feature points are extracted. Secondly, feature vectors are constructed using the MLDB descriptor and matched by computing the Hamming distance. Thirdly, wrong matches are eliminated by RANSAC and the global homography transform, and then a local projection transform is estimated using moving direct linear transformation in the overlapping regions. The image registration is achieved by combining the two transforms. Finally, the weighted fusion method fuses the images. A performance comparison test can be conducted aiming at KAZE, SIFT, SURF, ORB, BRISK. The experimental results show that the proposed algorithm has better robustness for the various transform, and the processing time is greatly reduced.

Abstract:The concentration of nitrogen oxides (NO2, NO, N2O, etc.) in power plant is an important index of environmental protection. Aiming at the problem that the detection accuracy of nitrogen oxides concentration based on spectral analysis could be interfered by all kinds of factors, such as temperature, moisture content, tar, naphthalene, noise of electric devices, optical lens aging, interference at spectral absorption characteristics of polluting gases etc, it is difficult to improve in a single way. At first, the hardware modification is favorable for gas purification and filter. And then, the selflearning and selftraining ability of RBF neural network can save the traditional model for the study of interference factors, and make the data processing more efficient. On the basis of a large thermal power plant’s real data in 2015, the computer simulation and analysis show that this method can improve the accuracy effectively. The overallaverage deviation is 0.841%.

Abstract:A simple, rapid and efficient retinal vessels segmentation method is proposed. After a general analysis on gray value distribution and contrast changes of fundus images, the standardizing fundus images are obtained by using the matched filtering technique to overcome the interference of background and noise. Then, a threshold can be automatically selected to achieve the effective segmentation of blood vessels in the fundus images by estimating the proportion of the background pixels. A lot of tests show that the good performance is achieved in the public fundus images database. The experiment shows that the proposed method based on matched filtering and automatic threshold has strong practicability and high accuracy. It is useful for computeraided diagnosis of ocular diseases．

Abstract:The triaxial array induction logging tool is an important tool for measuring complex formation information, but the key of design and application is to eliminate borehole effects of coplanar coil systems. The response characteristics of coplanar coil systems and coaxial coil systems of the triaxial array induction are analyzed numerically in the center hole by the FEM, and field experiment results are compared with modeling simulation on the coplanar coil systems. The results show that the borehole effects of coplanar coil systems are significantly different from that of coaxial coil systems besides the shortest subarray. The borehole effect causes the negative logging response. For the longest subarray, although the borehole effect is week, its skin effect is serious and the measurement range of formation conductivity is smaller than that of the coaxial coil systems. It reveals the reason that the greater difference is occurred between the coplanar coil systems and the coaxial coil systems through the eddy current distribution of coplanar coil systems around the borehole at different mud and formation conductivity. A new coil system which has two transmitters and one receiver is raised and its advantages are analyzed on the decreasing borehole effect and reducing negative response from the geometrical factor. The research lays a theoretical basis for the development and application of the triaxial induction logging tool.

Abstract:In disaster rescue and emergency situations, node energy in sensor network is especially limited. In order to reduce unnecessary forwarding consumption, this paper presents a MANET multicast routing tree algorithm with least forwarding nodes, which is based on shortest routing tree and subtree deletion. The algorithm is proved and analyzed in detail. Its practical distributed version is also presented. The simulation comparison shows that this distributed algorithm reduces the forwarding transmission in improved ODMRP, especially there are much more receivers in MANET. Minimum forwarding routing tree has the minimum network overhead. It is an effective way to extend the network lifetime.

Abstract:A fuzzy perception model is proposed to the directional sensor nodes based on the sensing characteristics of the nodes, and also the fuzzy data fusion rule is built to reduce the network uncertain region. Aiming at the problem of directional sensor network strong barrier coverage, a directional sensor network strong barrier coverage enhancement algorithm based on particle swarm optimization is proposed. The convergence rate of the algorithm is improved through the ndimensional problem be transformed into onedimensional problem. The simulation results show that, under random deployment, the perception direction of sensor nodes can be adjusted continuously. Compared with the existing algorithms, the proposed algorithm can effectively form strong barrier coverage to the target area, has a faster convergence rate, and prolongs the network lifetime.

Abstract:When congestion occurs in the network, VOQ routers still suffer a certain degree of headofline blocking (HoL) problem in the networkonchip (NoC).Aiming at this issue, we propose the loadbalancing AVOQ router architecture. Firstly, the VOQ mechanism is kept to deal with the HoL problem. Secondly, a flexible output port can be picked up in the routing computing module, making sure that the dada is ported out to the less congested road and a single virtual channel (VC) can read the packet adaptively, so that the less congested flow in the downstream can be transmitted. The experimental results show that, compared to the VC router and the VOQ router, AVOQ router reduces the average latency by 83.2% and 57.1% and improves the throughput by 72.7% and 33.3% at most, while the area overhead and power consumption are affordable. By the use of above adaptive mechanism, the network load is balanced and the congestion is relieved, and the appearing of the HoL is decreased. Moreover, the impact of HoL is eliminated as long as it appears,and the network performance is improved greatly.

Abstract:The key point of azimuthal propagation resistivity loggingwhiledrilling focuses on the structural design of the coil system. And the detection performance of azimuthal propagation resistivity LWD is mainly affected by the transmission frequency of electromagnetic wave signal, the transmitterreceiver spacing, the receiver interval, the coil’s angle and the formation resistivity. The testing method of measurements is determined with different inspection requirements of azimuthal propagation resistivity LWD. According to the various constraints of the coil system under the condition of different testing method, the structure of the coil system for azimuthal propagation resistivity LWD is designed by experimental simulation method. The results provide reference for the structural design of the coil system for azimuthal propagation resistivity LWD.

Abstract:Velocity smoothing is one problem which is proposed in high speed machining and coal mine safety production, the aim of which is to improve machining accuracy and equipment life. Aiming at this problem, this paper proposes a stage wise model and deduces the closedform expression solution for each stage based on the relationship of acceleration and velocity, and then deduces the general solutions of cubic equation in detail for the model. Finally, the solutions are applied to the velocity smoothing. The proposed schema shows the advantages of easy to program and smoothing in transition curve when being applied for velocity smoothing in coalmine. The result demonstrates that the proposed method adapts the highspeed scenarios well and has used in other several projects.

Abstract:In the field of integrated circuit testing, in order to improve the test data compression ratio and test generation, it is often necessary to do spectral analysis of the test set and test response and calculate their prominent spectral component. A method is proposed to enhance the prominent spectral component of test set by using KM (KuhnMunkras) algorithm. Based on the test set and its prominent spectral component, a bipartite graph and a weighting matrix are constructed. The problem of the enhancement of prominent spectral component is transformed into a bipartite graph matching problem, and then be solved by KM algorithm. After the order adjustment of test set according to the matching relationship, the correlation between prominent component and test set is increased, and the prominent spectral component is enhanced. In this paper, the experimental results about the test set of the ISCAS89 benchmark circuits show that the coefficient of the sorted test can increase by 19.05% on average, and the test set residue compression ratio basis on FDR (frequencydirected runlength) code can increase by 4.59% on average.

Abstract:Aiming at the shortcoming of the low accuracy of transformer fault diagnosis, the PSOSOMLVQ（particle swarm optimization,self organizing maps,learning vector quantization） mixed neural network algorithm is presented in this paper. Firstly, the weight of SOM neural network is optimized by the method of PSO algorithm to obtain the more effective topology. Based on that, LVQ neural network is combined to cover the shortage of unsupervised learning SOM neural network. The mixed neural network algorithm combined with PSO, SOM and LVQ can improve the accuracy and reduce the error of transformer fault diagnosis. Through simulation, the three algorithms of SOM, PSOSOM and PSOSOMLVQ are compared. The comparison result show that the PSOSOMLVQ mixed neural network algorithm has the highest accuracy, and the fault diagnosis accuracy rate is 100%. Thus it can be seen, the PSOSOMLVQ mixed neural network algorithm can enhance the performance of transformer fault diagnosis effectively.

Abstract:Musical simulation based on spectrum model is the use of acoustic theory that can achieve musical instrument’s sounds by sum of products of a series of basic functions and timevarying amplitude. A new digital piano sound simulation technique is proposed by analyzing piano string vibration and damping characteristics and investigating the resonance effect of resonance box. The simulation model consists of two parts: the excitation system and the resonance system. Based on the vibration equation of the strings, the envelope modification of time domain is carried out to simulate the natural attenuation of the strings, which can make music harmonious between the notes. Then, the filter group is modeled by spectrum envelope in frequency domain to achieve the simulation of resonance system. This new method can more effectively carving voice, has better performance timbre at the same time, therefore, it makes the sound more harmonious.

Abstract:A low power, small die size 14bit 125 MSPS pipelined ADC is presented. Switched capacitor pipelined ADC architecture is chosen for the 14bit ADC. In order to achieve low power and compact die size, the sample and hold amplifier is removed, the 4.5bit substage circuit is used in the first pipelined stage. The capacitor down scaling technique is introduced, and the current mode serial transmitter is used. A modified miller compensation technique is used in the operation amplifiers in the pipelined substage circuits, which offers a large bandwidth without additional current consumption. A 1.75 Gbps transmitter is introduced to drive the digital output code, which only needs 2 output pins. The ADC is fabricated in 0.18 μm 1.8 V 1P5M CMOS technology. The test results show that the 14bit 125 MSPS ADC achieves the SNR of 72.5 dBFS and SFDR of 83.1 dB, with 10.1 MHz input at full sampling speed, while consumes the power consumption of 241 mW and occupies an area of 1.3 mm×4 mm.

Abstract:In order to improve the multitarget tracking robustness and enhance the difference between the targets, this paper uses an energy minimization method for multitarget tracking. Different to the existing algorithm, the algorithm focuses on the representation of the complex problem in multitarget tracking as energy function model, which includes a better target segmentation strategy (similarity model). By assigns every possible solutions a cost (the “energy”), the algorithm transforms the multiple target tracking problem into an energy minimization problem. In the energy minimization optimization method, the algorithm uses the conjugate gradient algorithm and a series of jump moves to find the minimum energy value. The experimental results of open data demonstrate the effectiveness. And the quantitative analysis results show that this algorithm can improve the difference between targets or between target and background so as to obtain better robust performance compared with other algorithms.

Abstract:In order to solve the low efficiency and the influence of manual factors and many other problems existed in current water meter verification, the water meter verification system using machine vision technology is proposed. And the research keynote is how to realize the template matching algorithm for rapid location of plum blossom needle and the image morphological algorithm for eliminating the bubble of wet water meter dial. Harris algorithm is used to extract the corner points of the plum blossom needle template beforehand, and the corner points of the onsite image are extracted in real time. Then, the fast localization of the plum blossom needle is realized by the partial Hausdorff distance method. Finally, the effect of bubbles is eliminated by using the image morphological algorithm, and the count value of the rotating teeth of the plum blossom needle is completed. The experimental results show that the proposed system can shorten the verification time and improve the verification efficiency while ensuring the verification accuracy. The system solves the adverse effect of the bubble on the dial of the wet water meter, and it’s suitable for the verification of various types of water meters.

Abstract:Dongba pictograph has been known as "the only living pictograph in the world".In the aspects of image recognition, content interpretation,the current English and Chinese character recognition system often can not be applied to Dongba pictograph.Concerning the difficulties in the identification of Dongba pictograph, a new character recognition is proposed. Topological features processing and projection methodcompose thefeature extraction method,then, the character recognition method based on template matching is adopted.It is showed that the feature extraction method based on the intrinsic characteristic of the pictograph,and the Dongba character recognition method based on template matching,has high accuracy through the experiment.

Abstract:Rotating machinery is the most widely used mechanical equipment in production, and its transmission system is an important part of the rotating machinery. Whether the rotating machinery transmission system running in safe and stable state has an important impact on the economy and society, so, researched on running stability deterioration of rotating machinery is significant. Rotor test rig has typical transmission structure of rotating machinery, and feature parameter extraction method faced to running stability deteriorationof its transmission system has been proposed.This paper describes the evolutionary trajectory of stable running state deterioratedto the unstable running state, and deterioration evolutionary matrix has been proposed based on higherorder cumulant method. A feature parameter extractionhas been used to depict evolutionary track faced to running stability deterioration of rotating machinery based on characterization parameters of the state deterioration trend such as mean value of stability deterioration and varianceof stability deterioration. The vibration data of the actual operating state of rotor test rigare collected, and experimental verification is carried out based on the method.The results show that this method can describe the overall trend of the operation statesface to the transmission system of the rotating machinery.