To solve the trajectory tracking and attitude control problems of quadcopter aircraft during flight, a sliding mode variable structure nonlinear controller is designed based on an improved new reaching law and sliding mode function. According to the underactuation, strong coupling and multi-variable characteristics of quadcopter aircraft, the whole force analysis of its system is carried out, and the system mathematical model is simplified based on Newton's second law and Euler's momentum equation. Based on the traditional exponential reaching law and combined with the double-power term, proportional term and improved variable exponential term, a new reaching law sliding mode controller is designed for the inner and outer loops of the system. The output of the outer loop position controller is taken as the expected input of the inner loop controller to form a closed loop system in which the outer loop controls the inner loop, and the stability of the system is proved by the Lyapunov stability theorem. Simulation results show that the proposed method has shorter response time, better tracking performance, and smaller steady-state error than traditional reaching law sliding mode controllers in fixed-point levitation and complex three-dimensional trajectory tracking experiments.

In view of the problems such as unreasonable design, incomplete construction of barrier-free facilities, uneven guide AIDS, single functions and inability to meet users' outdoor travel needs, this study proposes a visual perception and detection method based on BlendMask algorithm and RGB-D dynamic obstacle targets for outdoor guide for the blind. The proposed method uses RGB-D data collected by Azure Kinect DK, compares six sets of instance segmentation algorithms, selects the best BlendMask instance segmentation model, and adds a depth data processing branch to obtain the depth information of a single instance, that is the distance from the target to the camera. To improve distance detection accuracy, outlier rejection operations are used in 3D, in the form of point clouds, to remove residual background information from the segmentation mask, where a statistically based outlier rejection method can more accurately detect the distance from the obstacle to the camera.Experimental results show that the proposed method is not only better at detecting dynamic outdoor targets, but also detects the distance from the target to the camera and improves the accuracy of distance detection.

The S-SSHI(Series Synchronous Switch Harvesting Interface Circuit), P-SSHI(Parallel Synchronous Switch Harvesting Interface Circuit ), and SECE(Synchronous Charge Extraction Interface Circuit) obtain higher output voltage and harvested power than the FB(Full-Wave Rectifier Bridge) in weakly coupled piezoelectric energy harvesting system. The S-SSHI, P-SSHI, and SECE interface circuits do not have an obvious effect gain on output voltage under strong coupling systems. To solve this problem, an asymmetric voltage interface circuit is proposed in this study. The circuit consists of a single voltage detection switch and a full wave rectifier bridge. When the charging capacitor voltage approaches the open-circuit voltage of the PEH(Piezoelectric Energy Harvester), the PEH voltage begins to be biased. The maximum output voltage of the asymmetric voltage interface circuit is higher than FB because of the bias voltage, so the asymmetric voltage interface circuit can improve the harvested energy. The simulation results of asymmetric voltage interface circuit are consistent with the experimental results. The captured energy of the asymmetric voltage interface circuit is 0.532 J under the strong coupling system. The captured energy of the asymmetric voltage interface circuit is 147% higher than FB and 75% higher than other improved interface circuits.

Human tissue pathology examination is mainly used for the diagnosis and treatment of various tumors. Immunohistochemical technique has important clinical significance in the early screening of colorectal cancer. In order to accurately determine the expression level of the tumor suppressor gene p53 in colorectal cancer, this study proposes a grading diagnostic method based on transfer learning with block-wise fine-tuning strategy. The parameters of the cell nucleus segmentation model are transferred to the diagnostic framework through image preprocessing, supervised model pre-training, and block-wise fine-tuning. The generated cell nucleus segmentation mask is subjected to PCA(Principal Component Analysis) dimensionality reduction and SVM(Support Vector Machine) multivariate classification to obtain the final image diagnosis result. The proposed method is verified on colorectal cancer p53 protein IHC(Immunohistochemistry) image dataset. Dice value of the model reaches 0.887 6 and classification accuracy reaches 90.28%. The results show that the proposed method can effectively grade the immunohistochemical images of colorectal cancer, and provide reliable auxiliary information for doctors to read the film.

In view of the single-link manipulator system with quantized input, a predefined performance adaptive neural network control method based on Funnel control is proposed in this study. Different from the traditional Funnel control scheme, this method can ensure that the system reaches the predefined performance index within the predetermined time by constructing a new performance function, and the neural network and dynamic surface control technology are used to solve the unknown nonlinear items in the system and the differential explosion problem in the traditional backstepping control method. Theoretical analysis shows that the proposed control scheme not only eliminates the negative effects caused by quantized inputs, but also ensures the stability of the system. By adjusting the design parameters in the MATLAB simulation experiment, the predetermined time convergence of the tracking error is realized, which verifies the effectiveness of the proposed control method.

Digital microwave relay communication system is a multichannel communication system in which digital signals are transmitted by microwave relay. In order to improve the signal-to-noise ratio and reduce the bit error rate in this kind of communication, a C-band sharp roll down broadband metamaterial absorber is designed in study. The quasi-periodic ring resonant structure is used to realize the polarization insensitive broadband absorption effect. The absorption efficiency is more than 90% in the range of 7~8 GHz, and the rolling down coefficient of the absorption line can reach more than 0.8. Based on the common structure, a dielectric layer is added to the proposed metamaterial absorber, which protects the ring structure of the surface of the metamaterial absorber and increases the impedance matching ability of the absorber. The proposed design has the characteristics of simple structure, thin thickness, common material and sharpe roll down absorption, and has important application potential in the fields of detection, stealth, sensing, communication and so on.

In recent years, surface plasmon polaritons has been developed rapidly and applied in many fields. Spoof surface plasmon polaritons, which is inspired by the surface plasmon phenomenon polaritons, is proposed in the form of constructing periodic grooves on the metal surface, and in turn achieving the surface plasmon polaritons phenomenon on the microwave or terahertz band. Based on the artificial surface plasmon theory, a planar asymmetric leaky wave antenna is proposed for indoor communication. The antenna has the characteristics of beam scanning and can achieve 67° beam scanning within 11.5~15.0 GHz. At the same time, a tunable gap and a matching branch are added into the periodic unit to control the direction of the antenna main beam by changing the gap size. The gain of the antenna at all frequency points is higher than 8.5 dBi, and the efficiency is higher than 75% in the full frequency band. In this study, the performance of the antenna in indoor communication is discussed, and the optimal placement area of the antenna is discussed by analyzing the field distribution of the antenna in different positions, which provides theoretical guidance for the application of artificial surface plasmon tunable leaky wave antenna in indoor communication.

In view of the problems such as low accuracy and slow recognition speed of existing steel surface defect detection methods, a defect detection method based on improved YOLOv5s(You Only Look Once version 5s) is proposed in this study. CBAM(Convolutional Block Attention Module) attention mechanism is introduced into the backbone feature extraction network to pay attention to the information of important regions and improve the model's learning ability of target defects. In order to improve the regression speed and localization accuracy of the target frame, the EIoU(Efficient Intersection over Union) boundary frame loss function combining distance loss and width-height loss are used to calculate the loss value. Transfer learning is used to accelerate the convergence speed of the model to improve the accuracy of defects detection. The experimental results on the data set NEU-DET show that compared with the original YOLOv5s network, the improved YOLOv5s network of the accuracy rate of the data set increased by 6.3 percentage point, the recall rate increased by 9.2 percentage point, and the mAP(mean Average Precision) reached 81.7%, which indicates the improved method has good performance for the detection of steel surface defects. The size of the steel surface defect detection model with the improved YOLOv5s algorithm is only 13.8 MB, which improves the detection accuracy on the basis of real-time performance and facilitates the deployment of the model in practical applications.

Direct torque control is a common control method for SynRM(Synchronous Reluctance Motor) because of its simple control structure, small motor parameter dependence and good robustness. The traditional direct torque control system uses encoder to obtain the information of motor speed and rotor position. The encoder is installed on the motor shaft, which is not good for the miniaturization of the motor, but also increases the production cost and reduces the system reliability. In this study, a piecewise saturation function is proposed to replace the traditional symbolic function, and an adaptive sliding mode gain design is proposed to improve the sliding mode observer, which is applied to SynRM position sensorless direct torque control. The CORDIC(Coordinate Rotation Digital Computer) algorithm is introduced to optimize the phase-locked loop, so as to estimate the motor speed. In order to verify the correctness and validity of the proposed theory, a special experimental platform is set up to test its start-up characteristics and anti-load disturbance performance. The experimental results show that compared with the traditional method, the sensorless control method can quickly and accurately estimate the motor speed, control the steady-state error within 0.5%, and effectively reduce the system cost.

In view of the massive complex and high dimensional stone image data and the uncertain, fragmented and heterogeneous stone information caused by different imaging techniques, deep learning is an efficient big data processing tool with excellent feature extraction and nonlinear recognition capabilities, and has become an intelligent solution to improve the level of kidney stone management. The application of deep learning in kidney stone image data analysis can adapt to the characteristics of different types of stone data, and use multi-variable comprehensive assessment of stones to achieve accurate quantification, improve the efficiency and accuracy of diagnosis and evaluation, and assist doctors to develop better stone treatment plan. This study reviews the progress of deep learning in kidney stone imaging in recent years, focuses on various deep learning algorithms, illustrates their novelty and advantages over traditional methods and machine learning algorithms, and points out their shortcomings, aiming to provide reference and direction for subsequent research work.

In view of the different requirements of stability, accuracy and rapidity of different control systems in supercritical CO₂ extraction process, an improved whale optimization algorithm objective function is proposed for the parameter tuning of supercritical CO₂ extraction. According to the control target of the controlled system, the weight relationship of absolute integral identification, overshoot, residual difference and adjustment time is determined by analytic hierarchy process. The performance index of the control system is normalized by the three fold line method, and the objective function of the whale optimization algorithm is constructed. Taking the pressure and temperature of supercritical CO₂ extraction as the controlled objects, the improved whale optimization algorithm objective function and the traditional whale optimization algorithm objective function are used to tune the parameters. The simulation results show that the objective function of the improved whale optimization algorithm can ensure no overshoot in the pressure control system of supercritical CO₂ extraction. In the temperature control system of supercritical CO₂ extraction, the adjustment time is reduced by 30.17 s, and the tuning optimization of control parameters in the process of supercritical CO₂ extraction is realized.

This study proposes a dynamic face recognition system to address the problem of requiring the recognized individual's cooperation in existing static face recognition processes. The system uses the RetinaFace and FaceNet algorithms for dynamic face detection and recognition, respectively, and is optimized for high recognition accuracy and real-time performance. In particular, GhostNet is used as the backbone network for RetinaFace detection, and Adaptive-NMS(Non Max Suppression) non-maximum suppression is used for face bounding box regression. For FaceNet recognition, MobileNetV1 is used as the backbone network, and a joint loss function combining Triplet loss and cross-entropy loss is used for face classification. The optimized algorithm has excellent performance in detection and recognition The improved RetinaFace algorithm achieves detection accuracies of 93.35%, 90.84%, and 80.43% on the WiderFace dataset, with a frame rate of 53 frame·s^-1. For dynamic face detection, the average detection accuracy is 96%, with a frame rate of 21 frame·s^-1. When the FaceNet threshold is set to 1.15, the highest recognition rate is 98.23%. The average recognition accuracy of the dynamic recognition system is 98%, with a frame rate of 20 frame·s^-1. The experimental results demonstrate that the proposed system fully addresses the problem of requiring cooperation from the recognized individual in static face recognition and achieves high recognition accuracy and real-time performance.