Abstract:Aiming at the solving the problem of measurement error from the geometric error calibration in equipment assembly. A calibration algorithm based on improved particle swarm is proposed for the calibration of geometric parameters of 3D laser scanner with orthogonal axes. First, the geometric error parameters to be optimized are constructed based on the improved DH parameter model of the 3D scanning equipment, and the particle swarm algorithm is used to perform iterative optimization within the constraints to determine the calibration value. Then, on the basis of traditional particle swarm optimization, the optimization of dynamic parameters (inertia weight, dynamic adjustment of learning factor and fitness function improvement based on dynamic plane fitting of global least squares algorithm) is carried out to solve the problem that the algorithm falls into local optimum. At last, three different methods are used to carry out the calibration experiments based on the standard checkerboard calibration board. The comparative analysis of the experimental results shows that the dynamic plane fitting-parameter improvement PSO algorithm proposed in this paper greatly improves the convergence speed of the calibration algorithm and the reliability of the fitness calculation, and can quickly calibrate the equipment. The accuracy is also greatly improved. This calibration method provides a reference for the calibration of geometric parameters of other orthogonal axis systems.

Abstract:Aiming at the problem of energy inconsistency during the charging and discharging process of lithium battery packs, this paper proposes a two-level equilibrium topology, which is divided into inner and outer groups. The inductance-based ring structure equalization circuit is used in the battery pack, which realizes a new type of active equalization of bidirectional ring transfer of energy between adjacent single cells and the head and tail cells. A centralized balance topology based on a single inductor is used outside the battery pack, which can achieve balance between any battery packs between the packs. In terms of balancing control strategy, taking the battery state of charge as the balancing variable, a fuzzy logic control algorithm is designed to dynamically adjust the balancing current to reduce the balancing time and improve the balancing efficiency. Using MATLAB/Simulink software to build and simulate the model, the experimental results show that the energy transfer topology proposed in this paper reduces the equalization time by 24.46% compared with the traditional Buck-Boost circuit energy transfer topology between adjacent cells. In addition, compared with the fuzzy logic control algorithm using the fuzzy logic control algorithm under static and charge-discharge conditions, the standard deviation of single cells after equalization decreased by about 11%. The feasibility of the equalization scheme is verified.

Abstract:The ground penetrating radar technology has been widely used in the rapid and precise detection of hidden diseases in urban roads and underground spaces. However, due to the complex interference of urban environment, the ground penetrating radar data is mixed with noise and clutter, resulting in low signal-to-noise ratio of data and affecting the processing and identification accuracy. In order to improve the signal-to-noise ratio of ground penetrating radar data and obtain high-quality detection data, this paper proposes an improved wavelet threshold denoising algorithm based on particle swarm optimization algorithm on the basis of the traditional wavelet threshold denoising algorithm. Through the use of MATLAB and gprMax2D tools to carry out the denoising simulation experiment. The experimental results show that compared with the traditional soft and hard threshold denoising algorithms, the signal-to-noise ratio is increased by 28.02% and 6.97% respectively, and the mean square error is reduced by 71.86% and 31.88% respectively, which has better denoising effect. Applying the algorithm proposed in this paper to the data processing process of ground penetrating radar can provide technical support for the safety of urban roads and underground space.

Abstract:In order to solve the problem of RFID communication security, this paper proposes a new physical layer tag ant-counterfeiting technology. This technology separates the original signal to obtain the expected, noise and normalized signals, and extracts the spectral entropy features of the three signals.Finally, feature selection is used to achieve true and false label classification. In addition, this paper proposes a new cross-validation to objectively test the performance of the physical layer method. The results show that the accuracy of the method in this paper is nearly 4% higher than that of the traditional method. Under the new cross-validation, the classification accuracy of the physical layer method will drop by 8~10 percentage points. From this, we get an important conclusion, it is of great significance to use the spectral entropy feature in the physical layer identification method to realize the label anti-counterfeiting.

Abstract:Most of the existing laser inertial navigation odometers adopt the filtering loose coupling fusion method, and there will be a certain motion estimation drift in large scene mapping, which will lead to the reduction of positioning and mapping accuracy. Aiming at this problem, a close-coupled odometer and mapping method of laser inertial navigation system based on graph optimization is proposed. At the front end, point cloud distortion compensation, point cloud clustering segmentation, ground and feature extraction are carried out in turn. At the back end, the map optimization method is used to integrate IMU pre-integration, laser odometer and loop detection information to complete the map construction. Finally, Kitti data set and self-collected data are used to compare and analyze LOAM, LeGO-LOAM and the method of this paper in odometer accuracy and loop detection effect. Experimental results show that compared with LOAM and LeGO-LOAM, the positioning and mapping accuracy of this method is improved by 45% and 35% respectively, and it has better robustness.

Abstract:The super regeneration receiver model is upgraded to 433 MHz in ISM band to solve the problem that the traditional dynamic model of super regeneration oscillator works at a low frequency and has no practical reference value. The chaotic dynamic characteristics of the improved model are studied by using the numerical simulation method of the software. The Lyapunov exponent method is used to quantitatively detect the chaotic state of the model, and the detection ability of receivers in different states at this frequency is studied. The simulation results show that the 433 MHz super regeneration receiver can produce chaos when controlled by a low frequency quench signal. The sensitivity of the chaotic super regeneration receiver in this frequency band is about 3 dB higher than that in the periodic state. This proves that chaos can be generated and applied to the super regeneration receiver in the actual ISM band. This provides a feasible scheme to improve the weak signal detection capability of the super regeneration receiver.

Abstract:A method for estimating line ice cover weight based on image texture features is proposed to address the issue that different types of ice cover formed under different environmental conditions can pose different degrees of threat to transmission lines. The line ice cover texture features are first weighted to create a fusion feature, which is then combined with spatial neighborhood information to estimate the thickness of the line ice cover. Next, a weight recognition model is built while taking into account how meteorological factors affect the type of ice cover formation. The results demonstrate that even when the environmental conditions change significantly and the type of generated overburden changes, the model can still estimate the overburden weight more accurately with an average absolute percentage error of only 2.246%.

Abstract:Aiming at the impedance mismatch between the piezoelectric ultrasonic transducer and the excitation source in the rail damage detection system, as well as the resonant frequency shift of the piezoelectric ultrasonic transducer after loading, this paper studies the broadband impedance matching of the piezoelectric ultrasonic transducer. Firstly, the parameters of the piezoelectric ultrasonic transducer BVD model are optimized. Secondly, four matching networks are designed based on the above models, and their performances are compared and analyzed to determine the optimal network topology and parameters. Finally, based on experimental platform of pitch-catch, the received signals of piezoelectric ultrasonic transducers before and after matching are tested and compared. The test results show that the impedance matching network designed in this paper can achieve efficient matching of piezoelectric ultrasonic transducers, and the active power of the transmitter transducer is increased by an average of 197.38% after matching, and the operating bandwidth is increased from 1.44 kHz to 2.22 kHz. Compared with the unmatched and traditional L-shaped matching network, the broadband impedance matching network designed in this paper increases the emission intensity by 61.07% and 18.05% respectively, and the piezoelectric ultrasonic transducer can achieve a good working condition within the working bandwidth.

Abstract:Aiming at the problems of too many detection and routing nodes and large environmental fluctuation in the large-area application of greenhouse environment monitoring and control system, the greenhouse environment monitoring and control system was designed by using an inspection vehicle and LoRa remote transmission network. The number of nodes was reduced by combining fixed-point and mobile acquisition. When the system adjusts the environment, it first determines the growth stage of crops, and then dynamically adjusts and controls them by the Fuzzy-PID algorithm, so that the environment is always maintained in the most suitable state, thereby enabling high quality and high yield of crops. To verify the superiority of the system, taking the temperature control as an example, the simulation model is constructed by MATLAB. Compared with the traditional PID control, the overshoot is reduced by 15.3% and the adjustment time is reduced by 21%. The field experiment results show that compared with the switch control and the traditional PID control, the system has higher control accuracy and smaller environmental fluctuation, which can effectively keep the environment in the shed stable.

Abstract:In order to compensate for the end effector attitude error caused by insufficient torsion of the robot joint and the end effector connection, a method based on inertial measurement unit to obtain the end attitude online is proposed. First of all, the motion process of the entire robotic arm system is divided into static and dynamic processes. At static, due to the small external acceleration noise, a method for estimating the attitude angle of the end effector based on local gravity using an accelerometer is proposed. In dynamic time, an adaptive extended Kalman filtering algorithm based on noise statistics is proposed for the problems of external acceleration noise, gyroscope zero drift, and scale factor error that affect the measurement accuracy. Based on the measurements of the accelerometer, the weights of the observed noise variance array are updated to adjust the Kalman gain and reduce the effect of acceleration noise on the measurement accuracy. Experimental results show that the average attitude angle error estimated by the static algorithm is 0.07°, 0.05°, 0.2°; In dynamic time, the proposed algorithm can compensate well for the influence of external acceleration on attitude, and can effectively improve the attitude measurement accuracy, compared with the EKF algorithm, the average error of attitude angle is reduced by 2.69°, 1.01°, 0.5°.

Abstract:Many existing signal recognition methods for steady-state visual evoked potential (SSVEP) do not pay sufficient attention to the importance of the phase features. In this paper, an improved extended canonical correlation analysis (eCCA) method is proposed for SSVEP signal identification. The phase parameter in the stimulus paradigm of joint frequency-phase modulation coding is added to the reference signal constructed from subjects′ training data as a way to achieve phase constraints on eCCA, thus improving the recognition performance of the eCCA method for SSVEP signals. Thus the eCCA-based SSVEP signal recognition performance is improved. To verify the effectiveness of the proposed method, SSVEP signal recognition experiments are conducted on a publicly available dataset and compared with the existing signal recognition methods. The experimental results show that the average recognition rate of the proposed method is improved to 82.76%, and the information transmission rate is reached to 116.18 bits/min with better stability.

Abstract:In the process of three-dimensional measurement and reconstruction, the extraction of the center of line laser stripe is particularly important. Steger algorithm can not meet the real-time requirements of industry due to its huge amount of calculation. In order to solve the problem that Steger algorithm has a large amount of computation and low extraction efficiency, a line laser center extraction method based on the improved Steger algorithm flow is proposed in this paper, which is mainly used to improve the extraction efficiency of light bar center. Firstly, the image is preprocessed by filtering and morphological methods to remove noise. Secondly, the gray image is thresholded to reduce the complexity of the image. Finally, the redundancy of the algorithm is reduced by extracting the region of interest, roughly estimating the width of the light bar, and then the Steger algorithm is used to extract the center of the light bar in the region of interest. Through the analysis of experimental data, compared with the traditional Steger algorithm, the improved Steger algorithm not only inherits the excellent stability of the traditional Steger algorithm, but also effectively improves the efficiency of light stripe center extraction, laying a foundation for real-time 3D reconstruction and measurement.

Abstract:The cleaning of the atmospheric electric field is the key step of pretreatment, which is of great significance to the subsequent excavation research. In view of the shortcomings of traditional anomaly detection algorithm, which needs to specify the corresponding parameters and fail to use the relevant information between time series, a new outlier detection and correction method based on the combination of isolation forest and Chen-Liu algorithm is proposed. The method uses ARIMA model to combine the atmospheric electric field to get the fitting residual. The isolation forest model is constructed based on residual sequence to determine the location of the outliers. Finally, the Chen-Liu algorithm is used to correct the outliers. The reliability of the proposed method is verified by simulation series and the atmospheric electric field test. Compared with the original prediction, the results of the prediction of the series of thr atmospheric electric field after cleaning are improved by 27.8% and 34.98% respectively in root mean square error and mean percentage error.

Abstract:Aiming at the problem that the structure from motion algorithm needs to change the camera position in the process of camera photographing, the relative motion relationship between the target object and the camera is used to obtain image sequences containing multi angle views of target objects. A structure from relative motion algorithm is designed and implemented by locating the region of the target object in the image sequence and constraining the applicable range of feature point extraction. The experimental results show that SFRM algorithm is suitable for the photographing mode of fixed position camera. SFRM can increase the proportion of matching pairs located in the target object area from 32% to 84% of the total matching pairs. Compared with the traditional SFM algorithm, the time efficiency is improved by 38.29%. The available scenarios of SFRM algorithm is successfully expanded.

Abstract:In hybrid energy storage systems, state estimation is the basis for power allocation and control strategy adjustment among the storage elements. In order to reduce the impact of state estimation errors on the subsequent energy management, a hybrid energy storage system "li-ion battery-supercapacitor" is used for modeling and simulation and accurate state estimation. Firstly, after identifying the basic component parameters, a simulation model of the hybrid energy storage system is built in MATLAB/Simulink environment based on the physical structure of the circuit for fast and accurate simulation of the physical experiment. Then, considering the complexity and the accuracy of estimation, the battery pack and the supercapacitor pack are modeled in a holistic form, and the joint estimation of charge state and health state is carried out with the "EKF-two-point method". Finally, the experimental verification shows that the errors of the state estimation are within 5%, i.e., the simulation model established in this paper can accurately reflect the operating characteristics of the components and the accuracy of the joint state estimation algorithm is high.

Abstract:Traffic sign is an important guide for vehicles in the process of standardized driving. Traffic sign recognition is an essential and important content in the environmental perception of driverless vehicles. Based on YOLOX-S algorithm, this paper strengthens the features obtained from the feature extraction network by adding CBAM attention mechanism module at the end of the backbone network. Utilizes Focal Loss function to better eliminate the imbalance between positive and negative samples and focuses on samples difficult to classify. Using the GIOU Loss function, the problems of inconsistent optimization and scale sensitivity of the original loss function are solved, and the recognition accuracy of the model is further improved. In this paper, the proposed algorithm is tested based on TT100k data set, and the recognition effects are compared with which of several mainstream algorithms. Experimental results show that under the premise of high FPS, the detection accuracy of most target categories is improved. Compared with the YOLOX-S model, the coco accuracy evaluation index Map_50 of the proposed model increased by 1.9%, Map_50:95 increased by 2.1%, and FPS is 35.6. The effectiveness of the improvement is proved.

Abstract:This paper presents a target detection algorithm based on PointRCNN. This method is aimed at vehicle targets. Aiming at the problem that the original PointRCNN is poor in vehicle detection at a distance, the method is optimized and the average accuracy of target detection is improved. In the first stage, the lidar point cloud is processed by pseudo-image structure and dimensionality reduction to 2D, and then processed by Point-Focus structure and restored to 3D point cloud. Then it will be sent into the backbone of PointNet++ for feature extraction, classification and regression. In the second stage, 3D frame is optimized and selected, and Point-CSPNet structure is introduced to further improve network learning ability and robustness. In this paper, the Focus and CSPNet structures of YOLO series algorithms are used for reference. The effective information in the original point cloud is fully extracted and the feature, gradient changes in the network operation are effectively integrated to improve the detection accuracy of the network. The average accuracy of the improved algorithm is improved from 81.10% to 81.74% in 3D scenes of KITTI dataset; and it is improved from 86.87% to 88.20% in BEV scenes of KITTI dataset, and the detection effect of vehicle targets in the far distance of visual effect has also been optimized to a certain extent, which has certain positive significance for further optimization and improvement of unmanned driving technology.

Abstract:Wireless network control is a favorable factor to promote the intelligence of high-speed trains. As a typical time series, wireless network delay has strong randomness, large volatility and other problems leading to difficult prediction. In view of these problems, a wireless network delay prediction model with singular spectrum analysis-improved particle swarm optimization and LSSVM is proposed. The length of the window was first determined by the Cao method, the delay sequences were analyzed by singular spectral analysis to obtain a series of subsequences. Each subsequence was predicted using the LSSVM model optimized for the chaotic particle swarm. Finally, all the subsequence predicted values were superimposed to obtain the final prediction results, the simulation results show that the average absolute percentage error (MAPE), mean squared error (MSE) and average absolute error (MAE) are 2.8%, 1.055 and 0.44 lower respectively compared with the wavelet decomposition model. Compared with the EMD decomposition model, 7.4%, 3.377 and 1.118 decreased, respectively. Compared with the CEEMD decomposition model, it was reduced by 6.2%, 2.568, and 0.974, respectively. The accuracy was significantly higher than that in the other models.

Abstract:Junction field-effect transistor has a large input impedance and low noise, which has been widely used to design high performance transient electromagnetic sensor preamplifier circuit. However, the parameters fluctuate greatly between JFET devices of the same type, which makes it difficult to match when the JFET devices are differentially amplified. Aiming at this problem, a low-noise JFET differential amplification technique based on multi-discrete current sources is proposed in this paper. Combined with the characteristics of each JFET amplifier branch circuit, the constant current source is designed separately and adjusted to the optimal static operating point to eliminate the problem of uneven current flow and unreliable operation of the amplifier branch caused by the discreteness of JFET parameters. Finally, the LTspice simulation and actual circuit test results show that the differential amplifier circuit designed based on this technology can work reliably, and the current of each JFET branch is 4.68 mA, and the gain of the amplifier circuit designed based on this technology reaches 40.00 dB and the noise floor is 0.51 nV/Hz@1.10 kHz. The new method proposed in this paper can effectively eliminate the discrete problem of JFET parameters, and lay a good technical foundation for designing a differential parallel low-noise TEM sensor preamplifier circuit.

Abstract:In order to adapt to the increasingly complex wireless communication environment and make full use of the scarce spectrum resources, this paper designs and implements an electrically tunable microstrip bandpass filter based on centrosymmetric structure and varactor diodes. The electromagnetic simulation software HFSS is used to conduct simulation experiments. On the split ring resonator structure loaded with branches, an interdigital structure and a spur-shaped coupled feeder are introduced to complete the prototype design of the filter. On the basis of the original, a variable capacitance diode with adjustable capacitance value is added to adjust the low-frequency transmission zero point to realize the reconfiguration of the passband bandwidth, so as to achieve the purpose of flexibly controlling the filtering performance. The actual test shows that the initial relative bandwidth of the filter is 5.1%, the absolute bandwidth is 170 MHz, the absolute bandwidth adjusted by the varactor diode is in the range of 140~200 MHz, that is, 82.4%~117.6%, the center frequency is in the range of 2.70~2.76 GHz, and the transmission zero point tuning range is 2.57~2.63 GHz, the insertion loss in the passband is in the range of 0.9~1.5 dB, and the return loss is in the range of 10~35 dB. The test results are basically consistent with the simulation, and it has certain application prospects in the direction of precious bandwidth control in the S-band.

Abstract:Aiming at the problem that the performance of timing synchronization algorithm is poor under low SNR and is sensitive to frequency offset in the OFDM system, an enhanced algorithm is proposed. First, based on the property that the cross-correlation value of the CAZAC sequence with different root values is close to zero, a preamble sequence with centrosymmetric characteristics is designed, then a timing metric function insensitive to frequency offset is designed using symmetric correlation instead of delayed correlation based on the structure of the preamble sequence. Finally, timing synchronization is realized by capturing the peak value of the timing metric function. The results indicate that, compared with the existing algorithms, the proposed algorithm has a strong anti-frequency offset capability and better timing synchronization performance under low SNR conditions, and is suitable for communication systems with low-SNR and frequency offsets.

Abstract:Smart phone based indoor positioning has great potential usage in LbS (Location based Service) applications, due to the abundant sensors embedded in smart phones. This paper proposed a hybrid positioning method, which adopted a particle filter to fuse three types of independent positioning results, including image matching positioning result, WiFi fingerprinting results and PDR positioning results, which achieve continuous and accurate positioning results. According to real scenario experiments, the proposed method has an accuracy improvement of about 48%, 24% and 4% over pure PDR, WiFi fingerprinting and image matching based methods. Compared with classical WiFi and image matching fusion method, our method can provide a decrease of 92% average positioning time. Compared with another classical fusion method, the accuracy has increased by 0.7 m and the positioning time has decreased by 87%.

Abstract:At present, garbage classification has been widely concerned by the government and society. Facing the demand for real-time and accurate judgment of waste classification in the sorting process, a Yolo_ES target detection algorithm was proposed. The algorithm takes Yolov4 as the basic network. Firstly, EfficientNet is used as the backbone feature extraction network to realize the lightweight of the algorithm; secondly, the MBConv module is reconstructed by the attention mechanism ECA to filter out the highquality information, enhance the feature extraction ability of the model and reduce the number of parameters; at the same time, aiming at the problem that it is easy to lose detailed information in the max-pooling, the SoftPool is used to replace the MaxPool layer in the SPP module to retain more fine-grained feature information. The experiments are conducted on the self-made HPU_WASTE garbage classification dataset, and the results show that compared with the Yolov4 basic network, Yolo_ES model increases the map from 91.81% to 96.06%, and the model size is compressed by 75.45%. Meanwhile, the processing time of each image is 58ms; Compared with other target detection networks, this model has better robustness and better detection performance.

Abstract:Aiming at the problem of multiple extraction or incomplete extraction of the center position of the light strip caused by overexposure and underexposure of the fringe image in the three-dimensional measurement of complex objects, a light strip center line extraction method based on deep learning semantic segmentation technology is proposed. The method uses the improved UNet++ network for semantic segmentation, roughly extracts the center area of the light strip, and obtains the light strip center line with one to two pixel width, Then the sub-pixel center is accurately extracted by the gray center of gravity method. Experiments show that this method can effectively overcome the adverse effects caused by uneven exposure of light bar images, and accurately extract the complete and smooth subpixel light bar centerline of complex surface objects.

Abstract:In order to improve the accuracy of meter reading detection and recognition, an improved target detection network is proposed based on the lightweight and efficient YOLOv5s network. Firstly, in the feature extraction stage, CBAM attention mechanism is added to learn the important features of the image, and a feature fusion network D-BiFPN is designed to enhance the extraction of deep features; Secondly, the CIOU loss function is introduced to make the regression of the target box more stable. The backbone network of CRNN text recognition algorithm is improved. The model maintains the characteristics of lightweight, and has a good prospect in the deployment of mobile terminals. Finally, the test on the meter data set shows that compared with the original YOLOv5 algorithm, the average accuracy of the proposed algorithm is improved by 5.13%; Compared with the original CRNN algorithm, the accuracy of the proposed text recognition algorithm is improved by 7.4%. The experimental results show that the proposed text detection algorithm has high detection accuracy and stability, and the text recognition algorithm has high accuracy and speed, which can meet the application requirements of meter reading recognition.

Abstract:The existing underwater enhancement algorithms have some problems such as color distortion and bad defogging effect. Therefore, this paper proposes an underwater image enhancement algorithm based on dual attention mechanism and improved U-Net. Firstly, the color correction module is used to process the red, green and blue channels to reduce the influence of color deviation. Then, the channel attention and spatial attention are fused with the U-Net network, and the images after color correction are defogged and denoised to retain the texture details of the images and enhance the contrast. Finally, the pyramid fusion module is used to fuse the image features with different resolutions to obtain a clear visual image. The experimental results show that based on UIEBD and UFO-120 test sets, the average values of UCIQE, NIQE, SURF and information entropy are 0.608 1, 4.440 3, 31.5 and 7.649 5, respectively. The proposed algorithm is superior to other classical and novel algorithms in subjective visual quality and objective evaluation indexes. The enhanced underwater image has good defogging effect and obvious advantages in color correction, which significantly improves the visual quality of underwater images.

Abstract:For problems such as small size, low resolution and not obvious features of mobile phones in the monitoring images, it has brought difficulties to the study of the detection algorithm. This article proposes an improved YOLOv5 network model method to identify the use of mobile phones. The improved detection algorithm introduces the lightweight network GhostNet as the main extraction network, and the GhostConv module, the C3Ghost module instead of the Conv basic convolution module and the C3 module in the main network to reduce the network parameters and complexity; at the same time, the attention mechanism CBAM is introduced into the main network, reducing the effects of redundant characteristics after fusion, and extraction of more critical feature information in the target area; using four scale feature detection, the corresponding increase of the detection layer on the basis of the algorithm, to improve the detection accuracy of smaller targets. The experimental results show that, the accuracy of the improved YOLOv5 algorithm is 95.7%, and the mAP is 97.1%, the accuracy and mAP of the training increased by 2.5% and 1.8%, the calculation and parameters were reduced by 14.3% and 24.5%. The improved YOLOv5 algorithm not only has the advantages of lightweight, but also ensures the mAP and accuracy. This method provides theoretical basis and technical reference for the use of mobile phones in the intelligent monitoring technology industry.