Search

This paper describes a work of EMI treatment for a large electrical vehicle (EV), and the practical debugging methods, used materials and tools are presented. Being different from the EMI treatment for consumer and IT products, usually there are many diverse electronic and electrical facilities being equipped in a large electrical vehicle, say, an electrical bus, and that makes the EMI environment very complex and hard to be solved.

Specific emitter identification (SEI) enables the discrimination of individual radio emitters with the external features carried by the received waveforms. This identification technique has been widely adopted in military and civil applications. However, many previous methods based on hand-crafted features are subject to the present expertise. To remedy these shortcomings, this paper presents a novel SEI algorithm using deep learning architecture. First, we perform Hilbert-Huang transform on the received signal and convert the resulting Hilbert spectrum into a grayscale image. As a signal representation, the Hilbert spectrum image has high information integrity and can provide abundant information about the nonlinear and non-stationary characteristics of signals for identifying emitters. Thereafter, we construct a deep residual network for learning the visual differences reflected in the Hilbert spectrum images. By using the residual architectures, we effectively address the degradation problem, which improves efficiency and generalization. From our analysis, the proposed approach combines high information integrity with low complexity, which outperforms previous studies in the literature. The simulation results validate that the Hilbert spectrum image is a successful signal representation, and also demonstrate that the fingerprints extracted from raw images using deep learning are more effective and robust than the expert ones. Furthermore, our method has the capability of adapting to signals collected under various conditions.

We have developed a voice-controlled assistive system to improve the independence of people with Cerebral Palsy (CP) to do art painting. The system allows users who have severe motor deficiency in limbs to use their voice command to control a painting canvas to be within their reachable painting zone, and to dispense colour material on a rotatory palette for painting. Trials conducted at a CP Welfare Home show that the system allows the CP user to paint more independently and with minimum assistance from the care-worker.

This article presents our initial results in deep learning for signal modulation recognition in satellite communication system. In this article, we propose a signal classification model based on the recurrent neural network. Different from existing methods that need estimating the signal parameters or design the features by the experienced experts, we use LSTM (Long Short-Term Memory) to extract deep features of signal sequence from time domain. Using this architecture, we realized the modulation recognition for target signals end-to-end finally. And the method is insensitive to frequency offset and timing deviation, which has a strong application prospect in engineering. From our simulation results, the deep learning based approach can achieve considerable performances that the signals recognition rate is close to 98% in 6dB. Furthermore, the deep learning based approach is more robust than conventional methods when there is frequency offset, timing deviation and low Signal-to-noise ratio.

Specific emitter identification (SEI) provides the capability to distinguish radio emitters with the external features carried by the received waveforms. Since the differences between emitters are very subtle, the fidelity of signal processing may affect the identification accuracy. In this paper, we propose a high-fidelity symbol synchronization approach to depress the processing errors introduced by demodulation. First, two existing problems of timing recovery are pointed out in SEI scenario. Thereafter, we present a window-based interpolator and a feedforward decision-directed timing estimator to address these problems. Simulation results show that our approach effectively depresses the processing errors compared with the existing methods. Furthermore, it achieves a better performance of SEI.

A new nonlinear activated complex-valued gradient neural network is proposed, which is used for complex matrix inversion in the complex area. This work focuses on the inversion of complex-valued matrices in complex domains rather than finding the inverse of real-valued matrices. Compared with the traditional linear complex gradient neural network (GNN), the main contribution of this paper introduces a nonlinear activation function, which can effectively improve the convergence rate of the GNN model. The computer simulation substantiate the effectiveness and superiority of nonlinearly activated complex-valued gradient neural network (NACVGNN) to the matrix inversion, as compared with the linear complex gradient neural network.

It is difficult to distinguish the mixed signal received by the single channel receiver from single signal simply based on the waveforms and the spectrum features. In this paper, the expression on the observed and the valid signal-to-noise ratio (SNR) of single-channel mixed signal is derived and compared with that of single signal(just one signal component). We find that the observed and the valid SNR of single signal are identical, but for the mixed signal, the observed SNR is significantly greater than valid SNR. An appropriate threshold is set under the condition of minimum the single signal leaking-detecting probability to obtain best detection performance. Numerical results show that for both symmetric and asymmetric mixed signal, the proposed algorithm performs well without timing recovery, thus making proposed algorithm more practical.

A strong fault tolerance algorithm is proposed for blind identification of LDPC codes to solve the problem that the parity-check matrix of LDPC codes is hard to be reconstructed. First, the codes are arranged by reliability to construct data matrix and part of the parity-check vectors can be obtained by using Gaussian-Jordan column elimination operation. Then, codes with errors need to be eliminated and some codes need to be replaced to obtain most of the parity-check vectors. Finally, by using the parity-check vectors, some codes with high reliability can be decoded to obtain all parity-check vectors with the repetition of previous procedures. Compared with the existing blind recognition algorithms, the proposed method can implement blind identification of LDPC codes under the higher BER, which has a relatively low complexity and can be better applied to the actual environment. In the simulation experiments which mainly use (576,288) LDPC codes, the recognition rate of this method is above 90% under BER 0.00235.

Matched mode processing (MMP), which introduces the acoustic propagation model in the complex underwater waveguides, has been one of the promising directions in source localization. The well-known Capon method is often used for MMP due to the capability of high resolution; however, its performance may degrade significantly in the presence of mismatch. There are various mismatches such as environmental uncertainties, array calibration errors or signal random distortions that make it quite difficulty for source localization in practice. In this paper, we develop a robust Capon MMP method based on the optimization of worst-case performance. The weight vector is obtained with a convex optimization based algorithm using the second-order cone programming (SOCP). Simulations with the benchmark cases of general mismatch show that the proposed robust Capon method for MMP provides certain robustness against mismatches as well as effective capability of sidelobe suppression.

In order to improve the convergence rate of the blind equalizer for sparse multipath channel, a l 0 -norm constraint blind sparse adaptive algorithm for sparse channel equalization is proposed in this paper. The constant modulus characteristics of multiple phase shift keying (MPSK) signal and the sparse property of equalizer are firstly taken into consideration to construct a new blind equalization cost function with the l 0 -norm penalty on the equalizer tap coefficients. Then the update formula of the tap coefficients is derived according to the gradient descent algorithm. Moreover, the iteration step is updated adaptively by drawing upon the normalized proportionate factor. Theoretical analysis and simulation results show that the proposed algorithm outperforms the existing blind equalization algorithms for sparse channel in reducing the residual inter-symbol interference (ISI) and improving convergence rate.

In I(L)-topological vector spaces (I(L)-tvs) that are L-topologically generated (L-tg), we investigate the generalized local boundedness in this article. We prove that an L-tg I(L)-tvs is generalized locally bounded iff the underlying L-topological vector space(L-tvs) is. We further verify that an L-tg I(L)-tvs is generalized L-fuzzy normable iff the underlying L-tvs is.

In the wireless position techniques for cellular networks, the traditional geolocation algorithms have been researched because of the advantage low complexity, however the representative LLOP algorithm does not well performed in the cellular networks since it is constrained with the geometry relationship between mobile station and the base stations. For the low accuracy problem of LLOP algorithm, we have proposed a TOA-based geolocation algorithm utilizing convex combination. The method introduces convex combination theory, which the convex combination of a set of coarse position estimations can be used to generate position estimation with higher accuracy, and proposes a linearized procedure to calculate the convex combination coefficients. Simulation results verify that the proposed algorithm decreases the position error by 6.19%, 7.51% and 40.97% compared with LLS- MLE algorithm, LLOP-LS algorithm and LLOP algorithm respectively when the number of base stations is 5 and the measurement noise standard deviation is 50m.

To reduce implementation cost and power consumption, a sparse minimum mean square error (MMSE) beamformer is designed using the orthogonal matching pursuit algorithm to compute the locations and weights of few selected active antenna elements and estimate the direction-of-arrival (DOA) angles and the number of sources. The estimated DOAs are used to reconstruct the received signal covariance matrix and improve robustness to small sample size effects. Simulation results demonstrate the superior performance of the proposed sparse beamformer designs compared with state-of-the-art algorithms in the literature.

Mimicry is an important social skill for every person. A humanoid robot is used to play with autistic children in order to attract their attention and want to improve their concentration. Music therapy plays also an important role in this experiment. Robot plays with autistic children with music. The robot will imitate the notes which the child has played, and then the child has to imitate the robot in return. The purpose of the treatment is to improve their social skills including interpersonal synchrony and concentration can help them to adapt this society easier and make a better life in the future.

The Full Matrix Capture (FMC) method can provide the original complete information which is detected by the ultrasonic array and can recover with the Total Focusing Method (TFM). This technique has better detection sensitivity and resolution relative to conventional scanning method and is valuable for improving the quality and reliability of industrial non-destructive testing. However, it has low computational efficiency and cannot meet the demands for industrial real-time detection. Here, we present a new way to improve the detection efficiency of the ultrasonic FMC imaging detection based on increasing the effective aperture. Firstly, we discuss the factors which impact the effective aperture of ultrasonic array and the detection performance impacted by the aperture. Then, we present the method to effectively improve the efficiency with increasing array aperture and maintain its good detection performance. The simulation and experimental results show that the efficiency of FMC method is greatly improved.

An arbitrary sampling rate conversion algorithm based on frequency domain filtering (FDF-ASRC) is proposed. The number of Fourier coefficients in FDF-ASRC is irrelevant to the symbol rate of interested signal. Applying the resampling algorithm to non-uniform channelization, a few modules of Fourier transform (FT) and inverse Fourier transform (IFT) can guarantee infinite channels in channelization. Meanwhile, the sampling rate of output signal can be adjusted to an integer multiple of symbol rate, which can be passed to demodulator directly. By comparison with seven architectures of non-uniform channelizers, the computational complexity of the proposed channelizers is equal to the best.

Total focusing method (TFM) is a post-processing algorithm for ultrasonic imaging, usually used for processing linear array data. In this paper, TFM is extended to obtain three-dimensional (3D) image based on two-dimensional array data. However, it takes a long time to get 3D information from large amount of 2D array data by TFM. So Sub-aperture matrix capture (SMC) and sparse array are taken into account to solve this issue. The experimental results prove that TFM precisely reconstruct 3D information of defects through the optimized method. In a small angle imaging range, SMC can reach the same image quality and the same array performance indicator (API) as full matrix capture (FMC). In addition, it won't take more than one quarter of the time token by FMC for ultrasonic imaging. Although the spurious image is enhanced with the reducing sparseness, the processing time is reduced. Considering both image quality and real-time imaging, the imaging efficiency might be improved by 50 times through the combination of the two methods. The paper provides an optional approach to achieve real-time imaging system.

In this paper, a design concept of sparse array is applied to total focusing imaging technique. During the process, the elements of a phased array are partially selected according to the theory of effective aperture, in order to reduce the amount of captured data and shorten the processing time of TFM imaging. Simulated imaging with Field II program and experimental verification on two standard phased array specimens are performed, based on three selected arrays which are sparsely as 1/2, 1/4 and 1/8. The results compared with the full array in terms of image quality and computing efficiency, show that the computing efficiency of TFM imaging can get a rise of 40 percent by adopting the sparse array design. Meanwhile, an approximate resolution can be achieved at the lowest array sparse rate of 1/4. Affected by the side-lobes of the ultrasonic beam, contrast of the image may suffer little loss.

This research aims to create a medical drilling device where surgeons can receive the signal and data on the computer screen. We analyze the current signal extract from hall effect-based linear current sensor using a Raspberry Pi engineer board to detect the breakthrough signal.

In wireless location area, the estimation precision of the TOA (the time of arrival) is low because it underutilizes the time-frequency characteristics of the OFDM signal in the OFDM system. To solve this problem, in this paper, a multipath delay estimation model based on frequency domain phase offset (FDP-MDE) is proposed. And then, a grouped joint time delay estimation algorithm (GJ-TDE) is proposed according to features of the LTE system. Simulation results show that the root mean square error (RMSE) of the GJ-TDE algorithm is 5.5034m smaller than that of the synchronization algorithm under the condition that the signal-to-noise ratio (SNR) is 0dB and the sampling interval is 8.

Using a robot playmate to help autistic children is a relatively new idea for therapy. A Robot can repeat the same procedure precisely and does not get tired or bored. Moreover, earlier experiments indicate that autistic children show an open behavior to robots. In this paper, we report about experiments with a humanoid robot in which the robot plays a scenario-role for autistic children and is later also interacting with them. This method origins from puppetry play.

PAC-Bayes bound provides a formal framework for deducing the tightest risk bounds of the classifiers. After formulating the concept space as a Reproducing Kernel Hilbert Space (RKHS), the Markov Chain Monte Carlo (MCMC) sampling algorithm for simulating posterior distributions of the concept space can realize the calculation of PAC-Bayes bound. A major issue is the computational complexity in geometric growth when the dimension of concept space increases. In this paper, we store a portion of the sampling data and calculate its variance, after which the variance minimization method is proposed to investigate the support vectors. Finally, we optimize the support vectors coupled with their weight vectors, and compare the PAC-Bayes bounds. The experimental results of our artificial data sets in low-dimensional spaces show that the optimization is reasonable and effective in practice.

The quality of wheel is especially important for the safety of high speed railway. In this paper, a new ultrasonic array inspection method, the Full Matrix Capture (FMC) and Total Focus Method (TFM) algorithm, has been studied and applied to the high speed railway wheel inspection, especially in the wheel web from the tread. The Total Focusing Method (TFM) which processing image with the data set captured by the approach of Full Matrix Capture (FMC) has been adopted in the field of industrial non-destructive evaluation from medical field since 2005. In this paper, the model is introduced to discuss the principle of FMC and the signal processing method of Hilbert transform which to solve the analytic signal to recover the instantaneous phase information for high resolution imaging. Then the TFM as the reconstruction algorithm is discussed and simulated with different size of aperture to evaluate the performance of TFM. The experiments on the PA block and railway wheel are carried out based on the conventional Phased Array (PA) and FMC method. The inspection result, such as the reliable detection, SNR and image resolution, has been analyzed and discussed according to the different active aperture size based the TFM algorithm, comparison to the conventional phased array UT. It is demonstrated that this TFM algorithm is with good potential for the wheel inspection.

In order to improve the location accuracy in “urban canyons” and indoor environments, a new algorithm based on MIMO-OFDM technique is proposed. Firstly, the error rate of the subcarriers in the OFDM signal is reduced by the transmitter diversity and maximum ratio combining in MIMO; and then the operation of cross-correlation between the received signal and reference signal is processed to obtain the time delay estimation; finally, we use the time delay to realize the location. The simulation results show that the location accuracy in proposed algorithm is improved about 25 m compared to the reference algorithm in the cumulative distribution probability of 90%.

The ESD robustness of gate-driven ESD clamp circuit in a 16-V CMOS process was investigated by the stresses of transmission line pulse (TLP), human-body-model ESD test, and machine-model (MM) ESD test. After TLP stresses of different voltage steps, the same ESD clamp circuit got different secondary breakdown currents (It2). In order to understand such unusual phenomenon, the failure analysis on the TLP-stressed ESD clamp circuits was performed to find the failure mechanism.

In recent years, learning from imbalanced datasets has attracted much attention both in academic and industrial fields. The kernel modification method based on Riemannian metric is an effective method to handle the class imbalance problem. But it cannot deal with the outliers and noise in the imbalanced datasets. However, Fuzzy Support Vector Machine (FSVM) can deal with the outliers and noise in the balanced datasets. In this paper, we combine the FSVM with the kernel modification method based on Riemannian metric together to handle the imbalanced datasets in presence of outliers and noise. Experimental results on four UCI datasets show this method to be effective in improving class prediction accuracy.

A clock skew-compensation and/or duty-cycle correction circuit (CSADC) is indispensably required to maximize the performance of synchronous double edge triggered systems. Most conventional CSADCs adopted a cascade structure that inherits a lower performance property so as to slower the locking procedure, meanwhile the dual loop design results in more power consumption and design complexity. A compact delay-recycled CSADC is proposed in this work. Compared to conventional output feedback tracking CSADC, the proposed circuit achieves a 24.3 times reduction in power, a 2.9 times reduction in lock-in cycles, and a 169.2 times reduction in power-to-bandwidth ratio. Furthermore, even compared to conventional open loop CSADCs, the proposed circuit still achieves at least a 2.81 times reduction in power and a 5.47 times reduction in power-to-bandwidth ratio.

A clock skew-compensation and/or duty-cycle correction circuit (CSADC) is indispensably required to maximize the performance of synchronous double edge triggered systems. Most conventional CSADCs adopted a cascade structure that inherits a lower performance property so as to slower the locking procedure, meanwhile the dual loop design results in more power consumption and design complexity. A range extending delay-recycled CSADC is proposed in this work. Compared to conventional CSADCs, the proposed circuit achieves at least a 2 times extension in bandwidth ratio, a 2.81 times reduction in power, and a 12 times reduction in power-to-bandwidth ratio.

New Event Detection is the task of automatically detecting novel events from a temporally-ordered stream of news story documents. Traditionally, on-line new event detection system determines whether the incoming document contains a new event based on the history of all processed documents. With the improvement of processed documents, the efficiency of on-line new event detection will decrease. In this paper, we apply a time window strategy to new event detection. A group of new incoming documents in the time window are firstly clustered to obtain candidate topics. Next, these candidate topics are compared with the previously identified topics to determine whether a new topic is detected. The first story of the new detected topic, in temporal order, is regarded as a new event. By analyzing the news story documents within the time window in group, the new event detection efficiency is improved. Furthermore, the evaluations show that the time window processing strategy is helpful to improve the accuracy of new event detection.

The double layer open-web roof structure is a type of heat preservation and insulation roof structures, in particular suitable for medium and large span roof structures. However, the construction procedures for traditional open-web roof structures are rather complicated because of the built-in molds in the cavity. On the other hand, passive ventilation is an effective route to enhance the thermal performance of the building envelope. In this paper, an improved open-web roof structure is presented. The steel tube shear connectors are employed to replace the tubby reinforced concrete shear connectors. The adopted shear connectors have small holes in the middle part of their heights. This improvement not only simplifies the construction procedures, but also enhances the thermal performance for open-web roof structures. In order to verify the effectiveness of proposed new open-web roofs, experiments with temperature measurements were performed for a real roof structure. The measured results show that nearly half of the heat flow is drawn out from the cavity due to the induced ventilation. The proposed new open-web roofs have significant improvement and effectiveness on thermal performances, when comparing with traditional open-web roof structures.

The thermal performance of building envelope is a major requirement for generating a comfortable and hygienic interior environment. A roof is a main part of building envelope and provides protection from thermal damage caused by the excessive hot sunshine. In order to improve this thermal roof performance, the use of the air gap in the roof can be considered. This study presents a new type of open-web roofs, which integrates the structural functions and thermal behaviours for building roofs, in particular large span roofs. The conceptual analyses for the proposed open-web roofs on both mechanical properties and thermal performances are carried out. The comparison of material usages and thermal performance between the open-web roof and traditional roof structures is presented for a successfully constructed building. The results from the real case study show the advantages of the new open-web roofs over traditional roofs.

The deployment problem of police patrol cars is more complicated than ordinary set cover problem due to the motility of vehicles and the multiple objectives, such as cost, response capability, deterrent capability and so on. This paper puts forward a new model for the solution of the patrol car deployment problem, which can be used to optimize the number of police cars and determine the patrol path. Road discretization treatment is introduced, simplifying the calculations and reducing the solution difficulty. Two criteria are set up in terms of response capability and deterrent capability for the judgment of a patrol scheme. A modified greedy algorithm is proposed to solve the multi-objective model. As the test show, the modified greedy algorithm can stably produce reliable and cost-effective deployment scheme, which is helpful to find an equilibrium point between patrol cost and patrol effect and improve the level of public security.

This paper presents an analysis of the accuracy of urban railway train's parking together with an elaboration of the system structure and principle of parking control in ATO mode. Through the construction of the model of pneumatic brake unit, the non-linear character of pneumatic brake unit is analyzed, and thereby the strategy of feedback control based on the train's position signal is given. Further on-vehicle tests have substantiated the feasibility of this method; meanwhile, the superiority of parking accuracy after the adoption of this method is also illustrated.

A turbo equalizer can benefit from channel decoding gain by iteratively performing equalization and channel de-coding. Turbo equalization algorithms always need perfect knowledge of the channel response. In this paper, we propose a blind turbo equalization and iterative channel estimator scheme. A fractional super-exponential blind equalizer is used to provide timing recovery and the a priori information that the SISO equalizer and soft-decision directed channel estimator need in the first turbo iteration. In order to reduce complexity, a simplified computation of the equalizer’s extrinsic information for MPSK modulation is used. Simulation results show that the proposed algorithm works effectively for multi-level modulations over severe frequency selective channel.

creep is an extremely harmful phenomenon which can not be neglected at high temperature. As metal is increasingly used in high temperature structure, investigation on its creep at high temperature has actual significance. In this paper, the creep constitutive relationship of LC4 alumunium alloy under constant stress has been investigated by using creep testing data of LC4 aluminium alloy obtained under the same stress level and different temperature as knowledge base, and combining the characteristics of rheological forming of materials and Based on Clustering Radial Basis Function Neural Network (BC-RBFNN). A creep model based on BC-RBFNN has been obtained and compared, which validates that this model has higher precision and generalizing ability.

This paper studies the application of the adaptive control method in lateral control of the autonomous vehicle. Based on the classical PID control, the design scheme accomplishes online parameter self-tuning of PID controller through the single neuron. It designs a non-model and non-parametric adaptive control method. On the basis of the preview follow theory and two degrees of freedom linear model, we establish the vehicle control model and verify the control method of the lateral control with the MATLAB simulation. The simulation results show that the control method can more effectively control the longitudinal stability of a vehicle and have better real-time performance than methods which need a precise mathematical model or need to define a model and its parameters.

The performance of TCP suffers from degradation in satellite network due to long latency, high bit error rate. A design of TCP performance enhancing proxy (PEP) over satellite network is proposed. The network architecture is given based on the PEP, and the key technique is introduced. At last the PEP is validated with OPNET in this paper. The result shows that this PEP can improve the transmission efficiency significantly over satellite network, and the fairness is also good when the bandwidth is shared among multi-connections.

In the application of Particle Swarm Optimization (PSO) for the problem of economic load dispatch (ELD) in power plant There are some problems such as premature convergence of PSO, how to deal with constraints of the problem are need to be addressed. An improved PSO with the constraints partially solved combined with penalty function were introduced in this paper, the improved PSO refined in constraints management, swarm initialization method of PSO with respect to the features of ELD in power plant. Based on above improvements, the improved PSO solved the problem of premature convergence in PSO. In the application to a three and a five units power plant respectively, the simulation results shows that the optimization success rate can reach 100% with the improved PSO, which has the advantage of few parameters to adjust, easy to implement, with high computational efficiency and high accuracy.