Suffering from complex sideslip angles, path following control of sailboat becomes significantly challenging. In this article, double finite-time observers-based line-of-sight guidance and finite-time control (DFLOS-FC) scheme is presented for path following of sailboat. Double finite-time sideslip observers (DFSO) are employed to observe the time-varying sideslip angle caused by external disturbances, which improves the accuracy of line-of-sight guidance. To avoid differential explosion problem caused by virtual control law, we designed a finite-time filter. The finite-time disturbance observer (FDO) is designed to accurately observe unknown external disturbances, which enables the controller to have excellent tracking accuracy and precise disturbance rejection. Considering the rotation limit of actuator rudder angle, we limit the rudder angle. The finite-time stability of the integrated guidance and control system is strictly guaranteed by Lyapunov method. Finally, the effectiveness of this method is verified by simulation and comparison with the traditional backstepping method.
In this paper, a novel disturbance observer-based fixed-time formation control method is proposed for autono-mous underwater vehicles with actuator faults, model uncertainties and unknown disturbances. Firstly, a fast fixed-time stable system is improved, which has a faster convergence rate than existing stable system. Then, the leader-follower formation control strategy is combined with the path planning strategy based on the artificial potential field to obtain a collision-free formation configuration. Besides, a fast fixed-time disturbance observer is designed to compensate for unknown composite disturbances, which enhances the robustness of the system. Further, based on the disturbance estimation information and the improved nonsingular fast fixed-time terminal sliding-mode surface, a robust fast fixed-time formation control method is presented, which achieves the desired formation configuration. Finally, simulation and comparison results show the effectiveness of the proposed control strategy.
The reasonable placement of lamps in the classroom will play a vital role in the classroom lighting environment of the classroom, in which the classroom lights and blackboard lights, as the two most important parts of the classroom lighting environment, their importance is self-evident. The illumination level in the classroom can be improved by adjusting the position of the blackboard lights and the classroom lights. At the same time, the influence of the lamps on the illumination level in the classroom is analyzed, and the best parameters are selected by changing the installation position and rotation Angle of the lamps, so as to put forward optimization suggestions for improving the illumination level in the classroom.
What types of shoes that the driver should be wear on driving is under the clear regulation. Non-standard shoe-wearing such as wearing high-heeled shoes, platform shoes, slippers or with bare feet will bring great safety risks and may lead to traffic accidents. According to statistics by traffic department, many traffic accidents are caused by irregular shoe-wearing. Although computer vision has been applied for monitoring driver’s behavior in many aspects, such as driver’s face, eye, and hand, there is still no computer vision-based method or hardware device on the vehicle that to monitor the driver’s shoe-wearing. Therefore, if the driver’s illegal shoe-wearing behavior can be identified before driving, it can play an important role in reducing the incidence of traffic accident. The main difficulties in drivers’ shoe-wearing detection lie in the diversity of shoe types, the variety of foot postures, and the complexity of the detection environment. Therefore, the traditional computer vision method will be high error detection rate in such scenes. In this paper, a deep learning-based method for detecting abnormal shoe-wearing of drivers before driving is proposed. Different models such as SVM, Fast-RCNN, Faster-RCNN, YOLO v2, YOLO v3, YOLO v4 are used to identified drivers’ shoe-wearing. To the best of our knowledge this is the first work that using the computer vision technology for automatic monitoring on drivers’ shoe-wearing. The experimental results show the proposed method can identify the drivers’ shoe wearing effectively and efficiently. It has high application value for improving traffic safety.
People's requirements for office lighting environment are becoming higher and higher, not only for its comfortable and healthy, but also to ensure that people's work efficiency. Aiming at the problem of visual comfort of office light environment, this study carried out subjective and objective evaluation experiment of lighting comfort of office reading environment. This paper studies the coupling influence of lighting elements on paper reading and VDT reading in the office light environment, conducts experiments on the fixed brightness of the screen, and analyzes and discusses the influence of lighting source on visual fatigue degree of reading. In this study, the visual comfort of paper reading and VDT reading was discussed from the perspective of correlation analysis and multiple regression analysis based on the psychophysical data of 30 observers. The experimental results show that visual comfort of paper reading and VDT reading is significantly correlated with illumination and color temperature. The development of this study will provide effective theoretical reference for the design of the current office light environment and has certain guiding significance for the design of light source in various office places.
Driving behavior is crucial to the energy consumption analysis of electric vehicles. This paper proposes an unsupervised learning method to classify driving behavior for three typical road conditions. First, three specific road conditions are selected from the open access data, including characteristic information such as speed and acceleration. Besides, the characteristic data is processed, so each distinct value has the same weight. Second, two unsupervised learning clustering algorithms are introduced and compared in typical working conditions. Finally, the clustering results under three working conditions are obtained. Specifically, we can classify driving styles in high-speed conditions into aggressive, standard, and calm; besides, the classification method of K-medoids is more advantageous. In intersection conditions, driving styles are usually divided into standard and calm. Considering the calculation time and other factors, the K-means algorithm shows superior effects compared to the K-medoids algorithm. The driving style can be divided into standard and calm in campus conditions. In this case, K-medoids have a more significant advantage. The research results have implications for the classification of driving styles under different road conditions.
In this paper, the current tracking control problem of direct-drive wave power generation system disturbed by complex environment is solved by designing finite-time controller(FTC) and backstepping sliding mode controller (BSMC). In order to realize the maximum power tracking control of the directly driven wave energy converter (DWEC), a fixed-time observer (FTDO) is designed to quickly compensate for environmental disturbances, and a backstepping sliding mode control strategy based on FTDO (FTDO-BSMC) is proposed to accurately track the current. Finally, Numerical simulation and comparison demonstrate the feasibility and superiority of the proposed scheme.
The security of cyber physical systems (CPSs) is the premise of resource sharing in modern industry, which has attracted considerable attention of researchers. Many effective methods have been proposed to defend CPSs, among which physical watermark is one prevailing approach which can enhance the replay attack detection capability of CPSs. In order to make the watermarks work more effective, a novel stochastic event-based feedback physical watermark is proposed in this paper to detect replay attacks. We formulate first the problem taking into account the Kalman filter, the linear quadratic Gaussian (LQG) optimal controller and the χ ^2 detector. Then, we characterize the LQG performance loss and the probability of adding a physical watermark in two different scenarios: the system operates with and without replay attacks. It is proved that the probability of adding a watermark signal will increase when replay attacks exist. Furthermore, we discuss the performance of the χ ^2 detector under the framework of our approach. Finally, numerical simulations are verified the theoretical results.
In order to solve the problems of not mastering the experimental process and not knowing the results clearly when using the full space distribution photometer for near-field measurement of lamps, a virtual simulation measurement of lamp intensity distribution based on near field Goniophotometer is proposed. The virtual simulation system builds the experimental platform and simulates the basic process of near-field measurement of lamps by the full-field Goniophotometer. The scheme includes the use of 3D Max and Blender to build models, unity 3D to simulate animation and scene development. According to the light intensity distribution of different types of lamps, the IES file is used to draw the light distribution curve of different lamps. The design results also prove that the system has good operability, interactivity and stability, realizes low-cost, safer and more convenient experiments, and improves the efficiency of lamp experiment of near field Goniophotometer.
On the basis of successive way points (WPs) and piecewise reference speed, a trajectory generator (TG) is elaborately devised to generate a smooth trajectory for an unmanned surface vehicle (USV) in this paper. Firstly, the seventh-order polynominal is exclusively deployed to fit foregoing WPs, resulting in the parameterized path with sufficient differentiability. Then, the reference speed passes through the second-order filter, such that the speed command is uniformly continuous. Together with the parameterized path and speed command, the path speed is readily designed, and thereby contributing to the continuously differentiable trajectory. Certainly, the desired trajectory is much easier for users to predesignate in the electronic chart, and is more favorable to be exactly tracked by the USV in practice. Eventually, the model-based controller for the USV is synthesized to asymptotically track the generated trajectory in the absence of unmodeled dynamics and external disturbances. Comprehensive simulations are conducted on the benchmark prototype CyberShip II, so as to validate the effectiveness of both the proposed TG and trajectory-tracking controller.
Partial least squares (PLS) is a widely used and effective method in the field of fault detection. However, due to the fact that the standard PLS decomposes the process variable space into two subspaces which are not completely orthogonal, it is insufficient in quality-related fault detection. To solve this problem, principal component regression (PCR) is used to decompose the quality variables of PLS model and realize the reconstruction of the process variable space. In this way, the process variable space is decomposed into highly correlated and highly irrelevant parts of quality variables, and the two are monitored by designing statistics respectively. Furthermore, an adaptive threshold based on the idea of exponential weighted moving average (EWMA) is introduced to reduce the false positives and missed positives caused by the traditional fixed threshold, and this method is named as improved regression partial least squares (RPLS). Finally, linear and nonlinear numerical examples and Tennessee Eastman (TE) processes are used to verify the effectiveness of the proposed method which named improved regression partial least squares (IRPLS). Finally, linear and nonlinear numerical cases and Tennessee Eastman (TE) processes are used to verify the effectiveness of IRPLS. The results show that the proposed method can effectively improve the fault detection rate and algorithm follow-through performance, and reduce false positives.
Aiming at solving the problem of low marine vessel detection accuracy in the sea fog environment, a deep learning-based anti-fog marine vessel detection method is proposed in this paper by combining defogging preprocessing with marine vessel detection model. Firstly, gated context aggregation network (GCANet) network is used to process the marine vessel image. Then, the processed image is sent to a modified SSD network, wherein anchors are tuned by statistical characteristics of the shape of marine vessel to detect the position of the marine vessel. Furthermore, to alleviate the loss of feature information due to defogging processing, channel attention mechanism based on the squeeze and excitation module (SE) is added to base convolutional layer of SSD. The comprehensive experiments and comparison results show that the proposed G-SEMSSD network is more suitable for marine vessel detection under sea fog environment.
Extended state observer(ESO) is a class of high-gain observers and plays a significant role in output feedback control theory for uncertain nonlinear systems. However, it faces a challenge in industrial applications when the output is corrupted by high-frequency measurement noise, which means the core parameter, referred to as bandwidth, cannot be too large. It faces a trade-off between convergence rate and noise suppression. In this paper, A new cascade finite-time nonlinear ESO(FTNESO) combining cascade structure and finite-time nonlinear observer, which provides extra degree of freedom to design the controller, is proposed to improve the estimation performance in the presence of measurement noise. Based on the proposed cascade FTESO, an active disturbance rejection control scheme(ADRC) is established, and the controller is applied to the nonlinear gas turbine model to show its effectiveness.
Marine organism detection is crucial for the intelligent construction of open-sea farm. Suffering from low-contrast, color-deviation and detail-blurry underwater environment, a coordinate attention and transformer neck-based benthonic organism detection (CATNBOD) scheme has been devised. Main contributions are as follows: 1) The coordinate attention (CA) module is designed in the feature extraction network to obtain meaningful features, such that the small-scale benthonic organisms can be accurately detected. 2) To efficiently address the challenge derived from intra- and inter-class occlusions of benthonic organism, the rotation window-based swin transformer (ST) module is devised in the neck structure. Combining with CA and ST modules contributes to the proposed CATNBOD scheme. The effectiveness and superiority have been sufficiently demonstrated on publicly available UDD dataset.
In this paper, to effectively strengthen quality of underwater image enhancement from both channel and spatial viewpoints, an adaptive channel attention-based deformable generative adversarial networks (ACADGAN) framework is established. Main contributions are as follows. 1) By virtue of multi-branch convolution architecture with dilated convolution mechanism, the adaptive channel attention (ACA) is devised, such that channel weight can be adaptively recalibrated, and thereby significantly contributing to preserving content features from channel viewpoint. 2) By augmenting offset position of sampling point with respect to convolution kernel, the deformable convolution network (DCN) is created, such that detailed information of underwater image can be dramatically retained from spatial aspect. 3) The ACADGAN scheme is eventually proposed by integrating ACA and DCN modules with a deep generative adversarial network. Comprehensive experiments demonstrate the remarkable effectiveness and superiority of the developed ACADGAN scheme.
Concept drift detection has attracted considerable attention due to its importance in many real-world applications such as health monitoring and fault diagnosis. Conventionally, most advanced approaches will be of poor performance when the evaluation criteria of the environment has changed (i.e. concept drift), either can only detect and adapt to virtual drift. In this paper, we propose a new approach to detect real-drift in the chunk data stream with limited annotations based on concept confusion. When a new data chunk arrives, we use both real labels and pseudo labels to update the model after prediction and drift detection. In this context, the model will be confused and yields prediction difference once drift occurs. We then adopt cosine similarity to measure the difference. And an adaptive threshold method is proposed to find the abnormal value. Experiments show that our method has a low false alarm rate and false negative rate with the utilization of different classifiers.
In this article, double finite-time observers-based line-of-sight guidance (DFLOS) and adaptive finite-time control (DFLOS-AFC) scheme is presented for path following of sailboat. The time-varying sideslip angle is accurately observed by double finite-time sideslip observers (DFSO), which improves the robustness and accuracy of line-of-sight guidance. For heading controller, the finite-time filter is used to solve the differential explosion problem of the virtual control law. The second order integral terminal sliding mode is devised to ensure the precision of heading control. Finite-time auxiliary system is introduced to compensate for actuator saturation constraints. Model uncertainties and external disturbance are compensated through three adaptive updated parameters, which reduces the calculation complexity of heading control system. Finally, simulation and comparisons results demonstrate the effectiveness of DFLOS-AFC scheme.
Speech recognition in smart home systems has become popular in both, research and consumer areas. This paper introduces an innovative concept for a modular, customizable, and voice-controlled smart home system. The system combines the advantages of distributed and centralized processing to enable a secure as well as highly modular platform and allows to add existing non-smart components retrospectively into the smart environment. To interact with the system in the most comfortable way - and in particular without additional devices like smartphones - voice-controlling was added as the means of choice. The task of speech recognition is partitioned into decentral Wake-Up-Word (WUW) recognition and central continuous speech recognition to enable flexibility while maintaining security. This is achieved utilizing a novel WUW algorithm suitable to be executed on small microcontrollers which uses Mel Frequency Cepstral Coefficients as well as Dynamic Time Warping. A high rejection rate up to 99.93
This paper presents a P300-based Brain Computer Interface (BCI) for the control of a mechatronic actuator (i.e. wheelchairs, robots or even cars), driven by EEG signals for assistive technology. The overall architecture is made up by two subsystems: the Brain-to-Computer System (BCS) and the mechanical actuator (a proof of concept of the proposed BCI is shown using a prototype car). The BCS is devoted to signal acquisition (6 EEG channels from wireless headset), visual stimuli delivery for P300 evocation and signal processing. Due to the P300 inter-subject variability, a first stage of Machine Learning (ML) is required. The ML stage is based on a custom algorithm (t-RIDE) which allows a fast calibration phase (only 190 s for the first learning). The BCI presents a functional approach for time-domain features extraction, which reduces the amount of data to be analyzed. The real-time function is based on a trained linear hyper-dimensional classifier, which combines high P300 detection accuracy with low computation times. The experimental results, achieved on a dataset of 5 subjects (age: 26 ± 3), show that: (i) the ML algorithm allows the P300 spatio-temporal characterization in 1.95 s using 38 target brain visual stimuli (for each direction of the car path); (ii) the classification reached an accuracy of 80.5 ± 4.1
Battery life and power consumption have been a challenging real-world problem for the internet of things (IoT). IoT applications in biomedical, agriculture, ecosystem monitoring, wildlife management, etc., need an accurate estimation of average battery life based on the environment and application. In this paper, we opt for an experimental approach and use various types of real-world environmental conditions such as the presence of interferences and high-intensity lights, to determine the actual power consumption of IoT nodes with a new set of off-the-shelf AA batteries for each scenario. We took readings in each of these environments such as an indoor Basketball Court, an Auditorium, and a room (our lab) and to verify results in outdoor conditions we chose parking lot as one of the testing environments. Further analysis and experimentation were performed to get detailed results. Results were obtained using widely used Zolertia Z1 hardware motes arranged in a specific and consistent pattern. We have compared our experimental results with simulated results in the Cooja simulator.
