Abstract:Phase-locked loop (PLL) plays an important role in wireless communication navigation, radar and other fields. It affects the data processing rate of the whole system and the extraction of clock phase information. The optical phase-locked loop (OPLL) has the advantages of high-speed real-time reception, low power consumption, strong ability to resist background light interference and Doppler frequency shift, and is the key circuit module of coherent optical communication system. Its performance directly affects the performance of optical communication system. The research of optical phase-locked loop is mainly aimed at optimizing and improving its parameters. This paper reviews the research progress of the technology from electric PLL to OPLL. Based on an introduction to the principle and key components, the classification and parameters of the electric PLL and optical PLL are introduced in detail, and the performance of various optical PLLs is compared. In addition, the OPLL technology applied in coherent optical wireless communication system is summarized, and the future development direction is prospected, which provides a benchmark for research and development of PLL technology in different areas.

Abstract:Time-interleaved analog-to-digital converters ( TIADC) and quantization-interleaved ADC ( QIADC) techniques are the effective solutions to improve sampling rate and resolution of acquisition systems. The hybrid parallel acquisition system based on TIADC and QIADC offers the capability to provide a blend of varying sampling rates and resolutions, however, high sampling rate is not compatible with high resolution. In this paper, the focus is on investigating the optimal acquisition performance of the system. To this end, we analyze the effects of the sampling rate and resolution on both input noise and quantization noise in the system. It is concluded that it exists an optimal sampling rate and resolution combination for a given number of ADCs present in the system, which makes the system achieve the best acquisition accuracy. Additionally, we also analyze the impact of offset mismatch and conclude that the offset error degrades the performance of QIADC, and establish the relationship between offset error and increased bits. Theoretical analysis and simulation results demonstrate that the optimal performance of the system is achieved when the input noise power equals the quantization noise power, leading to an excellent combination of sampling rate and resolution. The maximum output signal-to-noise ratio is attained at this point.

Abstract:Accurate prediction of corrosion state is very important for storage and transportation of oil and gas and safe and reliable operation of chemical equipment. Due to the complex corrosion process and many influencing factors, the prior model in the conventional corrosion prediction method is highly dependent on the environment and the medium and long-term prediction effect is poor. In this paper, a digital analog fusion driven Kalman filter corrosion prediction method integrating fuzzy reasoning and deep learning is proposed. Firstly, based on the long-term corrosion physical model and the actual short-term monitoring data, the fuzzy rules of corrosion velocity were established to obtain the modified corrosion velocity based on the field environment. At the same time, aiming at the prediction lag of fuzzy reasoning results and considering the long-term regularity of corrosion monitoring data, deep learning is used to predict the corrosion rate. Then, the fuzzy strategy and deep learning prediction results are fused to realize the digital analog fusion corrosion prediction based on Kalman filter. Finally, using the actual corrosion monitoring data of natural gas pipelines, this prediction method is compared with GPR, GM, PSOGM, FR, MLP and Kalman filter. The results show that the proposed method has good prediction effect. The relative prediction error of corrosion state in two years is within 1%, the root mean square error is 0. 000 49 mm, and the average absolute percentage error is 0. 34%.

Abstract:In order to solve the problems of inadequate feature fusion, low detection accuracy and poor robustness in current lane detection, this paper proposes a lane detection model called fusion of multi-branch structure and attention mechanism network (FMANet). In the image coding part, fusion of multi-branch structure and attention mechanism is adopted. swish is selected as the activation function, and the image decoding part adopts the jump connection structure to achieve cross-layer feature fusion. In this paper, TuSimple public dataset was used to evaluate and verify the FMANet model. The experimental results show that the mAP index of the FMANet model proposed in this paper is close to 97. 25%, and the lane detection accuracy reaches 98. 15%. In addition, CULane dataset verifies that the FMANet model has better robustness in different scenarios.

Abstract:In view of high noise interference and the difficulty of useful signal extracted for direct-driven permanent magnet motor in coal mine, a denoising method of integrating improved VMD and wavelet soft threshold is proposed. Firstly, particle swarm optimization is used to optimize the decomposition layers K and penalty factor α of the variational modal decomposition algorithm to obtain the optimal parameter combination. Based on the optimal parameter combination, K eigenmode components of abnormal sound signal for directdriven permanent magnet motor in coal mine are obtained. Secondly, the weighted margin index is used to screen out the effective signal components and the noisy components that need further decomposition. The wavelet soft threshold is used to further denoise the noisy components. Finally, the effective signal component and the wavelet soft threshold denoised component are reconstructed to obtain the final denoised signal. This method is used to denoise the simulation signal and the abnormal sound signal of direct-driven permanent magnet motor in coal mine respectively. In order to prove the validity of the method, we conduct the comparative test. The test results show that this method can increase the SNR of simulation signals to 27. 524 7 dB, and reduce the root mean square error to 0. 085 5. The SNR of measured signals is improved to 34. 715 3 dB, and the root mean square error is reduced to 0. 006 7. The method proposed can denoise effectively and provide data basis for subsequent fault feature extraction and fault diagnosis.

Abstract:In order to fully exploit the performance of the MEMS gyroscope and improve the accuracy of the MEMS gyroscope in practical applications,the output signal of the gyroscope array is denoised by constructing a four-gyroscope array combined with the improved SageHusa filtering algorithm. The actual performance of the MEMS gyroscope is effectively improved without changing the gyroscope processing technology and significantly increasing the production cost. By analyzing the systematic error and random error of MEMS gyroscope, the error model is built. The traditional Kalman filter, moving average filter, wavelet threshold denoising and the improved Sage-Husa filtering algorithm are used to denoise the single gyroscope and gyroscope array. Experimental comparison shows that the improved Sage-Husa filtering algorithm combined with the gyroscope array can effectively reduce the output noise of the gyroscope. The random error of the gyroscope array filtered by the improved Sage-Husa algorithm is analyzed by Allan variance. The angle random walk of the four gyroscope array is reduced from 0. 40°/ h to 0. 03°/ h , and the bias instability is reduced from 71. 11°/ h to 5. 83°/ h, which effectively improves the performance of MEMS gyroscope in practical applications.

Abstract:As the scale of integrated circuits continues to increase and their test cost increases with test time, the optimization of the test terms is an important topic. The existence of nonlinear implicit dependencies among test parameters in analog integrated circuit testing makes it very difficult to directly reveal their interrelationships and perform test method optimization. In this paper, we propose a method to optimize the number of test parameters of analog integrated circuits based on the XGBoost decision tree model, which explores the mutual characterization ability among different test parameters and sequentially deletes the test parameters that can be well expressed in the test sequence to achieve the purpose of shortening the test time under the condition of ensuring a certain escape rate. The paper discusses three metrics to evaluate the inter-representation capability of test parameters, such as the number of faults, feature importance and SHAP value, and conducts experiments on two types of analog IC test datasets. The results show that the number of faults is an excellent evaluation metric that can achieve 25% test parameter optimization under the condition that the test escape rate does not exceed 20 PPM.

Abstract:Aiming at the problems of false alarms, missing alarms and abnormal working status of sensor nodes in high temperature environment, this paper proposes a fire escape path planning research method combining LSTM and improved A ∗ algorithm. According to the LSTM, the real-time fire situation information was adaptive learned, and the abnormal node data prediction model was established to predict the threat situation of abnormal nodes, such as temperature and carbon monoxide concentration. Based on the real-time situation information of indoor fire, the fire threat situation spread model was built, and the improved A ∗ algorithm was used to dynamically plan the escape path to obtain the best safe escape path under abnormal conditions. The results show that this method can plan the best escape path in different fire periods, and gain valuable time for the evacuation of personnel, which has practical application value.

Abstract:The primary factor for train derailing is the turnout’s aberrant geometric location, thus, it is crucial to monitor it in real time to effectively avoid derailing. This work develops a set of online in-situ monitoring systems for crucial turnouts parameters based on binocular vision to meet this need. First, an online self-calibration method based on straightforward labels is proposed to address the issue that the vibration of railroad traffic readily changes the exterior parameters of the visual measuring equipment. Additionally, laser marking is utilized to strengthen the texture characteristics in turnout monitoring features in order to properly detect them. This can resolve the challenging issue of monitoring features locating following the change of imaging viewpoint produced by the swing of the sharp rail. The Gaussian-weighted grayscale center of gravity approach is proposed to extract the center of the light strip for outdoor light interference. Our method successfully overcomes challenges such easy diffuse reflection on the metal surface of the rail and can accurately find monitoring features. Binocular stereo collaboration is accomplished, and then the monitoring of turnout parameters is finally completed by calculating the spatial three-dimensional coordinates of monitoring features. The field real measurement demonstrates that this device has low hardware cost, high resilience, and rapid speed, with an inaccuracy of approximately 0. 3 mm.

Abstract:In response to the urgent demand for high g-value accelerometers in aerospace, vehicle collision analysis and other fields, a new multi-beam groove piezoresistive accelerometer was designed, which used variable section beams as sensitive beams to amplify the equivalent inertial force of the mass, and effectively improved the structural stiffness through the combination of auxiliary beam and back groove to achieve high sensitivity output and good bandwidth characteristics. The mechanical model of the accelerometer was established, and the structure and size of the accelerometer were optimized through ANSYS finite element software. Using the surface path of the sensitive beam, the stress distribution law was analyzed and summarized, and the varistor position was determined. The simulation results demonstrate that the designed sensor measurement range was 150 000 g, and the sensitivity was 1. 32 μV/ g, and the natural frequency was about 268 kHz. Based on SOI technology, the processing flow of the accelerometer was finally designed.

Abstract:Emission source microscopy ( ESM) based on synthetic aperture imaging principle has obvious technical advantages and application value in the field of electromagnetic interference (EMI) detection of electronic equipment, but it still needs further research and improvement in spherical imaging algorithm and system design. In view of the above requirements, this paper proposes the design method and control strategy of the manipulator system under the spherical scanning imaging task. On the system model, the manipulator configuration, spherical scanning boundary and detection antenna attitude are modeled, and the optimal spherical aperture solution method under the above constraints is explored. In terms of control strategy, manipulator control strategy based on joint angle search and the microwave microscopic imaging focusing algorithm are designed and implemented. The above models, methods and strategies achieve a 60. 21% reduction in the angle of the average joint change in the simulation experiment. In the verification experiment built by the sixaxis manipulator and the reference microwave radiation source, the automatic focusing and accurate imaging are realized ( the focusing accuracy is 0. 1 mm, and the imaging accuracy is not less than 3. 14 dBm), which is expected to be applied to EMI detection and positioning of electronic systems.

Abstract:The estimation accuracy of the instantaneous angular speed ( IAS) signal is easily affected by the encoder’ s error, which affects its fault detection results. Because of such problems, a local polynomial differential estimation method was proposed to estimate the IAS signal. First, the local polynomial differential was used to suppress the sawtooth noise generated by the encoder subdivision errors and the signal discontinuity caused by the estimation errors, so high-precision acquisition of the IAS signal was achieved, and finally, the accurately estimated IAS signal was used to extract the sun gear fault characteristics by local synchronous fitting technique. Through experiments and simulations, it is shown that the local polynomial differential estimation proposed in this paper can effectively suppress the IAS signal noise caused by encoder errors, improve the signal to noise ratio of the IAS signal, and combine local synchronous fitting to effectively extract the features of the sun gear tooth surface spalling fault.

Abstract:Aiming at the problems of high management cost and low production efficiency in traditional production methods, in order to realize intelligent production, the application framework of production line lifecycle management based on digital twinning technology is designed. First, based on the multiple data of physical production line, the digital twinning model of production line is built through modeling technology, including the planning stage, commissioning stage and operation stage of production line, provide simulation preview analysis, connection debugging, production status monitoring, data visualization and other services for the physical entity production line. The twin model was used to guide and improve the planning and operation of the actual production line. The balance rate of the production line increased by 6. 84% and the annual output increased by 792 pieces. The results showed that the method was effective.

Abstract:A single shot of a light field camera can record the intensity and direction information of light at the same time. Compared with the RGB camera, it can better reveal the three-dimensional structure and geometric characteristics of the scene and has unique advantages in the field of object 6D pose estimation. Aiming at the problems of low detection accuracy and poor robustness in complex scenes in existing RGB pose estimation methods, this paper proposes an end-to-end convolutional neural network object pose estimation method based on light field images for the first time. In this method, the dual-channel EPI encoding module is used to process highdimensional light field data. By reconstructing the light field EPI image stack and introducing horizontal and vertical EPI convolution operators, the modeling ability of the spatial angle information association of the light field is improved. Two-branch siamese network for shallow feature extraction of light field images. Secondly, a feature aggregation module with skip connection is designed to perform global context aggregation on the concatenated light field EPI shallow features in the horizontal and vertical directions, so that the network can effectively combine global and local feature clues when predicting pixel-by-pixel key point positions. To solve the problem of insufficient light field data, this paper uses the Lytro Illum light field camera to collect real scenes and constructs a rich and complex light field pose dataset—LF-6Dpose. The experimental results on the light field pose dataset LF-6Dpose show that the average pose detection accuracy of this method is 57. 61% and 91. 97% under the ADD-S and 2D Projection indicators, which surpasses other advanced methods based on RGB and can better solve the target 6D pose estimation problem in complex scenes.

Abstract:Aiming at the problem of multivariable and high nonlinearity of switched reluctance motors and the inability of traditional design process to obtain the optimal solution quickly and accurately, a parameter optimization strategy based on Kriging model and improved particle swarm algorithm is proposed. Firstly, a multi-objective optimization model is established, and Taguchi orthogonal method is used for sensitivity analysis, and the optimization variables are divided into two subspaces according to the sensitivity magnitude. Secondly, in order to improve the convergence speed and global optimization accuracy of multi-objective particle swarm optimization algorithm, the environmental induction mechanism in beetle antennae search algorithm and the crossover and mutation strategy in genetic algorithm are introduced. Finally, Kriging model is established and improved particle swarm algorithm is used to iteratively optimize the two subspace parameters. The experimental results show that the optimized torque ripple is reduced by 23% and the average torque is increased by 2. 3%, maintaining a large average torque with a significant reduction of torque ripple. The conclusion is that the combination of improved particle swarm optimization algorithm and Kriging model is suitable for optimization process of switched reluctance motor, which can significantly improve optimization efficiency and ensure solution accuracy.

Abstract:Aiming at the problem of class determination conflict in vehicle recognition technology, a multi-sensor fusion target recognition method based on compatibility coefficient is proposed. Firstly, multi-hypothesis thinking is used to realize multi-sensor heterogeneous data clustering and merging to obtain single-frame fusion detection results; then, the compatibility coefficient between evidences is calculated to redistribute evidence conflicts, and the single-frame processing is combined with Dezert-Smarandache theory ( DSmT) conflicts of evidence that may arise after fusion, and obtain accurate identification results of the target category. The results of actual measurement data show that this method can overcome the problem of limited sensor detection range and conflict of category determination, and the target recognition accuracy rate can reach more than 93%, which reduces the missed detection rate and false detection rate of target recognition, and can achieve good target recognition accurate performance.

Abstract:Due to the complex structure, poor rigidity and high accuracy requirements of thin-walled parts, the assembly of thin-walled parts is mainly completed manually. There are problems such as low degree of automation, low product yield, poor efficiency and heavy reliance on the technical level of workers. In order to improve product quality and production efficiency, automatic assembly equipment is urgently needed. The assembly unit of thin-walled parts needs to be classified and matched according to the online measurement data. The traditional industrial robot takes up a large working space. The parallel robot has a flexible structure and takes up a small space. In this paper, a design scheme of thin-walled parts automatic sorting assembly unit is designed to replace the traditional manual sorting and assembly production mode. Firstly, according to the functional requirements of thin-walled parts classified assembly, the parallel structure is designed, and the dynamic model and kinematic model of the parallel structure are established. The change law of velocity and acceleration of parallel structure during rapid movement is analyzed. The angle data curve of three active arms and intermediate branch chain changes smoothly without sudden change. The angular velocity and angular acceleration are relatively stable, which ensures the working accuracy of the measuring camera. The stability of parallel mechanism meets the design requirements. The displacement curve of the moving platform coincides with the predetermined trajectory, and the results show that the designed mechanism can meet the functional requirements. The structure has small working space, can realize the automation of thin-walled parts assembly, and ensure the quality of products.

Abstract:Because of the limitations of wireless Internet of Things technology and battery life, the development of smart manhole cover technology is slow. In this paper, a self-powered smart manhole cover is proposed. Thermal power generation technology is used to convert the temperature energy under the manhole cover into electrical energy. To increase the temperature difference near the bottom of the manhole cover, the heat insulation layer and the radiator are added, and the temperature gradient is increased by 5 ℃ ~ 10 ℃ . The experimental equipment of thermoelectric generator module was built, and the thermoelectric generator was tested and analyzed. The results show that the thermoelectric module generates 7. 92 J electric energy with an average power of 2. 2 mW under the temperature difference of 13 ℃ in one hour. Based on NB-IOT technology, smart manhole cover monitoring node is designed to monitor the tilt angle of manhole cover. Through the dormant standby method, the power consumption of the node system is reduced. The node acquires and transmits data once and consumes 0. 185 J power. The node standby current consumption is only 87. 45 μA. The combination of selfpower generation and low power consumption Internet of Things has verified the feasibility of thermoelectric power generation as the energy supply of the smart manhole cover.

Abstract:For the purpose of tackling the difficulties of the low accuracy of transformer fault diagnosis, a transformer fault identification method based on multi-strategy improved ocean predator algorithm ( MPA ) and hybrid kernel extreme learning machine ( HKELM ) has been proposed. Firstly, kernel principal component analysis ( KPCA ) is applicable to decrease the dimension of high-dimensional linear inseparable transformer fault data and it is also used to obtain feature support data. Then, the MPA is comprehensively improved through strategies such as Bernoulli chaotic mapping, improved stage transition criterion, and best candidate to strengthen the global development ability. Finally, the improved IMPA algorithm is used to optimize the parameters of HKELM and construct the transformer fault diagnosis model. Aiming to validate the validity of the model, four transformer fault diagnosis models of HKELM optimized by common algorithms are analyzed and compared. The diagnostic accuracy of IMPA-HKELM is 94. 7%, compared with the other three basic algorithms, the diagnostic accuracy is improved by 5. 4%, 8% and 10. 7% respectively. The results show that the proposed model effectively improves the classification performance of fault diagnosis and achieves higher fault diagnosis accuracy.

Abstract:Aiming at the problem of large positioning error of dual Mach-Zehnder interferometric optical fiber vibration sensor system, a vibration positioning method based on Burg algorithm is proposed. The spectral analysis method is used to compare the energy characteristics of the Burg algorithm and the fast Fourier transform algorithm at different frequencies. The optimal frequency is determined by calculating the energy ratio, and the value selection analysis of the optimal order of the Burg algorithm is carried out. Under the conditions of optimal frequency and order, feature data frames are extracted, and the time delay between two vibration signals is obtained through cross correlation calculation, so as to obtain the vibration position. In the dual Mach-Zehnder interferometric optical fiber vibration sensor system, the experimental research on vibration location is carried out. The experimental results show that the method can successfully extract the characteristic data frame of vibration signal frequency on the 2. 2 km sensing optical fiber, and the absolute error of vibration location is 7. 3 m, which provides a new method to improve the positioning accuracy of the dual Mach-Zehnder interferometric optical fiber sensing system.

Abstract:With the cross construction of power grid and pipe network, the dynamic DC interference and AC interference on oil and gas pipelines show a mixed development trend, which aggravates the pipeline corrosion rate. In order to separate and detect dynamic DC and AC interference on oil and gas pipelines, the spectrum characteristics of dynamic DC interference and AC interference in the mixed interference signal are extracted using FFT with different resolutions. Then, the dynamic DC interference is approximately treated as DC signal by using the combination of subsection aggregation approximation and FFT to realize the separation of dynamic DC interference and AC interference and the calculation of characteristic parameters. The test results show that the detection progress of dynamic DC interference and AC interference meets the expected requirements. The deviation between the original AC interference potential effective value and the AC interference potential measured after separation is not more than ± 2. 5‰, and the difference between the dynamic DC interference potential effective value and the dynamic DC interference potential effective value measured after separation is less than 1 mV (not more than ± 0. 5‰). The work provides data support for interference source troubleshooting and targeted pipeline corrosion protection

Abstract:To locate the fault point of twisted pair in voice communication is of great significance for the detection of open circuits, short circuit faults, and the prevention of eavesdropping on confidential information. In the fault detection and localization of voice communication twisted pair based on time domain reflectometry (TDR) and spread spectrum time domain reflectometry (SSTDR), due to the low-frequency transmission characteristics of twisted pair, the reflected signal is seriously attenuated, and long-distance fault point location cannot be realized. Given this problem, a time-frequency domain reflectometry ( FRFT-TFDR) based on fractional Fourier transform is proposed. First, the chirp signal (LFM) is used as the detection signal, the received signal is windowed to improve the resolution of the signal, then the fractional order Fourier transform is performed to transform the received signal into the fractional domain. Finally, the position difference of incident signal and reflected signal spectral peak in the u domain is analyzed and converted into the time domain, so that the location of errors can be determined. The experiment verifies that the method can effectively locate different circuit-breaking fault points, and its positioning error at 1 557 meters is less than 7%, meeting the requirement of remote fault detection of twisted pair in voice communication.

Abstract:As a large number of lithium-ion batteries are being retired from electric vehicles, research on rapid screening of retired batteries is urgent. Aiming at the problem that the consistency screening time of batteries before secondary utilization is long due to the difference in initial state of traditional methods, this paper proposed a quick test method based on battery charging curve. By charging the battery to the cut-off voltage to ensure that the battery has the same initial state, instead of emptying the battery, the test time is only 12. 5% of the complete battery charging and discharging time, and the screening efficiency was largely improved. After the features were extracted, the K-means++ clustering algorithm combined with Canopy was used to verify the results on NASA data sets and laboratory batteries. The clustering accuracy reached 80. 5%, which proved the feasibility of the designed rapid test method.

Abstract:When recognizing images with complex scenes and obscure features such as weather images, there are often problems such as low recognition rate and feature redundancy. Based on this, an image classification algorithm based on deep transfer learning is proposed in this paper. The algorithm uses the model parameters of ImageNet dataset to construct three network models, ResNeXt, Xception and SENet, to extract image features, and uses a domain-adaptive discriminative joint distribution adaptive algorithm to resemble the feature vectors to complete a high-quality feature representation, and uses the result as a criterion to fuse the model features, and trains the fused features through a multilayer perceptron to achieve image classification with high accuracy recognition. The experimental results show that the algorithm outperforms the traditional single network model and further improves the upper limit of image classification accuracy.

Abstract:Optical element defects will directly affect the performance of the entire optical system. In the detection of optical element defects, scratch defects are undoubtedly the difficulty of detection. The scratch defects have the problems of small size, large aspect ratio, and easy to be affected by impurities. In this paper, depth learning algorithm is applied to optical element defect detection, and according to the characteristics of scratch defects, the Mask R-CNN network model is improved. The algorithm also has a better detection effect on scratch defects. First, the original ResNet is replaced by CSPRepResNet proposed in this paper, and ESE attention mechanism is added to improve the ability of feature extraction and reduce the amount of computation. Secondly, K-means algorithm is used to recluster the length width ratio of anchor boxes. Thirdly, the loss function of target detection is changed from Cross Entropy to gradient balanced Focal Loss, which solves the problem of imbalance between positive and negative samples, is more conducive to the detection of difficult samples, and can also eliminate the influence of outliers. In general, the tested mAP@ . 5 The original 52. 1% is increased to 57. 3%, an increase of 5. 2%, and the reasoning speed is almost unchanged. It can be seen that the improved Mask R-CNN has a better detection effect on optical element scratch defects.

Abstract:The rods produced by the high-speed assembly line are highly susceptible to various surface defect, but the defect identification method based on conventional image processing is unreliable and susceptible to environmental factors, while the defect identification method based on deep learning suffers from oversized models and recognition accuracy that is constrained by the quantity of samples. Therefore, this paper suggests an identification system of rods surface defect based on improved SqueezeNet. An acquisition device was designed to obtain the full surface image of the circumferential rods, and the attention module is introduced into the lightweight convolutional neural network SqueezeNet to improve the feature extraction effect of the model, data balancing is used to improve the recognition accuracy of minority samples, transfer learning is employed for deep learning training to minimize the impact of insufficient samples on the training effect. Taking the cigarette on the production line as the research object, the circumferential surface image of the cigarette is collected for experiment, the results show that the classification accuracy of the improved method under the condition of few samples reaches 94. 49%, with the F1 score for minority samples being improved by 31. 19%, and the detection time of a single image being approximately 1. 66 ms. Additionally, the model is lightweight, meeting the need for rods in industrial production lines to have real-time defects recognized.