Abstract:To design a high-voltage solid-state switch to protect the protect the electron cyclotron tube of J-TEXT Tokamak, the new wide-band-gap semiconductor SiC-MOSFET plays a central role with outstanding electrical characteristics such as low switching loss, high voltage withstand and switch frequency at MHz level. However, the extremely fast switching speed aggravates the switching oscillation, which causes the decreasing quality of the switching waveform when the devices are in straight series. Therefore, it is necessary to develop a precise model to provide guidance for the application of SiC-MOSFET in high-voltage switch. Based on the commercial model developed by CREE, an optimized dynamic characteristic SiC-MOSFET model that can better simulate the influence of nonlinear junction capacitance is proposed. The optimized model comprehensively considers the factors of gate-source voltage and drain-source voltage. A variety of functions are used to reproduce the nonlinear junction capacitance, and the key parasitic parameters such as stray impedance are validated. Through the simulation and 300 V/3 A double pulse test, it is proved that the optimized model has higher accuracy in the change rate of drain-source voltage, change rate of drain-source current, oscillation frequency and turn-off peak voltage. The optimization model is also applied to design the topology of voltage sharing and overvoltage suppression for high-voltage solid-state switch, and the switching module is tested under the condition of 1400 V/700 A. On the key indicators such as turn-off time and peak voltage, the simulation waveform is highly consistent with the experimental waveform. The measured turn-on and turn-off time is less than 120 ns, which meets the design requirements of switching module, and the practicability of the optimization model is reflected.

Abstract:Light buoy is a navigation aid sign to guide the navigation of ships, which plays an important role in ensuring the navigation safety of vessels. In order to predict the light buoy's offset position, and provide accurate buoy's position for shipping, a mathematical model of light buoy's offset prediction is established using multiplicative seasonal ARIMA model based to the seasonal time series characteristics of buoy's position data. Taking the No. 1 light buoy in Meizhou Bay as an example, the model is used to predict the offset position of the light buoy and compare with the actual position data. The results show that the multiplicative seasonal ARIMA model performance well, and the average absolute percentage errors of azimuth and shift distance are 1.79% and 7.37% respectively. This provides a new idea for the prediction of light buoy offset position.

Abstract:A single-channel fetal electrocardiogram (ECG) monitoring system based on Android mobile smart terminal is designed to realize single-channel fetal ECG extraction algorithm using twice Singular Value Decomposition (SVD), solve the problem of missing waveforms in fetal ECG extraction and obtain real-time fetal ECG signals with high signal-to-noise ratio. Using STM32 single-chip microcomputer to control the 24-bit sampling chip ADS1298, the single-channel pregnant woman's abdomen signal is collected. Then the collected data is transmitted to the mobile intelligent terminal via Bluetooth. On the Android-based mobile smart terminal, the functions of real-time extraction, display, storage and analysis of fetal ECG, calculation of heart rhythm variability, and control of the entire monitoring system are completed. The test results show that the system can accurately extract fetal ECG signals from single-channel pregnant women's abdominal signals, with an improved signal-to-noise ratio of 29.43dB, an accuracy of 98.01%, and a positive prediction rate of 97.81%. The system has good stability and compatibility, and takes about 3.671ms to continuously process 5 maternal cardiac cycles, which is much less than the time of a maternal heartbeat cycle (about 0.8s), and is suitable for maternal or fetal arrhythmia. The fetal ECG monitoring system has the characteristics of simple connection, easy operation, high accuracy, strong portability, etc., and is suitable for primary hospitals and families.

Abstract:To improve the performance and applicability of license plate recognition, a license plate recognition method based on foreground polarity detection and improved convolutional neural network (CNN) is proposed. The proposed method consists of two main modules: character segmentation module and character recognition module. In the character segmentation module, foreground polarity detection based on RGB color is used for binarization and ROI segmentation, and then character height estimation and skew correction are performed. In the character recognition module, the depth features are extracted by the multi-channel deep CNN framework including the aggregation module to improve the representation ability of the output features. The experimental results show that the proposed method has good recognition accuracy, and the recognition rate is 92.2% and 94.1% respectively on the more difficult SSIG test set and aolp data set, and it is superior to the commercial vehicle access control system in some extreme cases.

Abstract:In order to ensure the operation safety of industrial machinery and equipment, avoid vibration damage, realize the real-time monitoring of mechanical equipment vibration signals, and improve the real-time performance of vibration monitoring, a vibration acquisition and monitoring system was designed, which took the FPGA of Altera EP4E115F29C7N as the digital control unit, and carried out data transmission through Ethernet UDP protocol. The system adopts multi-point vibration signal acquisition, the collected signal is pre-processed and stored to external SDRAM, and then the stored data is packaged into UDP data packet format and uploaded to the computer through the network cable. Finally, LabVIEW is used to upload data filtering to the lower computer system and display the vibration waveform of mechanical equipment. Ethernet communication technology to reduce the vibration data in long distance transmission delay loss, in the vibration of mechanical equipment in the running state of the system can realize high speed data transmission, real-time monitoring, vibration signal to realize the remote on-line monitoring vibration signal, will help the staff for the remote real-time monitoring of machinery vibration state, It is suitable for inconvenient collection and complex working conditions.

Abstract:Aiming at the problem that the smooth metal furnace edge of the test sample support platform has extremely strong scattering during the reflectance test of high-temperature materials, resulting in a greatly reduced reflectivity test accuracy, a method of structural transformation of the heating furnace edge is proposed, and different depths and widths are designed. RCS simulation of the scaled model of the designed structure is carried out through the FEKO software on the V-shaped and U-shaped furnace edges of the furnace. The simulation results of three incident angles show that the V-shaped and U-shaped groove edges of depth 2 have obvious RCS reduction effects at the frequency points in the frequency band around 24-40 GHz. Compared with the flat furnace edges, the V-shaped and U-shaped groove edges have a significant RCS reduction effect. The furnace edge has been reduced by about 8.6dB and 9.8dB respectively, which indicates that the reflectance test of the material after the structure optimization can improve the test accuracy.

Abstract:Aiming at the problems of high failure rate, low power density, and poor economy in parallel configuration of multiple independent DC-DC converters in power generation systems containing new energy, an intensive high-gain three-port DC-DC circuit is designed. The converter is respectively connected to the power supply unit, the energy storage unit and the load, and can realize energy flow to any two ports. The secondary boost circuit is used to obtain a high-gain voltage output. First, analyze the principle of the three working states of the proposed three-port converter, then calculate the voltage gain of the converter and the stress of the switching device. Finally, an experimental prototype with a power of 200W was built to simulate and verify the theoretical analysis, the experimental result is that when the input voltage is 25V, the inverter can get an output voltage of 120V in three working modes. The charging current of the energy storage battery is 3A and the discharging current is 4A. The experimental results show that In the new energy power generation system connected to the energy storage unit, the converter can ensure stable and high-gain voltage output.

Abstract:The radiated immunity measurement of ICs has become a major factor limiting the performance of electronic equipment. Up to now, the radiated immunity measurement methods of IC mainly include TEM cell method, GTEM cell method, IC stripline method and near-field scanning immunity method, but how to choose these methods is a dilemma. Moreover, with increasing frequency spectrum of disturbance sources to GHz range, some methods present their problems as well. In aspects of test set-up and pros, four IC radiated immunity measurement methods are described in this paper, with their frequency range, interference field strength and cost compared and analyzed. From test set-up to adjustment of test condition, the whole steps of IC radiated immunity measurement are induced. In addition, the key problems and solutions of the four methods are listed. Finally, this paper puts forward the principles and suggestions for methods selection. Being the reference for testers, the research results can be used for the selection and improvement of IC Radiated Immunity measurement.

Abstract:With the popularization of new energy vehicles, in response to the market's increasing requirements for battery management systems, BMS (Battery management system) applied to pure electric vehicles has been designed. The system takes both cost and function into consideration, with STM32 single-chip microcomputer as the main control chip, using analog front-end chip, real-time detection of monomer and overall voltage. The thermistor is matched with a multiplexer to realize single point temperature collection. The Hall element can collect current information in a non-contact manner. The equalizing circuit automatically equalizes the capacity difference between the batteries according to the information collected by the system. In addition, the system also has insulation detection and communication functions. Use the extended Kalman filter algorithm to estimate the system SOC. Research experiments show that the system has high accuracy and reliability for voltage, current and temperature detection and SOC parameter estimation, which improves battery safety and efficiency. The system has complete and reliable functions, and has certain practical engineering significance.

Abstract:Grounding resistance is an important parameter to measure the stability of power system. If the grounding system is buried in the soil with poor contact with the electrode, the resistance reduction effect of simple graphite grounding device in complex soil environment is general, which is difficult to meet the engineering requirements. conditions. Based on the principle of reducing resistance of grounding device, this paper puts forward a method of reducing resistance of grounding cloth to increase the area of stray current, and designs a grounding scheme of grounding device plus grounding cloth. Firstly, the simulation analysis is carried out, and the grounding experiment of graphite grounding device with grounding cloth is carried out in the simulation test platform. Several factors such as the location of grounding cloth, the depth of embedding and the amplitude of impulse current are mainly measured. Through the comparison of simulation test and simulation, the change of grounding resistance is measured, the influence law of various factors on grounding resistance is summarized and analyzed, and the auxiliary current dispersion scheme of using grounding cloth to reduce grounding resistance is proposed, which provides reference for reducing grounding resistance.

Abstract:The hand-held 3 D laser scanner is affected by a variety of error factors, which makes the collected point cloud and the accuracy of the final measurement results, so the result needs to evaluate the uncertainty of the measurement results. According to the working principle of the scanner, the source analyzed the uncertainty, establish the uncertainty model and evaluate GUM, MCM and adaptive AMCM. The uncertainty evaluation results of the measurement plate planness experiment show that the MCM method fully considers the distribution type of input quantity, and the uncertainty evaluation result is more accurate than the GUM method; the AMCM method can effectively balance the contradiction of sampling number M too large or too small, and the process of measurement uncertainty evaluation is more convenient and efficient than the GUM and MCM methods.

Abstract:In order to achieve high accuracy in hybrid fault diagnosis of rotating parts, a hybrid fault diagnosis method based on information fusion of two channels under variable working conditions is proposed in this paper. The signal includes both rolling bearing and gear vibration signal. The vibration signal of channel 1 is generated by generalized S transform, and the feature map is used as the model input of channel 1. In channel 2, the time-domain signals of the rotating parts are taken as the characteristic input, and the two-channel output layer is randomly fused with the features. By fine-tuning the parameters of the whole two-channel convolutional neural network (CNN) model, the diagnosis and identification of the mixed fault state of the rotating parts under varying working conditions are realized. The results show that the proposed method can be effectively applied to the mixed fault identification and diagnosis of rotating parts. Compared with the one-dimensional and two-dimensional convolutional neural network and other machine learning methods, the proposed method has the highest fault identification accuracy, reaching 98.18%.

Abstract:When the cascaded H-bridge converter operates unbalanced under active thermal control, the clamping H-bridge module generates a large number of harmonics around twice the carrier frequency. Aiming at this problem, a harmonic performance optimization strategy under active thermal control of the cascaded H-bridge converter is proposed. The harmonic distortion of the converter output voltage with the clamped H-bridge module under thermal control is analyzed in-depth, and an unconventional PSPWM based on the carrier angles correction of the H-bridge module is designed to eliminate the sideband harmonics of the double carrier frequency. Experiments were carried out using a prototype of the cascaded H-bridge converter, and the test result is that after adopting the harmonic performance optimization control, the dominant harmonics around twice the carrier frequency were eliminated, and the temperature of the clamping H-bridge module was reduced by 0.9°C. The test results show that the novel harmonic performance optimization strategy under the active thermal control can effectively improve the output harmonic performance of the cascaded H-bridge converter system and realize the thermal equilibrium of the H-bridge modules.

Abstract:In order to solve the problem that continuous wave active sonar is easily exposed by the sonar array when detecting targets, an active non-linear frequency modulated continuous wave ranging method based on digital delay and resampling technology is proposed. The continuous wave measurement signal with non-linear frequency variation is generated by the signal generator. After the signal is returned by the measured object, it is mixed and filtered with the original signal to obtain the measurement mixing signal. According to the digital model of the original excitation signal, a reference signal with digital delay is constructed along the way. The reference signal and the echo signal are mixed and filtered to get the reference mixing signal. The reference mixing signal is used for zero-crossing resampling and spectrum analysis of the measured mixing signal, the main frequency of the signal is extracted and then the distance of the measured object is calculated. The underwater resampled nonlinear frequency-modulated continuous wave ranging experiment system is built. and according to the laboratory measurement results, the repeatable standard deviation is only 0.083 m and the average error is better than ±8.5%, which verifies the feasibility of the ranging method, the main sources of ranging errors are further analyzed. Through the comparison and simulation of the generalized cross-correlation time delay estimation algorithm, it is confirmed that the propagation time delay cannot be accurately estimated by using this method, and the safety of ranging is improved.

Abstract:The measurement of liquid physical parameters is one of the most frequent measurement operations in many fields, which provides raw data for the analysis and decision-making of scientific research and production. However, most of the existing instruments are contact, manual, single parameter measurement, with low efficiency and the risk of contamination of test samples. Therefore, a non-contact automatic measurement device for liquid physical parameters is designed. The device is mainly controlled by STM32 microprocessor. Based on the principle of flat capacitance, flexible PCB and TI company's fdc2214 capacitance sensor chip are used as the core. The liquid volume measurement is realized by fitting calculation. At the same time, the liquid quality is measured by the resistive film pressure sensing circuit. The experimental results show that the errors of liquid volume and mass measured by the device are 2.31% and 1.07% respectively, which can meet the requirements of liquid measurement in most cases. It provides a new method for automatic liquid measurement in biochemistry and other occasions.

Abstract:The imaging results of positron emission tomography (Positron Emission Tomography, PET) equipment are often constrained by some factors such as tracer dose and scanning time, resulting in the image quality decline and affecting doctors’ diagnostic results. At present, improving the quality of PET imaging with artificial intelligence (AI) technology is a hot research topic. This paper aims at the problems of existing methods such as many training parameters, loss of shallow information, loss of texture details, etc., and proposes a method based on residual hybrid domain attention. The PET super-resolution reconstruction method of force network. This method designs a lightweight convolutional network, in which the residual learning structure is added and the mixed domain attention block is incorporated. While enhancing the interaction of the neural network, it also increases the attention to the high-frequency information area then quickly reconstruct the high-frequency details of the image. The data set includes open source data in the network and clinical data obtained from hospitals. As a result, a super-resolution data set of PET images is established for training and testing. The experimental results show that the test results of the algorithm in this paper and the comparison network are significantly improved. When the scale factor is 4, compared with CARN (Cascading residual network), the average values ​​of PSNR and SSIM are increased by 0.09dB and 0.0009, respectively. In addition, the number of parameters is reduced by 50.26%, which effectively improves the reconstruction efficiency of the model.

Abstract:The breakthrough of deep learning in the field of image makes the rapid development of feature learning. Aiming at the temporal correlation of consecutive frames in video sequences, a residual 3D convolutional network model based on attention mechanism is proposed for human action recognition. Firstly, residual 3D convolution network is used to learn the temporal correlation between consecutive video frames in video sequence. Then, each feature channel learned by residual 3D convolution structure is given different weights by using channel attention network which is extended to three-dimensional. Finally, the reweighted features are input into the classifier to get the final classification. Experiments are carried out on UCF-101 and HMDB-51 datasets, and the accuracy is 95.8% and 69.7%, respectively. The experimental results show that the proposed model has high recognition accuracy in video human action recognition.

Abstract:To overcome the shortcomings of existing methods in the segmentation of solar cell images, an improved U-Net structure for defect segmentation of solar cell images is proposed. First, the dense connection structure is introduced to alleviate the problem of gradient disappearance and make defect extraction more fully; at the same time, a batch normalization layer and Relu layer are added after each convolution layer to prevent the loss of defect details; then a dual attention mechanism is introduced. To enhance target features and suppress irrelevant features,improving the overall detection accuracy of the model. Finally, two different networks are used to be compared with the method in this paper. The experimental results show that the network can obtain more detailed feature information, which further improves the accuracy of solar cell image segmentation.

Abstract:Aiming at the fact that the importance of global features is not clear in the text detection of natural scenes, which leads to the misdetection and missed detection of text in the text detection process, a natural scene text detection method based on attention mechanism is proposed. Based on the CTPN network, this method uses the ResNet network and feature fusion technology to extract deeper multi-layer network text features; at the same time, the attention mechanism is introduced into the improved feature extraction network, which is enhanced by the same features gathered from all positions The original features, and the attention weight is obtained, the global attention is collected, and the features that need attention are clarified. Secondly, for the problem of low text positioning accuracy in natural scenes, GIoU (Generalized IoU) loss is used instead of coordinate loss, and the Focal Loss loss function is introduced to improve the original loss function. Experiments show that this method obtains a recall rate of 83%, a precision rate of 87% and an F value of 85% in the text image detection of natural scenes, which ensures the integrity of the text information in the text detection process.

Abstract:To improve the anti attack ability and adaptability of the watermarking scheme, a blind watermarking scheme based on GoogLeNet is proposed. Firstly, the proposed network is relatively simple, and the deepest path (that is, the path through preprocessing network, embedding network and extracting network) only contains 17 layers. The resolution of the host image is maintained by increasing the watermark resolution in the watermark preprocessing network, thus enhancing the transparency of the watermark. The average pooling is used in the watermark preprocessing network to combine the binary value of the watermark data with the host image properly, so it can enhance the transparency of the watermark. Finally, The extractor uses cross entropy as the loss function to achieve the training balance between the embeder and the extractor. The experimental results show that the performance of the proposed scheme is excellent, the watermark capacity is 0.0038, and the average PSNR in the dataset is 40.57 dB. The performance under meaningful attack is better than other advanced methods.

Abstract:Aiming at the problem that it is difficult to quantitatively detect the leakage pore size of pressure gas pipelines, a method for estimating pore size equivalent based on characteristic function is proposed. This method studies the data characteristics of the micro-hole leakage noise of the pressure gas pipeline under constant pressure and temperature conditions as the leakage signal source in the 40kHz ultrasonic frequency band, and extracts the characteristic parameters of the leakage signal in the time domain and frequency domain ,established the functional relationship between each characteristic parameter and the leakage aperture, and selected the optimal characteristic function through error analysis,Combined with MATLAB signal processing, the aperture size of micro-leakage is estimated. The experimental results show that the Hilbert relative energy of the leakage signal and the aperture size are fitted to the characteristic function of the third order, and the error between the estimated value of the leakage aperture and the actual value of the leakage aperture is within 0.05mm. The purpose of the study on the estimation of the leakage aperture equivalent is provided, and It can be used for reference in quantitative analysis of the equivalent of micro leakage holes in gas pipelines.

Abstract:Aiming at the problem that the bearing fault characteristic signal is susceptible to noise interference, which leads to the difficulty of extracting the bearing fault impact characteristic signal. A bearing fault diagnosis method using the combination of the Chaos Sparrow Algorithm (CSSA) and the Maximum Correlation Kurtosis Deconvolution Algorithm (MCKD) is proposed. First, construct the adaptive function of CSSA based on the principle of kurtosis. Then, the CSSA algorithm is used to find the optimal period T and filter length L. Finally, the optimized MCKD algorithm is used to extract the faults of the motor bearings. And compared with unoptimized MCKD, particle swarm optimization optimization maximum correlation kurtosis deconvolution algorithm (PSO-MCKD), and Sparrow algorithm optimization maximum correlation kurtosis deconvolution algorithm (SSA-MCKD). The experimental results show that the CSSA algorithm has a faster convergence rate and better global search ability when searching for MCKD parameters compared to the particle swarm optimization (PSO) and the sparrow algorithm (SSA) algorithm. The proposed CSSA-MCKD method can effectively enhance the fault extraction ability of the MCKD algorithm, and has a faster convergence speed and global search ability.

Abstract:Aiming at the problems of high dimensionality and too much invalid information in traditional signal feature extraction methods, this paper proposes a switchgear partial discharge pattern recognition method based on linear discriminant algorithm and radial basis neural network. This method combines the two algorithms to achieve the purpose of both the recognition rate and the recognition accuracy. First, establish three ultrasonic partial discharge (PD) models of the switchgear. Then, time-frequency analysis and wavelet decomposition are used to extract the time-frequency features and wavelet coefficient features of the signal, and the extracted feature vectors are reduced by linear discriminant algorithm (LDA). Finally, the radial basis basis (RBF) neural network is used to analyze the local The types of discharge defects are classified, and the recognition accuracy is above 90%, and the training time is reduced by more than 50%, which proves that the recognition method is practical.

Abstract:In order to facilitate the collection of multi-channel signals that do not share the same ground and prevent crosstalk between signals, an isolation design is added on the basis of the collection circuit: isolated power supply, isolated optocoupler and digital isolation channel, which greatly reduces noise interference and improves the collection accuracy of the system . Based on the system circuit design, the working principles of the isolated power supply LTM8068 and the linear isolation optocoupler HCNR201 are analyzed, and the ROM table is established to realize the chip selection of the acquisition module and the switching of the analog switch channel, which facilitates control and improves the versatility of the acquisition board. The experiment verifies the sampling accuracy of the circuit by inputting a 0~5V analog quantity. After practical verification, the sampling accuracy of the isolated acquisition circuit is better than 2‰, and it has strong anti-interference ability, good circuit versatility, and strong practical value.

Abstract:Satellite faults is one of the main problems in meteorological polar orbiting satellite data processing system. Satellite faults, which is difficult to be monitored and solved, always have serious impact. Current solution adopts threshold algorithm to analyze the telemetry downloaded by satellites. Threshold algorithm works poorly in analyzing volatile telemetry channels. This paper propose a telemetry spectrum analysis based faults diagnosis algorithm, which performs better on volatile telemetry channels. Firstly, some key telemetry channels are FFT transformed, the spectrums are analyzed and some certain frequencies are determined. The basic logic of the algorithm is proposed. Secondly, this paper discuss how to determine the telemetry channel length N based on the orbit characters and monitor requirements. Some typical satellite faults are simulated and diagnosed by this proposed algorithm. Lastly, this proposed algorithm is tested by an authentic satellite fault in history. Tests result shows, this proposed algorithm can diagnose meteorological satellite faults quickly.

Abstract:Aiming at the problems of uneven cutting, low accuracy and manual adjustment of discharge of traditional cascade chicken house feeding equipment, this paper proposes a new method to solve the problem.This paper designs a chicken house intelligent feeding control system based on Programmer Logic Controller(PLC) . The system uses a low-voltage brushless servo motor instead of the traditional AC motor to drive the traveling crane, and uses a stepping motor instead of the manual adjustment board to control the auger discharge.This paper first introduces the composition of the system, then introduces the hardware design and software design of the control system, and finally introduces the breeding process design and the experimental test of the control system.The whole control system takes the discharge control and data analysis as the core, realizes the nine axis feeding and growth parameter analysis function, realizes the process and data visualization through the touch screen, and finally forms a set of fully automated intelligent feeding system for chicken house.This control system improves the feeding accuracy of the equipment to within 10g, improves the feed uniformity, realizes the unmanned breeding in chicken house, and provides a reference method for the intelligent development of animal husbandry.

Abstract:In order to study and optimize the platform orienting sign system of high-speed railway station, particle swarm optimization (PSO) is introduced to build the platform model. A new "air sign" system is added to the original high-speed railway landmark system. The platform simulates passengers to find the corresponding location. After adding the new sign system, PSO is used to verify that the model has higher fitness and convergence speed, and its performance is better than that of the landmark only guidance system. The optimization of platform guidance sign will use esp8266 wireless module for information transmission and display screen to display information. The new "air sign" system can be combined with dispatching information to display the current platform car number information in real time. In the training of PSO, the iterative time of the optimized guidance identification system is reduced by 44.8%, and the effect is more significant.

Abstract:The nitrogen may cause eutrophication in water bodies. Monitoring and controlling the concentration of total nitrogen in water bodies is the key to ensuring the safety of the water environment. The total nitrogen content in the water body includes nitrogen in nitrite nitrogen, nitrate nitrogen, inorganic ammonium salt, dissolved ammonia and most organic nitrogen compounds. According to environmental protection requirements, total nitrogen online analyzer has been widely used in surface water and pollution source on-line monitoring. This paper summarizes the current status and future development of total nitrogen online automatic monitoring equipment, introduces the basic principles of total nitrogen detection, water sample treatment and digestion methods, analyzes and compares the technical indicators of water quality total nitrogen automatic online monitoring equipment. According to china's water environment governance development, the total nitrogen automatic online monitoring technology and instrument should be development. This paper provides reference for the development and industrialization of related instruments.