Search

Aiming at the uncertainty of external disturbance and internal parameters of unmanned surface vehicle during navigation, based on the mathematical model of unmanned surface vehicle response, an extended state observer is designed to estimate the ‘total disturbance’ and complete the design of active disturbance rejection control law. For the difficulty of parameter tuning of active disturbance rejection control, the immune particle swarm optimization algorithm is combined with active disturbance rejection control to optimize the parameters of active disturbance rejection control. The simulation results of heading control show that the active disturbance rejection control with optimized parameters has faster rise time and stronger disturbance rejection ability.

In this paper, a formation control method based on backstepping and control force saturation constraint is proposed to solve the formation control problem of surface unmanned vessel with actuator input saturation. According to the control strategy of the virtual leader vessel, it can prevent the trouble of the leader vessel and improve the performance of the controller. The formation controller is designed by using the virtual leader strategy combined with the backstepping and the generalized saturation function to avoid the insufficient force of the actuator. The stability of the closed-loop system is analyzed by Lyapunov method, and it is proved that the whole system is uniformly final bounded, and the tracking error can be arbitrarily small by choosing reasonable design parameters. The simulation results show that the formation controller designed in this paper has strong anti-saturation ability and can effectively complete formation control.

With the continuous development of marine economy, marine ship detection has a very wide application prospect. To solve the problems of low ship detection accuracy and easy missing detection of small targets in complex marine environment, an improved YOLOv3 target detection algorithm is proposed. Firstly, before data training, the methods of image blending and small target enhancement are adopted to overcome the influence of complex environment background on target detection and the imbalance of small target data, and improve the network's ability to extract image features and detect small targets during the training process. Then, the convolution in the original network is replaced by Octave convolution, which reduces the amount of calculation during network training and improves the detection speed of the network through image frequency division. At the same time, GIOU loss function is introduced to alleviate the problems that the gradient is zero when the detection frames do not overlap, and IOU cannot distinguish the alignment of prediction frames and annotation data in different ways, thus improving the accuracy of detecting the target position. The experimental results show that the mAP of the improved YOLOv3 algorithm on the self-made ship data set can reach 84.07%, and the improved algorithm is improved by 3.52% compared with the original algorithm, and the detection speed can reach 31.8 frames/s, which effectively improves the accuracy and speed of detecting ships at sea.

The traditional theoretical teaching and practice ways of circuit analysis basic course are relatively simple, which are not in line with the requirements of the current discipline construction. In this paper, the teaching content, teaching methods, practical links and assessment methods have been appropriately improved to achieve the student-oriented education reform mode, stimulate students' learning enthusiasm and subjective initiative, exercise students' autonomous learning ability and cross-border integration ability, and improve the quality of teaching.

The paper presents an improved algorithm of P wave peaks, onsets and offsets detection in Electrocardiogram (ECG). By reducing the impact of information loss during signal processing and detecting the effects of node errors, the algorithm can detect the P peaks, onsets and ends more quickly and accurately. For ECG signals, a denoising algorithm based on the Generalized Minmax Concave (GMC) penalty is used to remove power frequency interference, baseline drift and myoelectric interference. And then, the R wave peaks and the P wave peaks, onsets and ends are detected using the wavelet transform and the adaptive threshold, and the false detection rate is further reduced by the differential correction. The performance of the proposed algorithm is evaluated on QT database and sensitivity and positive predictivity of P wave detection reached 99.12% and 91.14%, respectively. Moreover, the proposed algorithm makes the detection of features such as P wave dispersion and P wave duration more accurate, which is beneficial to the diagnosis of diseases such as atrial fibrillation.

In order to improve the accuracy of the ship ’s rolling motion prediction when sailing at sea, a combined prediction algorithm of ship rolling motion based on Complete Ensemble Empirical Mode Decomposition(CEEMD)-Long Short Term Memory Network(LSTM) is proposed. The traditional prediction method ignores the relationship between the data. LSTM has the memory of data, it does not need a lot of data processing or complex modeling process,. The CEEMD algorithm decomposes the original sequence of ship rolling into several modal functions with different characteristics, which makes the non-stationary time series stable and periodic. The LSTM prediction model is established for each component, and the prediction results of each component are superimposed to obtain short-term rolling prediction value. In order to verify the effectiveness of the algorithm proposed in this paper, the real ship data is used for verification. The first 85% of the data is used as the training set, and the last 15% is used for testing. Finally, the effectiveness of the proposed model is verified by experimental results.

This paper presents a graphene film-based rectenna for wireless power transmission. In the rectenna design, a $\Gamma$ shape monopole is fed by a micro-strip line and the antenna is placed above a RT 5880 substrate. A reflector is added to increase the antenna gain up to 6.5 dBi at 2.45 GHz. Furthermore, a class-F rectifying circuit is connected to the antenna, and the results show that the rectenna can achieve the highest conversion efficiency of 82% at 2.45 GHz with the input power of 34 dBm.

Graphene film based circular shaped monopole antennas with ultra- wideband (UWB) performance is presented. Proposed antennas consist of a circular patch radiator made by graphene films and a ground plane made by copper and graphene film, respectively. This proposed nonmetal antenna can achieve a bandwidth of120% and covers the frequency range from 2.5 to 10.5 GHz for a return loss greater than 10 dB. In addition, the antenna is implemented and experimental results are in good agreement with the simulated ones.

This paper presents a 2.45 GHz flexible wearable rectenna based on conductive graphene films for wireless energy harvesting. In the antenna design, the graphene patch is fed by a micro-strip line and the antenna is placed above a soft substrate. The antenna dimension is optimized to adjust the antenna operation frequency. In addition, the antenna has broad beamwidth to capture ambient power. Furthermore, a class-F rectifying circuit is designed by using ADS software, and the simulated results of the rectifying circuit has been produced. The efficiency of rectifying circuit is 72% at 0 dBm. The antenna and rectifying circuit is connected with a matching network and the simulation results show that the rectenna can achieve the highest conversion efficiency of 70% at 2.45 GHz.

The problem of time-varying uncertain of system dynamics model parameters for the omnidirectional mobile robot is mainly discussed in this paper when it is moving with rotating. A double closed-loop Linear Active Disturbance Rejection Control (LADRC) method is proposed. Firstly, the kinematics and dynamics model of the omnidirectional mobile robot are established. Then, two closed-loop controllers are employed in the omnidirectional mobile robot. The low-level controller is running on the robot with feedback from motor encoders. Simultaneously, the robot trajectory tracking high-level controller design method is given. The Linear Extended State Observer (LESO) is used to measure and compensate the disturbance in the trajectory tracking control system as well as the model parameter variation. Finally, the experimental results prove the theories and using the LADRC controller cause more precise motion of the omnidirectional mobile robot.

The prediction of the future climate of the Earth is called phenological prediction, and accurate prediction of climate can be beneficial to avoid harm. The traditional phenological prediction method is not thorough enough to analyze the data, and it is time consuming and labor intensive. But the recent advancements in the field of artificial neural networks, especially in deep learning, have paved a new way for applying computer vision methods to phenological prediction. In this paper we proposes a deep learning algorithm combining FCN and Lstm to improve the accuracy and end-to-end generalization ability of the phenological prediction model. The algorithm use the dataset provided by DataCastle Big Data for training and verification. The data consists of vegetation density maps in a certain area in recent years. The algorithm aims to predict the vegetation density of the region in the future and then infer the future climate based on the future vegetation density. The experimental results show that the algorithm can well study the spatial information and time series information of the data, and accurately predict the vegetation density in different regions, so as to accurately predict the future climate.

China has abundant marine resources and a vast sea area. In military and civilian applications, the detection of targets for marine ships has important research significance. Complex and variable sea conditions make ship detection more difficult. In order to more accurately detect the ship's target, this paper proposes an improved YoloV3 algorithm to realize the end-to-end ship target detection system. Our algorithm achieves high-accuracy fast ship identification by introducing CFE modules, improving loss functions, and augmenting data for small targets. The experimental results show that compared with the traditional machine learning target detection algorithm, our method has greatly improved the detection accuracy and detection rate. It achieves an accuracy of 74.8% and a detection rate of 29.8 frames. The ship detection system of this paper is excellent in detection accuracy and speed.

In order to solve the nonlinear ship course control problem with uncertain sea wave disturbance, this paper designed a ship course control method of linear active disturbance rejection control (LADRC) based on linear extended state observer (LESO). Combined with the ship motion kinematics and dynamics equations, the LESO is used to come over the influence of the model parameter perturbation and external sea wave disturbance caused by the change of the ship motion model and indicated how to design the ADRC parameter. The simulation experiments are compared ship course ADRC with the traditional PID controller under different speed conditions. Simulation results show the ship course ADRC controller is simple and intuitionist, and exhibits good control performance under disturbance of sea wave and model parameters. It has better disturbance rejection ability and robustness than PID controller.

In order to improve the disturbance rejection performance of the omnidirectional mobile robot control system, an attitude and position control system of omnidirectional mobile robot based on linear active disturbance rejection controller (LADRC) is designed. The kinematics and dynamics model of the omnidirectional mobile robot is established and the overall design scheme of omnidirectional mobile robot is given. The perturbation of “internal” model parameters of omnidirectional mobile robot system and “external” environmental interference and randomness of load variation are taken as “total disturbances” and the “total disturbance” is estimated and compensated in real time by constructing an extended state observer (ESO). The final experimental results show that the LADRC has a strong ability and good control accuracy on controlling uncertainty for the attitude and position of the omnidirectional mobile robot with the external disturbances and the nonlinearity of the carrier itself.

For the problem of low recognition rate of traditional digital image processing in actual maritime target detection, an improved maritime target detection method based on deep convolution neural network is proposed. The maritime targets (mainly including fixed-wing aircraft, surface ships and helicopters under the background of sea and sky) are marked to establish the target datasets. The detection framework of maritime target based on Faster RCNN is given and improved by using the Resnet to extract the feature of target, and using the batch normalization layer to optimize the Faster RCNN network, using online hard-examples-mining algorithm to improve the training process. The training and test experiments of the improved target detection method were completed. The experimental results show that improved method can identify the common maritime target effectively and has better detection accuracy and robustness compared with Faster RCNN.

In this paper, the problem of adaptive robust model tracking for nonlinear dynamical systems with parameter uncertainties and unmatched disturbances is studied. It is assumed that the parameter uncertainties and the bounds of the unmatched disturbances are unknown. For tracking reference model, a kind of adaptive robust controller is proposed. The uniform ultimate boundedness of the error between state $x(t)$ and the state $x_{r}(t)$ of the reference model can be guaranteed by the proposed controller in the presence of parameter uncertainties and unmatched disturbances. Finally, a numerical example is provided to illustrate the effectiveness of the proposed control scheme.

Compared with the conventional orthogonal frequency division multiplexing (OFDM) system, the orthogonal condition of the orthogonal frequency division multiplexing with offset quadrature amplitude modulation (OFDM/OQAM) system only holds in the real field. Due to the relaxation of the orthogonal condition, OFDM/OQAM systems could remove cyclic prefix (CP) and obtain higher net bit rate. However, for OFDM/OQAM systems, the intrinsic imaginary interference (IMI) seriously deteriorated the system performance. In order to combat IMI, the prototype filter with well time frequency Iocalization (TFL) property is highly desired for OFDM/OQAM system. In this paper, we analyzed the IMI in the optical OFDM/OQAM system, and discussed the TFL property of the filter bank for this kind of system. Then we compared the performance of the optical OFDM/OQAM system with the using of the isotropic orthogonal transform algorithm (IOTA) filter, square root raised cosine (SRRC) filter and extended Gaussian function (EGF) filter respectively. As demonstrated in the TFL property analysis and simulation results, the system robustness against IMI can be improved significantly by the using of the IOTA filter thanks to the optimal TFL property.

In order to improve the accuracy of intelligent recognition of the marine ships besides the realization of the algorithm, a deep learning technique is introduced applying Faster RCNN algorithm to the recognition of marine vessels. The marine target recognition system structure based on Faster RCNN is established. The effectiveness of Faster RCNN algorithm is verified on the VOC2007 dataset, and the marine vessel target recognition experiment is carried out. On the basis, the target database of ships in different marine environments is established and marked, and the new dataset is used to train Faster RCNN network model, the new trained Faster RCNN network model is compared with the old one by taking the marine ship target recognition experiments. The experimental results show that the Faster RCNN model trained on the new created dataset has good recognition ability and robustness for the targets of marine ships while it can be more accurate giving the target area with the low missed detection rate.

The problems of robust stabilization and adaptive robust controller design are considered for a class of nonlinear systems with time-varying delays. It is supposed that the nonlinear uncertainties with delayed states can be approximated by neural networks which are not used in the actual control process, and that the time-varying delays are bounded continuous functions. Particularly, the derivatives of the time-varying delays are not required to be less than one. For such uncertain systems, a class of redesigned adaptive robust controllers is synthesized. Furthermore, it is also shown that the proposed control schemes can guarantee the uniform asymptotic stability of such uncertain systems. Finally, a numerical example is given to demonstrate the validity of the results.

This paper focuses on the problem of adaptive robust output tracking for delayed chaotic systems with sector nonlinearity in input function. Based on Lyapuvov-Krasovskii function, an adaptive robust controller is proposed. It is also shown that the proposed feedback controller can guarantee that the error between state x(t) and delay state x(t — τ) is uniformly ultimately bounded. Finally, a Chua's circuit is given to illustrate the validity of the proposed control scheme.

Paper proposes a kind of multiple environmental parameters monitoring system for granary that is based on ZigBee network and takes ARM controller as core. With regard to hardware, monitoring circuits for temperature, humidity, CO 2 concentration and grain water content are designed. With regard to control program, fuzzy intelligent algorithm is designed and transplanted to ARM controller. The test result shows that the maximum data transfer error of ZigBee network is 5%, the data transfer successful rate of sending 10000 control commands with acknowledging communication strategy is 100%, the maximum error of temperature control is ±1.3 °C, the maximum humidity control error is ±1.8 %, the maximum CO 2 concentration control error is ±1.7 % and the maximum error of grain water content control is ±1.2%, such that the system fully meets practical requirements for use. The system is designed with high performance ARM controller, which is beneficial functional expansion such as video monitoring.

In order to improve the eliiciency of apple's grading and ensure the scientilicity and objectivity of grading, this paper presents a three-stage grading system, namely weighing grading, color grading and shape grading. Hardware design uses high-performance S3C2440 embedded controller, DSP controller, single-chip micro-processor controller, pressure load sensor, color recognition sensor, graphic recognition sensor and motor. Software design includes the realization of apple weighing DSP monitoring procedures, color grading SCM C51 procedures and shape grading embedded system procedures. Test results show that the maximum relative error of single stage is 5.6%; therefore system can fully meet the practical demands of apple grading.

The method of using electronic stability to remove the video image dithering is currently one ofthe main direction of image stabilization technology. SIFT algorithm is used to extract feature points, solve the rotary motion between the video frames. KLT algorithm is used to track SIFT feature points and through Ransac algorithm for further filter, then estimate the affme transformation to compensate. The simulation results showed that the method can effectively remove randomjitter video and get high image stabilization accuracy.

Aiming at complex characteristics of magnetic flux leakage signal, and it is hard to describe geometric features of pipeline defect. A novel approach based on improved particle swarm optimization and Least Squares Support Vector Machines (LSSVM) algorithm was proposed to reconstruct pipeline defect in this paper. The nuclear function parameter and penalty parameter are vital factors which determine performance of Least Squares Support Vector Machines. Use improved particle swarm optimization optimize the parameters of LSSVM automatically, insuring the accuracy of parameter choice. This method was used to pipeline defect reconstruction, simulation results demonstrate that method can break through the difficulty of magnetic flux leakage signals described defect geometrical characteristics, improving the reconstruction accuracy, with a highly practice value.

A novel image segmentation for infrared small target of agriculture and forestry fire is proposed in this paper. Usually, Maximum Variance Image Segmentation method (Otsu) is a popular non-parametric method in image segmentation. However, it needs a lot computation and has poor real-time quality. Thus it is hard to be wide applied in many situations. To over come this issue, a constructive approach to obtain optimal threshold of between-class variance as fitness function for Otsu by particle swarm optimization (PSO), reduce the amount of computation and improve real-time performance. The performance of the proposed method is evaluated through infrared small target of agriculture and forestry fire. The experimental results demonstrate the effectiveness of the proposed method.

this paper presents a disturbance rejection control methodology for the course keeping of a fully-submerged hydrofoil craft. The course keeping dynamics of the fully-submerged hydrofoil craft with parameter perturbation is firstly analyzed. Then an extended disturbance observer (EDO) is developed to make real time estimate of the disturbance as well as the model uncertainties. A nonsingular terminal sliding mode controller is proposed to compensate the residual error of the EDO. Stability of the overall closed-loop system is evaluated to be ultimate uniformly bounded. Numerical simulation indicates the effectiveness of the proposed control method.

The Least Squares Support Vector Machines (LSSVM) is a promising artificial intelligence technique in which the regression algorithm has already been used solve the nonlinear function approach successfully. The nuclear function parameter and penalty parameter is a pivotal factor which decides performance of LSSVM. Unfortunately most users selected parameters for an LSSVM by rule of thumb, so they frequently fail to generate the optimal approaching effect for the function. This has restricted effective use of LSSVM to a great degree. To solve these problems, a new approach based on an adaptive genetic algorithm (AGA) was proposed, which automatically adjusts the parameters for LSSVM, this method selects crossover probability and mutation probability according to the fitness values of the object function, therefore reduces the convergence time and improves the precision of genetic algorithm (GA), insuring the accuracy of parameter selection. This method was applied to ship rolling prediction, and simulation results showed it can effectively improve prediction accuracy.

Accurate modeling and simulation of submersible vehicle is essential for autonomous control and maneuverability research. In this paper, a variable direction rotatable axis of variable vector propeller (VDRA-VVP) is proposed and researched innovatively. The structure and working principle of VDRA-VVP of submersible vehicle are described and the nonlinear mathematic model of the submersible vehicle in spatial motion was derived based on momentum theorem. The forces acting on submersible vehicle were resolved to several modules which were expressed in matrix form. Based on the motion model and combined with three degrees of freedom model of vertical motion, a motion simulation system was constructed. Considering the characteristic of VDRA-VVP, the depth control and pitch angle control are simulated by adopting both robust PD control method and adaptive robust PD control method. The comparison result shows that the control effect of adaptive robust PD controller is obviously better than robust PD controller in vertical motion. The simulation results show that the VDRA-VVP of submersible vehicle has good spatial maneuverability, and verify the feasibility and reliability of control method.

Fault diagnosis is very important for development and maintenance of safe and reliable electronic circuits and systems. This paper describes an approach of soft fault diagnosis for analog circuits based on least square support vector machines (LS-SVM) and adaptive genetic algorithm (AGA), known as AGA-LS-SVM. AGA is applied to optimize the parameters of LS-SVM, and fault features are extracted from the frequency domain response of circuit under test (CUT) and the LS-SVM which trained by the fault features is used to recognize the unknown faults. The experimental results demonstrate that LS-SVM optimized by AGA performs better forecast accuracy and successful modeling of diagnosing analog circuits fault.

In odder to improve positioning precision and decrease the exceed initial value of solar cell transfer robot, we design a kind of PID controller with fuzzy self-adjusting function based on fuzzy control technique, solve the problem that the PID-controller can not adjust parameters itself on line. The result of simulation in fuzzy logic toolbox of Matlab proves that the controller possess both the precise character of PID-controller and the flexible advantage of fuzzy controller. Fuzzy PID control be used into transfer robot with solar cell, the positioning experiment has been done. The result show that the positioning error can be controlled within 0.028% of real valve and the regulation system has good dynamic and static quality. This system not only solves the fragments emerging in the procedure but also enhances the efficiency of the test. it has amplitude application foreground in the solar cell field

Camera calibration is one of fundamental issues in computer vision and required for achieving accurate visual measurements. Least Squares Support Vector Machines (LS-SVM) are used to achieve the camera calibration. It doesn't need to confirm the intrinsic and extrinsic parameter of the camera. For a little more accurate calibration, acquired image is divided into two regions according to radial distortion of lens and Least Squares Support Vector Machines is applied to each region. A simple and flexible camera calibration using divided Least Squares Support Vector Machines (DLS-SVM) is proposed in this paper. Experiment results and comparison with BP neural network and Least Squares Support Vector Machines prove the validity of the proposed camera calibration.

Two kinds of novel zinc/CNT composites were prepared by composite electrodeposition. The codeposition of carbon nanotubes and zinc was verified by the Raman spectrometry analysis. Scanning electron microscopy images illustrated the continuous and uniform zinc coverage on CNT, which were unique interfacial features in the currently known metal/CNT composites.