Abstract:Once the high-speed train is unstable, the operation safety of the train will be seriously threatened. Before the high-speed train appears to be unstable, it will enter a small hunting divergence state. Therefore, monitoring the train′s evolutionary trend of small hunting can predict the running status of trains. However, the existing literature rarely studies the evolution characteristics of small hunting, therefore, this paper proposes a high-speed train feature extraction framework based on EEMD-SVD-LTSA method, to identify whether the evolution trend is small divergence or small convergence, and then predict the train running status. The verification of online experimental data shows that the framework proposed in this paper can successfully extract the small convergence and small divergence operating characteristics of high-speed trains, and the recognition rate of using LSSVM can reach 100%, so as to predict the running status of high-speed trains in time and ensure the safety of trains.

Abstract:In order to improve road traffic safety, an early warning algorithm based on Multi-Layer Perceptron Neural Network is proposed to solve the problem of low prediction accuracy caused by deterministic parameters in the current traffic safety early warning algorithm. The algorithm is based on artificial neural network (ANN). The relative distance, relative speed, driver′s driving style, the acceleration of preceding vehicle, the acceleration of following vehicle and the speed of following vehicle are used as the input of the system, and the warning level of traffic safety is the output of the system. The prediction value of traffic safety early warning level is obtained by training with sample data, and compared with the two early warning models of the traditional collision time algorithm and the stop distance algorithm. The experimental results show that the multilayer perceptron neural network early warning algorithm is superior to the traditional warning algorithm in the effectiveness and accuracy of early warning.

Abstract:Due to the difficulty of the frequency coordination in high voltage interconnection network and the problems of poor tracking performance, difficult parameter optimization and poor optimization effect, a closed-loop feedback automatic generation control based on simulated annealing and particle swarm optimization algorithm is proposed. The improved optimization algorithm combines the advantages of simulated annealing and particle swarm optimization, and optimizes the key parameters of the controller to minimize the value of area control error, so that the system can recover stably quickly. The simulation results of MATLAB/Simulink verify that the closed loop feedback AGC controller with improved parameters can restore the stability of the system quickly and achieve the control effect of frequency coordination when the tie-line of the four-machine two-area model fails.

Abstract:Focus on the basis of the mathematical model for the electronic throttle, a sliding mode control strategy based on the nonlinear terminal sliding surface is designed on the basis of the nonlinear characteristics of the electronic throttle system which used in hybrid electric vehicle and the disturbance in the control system. At the same time, in order to realize the electronic throttle of the hybrid power vehicle under the complicated driving condition. The non measurable variable and uncertain disturbance of the gate control system are accurately estimated. The nonlinear expansion state observer is designed. The experimental verification of the angle following of the electronic throttle under three different operating conditions is tested. The experiment proves that dual loop sliding mode control strategy is used to improve the convergence speed of the valve system and the phenomenon of no-overshoot. It meets the target that hybrid electronic vehicle′s electronic throttle has high precision angle following performance.

Abstract:Various new modulation modes have been introduced into the signals with different frequency bands of Galileo satellite navigation system, which is a typical representative of the new generation satellite navigation system. The limited number of visible Galileo satellites has brought some difficulties to further exploration. In this paper, a Galileo+GPS satellite navigation signal simulator is designed and implemented based on the architecture of "PC+FPGA+DSP". The algorithm of FPGA in satellite navigation signal simulator is mainly studied, doppler frequency shift simulation, subcarrier generation, offset carrier modulation and pseudo range signal generation are completed, key technology of signal alignment of all modules is solved,and the intermediate frequency signals of Galileo E1 OS and GPS L1 C/A are generated. The hardware receiver is used to receive and verify the generated signal, the results of experiments show that the receiver can correctly capture, track and fix positions, and the location coordinates are consistent with the preset coordinates, which prove the correctness of the simulation signal.

Abstract:Nowadays, the network era is full of tamper-evident images. At present, the detection method such as local invariant characterization and the Harris corner algorithm has low accuracy in copying and pasting tampering. In this paper, a complete grayscale partial image is obtained by color image segmentation, color space edged extraction, and image grayscale. By extracting, marking, matching and normalizing the feature vector set in different image blocks, the feature vector is successfully matched after the Euclidean distance reaches a certain threshold, that is, the image has the trace of copying and pasting tampering. Finally, three different types of photo simulation tests are selected, which shows that the algorithm can effectively improve the detection success rate and detection rate of copying and pasting tampering images.

Abstract:As a vital part of the convolutional neural network model, the pooling model has the functions of dimension reduction and generalization of the model. In order to further improve the accuracy of the convolutional neural network model and optimize the learning performance of the model, this paper proposes an improved pooling model based on maximum pooling and average pooling, and the global handwritten digital datasets MNIST and CIFAR-10 data. The effectiveness of the improved pooling model was verified on the two dataset. Comparing with the common pooling model, it is found that the learning performance of the convolutional neural network with improved pooling model is better. In one iteration, the error rate decreases by 4.28% on the MNIST and decreases by 2.15% on CIFAR-10 datasets.

Abstract:Biometric recognition technology has higher reliability than traditional cryptography. As one of the important research directions of biometrics, voiceprint recognition method has more research significance to study more accurate voiceprint recognition methods. With the development of deep learning, the application of deep learning in voiceprint recognition technology has become the focus of research in voiceprint recognition field. In this paper, a speaker recognition method based on deep neural network and beyond triplet loss is proposed. The model extracts the acoustic characteristics of MFCC through Mel-frequency cepstral coefficients, and extracts the voiceprint characteristics of the speaker from the MFCC acoustic characteristics, and then carries out the beyond triplet loss model training. Experimental results show that DNN-BTL algorithm has better recognition effect in speaker recognition field than Gussian mixture model-hidden Markov model.

Abstract:When synthetic aperture radar (SAR) imagines a moving target with high accuracy, the radial velocity of the target causes the range migration and the azimuth velocity causes the azimuth defocus，which will lead to the decrease of the signal-to-noise ratio and detection probability of the target. In this paper, a method of the series connection of range migration correction filter bank and azimuth matched filter bank is proposed to optimize the two-dimensional matching imaging of the multi-channel clutter suppressed data, which can reduce the residual clutter power and improve the focusing effect of moving targets. Then the CFAR detection is performed using adaptive threshold after each filter is imaged, and the output of multiple test results can be judged and fused. The results of simulation and measured data processing demonstrate the effectiveness of the method.

Abstract:In order to improve the sub-filter filtering accuracy and optimize the information fusion algorithm, an adaptive Kalman filtering algorithm based on online adjustment factor is proposed. First of all, discuss the theoretical basis of using Kalman filter technology, and design SINS/GPS tight integrated navigation system.An improved adaptive Kalman filter algorithm is proposed. By constructing an adaptive parameter factor and using the ratio of the measured noise covariance matrix to the adaptive parameters, the online correction measurement noise covariance matrix is realized. Through the result of MATLAB simulation, its position error and speed error are significantly reduced, compared with traditional tightly integrated navigation systems based on standard Kalman filter algorithm, so as to improve the positioning accuracy of the integrated navigation system and optimizing information fusion algorithm.

Abstract:In order to design a support vector machine (SVM) classification model with better performance, the parameters and sample feature subsets that affect its classification performance are optimized, and the support vector machine theory and gravitational search algorithm (GSA) were studied. The optimal combination solution can be obtained by simultaneously optimizing the relevant parameters and effective sample feature subsets which affect the classification performance of SVM. Its effectiveness is compared and verified by experiments. The experimental results show that the proposed BGSA-SVM classification model can effectively improve the classification performance of support vector machines, which can be further extended to engineering applications.

Abstract:Multi-h CPM has great potential advantages in future space telemetry due to its high bandwidth efficiency and power efficiency, but the high complexity of demodulation technology limits its wide application. In order to simplify the demodulation structure under the premise of less performance loss, after a brief introduction of the characteristics and development status of Multi-h CPM system, the basic structure of the receiver for multi-h CPM signal is emphasized. Carrier synchronization, timing synchronization, modulation index synchronization and sequence are summarized in detail. The origin, development process and theoretical basis of detection technology are analyzed. The advantages and disadvantages of each method are analyzed. Finally, the further research direction of Multi-h CPM system is proposed.

Abstract:Considering the nonstationarity and information redundancy of living tree stem moisture signals in time domain, an approach of MP decomposition and reconstruction of living tree stem moisture signals based on Gabor atoms was presented. The experimental results showed that living tree stem moisture signals can be represented sparsely by Gabor atom library. The front of atoms reflected the main features of signal and the back of atoms reflected the subtle features of signal. The more the number of atoms was, the more the sparse signal can better represent the features of original time-domain signal. Compared with the original signal in time domain, the sparse signal had many advantages. Firstly, the length of sparse signal was reduced significantly. Secondly, the sparse signal can avoid information redundancy. So the approach of representing signal sparsely can achieve the purpose of data compression and save physical space to store a large amount of data. Under the condition of Gabor atom library being redundant, original time-domain signal can be constructed with high quality from the sparse signal. And reconstructive errors at the main feature points were larger than it at the subtle feature points.

Abstract:At present, the detection of the yield strength of cold-rolled strip steel mainly depends on the damage detection, which greatly increases the detection cost. In this paper, the BP neural network is introduced into the yield strength prediction of cold rolled strip steel based on pulse eddy current. Firstly, the time domain and frequency domain characteristics of the pulse eddy current response signal are extracted. The stability of the characteristics of each pulse eddy current signal is analyzed, and the BP neural network model for signal characteristics and material yield strength is established, and the yield strength of the material is predicted using the established model. Experiments show that yield strength prediction error is 6% or less using the BP neural network to predict the yield strength of cold-rolled strip steel. This method has certain practical value for reducing the detection cost of industrial production and improving the detection efficiency.

Abstract:Aiming at estimating the instantaneous frequency of FM signals in the complexα-Gaussian mixed noise environment, a new method called the relative random sample consensus (RANSAC) was developed on the basis of directional S transform. The new algorithm combined the searching strategy of the original RANSAC algorithm tightly with FM signal’s physical characteristics altogether to propose a new function that can ensure the mixed noises be depressed and the instantaneous frequency trajectory be accurately tracked simultaneously. With the quadratic smoothing ability of the directional S transform, this new algorithm managed to recover phase information of the target signals in mixed noise environment of -7 dB the signal to noise ration. Numerous simulations have proved the algorithm’s effectiveness in α-Gaussian mixed noise environment.

Abstract:A two wire infrared photoelectric sensor frequency measuring system based on NRF24L01 wireless communication is designed. In order to reduce the number of wires used in three-wire sensor and improve the stability of signal, a two-wire photoelectric sensor is designed. In the design process, the improved two-wire sensor circuit is designed after the in-depth study of the three-wire sensor circuit, and then the signal processing circuit is designed to convert the pulse signal of the sensor into frequency signal, which is transmitted to the wireless transmission module through RS-485 bus, and finally transmitted to wireless communication. The receiving module finally displays the measured data through the host computer software. It is proved by experiments that it can collect signals in a stable and accurate manner. After comparing the data obtained by the two sensors, it is concluded that the data measured by the two-wire sensor is more stable, less fluctuation, and only two lines are needed to achieve the purpose of the experiment.

Abstract:Aiming at the limitation of the existing hand gesture recognition methods, considering the sparse multi-channel feature of MYO armband, this paper proposes a hand gesture recognition scheme which combines surface electromyogram (sEMG) and acceleration (ACC) information. Firstly, 5 participants wore MYO armbands, and simultaneously collected 8 groups of sEMG and ACC data of different gestures. Secondly, a new data preprocessing method is proposed, which will segment multi-channel sEMG and ACC data through high-energy sliding windows to obtain short-term effective active parts. Finally, a parallel sequence LSTM network is designed to extract and fuse the features of the two types of data. The accuracy of gesture recognition is 96.87%. The results show that the proposed scheme is simple and feasible.

Abstract:Synthetic aperture radar (SAR) utilizes relative motion between the radar platforms and targets to form long aperture, which is designed for static targets and finally obtain high resolution images. There are moving targets in the ground stationary scene. Since the motion parameters are unknown, the moving targets are hardly focused in SAR image. On the other hand, inverse synthetic aperture radar (ISAR) technology has been known that moving targets can be focused depending on the motion of targets and static radar platform. The work of this paper combines the multichannel SAR-GMTI technology and come up with the image of non-cooperative moving target based on clutter suppression interference (CSI) and ISAR. And meanwhile moving targets are detected after CSI processing before ISAR processing which picks the main energy area of the targets. At the end of the paper, simulation results prove the effectiveness of the chain and success of moving targets defocus.

Abstract:In order to solve the difficulty in predicting the communication distance of ships that traveling in certain sea areas and based upon the analyzed results of Ultrashort Wave’s features, different types of ultrashort waves’ wireless communication models were established corresponding to different situations, including large-scale fading free space path loss model，log-distance path loss model, log-normal shadowing path loss model, as well as small-scale fading model’s Doppler principle effect. MATLAB was used to stimulate different situations and calculate different models ‘effect on radio station’s power path loss. Building on this base, methods of calculating link budget were brought up and methods of calculating the communication distances between ships‘ultrashort waves and airplanes’ radio stations were created. Strategies of deciding link budget were improved.

Abstract:In order to meet the requirement of wireless networking of curing barn controller in bulk curing barn system, a design scheme of curing barn controller based on ZigBee wireless communication is proposed. Using STM32 as the main control chip. Backup battery power supply mode can ensure automatic data storage when the system is powered off. Serial port touch screen makes the man-machine interface more friendly, intuitive and easy to operate. Serial port ZigBee wireless module meets the network design requirements of star network in roasting barn group. The stepless speed regulation of the recirculating fan driven by the frequency converter improves the precision temperature control and energy saving of the roasting. Finally, The experimental results show that the controller is stable and reliable, and effectively improves the management level of the roasting barn group.

Abstract:Wearable computing is widely used in the field of monitoring human health indicators, and monitoring energy consumption through acceleration sensors has broad application prospects. Accurate monitoring of college students′ exercise energy consumption is of great significance to help college students formulate scientific exercise and fitness plans and improve their physical health. This paper designs and implements the college students′ energy consumption monitoring system, and proposes a motion energy consumption monitoring algorithm based on geometric mean (GM), which is verified by experiments. The system obtains the motion parameters through the three-dimensional acceleration sensor, uses the GM algorithm to obtain the sports energy consumption data, and compares and analyzes the standard data of different sports types, so as to obtain the sports differences and suggestions for improvement of college students, and improve the standard and safety of college students′ sports.

Abstract:The complex indoor environment brings non-visual range error and multipath interference to the localization system, How to eliminate or reduce error becomes a hot spot in research to UWB indoor localization. A UWB indoor localization method based on IA-BP neural network is proposed, Which is that the error of training by BP neural network is as the antigen of calculating affinity to immune algorithm,the optimal weight and threshold of BP neural network are obtained through immune algorithm, so as to avoid the problem of slow convergence and getting into easily local optimal value of BP neural network, then get the minimum localization error. Simulation results show that the maximum error from training to 100 samples by IA-BP neural network was not more than 0.02, In the positioning scene composed of three anchors, the simulation output trajectory of the undetermined bit node was basically consistent with the actual motion trajectory.

Abstract:Aiming at the defects existing in the current wired monitoring system for snubbing operation, including complicated wiring, poor anti-interference and difficult maintenance, in this paper, we design a monitoring system of snubbing operation based on ZigBee technology. The system comprises a data base station and several multi-parameter terminal collection nodes, which can safely and effectively collect various data during the operation of snubbing operation equipment under wireless conditions. The numerical simulation results show that the packet loss rate is 0 when the communication distance is less than 150 m, and the packet loss rate is 0.4% when the communication distance reaches 200 m, which shows the system can fully meet the requirements of 50 m transmission distance in practical applications. The practical application results demonstrate that the system overcomes the constraints of traditional physical wiring, facilitates the efficient collection of data information, realizes wireless monitoring of snubbing operation and has wide application prospects in the snubbing operation system.

Abstract:In order to improve the ability to monitor the emissions of atmospheric pollutants, and for the shortage of the traditional atmospheric environmental monitoring stations, such as the lack of distribution, the limited monitoring scope and the high cost, a design of multi-parameter environment wireless monitoring node based on STM32F103 microcontroller and LoRa is proposed. The ADC and the serial port with DMA function inside of the STM32F103 is used to collecte the data of 6 electrochemical gas sensors, GPS location data, PM2.5/PM10 particle concentration data, the temperature and humidity. The collected data is transmitted to the LoRa gateway through the LoRa wireless module and displayed on the terminal. Test results show that the node can monitor 6 kinds of gas concentration, PM2.5/PM10 particle concentration, temperature and humidity and travel path in real time, and the error is within the allowable range. The node combined with UAV or unmanned vehicle has the advantages of wide monitoring range, high detection efficiency, high mobility and low cost; and it provides a new technical means for environmental monitoring in real time.

Abstract:Research on probes for transient electromagnetic detection technology. Firstly, the principle of transient electromagnetic technology is introduced, and the advantages and disadvantages of several kinds of focusing probes are analyzed. Then, a new type of transient electromagnetic focusing probe structure is proposed according to the design requirements, which solves the shortcomings of transient electromagnetic probes, such as non-focusing, low energy and heavy equipment. Finally, The finite element simulation is used to verify the focusing effect of the new focusing probe, and compared with the traditional coil simulation results, the results show that the new focusing probe focuses the main energy of the magnetic field at a radius of less than 0.3 m, which is 40% smaller than the traditional coil magnetic field distribution, and the magnetic induction intensity. B=1.499 8×10-5, which is 19.3% higher than the traditional coil induction intensity, and the focus probe has a larger gradient of magnetic induction intensity distribution than the conventional coil. It provides an effective reference value for optimizing the design of transient electromagnetic technology probes and accurately measuring the corrosion of metal pipes. It also provides a basis for the extended application of this technology in pressure pipes and heat pipes.

Abstract:In view of the feature extraction for analog circuit fault diagnosis, this paper introduces the convolutional neural network into this study field, and presents an analog circuit fault diagnosis method based on deep temporal convolutional neural network (TCN). Comparison experiments on a 4th-order Butterworth lowpass filter with a group of TCN with various depth prove that the model of deep TCN is effective in feature extraction of circuit fault. Moreover, comparison experiments with a group of models including TCN, SAE-SOFTMAX, deep belief network and long short-term memory network prove that the TCN based fault diagnosis method is more effective to extract features closed to essence of data, and capable to achieve a better accuracy in analog circuit fault diagnosis.

Abstract:Vacuum thermal environment test is an indispensable experimental project in the development of spacecraft. A large number of test cables need to be configured during the experiment, so the fault detection of test cables is of great significance. In order to effectively improve the positioning accuracy of test cable faults, this paper studies the application of wavelet transform and fractional Fourier transform in cable fault test. The transmit signals of the TDR and FDR cable fault detection methods are subjected to noise addition processing, and the corresponding time-frequency analysis methods are respectively used for denoising. The simulation results show that the wavelet transform and the fractional Fourier transform have good denoising effect, which is of great significance to improve the fault location accuracy of the test cable.

Abstract:Aiming at the fault of electronic communication equipment, which is difficult to reproduce and lacks test data and prior knowledge, a fault diagnosis method based on simulation model is proposed. First, the simulation model of the fault system is established, and the model is simplified as far as possible. Then the model is used to simulate the hypothetical fault causes one by one, and to observe the output of each key node at the same time. Finally, find the most consistent or similar results with the existing test data, and improve the simulation model. On this basis, the failure mechanism is analyzed, and the fault diagnosis is completed. This method can be used as a supplementary means of fault tree analysis which is a forward logic reasoning method and is opposite to the fault tree analysis method, and plays an important role in the fault diagnosis of a voice radio.

Abstract:Valid roller fault diagnose plays an important role in improving working efficiency and intelligent plant. In view of complex industrial environment and numerous noise types, first, a difference method is used to eliminate the influence of time trend in audio sequence data, and to extract the characteristics of the roller audio sequence. Secondly, K-Means and spectral clustering algorithms are used to have cluster analysis and roller faults identification. In order to evaluate advantages and disadvantages of proposed clustering model, an average ratio of sub-sequence same labels from an audio sequence is proposed to achieve the above aim. Experimental results show that local diagnostic accuracy can be improved by dynamic selection of parameter values. Roller fault can be effectively identified and diagnosed by two proposed clustering algorithms, but spectral clustering algorithm is superior to K-Means algorithm. By use of the proposed methods, one can see that the efficiency of coal preparation is improved, number of unplanned outages is reduced, and good economic benefits are also produced.