Abstract:Based on a coaxial resonator, a method for measuring the relative permittivity of low-loss flake dielectric materials is proposed. Combined with Maxwell's equations and electromagnetic field boundary conditions, the TEM resonant modes in the partially filled dielectric coaxial resonator are analyzed, and the eigen equations of the dielectric filled coaxial resonator are deduced. A polynomial fitting method was used to simplify the computational model. In this paper, a coaxial resonator with a cavity operating frequency of 1.8183 GHz is designed and analyzed by calculation and analysis. The simulation analysis was carried out in the HFSS electromagnetic simulation software to study the influence of the thickness of the filling medium material and the placement height on the measurement results. The simulation measurement results were consistent with the theoretical model results. In the experiment, a measurement system was built to realize automatic measurement. The experimentally measured resonant frequency of the coaxial resonator cavity is 1.8183 GHz. The relative permittivity of FR4 dielectric materials is measured. The stability of the measurement system is good after several measurements. The experimental results are in line with the actual nominal value, and the error with the simulation is less than 5%, which proves the feasibility and accuracy of the method.

Abstract:In the energy storage application of retired batteries, the large inconsistency between the retired batteries makes the battery pack more prone to over-charge and over-discharge during the charging and discharging process, which reduces the overall available capacity of the battery pack and even causes safety issues. In order to solve the above problems, an active equalization circuit based on an isolated dual half-bridge DC-DC converter is proposed in this paper. The equalizer circuit consists of a switch array of N+5 switches (N is the number of cells) and an isolated dual half-bridge DC-DC converter, which ensures the flexibility of the circuit. Based on the analysis of the working principle of the main circuit, a sub-state equalization control strategy based on SOC is proposed to achieve balancing of the battery pack by using the corresponding equalization strategy under three different states of battery pack: charging, discharging and resting. Finally, an equalization experiment was carried out on 5 series lithium-ion batteries. The experimental results show that the available capacity of this method is increased by 12%, 9.9%, and 17.5% in the resting, charging, and discharging states, respectively, compared with the battery pack without the equalizer, which proves the feasibility and effectiveness of this method.

Abstract:In recent years, orbital angular momentum has been a hot research topic in wireless communication. In this paper, an orbital angular momentum antenna based on dielectric resonator is proposed by combining orbital angular momentum with millimeter wave technology. The equivalent model of the antenna is established, the theoretical expression of its radiation field is derived, and the effect of the dielectric resonator radius on the vortex wave electromagnetic wave mode is discussed. The simulation results show that the antenna has four resonance points in band and can generate vortex electromagnetic waves with mode respectively. In addition, the antenna has a compact structure, low cost, good gain, high antenna efficiency, and the generated vortex electromagnetic waves of each mode have good rotational properties, which can obtain a strong anti-interference capability and provide some practical significance for the application of orbital angular momentum in the millimeter wave band.

Abstract:Spatial resolution is an important performance parameter in Brillouin optical time-domain reflection(BOTDR) technique. Affected by the acoustic phonon lifetime in the fiber, the spatial resolution of the Brillouin optical time domain reflectometer used in the engineering field is more than one meter, which can meet most engineering needs. With the continuous expansion of application scenarios, the requirements for the spatial resolution of Brillouin optical time domain reflectometers in fields such as special structural deformation monitoring and precise positioning of pipeline deformation are also increasing. This study introduces the structure of BOTDR and illustrates Two types of methods to improve the spatial resolution of BOTDR. Additionally, important progress in BOTDR sensing for decades are reviewed. On the basis of the relatively mature existing hardware, the method of improving the spatial resolution in the future will develop in the direction of improving the software processing algorithm and combining various signal processing algorithms.

Abstract:The abnormal state of the secondary circuit of metering current transformer will not only cause economic losses to power grid enterprises, but also lead to safety accidents. In order to realize the accurate on-line monitoring of the secondary circuit state, through the theoretical study of the circuit equivalent circuit under different states, a secondary circuit state monitoring method based on the 10kHz impedance characteristics is proposed. The circuit 10kHz impedance measurement technology based on micro transformer coupling and vector voltage current method is adopted, and the current transformer circuit state monitoring device is developed. The experimental results show that the developed device can accurately identify the normal connection of secondary circuit, short circuit of secondary terminal, secondary open circuit and primary bypass under the full working conditions of current transformer transformation ratio of 30/5~2000/5, ambient temperature of (-40~60)℃ and secondary current of (0~120%)I2n.

Abstract:In order to solve the potential safety hazards caused by transformer partial discharge fault, an image recognition method of transformer partial discharge signal based on Inverse Radon transform (Iradon)-Convolutional Neural Networks (CNN) was proposed. Partial discharge experiments were carried out for three kinds of faults. First, the partial discharge signal was decomposed by resonance sparse decomposition to obtain low resonance components, which were then converted into Iradon images. Finally, CNN was used to adaptively extract the feature information of Iradon images. The results show that, this method can accurately extract signal features, has powerful data processing and identification functions, and provides rich information for the identification of partial discharge states of transformers, and improves the learning effect and identification accuracy.

Abstract:According to the application requirements of SOC in battery management system, an ARWLS-AEKF joint algorithm was proposed to estimate the SOC of lithium ion battery. Based on the second-order R-C network model, the method introduced adaptive genetic factors through the weighted adaptive algorithm, to optimized the parameter identification method,. And combined with the adaptive extended Kalman Filter (AEKF) algorithm for online identification, to complete the estimate of SOC. The simulation results and experimental data show that the error of ARWLS-AEKF algorithm is within 2% under LA_92, UDDS and HWFET conditions, MAE is 0.45%, 0.74% and 0.87%, RMSE is 0.54%, 0.71% and 0.42%, respectively. ARWLS-AEKF algorithm has higher accuracy and stronger disturbance resistance to noise than off-line EKF method.

Abstract:In order to identify the structural damage status with a small amount of unlabeled monitoring data, an innovative damage identification method is proposed in this paper. Firstly, the dynamic response data measured by each long-gauge fiber Bragg grating sensor in the experiment is subjected to variational modal decomposition, and the time-domain and frequency-domain characteristics of each component signal are extracted. The sensitive features of each sensor signal are selected by Relief-F, combined into a sensitive feature set and input into the affinity propagation clustering for damage identification. Two different sets of experiments are used to verify the effectiveness and robustness of the method, and damage recognition rates of 100% and 98.7% are obtained in the two sets of experiments. The results show that the method has practical application value.

Abstract:To solve the problem of fault of followers in multiple unmanned aerial vehicles (UAVs) formation flight, an adaptive fault-tolerant control method based on observer is designed. Firstly, based on the leader-follower method, the formation model of UAVs and the fault model of followers in the formation are established, which are divided into the position subsystem and the yaw angle subsystem. Secondly, an observer is designed to observe the fault and the state in the position subsystem, and combined with the observed state and fault, a state feedback control law is constructed. Then, the control design scheme of yaw angle subsystem is given based on the adaptive method. The final bounded convergence of the tracking error of the system is proved based on the Lyapunov theory. Through simulation, compared with the method based on the robust fault estimation, the algorithm in this paper reduces the total tracking time and steady-state error of the system by 76% and 70.3% respectively, and both of them are significantly reduced compared with the traditional observer method, which proves that the algorithm in this paper can better overcome the adverse effects caused by the deviation fault and effectively realize the formation flight of quadrotor unmanned aerial vehicles.

Abstract:n recent years, a lot of oil-water sensor and oil-water detection systems for crude oil dehydration appear on the market, but there are problems emerged such as low accuracy, large error and so on. The performance indicators need to be further improved. Therefore, based on impedance measurement technology, an intelligent oil and water construction method based on AC impedance measurement was proposed. The principle and structure of the oil-water detection system are introduced, a nonlinear error compensation model based on the least square method is constructed, which solves the influence of the temperature and water content of the medium on the AC impedance are solved, and the accuracy of the intelligent oil-water detection effectively are improved. The experimental results show that the method proposed in this paper can effectively improve the detection accuracy of water content detection in crude oil. The detection range of water content in the oil-water detection system using this nonlinear error compensation method is 0 %-100 %. The detection accuracy can reach 0.3 %. The detection system has high accuracy, meets the design requirements, and the needs of related enterprises.

Abstract:In the construction industry, safety accidents caused by not wearing helmets account for a relatively large proportion. Aiming at the problems of strong interference and low accuracy of small targets in helmet detection, an improved algorithm based on YOLOX is proposed. Firstly, an ECA-Net attention mechanism is added to the enhanced feature extraction network to carry out cross-channel interaction, suppress the interference information according to the corresponding channel weight value generated, strengthen the model's attention to the target feature, and then fuse the recalibrated feature map more deeply to improve the expression ability of the target feature. Secondly, the CIoU is used to calculate the loss, the distance between the two boxes of center points and the aspect ratio are considered into the penalty term, and the loss function is constantly adjusted and updated to accelerate the model convergence speed. Finally, a small target helmet dataset in a real construction scenario is constructed. Experimental results show that the improved algorithm mAP reaches 91.7%, which is 1.2% higher than the original YOLOX calculation, the average accuracy of the detection of workers who have worn helmets reaches 93.9%, the average accuracy of detection of those who have not worn helmets reaches 89.5%, and the detection speed reaches 71.9 frames/s, which ensures that the real-time detection of helmet wearing has a high accuracy rate.

Abstract:This paper proposes a traffic sign detection method based on an SSD network. This method improves the existing SSD algorithm, which has low detection accuracy and weak generalization ability for small targets, and has problems such as false detection and missed detection. The ResNet-50 network is used as the backbone network of the SSD algorithm, and the BN layer is added to the additional layer to improve the training speed. Sub-pixel is used instead of upsampling to improve the resolution of the recognition target, and the MFPN model is added to fuse the low-level and high-level feature information to avoid the problem of missed detection. The experimental results show that the improved SSD algorithm improves the mAP value by 4.2% and 3.1% on the public datasets CCTSDB and GTSDB datasets, respectively, the FPS remains at 87.2f/s, and the detection accuracy is significantly improved. This work meets the requirements for real-time detection of traffic signs and has broad application prospects in the field of unmanned driving.

Abstract:Aiming at the time-consuming problem of establishing deep learning labeling data for unmanned lifting target detection, a cargo image detection generation admision network was designed to construct an accurate data set containing semantic labeling and key point labeling, which could be used for the training of supervised deep learning semantic segmentation model. The generative adversation network of StyleGAN and DatasetGAN was fused to improve the semantic feature deformation in practical applications. The sample normalization layer of generator was modified to remove the mean operation and modify the input mode of noise module and style control factor. To solve the problem of weak coding ability of spatial position of objects with single texture feature, the constant input of generating network is replaced by Fourier feature, and a module integrating nonlinear up-down sampling is proposed. Finally, WGAN-GP is introduced to improve the objective function. Using deeplab-V3 as evaluation network and DatasetGAN as baseline, the output mIOU value of Deeplab-V3 increases by 14.83% on average in semantic label generation task, and L2 loss decreases by 0.4×10-4 on average in key point label generation task. PCK value is increased by 5.06% on average, which verifies the feasibility and advance of the improved generative adversarial network generation semantics and key point annotation data.

Abstract:Image object detection is crucial for the automatic counting of railway tools. However, collected images of railway tools have the characteristics of low illumination, huge difference in the scale of different objects and complex backgrounds. Existing image object detection methods cannot detect railway tools efficiently. In this paper, we propose a novel object detection model which is able to enhance the object-detection ability by fusing multi-scale features. Based on the object detection deep learning model Retinanet, we construct a feature fusion enhancement module. By fusing features, our model can efficiently detect railway tools with different scales. Experiments were conducted based on real-world datasets. Results show that our method is more efficient than Retinanet, and the mAP is increased from 97.85% to 98.11%. By the accurate detection of railway tools in complex backgrounds, our approach can lay the technical foundation for intelligent railway operation and maintenance.

Abstract:In indoor environment, the Ultra-Wide Band (UWB) signal is often affected by the interference from the occluding objects, which leads to the anomaly of the measured distance data and spurious delay, resulting in the degradation of the positioning performance. Firstly, a K-means model is developed to reject the false observation values by preprocessing the Euclidean distance between data. Secondly, the conventional least-squares localization method is improved by using the L2 norm regularization method. Finally, in order to verify the effectiveness of the method, an application example of indoor 3D positioning of an unmanned aerial vehicle (UAV) is designed in this paper. The flight data of UAV are collected under the conditions of no interference and interference, and then the proposed data preprocessing method is applied to eliminate the abnormal data and estimate the position of UAV in both cases using the improved least squares method. The results of the practical example show the effectiveness of the proposed method, which can improve the UWB 3D positioning accuracy under the interference environment.

Abstract:The shape size of the nozzle in the water-jet loom plays an important role in the operation of the loom. In order to efficiently detect the shape and size parameters of the loom nozzle with high precision, a nozzle shape size detection system based on machine vision is designed. In this system, the rotation of the nozzle is controlled by stepper motor, the nozzle image is collected from multiple angles by CCD camera, the outer contour edge of the nozzle is obtained by sub-pixel edge detection method, and then the edge is fitted in a straight line. finally, the diameter, taper and tilt angle of the nozzle are calculated. The experimental results show that the relative error of diameter measurement of the system is less than 1%, and the angle measurement error is less than 2′. Multi-angle measurement can improve the measurement accuracy of taper and inclination angle. The system has high detection accuracy and can meet the requirements of practical production and application.

Abstract:In order to solve the problem of texture loss and image edge degradation, an improved anisotropic diffusion model is proposed in this paper. Firstly, the PM model and LCC model are combined. According to the changes of image gradient, the relationship between the gradient modulus of local image and the diffusion intensity is constructed. Different gradient modulus values are selected for different diffusion functions. Then, the -norm is used to determine the gradient threshold in the diffusion function, which further improves the generalization ability of the proposed model. Experimental results show that this model can not only solve the problem of outliers existing in the traditional PM model, but also effectively protect the integrity of image edge features and contour structure. Compared with the original algorithm, the image signal-to-noise ratio is improved by 1.47~1.57dB, and the structural similarity is improved by 17%, the denoising efficiency is improved while ensuring the denoising effect.

Abstract:The detection of electrical cabinet parts is an important part of the production of electrical cabinets. Machine vision is used to automatically identify the type and installation location of parts in the electrical cabinet, and to detect the assembly defects of the electrical cabinet in time. However, the existing object detection in-depth learning model has low timeliness, which makes it difficult to meet the online detection requirements of electrical cabinet parts. In this paper, the YOLOv4 object detection model is pruned and optimized, and a lightweight object detection model SlimYOLO is proposed. SlimYOLO improves the feature extraction network structure, compresses the redundant feature layer, and improves the detection speed of the model. At the same time, the Kmeans++ clustering algorithm is used to cluster anchor box parameters, which improves the detection effect of the model for electrical cabinet parts. Based on the self-built data set of electrical cabinet parts, an experimental study was carried out. The average detection accuracy of SlimYOLO is 98.08%, which is 0.58% higher than YOLOv4, the model volume is reduced by 9.8%, the parameter is reduced by about 7 million, and the detection speed is increased by 10%, which lays a foundation for the fast and intelligent detection of electrical cabinet parts in the actual industrial scene.

Abstract:To give the real-time prediction of accuracy of the ball mill soft sensor, and to solve the model mixing problem in data decomposition of soft sensor. this paper proposed a new Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN), Gaussian Mixture Model (GMM) and Gaussian Process Regression (GPR) based ball mill load parameters soft sensor method. The key features of this method are using CEEMDAN-GPR to decompose and classify vibration and acoustical signals time domain signals to a series of intrinsic mode functions (IMFs). GPR is used to provide the predicted values. Comparing to the other soft sensor method, the CEEMDAN-based method is largely avoiding the mode mixing issue coming with the original EMD method. Anomalous signals can be classified while feature clustering by giving a probability threshold to the GMM. The GPR-based predicting method will not only provide the data-driven predict values, but also provide their confidence intervals, and warn the operator if necessary. The experiment result shows that comparing to the other soft sensor method, the proposed method has improvements on mill load predicting accuracy and ability of abnormal warning.

Abstract:In order to facilitate the analysis of the disaster impact caused by hail on social production, it is necessary to make classified statistics on hail magnitude and quantitative analysis on hail magnitude, which can not only provide the basis for disaster assessment, but also give feedback to weather forecast and false report. In this paper, an improved complementary set empirical mode decomposition (CEEMD) reconstruction algorithm is proposed for hail sound signal. The reconstructed signal retains the original time domain characteristics to the greatest extent, and can also denoise hail sound signal. Secondly, a multi-domain feature fusion 1D-CNN model is designed. The reconstructed original data, time-domain features and frequency-domain features are used as the input of 1D-CNN,the features are spliced in the middle layer, and finally the classifier is output. The results show that the recognition rate of the multi-domain feature fusion 1D-CNN designed in this paper is as high as 99.58%, which is 8.75% higher than that of the original data and the traditional 1D-CNN model.

Abstract:Current transformer plays an extremely important role in AC measurement, relay protection, maintenance control of power equipment and other related fields of electric system. At present, electromagnetic current transformer has exposed many shortcomings gradually, such as poor insulation and weak anti-interference ability. while electronic current mutual inductance is expected to become the main equipment for detecting current in the future due to its good insulation, small size and digitalization. By describing the research and development status of electronic current transformers, this paper focuses on Rogowski coil current transformers (Rogowski coil current transformers), low-power current transformers (LPCT) and optical current transformers, by analyzing their respective structures and working principles, this paper summarized their advantages and disadvantages, and put forward some improvements. On this basis, it pointed out the challenges faced by electronic current transformers in equipment power supply, environmental adaptation, sensing methods, etc., and looked forward to its future trend from the sensing mechanism and sensing materials.

Abstract:Pipeline defect detection is a necessary means to ensure the safe operation of pipeline system, and this paper uses the method of full focus imaging to image and identify pipeline defects. The finite element software ABAQUS is used to conduct numerical simulation of the pipeline defect detection method based on L(0,1) modal guided wave, and the signal is preprocessed by median filtering, Hilbert transformation and signal envelope sharpening method, and then the processed signal is used to construct the full matrix data, and finally the defect imaging is realized by the full focusing algorithm, and the imaging results are finally displayed. Experimental results show that the signal is pre-processed and then imaged, which can effectively improve the resolution and achieve high-precision visualization of defects.

Abstract:The gas-insulated switchgear is always affected by the white noise during partial discharge detection. A stationary wavelet transform noise filtering method based on a new noise threshold rule is proposed to filter out the white noise in partial discharge. The lower control limit and upper control limit of the wavelet coefficient are determined by statistical process control theory, and they will be updated iteratively according to the statistical characteristics of wavelet coefficients in the method. Then the noise threshold level of the signal is obtained through the upper and lower limits, and the white noise in the signal will be reduced adaptively. Since the down sampling of traditional wavelet transform will not appear in the stationary wavelet transform, the feature of the partial discharge signal will be more complete. In this paper, the noise suppression of three 5dB noise-stained partial discharge signals is carried out. The signal-to-noise ratio reaches 19.1433 dB, and the root mean squared error is maintained within 0.03 after noise reduction. In addition, the noise rejection ratio is 17.1769 for signals from a laboratory. The noise in the partial discharge can be better suppressed by the proposed algorithm. In addition, the feature of the waveform is obvious and the distortion is low.

Abstract:The magnetic flux leakage detection of rail surface defects will be affected by the inspection speed and other factors, which increase the background noise and reduces the detection sensitivity. In order to enhance the defect signal characteristics and improve the signal-to-noise ratio of MFL signal, a MFL signal processing method based on minimum entropy deconvolution is proposed in this paper. Through the objective function method, the optimal inverse filter parameters are calculated, and the collected magnetic flux leakage signal is processed by filtering. In order to measure the filtering effect of the minimum entropy deconvolution algorithm, the pep-to-peak values of the processed defect signals and background noise signals were compared with the wavelet transform and median filtering. The experimental results show that the minimum entropy deconvolution algorithm plays a significant role in enhancing the weak defect signal, and its effect is better than that of wavelet transform and median filtering.

Abstract:The liner is the middle layer of the solid rocket motor bonding propellant, and its curing state affects the bonding quality of the interface. Therefore, online monitoring and judging the curing state of the liner is the premise to ensure the engine charge and safety quality. In this paper, using laser ultrasound under the ablation mechanism as a means to study the monitoring method and curing state characterization technology of butylated hydroxy lining during the curing process, the online monitoring of the curing process of the lining is described in detail. The time-domain and frequency-domain bispectral characteristics of the echoes are analyzed emphatically, and the propagation time, velocity and nonlinear harmonic third-order spectral peaks of the characteristic waves of the laser ultrasonic longitudinal wave in the lining of the multilayer bonded structure are extracted as the characteristic quantities. , which characterizes the state change law of the liner curing process. The results show that in the time domain information of the characteristic wave, the progress of the curing reaction will lead to a sequential decrease of the arrival time of the characteristic wave between different states by 0.02 µs, and the propagation sound speed of the laser ultrasonic longitudinal wave in the lining layer is successively increased by about 40 m/s; In the frequency-domain bispectral information, the third-order spectral peaks of nonlinear harmonics gradually increased with the progress of the curing reaction, and the ranges of different states were 0.1167, 0.9799, and 0.6360in proper sequence. In conclusion, the online monitoring experiments and laser ultrasonic parameters in this paper can effectively monitor and characterize the curing state of the lining, respectively.