Intelligent reflecting surface (IRS) composed of a large number low-cost passive elements has attracted significant attention from communication communities because its ability to substantially improve the performance. In this paper, aperture area and radiation pattern IRS are considered, novel geometry-based three-dimensional (3-D) wideband channel model is proposed for IRS-assisted unmanned aerial vehicle (UAV) systems. model, reflection phase designed by jointly considering element...

Dynamic gesture recognition using millimeter-wave radar has a broad application prospect in the industrial Internet of Things (IoT) field. However, existing methods random dynamic interference environment, such as throwing objects and waving easily cause wrong recognition. This article proposes method based on convolutional neural network (CNN)-Transformer to solve this problem. First, we reshape original echoes acquired by frequency-modulated continuous-wave (FMCW) into 3-D data blocks...

Abstract The triple exponentially weighted moving average (TriEWMA) controller integrates three distinct EWMAs to mitigate process dynamics and disturbances, enhancing system stability in the semiconductor manufacturing process. However, performance of TriEWMA is highly dependent on its weight parameters, which are challenging optimize manually. This work develops a composite strategy utilizing deep reinforcement learning autonomously these weights. Specifically, adjustment modeled as Markov...

Marine fish target detection technology is of great significance for underwater vehicles to realize automatic recognition. However, the complex environment and lighting conditions lead background collected image more irrelevant interference, which makes difficult. In order detect targets accurately quickly, a real-time network based on improved YOLOv5s proposed. Firstly, Gamma transform introduced in preprocessing part improve gray contrast marine image, convenient model detection. Secondly,...

Due to the large path loss in millimeter wave (mmWave) band, transmission between transmitter (Tx) and intelligent reflecting surface (IRS) is considered as a Rayleigh fading channel, physics-based channel model proposed for IRS-assisted mmWave communication system urban scenario. Also, horizontal vertical rotation angles of IRS relationship scattering gain unit its effective aperture incident direction desired reflection are model. For scenario, existing phases, which designed align virtual...

A non-stationary three dimensional (3-D) intelligent reflecting surface (IRS)-assisted unmanned aerial vehicle (UAV) multiple-input multiple-output (MIMO) channel model is proposed. The time-varying reflection phases of the IRS, which are designed based on optimal power received signal, investigated. Based proposed phases, effect number and geometric area IRS units speed UAV statistics also numerical results show that signal can be enhanced multipath fading phenomenon effectively reduced by...

The navigation accuracy of the ship can be improved by associating output tracks radar with that automatic identification system (AIS). traditional association method only uses track features (including position, speed, and heading angle) has low weak scene adaptability. We integrate through deep learning to associate AIS tracks. First, we employ autoencoder (AE) reduce noise original data. statistical distances two in are obtained as new map data high-dimensional space extract fully...

The millimeter-wave radar can sense the subtle movement of hand. However, traditional hand gesture recognition methods do not work well in dynamic interference scenarios. To address this issue, a robust method is proposed based on self-attention time-series neural networks. Firstly, original echo constructed terms frame, sequence, and channel at input terminal network. A one-dimensional network built, time-distributed layer used as wrapper to extract feature from each frame sequence...

A three dimensional (3-D) geometry-based channel model is proposed for intelligent reflecting surface (IRS)-assisted unmanned aerial vehicle (UAV) multiple-input multiple-output (MIMO) communication scenarios where the users are located on edge of IRS service area. In order to reduce interference between different reflection units, we first construct an optimization problem about statistical propagation phases and then derive optimal phase by solving designed problem. Finally, align with...

Abstract Millimetre‐wave radar offers high ranging accuracy and can capture subtle vibration information of the human heart. This study proposes a heartbeat prediction method based on transformer model using millimetre‐wave radar. Firstly, was used to collect data conduct normalisation processing. Secondly, position coding introduced assign sine or cosine variables input extract their relative relationship. Subsequently, encoder adopted allocate attention through multi‐head mechanism, mask...

Considering the shortcomings of low precision, response lag and control instability conventional PID controller in brushless dc motor speed system, an intelligent is designed, which combines RBFneural network with controller. This paper analyzed operating principle motor, deduced transfer function model The programs designed were also provided. At last, was validated through Mat lab software simulation. In process simulation, three parameters adjusted obviously by RBF-Neural Network. results...

Abstract In order to achieve precise localization in the underwater environment, an method based on hybrid invasive weed algorithm is proposed. First, considering insufficient local search capability of algorithm, salp swarm used pre‐process initial population, position and velocity update formulas are employed further improve depth, obtained. Then, error designed as objective function obtain value source, a new model built using values interference parameters. Finally, improved two‐step...

Abstract This paper focuses on the problem of lateral stability four hub-motor-in-wheels drive electric vehicle, 7 DOF (degrees freedom) vehicle simulation model which is verified by field test established based Matlab/Simulink software. On basis simulated model, BP neural network PID torque distribution controller proposed. The sideslip angle at mass center and yaw rate are selected as control variables, designed. result shows that proposed strategy can vehicle’s rate, avoid under steer...

Frequency diversity array (FDA) radar can provide full spatial coverage with stable gains within a pulse duration. Based on the FDA, integrated radar-communication system perform multi-directional communication and whole-space detection. However, embedded bits disrupt correlation of transmitting waveform each element. Correspondingly, range sidelobe level (SLL) multi-dimensional ambiguity function increases significantly. To address this issue, low-sidelobe for systems based FDA was...

The BP neural network PID algorithm torque distribution control strategy for electric vehicle with in-wheel motor is proposed, in which the constraint conditions of characteristic and road adherence are taken into concerned. model-matching adopted using double parameters yaw rate body slip angle adjusted according to weight factor. In addition, more effective 3-freedom model implemented determine desired by considering variation tire lateral stiffness affected wheel vertical load transfer...