Abstract:The control strategy based on surface electromyography (sEMG) is an important basis for humancomputer interaction and has important significance in human control of peripheral devices. The myoelectric control strategy recognizes the body′s motion intention through sEMG, and then converts it into control commands to achieve a precise and stable control of peripheral devices. The use of sEMG has overcome the limitations of traditional input devices in terms of portability, operating space and special groups. This paper discussed the development of the myoelectric control strategy, and compared the differences between sEMG and the Electroencephalogram (EEG) as the input of the control system. The application of the sEMGbased control strategy in rehabilitation and orthopedics was expounded, and the future improvement of these applications was pointed out. This paper analyzed the technical problems related to the acquisition of sEMG and the improvement of the control effect and robustness, which hindered the development of myoelectric control strategy based on pattern recognition. Finally, the paper listed some feasible improvement directions for sEMG control systems.

Abstract:In the traditional denoising method of the seismic data, it has poor effect because it depends on prior data too much. In order to suppress seismic data noise more effectively, a joint denoising method based on the respective characteristics of BP network and SVD algorithm is proposed. The proposed method has carried on the thorough discussion to the BP network structure and the experimental methods. The determined experimental method is as follows: firstly, the noisy seismic data are separated by BP network, and then the output noise is reconstructed by SVD algorithm, which is the output noise of the joint algorithm. Finally, the denoised seismic data can be obtained by subtracting the noisy seismic data from the output noise. Experiments on prestack and poststack seismic data demonstrate that the proposed method is feasible and effective. Compared with the traditional denoising algorithm, the MSE (mean square error) of the proposed method is lower and the SNR (signaltonoise ratio) is higher, which shows that it has better denoising effect on actual seismic data.

Abstract:It’s difficult to extract the rich fault information from vibration signal by the traditional feature extraction methods of time domain, frequency domain and time frequency domain parameter. In order to solve this problem, a multiscale permutation entropy is proposed to extract fault features and combine the improved multiclass relevant vector machine to fault diagnosis. Since the kernel parameters of the multiclass relevant vector machine do not have the adaptive ability, it has the great influence on the accuracy of fault diagnosis. The multiclass relevance vector machine is improved by a grasshopper optimization algorithm to realize the adaptive fault diagnosis. The experimental data from the University of Western Reserve in the United States show that the proposed optimized fault diagnosis model can realize the fault diagnosis of different types and the identification of different fault degrees. Compared with the fault diagnosis model of particle swarm optimization optimizes multiclass relevant vector machine, the accuracy of proposed fault diagnosis model is 100%.

Abstract:Radio fingerprint databased (RFDB) is the prerequisite for WLAN based indoor positioning of the utmost importance. In this paper, a RFDB surveying method is proposed without dependence on additional information such as external hardware or indoor maps. In the method, the inertial data and the fingerprinting data are fused through graph optimization (GO). Then the indoor positions along with the RFDB can be estimated with greatest consistency of all types of information. The acquired RFDB can then be adopted for WLAN based indoor positioning. Experiments are carried out to evaluate the proposed method. It is proved that compared with pin point method, the proposed method has a mean positioning error decrease of about 1.0 m and compared with the Gaussian process based method with sparse reference points, the number is 0.8 m.

Abstract:Mechanical equipment fault detection is of great significance in industrial applications. The traditional fault diagnosis method based on vibration signal processing and analysis relies on rich professional knowledge and artificial experience, and it is difficult to achieve accurate feature extraction and fault diagnosis. In this paper, the deep learning method can be used to automatically learn the characteristics of deep features from the data. A qualitative and quantitative diagnosis method for rolling bearing faults based on improved convolution deep belief network is proposed. First, in order to provide better shallow inputs, the original vibration signal is converted to the frequency domain signal by the fast Fourier transform. Secondly, in the process of model training, the Adam optimizer is introduced to speed up the model training and improve the convergence speed of the model. Finally, in order to make full use of the characterization capabilities of each layer, the model structure is optimized to come up with a multilayer feature and fusion learning structure is proposed to improve the generalization ability of the model. The experimental results show that the proposed improved model has better diagnostic accuracy than the traditional stack autoencoder (SAE), artificial neural network (ANN), deep belief network (DBN) and standard convolution deep belief network (CDBN). It has better diagnostic accuracy and effectively realizes qualitative and quantitative diagnosis of bearing faults.

Abstract:For spiral bevel gears widely used in various fields of industrial engineering, the vibration signal is greatly disturbed by environmental noise. When the fault occurs, the signal exhibits nonlinear, nonstationary characteristics, the fault feature information is weak, the fault feature extraction is difficult, and the diagnostic efficiency is low. Therefore, a spiral bevel gear state recognition method based on MPELPP and ELM is proposed. Firstly, construct multiscale entropy values as the original highdimensional feature vectors, then use LPP to obtain the optimal lowdimensional sensitive feature vectors by reducing the original highdimensional feature vectors, which can mine and preserve the nonlinear structural features of the original highdimensional features. The obtained sensitive feature quantity is input into the ELM for recognition diagnosis. The method is applied to the diagnosis of four kinds of fault state spiral bevel gears under three kinds of speeds, and compared with MPEPCAELM and MPEELM. The results prove the accuracy and superiority of the proposed method.

Abstract:For the problem of low positioning accuracy and low antinoise performance of traditional islanding detection methods, this paper proposes an islanding detection algorithm based on energy operator and improved maximal overlap discrete wavelet transform (MODWT). In order to effectively solve the boundary effect problem of MODWT algorithm, based on the traditional MODWT algorithm, the original wavelet coefficients are updated by the circular boundary coefficients. The energy is analyzed through the sliding window. Then the algorithm is applied to the island detection, and the detail coefficient and the approximate coefficient energy obtained by the processing are used to analyze the voltage disturbance signal characteristics of the common coupling point in the island state. The simulation results show that the algorithm can accurately detect the start time and amplitude variation of the voltage disturbance signal. And it is not affected by mother wavelet and decomposition layers and it has strong antinoise ability and small delay deviation in actual signal detection.

Abstract:Problem: A micromotion period estimation based on the Viterbi algorithm and the timefrequency squared difference sequence is proposed. Procedure and method: The micromotion states of the radar target is dependent on the dynamics, structure of the target, and they are usually different for different targets. As a result, the micromotion states reflect some important information of the target and can be used to aid the target detection and recognition, which has a great importance. In the research of the micromotion features, space targets such as missile are hot points. Numerous micromotion feature extraction methods for space targets are published. Usually, the motion of the space target is compound of the translation and the micromotion. The implication of the translation on the micromotion feature extraction should be considered. This work focuses on the space target with compound motion. A micromotion period estimation based on the Viterbi algorithm and the timefrequency squared difference sequence is proposed, in which the Viterbi algorithm can provide robust instantaneous Doppler estimation in low signaltonoise ratio (SNR) and the timefrequency squared difference sequence can remove the effect of the translation. Thus, the proposed method has a good performance. Results: Numerical results prove the effectiveness of the proposed method.

Abstract:Aiming at the problem of fault diagnosis of analog circuits with tolerance, a subband polymorphic spectra (SPS) approach of fault feature extraction is proposed in this paper, which is based on empirical mode decomposition (EMD). Firstly, the secondorder Volterra kernel sequences of the circuit under test are calculated. Then the Volterra sequences are decomposed based on EMD and the intrinsic mode functions (IMFs) are obtained. By calculating cepstrum (CS) and Hiltert spectrum (HS) from IMFs, the polymorphic digital fault features in timefrequency domain are extracted, the soft and nonlinear faults in analog circuits with tolerance are isolated and the fault diagnosis of analog circuit is achieved. The experiments results show that the proposed method in the paper can solve the fault aliasing problem effectively and improve the capability to locate and isolate fault components.

Abstract:In order to further improve the accuracy of facial expression recognition algorithm, this paper proposes a facial expression recognition method that combines the double coding local binary pattern (DCLBP) operator and the histogram of oriented absolute gradient (HOAG) operator. First, the method uses DCLBP operator to extract the local texture features of the face image and the HOAG operator to extract the local shape features of the face image. Then, the two extracted correlation features are fused by the canonical correlation analysis (CCA). Finally, using the support vector machine (SVM) to classify facial expression. Compared with the single feature recognition method and the cascade feature recognition method, the experimental results show that the proposed method achieves better recognition results, and the recognition rate on the CohnKanade (CK) and JAFFE data sets achieves 100% and 9905% respectively, the comparison with other related methods also verified its effectiveness.

Abstract:Aiming at the problems of nonlinearity and nonstationarity of analog circuits, and the ambiguity and randomness of diagnostics caused by the tolerance of circuit components, an analog circuit fault diagnosis method combining local mean decomposition (LMD) cloud model feature extraction and ABCLSSVM classifier is proposed. First, the LMD algorithm is used to decompose the initial fault signal, and the cloud digital feature of the decomposed signal is calculated by the cloud inverse generator, and the obtained cloud digital feature is constructed as a fault feature vector. Then, a part of the fault feature vector is input as a test sample into the LSSVM optimized by the artificial bee colony (ABC) algorithm, and the circuit faults are classified and identified to obtain the classification accuracy of each fault. Two international benchmark circuits, CTSV and Sallen_Key, are used as verification objects. The results show that the fault features extracted by this method can well reflect the fault status information of the circuit, and the fault diagnosis accuracy of the proposed method reaches 99%.

Abstract:Existing fog detection models have defects and the optimization of deep neural network is difficult. Based on imitating human cognitive model, transfer learning and closedloop control theory, this paper explores an intelligent cognitive method of fog level based on dynamic interleaved group convolution. Firstly, an interleaved group convolution layer is constructed to reduce the channel redundancy of convolution operation. Secondly, a discriminability measure index and a cognitive decision information system are constructed to obtain the spatial data structure of differentiated and simplified features of fog images. Thirdly, the deep stochastic conguration networks classifier is designed and the classification criterion with strong generalization ability is constructed. Finally, based on the generalized error and entropy theory, the cognitive model that mimes human repeatedly deliberates and compares to evaluate the credibility of the cognitive results of fog grades in real time. Based on the transfer learning mechanism, the selfoptimization reconstruction of the multilevel differentiated feature space of fog images and their classification criteria are realized, and the lowcredibility fog images are rerecognized again. The average recognition rate of 15,000 fog images is 9598%. The experimental results show that compared with other algorithms, this method enhances the generalization ability and improves the cognitive accuracy of the model.

Abstract:This paper put forward an adaptive threshold image denoising algorithm based on fitting diffusion to deal with the problem of texture loss and edge degradation. The algorithm will first improve the diffusion coefficient of the diffusion equation, establish fitting diffusion coefficient, avoid filtering incomplete due to the rapid convergence and the problem of image excessive smoothing. Then, the threshold will be designed and improved, and it will be automatically controlled by the maximal image gray value and iterative times, which can keep the image edge and detail features. Last, the designed algorithm will be simulated. The experimental result shows that the proposed algorithm can enhance the performance of denoising and protection of edge and detail information of texture, the peak Signal to Noise Ratio is promoted drastically. The new algorithm has excellent performance and is beneficial to practical application.

Abstract:Aiming at the problem of low precision and stability of traditional phase shifting methods, a high precision adaptive digital phase shifting method is proposed. The method is based on stochastic approximation algorithm and active RC phase shifting circuit to construct a closedloop control system. Configuring a highprecision digital potentiometer via FPGA, then detect the actual output phase and use it as an observation of the stochastic approximation algorithm. After many iterations, the potentiometer resistance is calculated and the output phase value gradually approaches the set phase value. The experimental results of phase shift accuracy and stability show that within the adjustable range of the digital potentiometer, the phase shift error does not exceed ±03°, and the phase shift error does not exceed ±025° for at least 11 hours. The phase shift error does not exceed ±03° during a rapid change in ambient temperature of approximately 20℃. Practical application experiments show that the method has good robustness and practicability. This adaptive digital phase shift provides a new method for high precision phase shifting.

Abstract:Strong supervisory recognition algorithm requires a large amount of annotation information and consumes a lot of manpower and material resources. In order to solve the above problems and meet the practical requirements, two image recognition methods based on weak supervisory information are proposed for finegrained vision classification (FGVC). One is the combination of ResNet and Inception network, which improves the ability of capturing finegrained features by optimizing the network structure of convolutional neural network. The other is to improve the Bilinear CNN model, feature extractor selects Inceptionv3 module and Inceptionv4 module proposed by Google, and finally gathers different local features for classification. The experimental results on CUB200-2011 and Stanford Cars finegrained image datasets show that the proposed method achieves classification accuracy of 883% and 942% on the two data sets, and achieves better classification performance.

Abstract:The multi-layer Morphin algorithm extends the prediction range of the unknown environment and overcomes the shortcomings of the traditional Morphin algorithm′s invisible search trajectory. However, the number of search arcs generated by each search node is fixed, and the time spent searching and evaluating varies with the number of search layers. The increase is exponentially increasing. Aiming at this problem, a variable search arc of Morphin (VSAMorphin) is proposed. The number of search arcs generated by each search node is adjusted so that it is no longer fixed, but decreases as the number of layers increases, thereby shortening the search and evaluation time. The simulation results of MATLAB show that the multilayer VSAMorphin algorithm is basically consistent with the path planned by the multilayer Morphin algorithm, but the running time is relatively less, thus verifying the validity and correctness of the multilayer VSAMorphin algorithm.

Abstract:Aiming at the bifurcation and chaos of Buck-Boost matrix converter (BBMC) inverter stage under the action of pulse input voltage, a piecewise delay feedback control method is proposed to study the stability domain of its control parameters. The discrete iterative mathematical model of the inverter stage is established. The control characteristics of the inverter stage under the piecewise delay feedback control method are analyzed. The function relationship between the control parameters k1 and k2 during the stable operation of the system is determined. The stability domain range is determined. Finally, the correctness of the above theoretical analysis is verified by simulation. The results show that the piecewise delay feedback control can effectively suppress the chaotic phenomenon generated by the BBMC inverter stage under the pulse input voltage, which is of great significance for ensuring the stable operation of the BBMC inverter stage.

Abstract:Aiming at the problem of bottle recycling counting, an intelligent and environmentally friendly device for bottle recycling is designed. This device counts through the sensor. During the process of placing the bottle, the vibration of the flip plate structure will occur, which will affect the accuracy of the bottle count. In order to reduce the vibration of the flip plate and improve the accuracy of counting, spring and damper were added into the counting device to establish the vibration mechanics model of the device. Then, MATLAB was used to carry out simulation analysis on the model to select and optimize the spring elastic coefficient k value and damping coefficient c value of the damper. Through the physical model test, the selected elastic coefficient k and damping coefficient c were further verified to meet the expectations. The test results show that the k and c values selected by MATLAB simulation analysis can achieve the ideal damping effect. The method provides reference value for the count of bottle recovery in environmental protection industry.

Abstract:Chinese character recognition is an important research content in the field of artificial intelligence and pattern recognition. Existing research still has problems such as difficulty in parameter adjustment, small number of training samples, and inability to identify all common characters. Aiming at these problems, we propose a Chinese character recognition method based on character encoding and convolutional neural network. First, we obtain all the character information by querying the font database, which are encoded and outputted by using UTF8 encoding method and various font encoding files to generate character images. Further, we apply various of image processing to obtain the new character image dataset. Then, we propose a deep convolutional neural network for Chinese character recognition. In the training procedure, data augmentation, batch normalization, RMSProp optimization are optimized, regularization and dropout are used to prevent overfitting for optimization. The experimental results show that the proposed method is simple yet effective, the recognition accuracy rate for Chinese characters is 9808%. Compared with Alexnet and LeNet5, we obtain a significant improvement by 937% and 2114%. A neural network with high recognition rate, strong feature extraction ability and generalization ability is realized.

Abstract:In order to study the aerodynamic characteristics of the model under high angle of attack ventilation, a special ventilation force measurement wind tunnel test device was developed. The device consists of a sixcomponent balance, two bellows and other parts. The bellows realize the ventilation and air tightness requirements of the model, and the balance realizes the force measurement requirements of the model. ANSYS software is used to simulate and analyze the characteristics of the balance and bellows. The detailed structure and theoretical results of bellows interference on the balance are obtained. The working formulas of the balance and the force measuring device are obtained by static calibration. The results show that the finite element analysis is consistent with the measured results. The measurement accuracy of the force measuring device is further verified by wind tunnel test. The results show that the force measuring device meets the design requirements and meets the requirements of this type of wind tunnel test.

Abstract:Accurate measurement of the acceleration sensor output value is a necessary prerequisite for the prediction of relevant data. In order to compensate the output error of accelerometer sensor caused by manufacturing process and measurement environmental impact and accurately predict the output value of accelerometer sensor, an acceleration sensor error compensation and numerical prediction method based on adaptive singular spectrum and neural network is proposed. Firstly, the cause of the output error of the acceleration sensor is analyzed. Secondly, an adaptive singular spectral method is proposed for the acceleration sensor error compensation according to the singular entropy order determination denoising method. Finally, the radial basis function (RBF) neural network is selected as the numerical prediction method for the acceleration sensor output data, and the particle swarm optimization algorithm is used to optimize the initial parameters of the RBF neural network. The experimental results show that the adaptive singular spectral method can effectively compensate the output error of the acceleration sensor, and different adaptive parameters can be selected to meet different error requirements, and the RBF neural network optimized by the particle swarm optimization algorithm can effectively predict the output value of the acceleration sensor.

Abstract:This paper proposes a synthetic aperture radar (SAR) target recognition method based on combination of global and local representations. Sparse representation over the global dictionary could effectively compares the relative description capabilities of different classes for the test sample. However, local dictionarybased sparse representation reflects the absolute description ability of each category on the test sample. Therefore, the two representations could complement each other to provide more information for correct decisions. The decision value vectors (i.e., reconstruction errors) from the global and local representations are fused by DempsterShafer (DS) evidence theory for robust target recognition. Experiments are conducted on public moving and stationary target acquisition and recognition (MSTAR) dataset to be compared with other SAR target recognition methods. The experimental results show the effectiveness of the proposed method.

Abstract:In order to solve the problems of node selfhealing when node failure occurs in mobile wireless sensor networks, such as large changes in network topology and limited data transmission performance, a selfhealing scheme based on feedback decision mechanism for wireless sensor networks is proposed. Firstly, according to the identity of network nodes and the cross characteristics of transmission links, with eliminating the same level cut points of failure nodes, the separation effect of cut points on network topology is avoided, the node failure decision model is constructed by traffic analysis, which can significantly reduce the phenomenon of node failure misjudgment. Secondly, based on node failure decision model, combining the traffic transfer characteristics of primary and secondary nodes, the node failure evaluation party is designed. In this case, through cutpoint decision and secondary node selfhealing reconstruction, the topological changes in the process of selfhealing are improved, and the flow limitation in the process of node failure is effectively avoided. Then, considering the difficulty of node selfhealing in the process of selecting standby repair nodes, an optimal selection scheme is constructed based on the characteristics of network topological cutpoint. The flow feedback mechanism is used to optimize the selection of standby repair nodes and reduce the risk of repair failure. The experimental results show that compared with the TVLLRS algorithm and the PFO scheme, the proposed algorithm has lower total number of selfhealing nodes, lower distance of node topological movement and higher network outlet bandwidth.

Abstract:The visibility meter calibration system is the standard equipment for visibility calibrating, and its uncertainty directly affects the calibration results of forward scattering visibility meter. In this article, the structure, specification and calibration method of the visibility meter calibration system of the national visibility meter calibration laboratory are introduced, the main uncertainty source of the calibration process is analyzed, according to the National Calibration Specification of JJF 10591-2012 Evaluation and Expression of Uncertainty in Measurement, Type A and Type B uncertainty of the calibration system are evaluated using the visibility data, meanwhile, three forward scatter visibility meters are calibrated using the calibration system. Results show that the uncertainty of the system below 3000 m visibility is 58%, after calibration, the biggest and smallest relative errors of three forward scatter visibility meters at 100，500，1 000，1 500，2 000，3 000 m are 78% and 18% respectively, which can meet the visibility meter calibration and operational measuring requirements.

Abstract:Reactive power metering circuit in existing power metering chips cannot meet the requirements of high metering accuracy and low hardware overhead at the same time. In order to solve this problem, a new method of reactive power metering is proposed, which uses three firstorder IIR filters, in which the phase shift filter in the voltage channel makes the voltage signal shift -90 degrees phase. The two amplitude compensation filters in the voltage channel and the current channel jointly compensate for the attenuation of the signal amplitude generated by the phase shift filter. The two amplitude compensation filters are required to be designed to have the same structure and parameters, thus their application in parallel mode will not change the phase difference between voltage and current. The design method of efficient reactive power metering system is proposed, especially the optimization method of amplitude compensation filter design. The simulation and actual product test results show that the reactive power metering method not only consumes less hardware resources, but also achieves high metering accuracy, which is better than the reactive power metering methods of the existing power metering chip.

Abstract:In order to carry out the modal analysis of the exhaust system, the modal parameters of a car exhaust system at 0~200 Hz are obtained by finite element calculation and experimental measurement respectively. By comparison, it is found that the errors of natural frequencies of each order obtained by the two methods are less than 10%, and the mode shapes are identical, which proves that the finite element model of exhaust system is correct and reliable. Afterwards, four hangers’ locations of exhaust system are preliminarily determined by means of ADDOFD method. In order to validate whether the hanger’s location is reasonable or not，a finite element model of exhaust system including power assembly is established for static simulation calculation of hanger lugs, it is found that the deformation of two hanger lugs exceeds the engineering limit of 35mm. After adding one hanger for optimization, the static deformation and stress improvement percentage of the five hanger lugs are more than 13%. The research method has certain reference value for the early development and the selection of hanger location of a car exhaust system.