Search

Rapid and accurate monitoring of the chlorophyll content of ginseng leaves to reflect the growth condition of ginseng is of great significance in guiding the cultivation of ginseng. To address the problem that the existing means of detecting chlorophyll in ginseng leaves are time-consuming and disruptive and cannot meet the demand for rapid detection of chlorophyll in ginseng leaves, this study firstly establishes a variety of prediction models for chlorophyll in ginseng leaves based on the hyperspectral reflectance data and the vegetation index, respectively, and determines the strengths and weaknesses of the models by comparing the RMSE and the MAE of the test sets; Secondly, the analytical model construction process used for ginseng leaf chlorophyll content prediction was obtained through comparison and summary, and a fast and non-destructive ginseng leaf chlorophyll prediction method based on hyperspectral imaging technology and combining vegetation indices with machine learning algorithms was proposed; The experimental results showed that the final VI-SPA-RFR ginseng leaf chlorophyll prediction model had the best prediction performance, which had an RMSE of 1.1568 and an MAE of 0.9936 in the test set. In summary, this study provides a method to achieve rapid and accurate monitoring of the chlorophyll content of ginseng leaves, which is expected to provide a scientific means of monitoring for the cultivation of ginseng and expand the application field of hyperspectral imagers in the field of agriculture.

The high charging efficiency of LLC resonant converters over a very wide range of load conditions in electric vehicle charging systems has long been a research hotspot. In this paper, a hybrid control method is proposed for LLC resonant converter by combining pulse width modulation (PWM) and burst control to address the issues of small excitation current and limited output voltage range under the light load condition. The resonant current energy in the burst-on period is used to supplement the missing excitation current energy from PWM control zero voltage switching to realize PWM-Burst control. Therefore, a hybrid control of PFM and PWM-Burst is employed for LLC resonant converters under full load conditions. The proposed hybrid control method focuses on steady-state operation, and its operating principles are introduced and analyzed. Finally, simulation and experiments were conducted through Matlab-Simulink to verify the feasibility of the proposed method, and a 1-kW laboratory prototype was constructed to validate the effectiveness and advantages of the proposed method.

Unmanned aerial vehicles have a wide range of applications. An intelligent optimization algorithm based on the traditional bat algorithm (BA) is investigated in this paper for UAV flight path planning in a static complex environment. The primary goal of this work is to develop a safer flight path while considering the feasibility of the UAV and the requirements for safe operation. This research proposes an improved spherical coordinate and truncated average stable strategy-based bat optimization algorithm (TMS-SBA). The algorithm uses the UAV’s motion space to encode the operator, and by substituting a new bat for the worst of the old one after each iteration to increase population diversity, the algorithm can converge quickly in a complex environment while maintaining stable operation. In addition, the flight path is smoothly generated by using B-spline curves to make the planned path suitable for UAV. MATLAB simulation experiments show that, compared with other traditional swarm intelligent algorithms, TMS-SBA can successfully generate feasible and effective optimal solutions in complex environments and plan shorter, safer, and more accessible flight paths for UAV.

Reconfigurable Intelligent Surface (RIS) has emerged as a promising technology in wireless networks to achieve high spectrum and energy efficiency. RIS typically comprises a large number of low-cost nearly passive elements that can smartly interact with the impinging electromagnetic waves for performance enhancement. However, optimally configuring massive number of RIS elements remains a challenge. In this article, we present a machine learning (ML) based modeling approach that learns the interactions between the phase shifts of the RIS elements and receiver (Rx) location attributes and uses them to predict the achievable rate directly without using channel state information (CSI) or received pilots. Once learned, our model can be used to predict optimal RIS configuration for any new receiver locations in the same wireless network. We leverage deep learning (DL) techniques to build our model and study its performance and robustness. Simulation results demonstrate that the proposed DL model can recommend near-optimal RIS configurations for test receiver locations which achieved close to an upper bound performance that assumes perfect channel knowledge. Our DL model was trained using less than 2% of the total receiver locations. This promising result represents great potential in developing a practical solution for the optimal phase shifts of RIS elements without requesting CSI from the wireless network infrastructure.

A detection model of Internet of Things encrypted traffic based on edge intelligence is proposed in the paper, which can reduce the communication times of distributed Internet of Things gateways in the process of edge intelligence as well as the encrypted traffic detection model establishment time, in order to solve the problems that it is difficult to carry out efficient classification and accurate identification of the encrypted traffic of Internet of Things. In this paper, four new classification and identification methods for encrypted traffic are put forward, namely time-sequence behavior analysis, dynamic behavior analysis, key behavior analysis and two-round filtering analysis. The experimental results show that when the sample size is 1600, the encrypted traffic detection model establishment time is less than 100 seconds, and the accuracy of all the four new traffic classification methods is more than 92% and the recall rates of them are more than 83%.

Integrated motor-transmission (IMT) system is prone to suffer from torsional oscillations due to its fast dynamic behavior and weak damping characteristics. Meanwhile, signal communication between sensors, controllers and actuators via controller area network (CAN) that introduces time-varying delays, could also stimulate torsional oscillations in IMT system. Thus, this paper is intended to develop a robust controller to suppress these oscillations for IMT system during vehicle speed tracking. Considering the coupling effects of CAN-induced time-varying delays and event-driven manner of the controller nodes, as well as possible sampling period change, a delay-free discrete time model is built for IMT system by using polytopic inclusion approach and system augmentation technique. Based on this model, an energy-to-peak performance based robust controller is then developed for IMT system over CAN. It can achieve stable as well as good torsional oscillations suppression performance in spite of CAN-induced time-varying delays, sampling period change and also measurement noises. Finally, with a detailed CAN model developed by using SimEvent, conventional proportional-integral (PI) controller and energy-to-peak controller that designed with fixed sampling period are utilized in the comparative tests to show the effectiveness as well as performance of proposed torsional oscillations controller.

In order to make more effective use of Wi-Fi fingerprint data to position an object, an improved adaptive genetic algorithm (IAGA) is proposed to optimize the BP (Back Propagation) neural network, namely, IAGA-BP. In this method, the selection, crossover and mutation operations of the genetic algorithm are used to optimize the weights and biases of the BP neural network. On the one hand, the proposed algorithm improves the selection operator in the adaptive genetic algorithm on the basis of preserving the optimal strategy. That is, the population of each generation will be sorted according to the adaptability from the highest to the lowest, then the highest 20% of the population will be directly inherited to the next generation while the worst 20% will be eliminated. The remaining 80% of the population will be selected by a roulette algorithm based on the selection probability of each individual, as to ensure the population volume unchanging. On the other hand, the crossover and mutation probability formulas in the adaptive genetic algorithm are improved. The crossover and mutation rates will be adjusted to preserve superior individuals and genes according to the level of individual adaptability and the current evolution stage of the population. The simulation results show that compared with the traditional Wi-Fi positioning method, the proposed Wi-Fi positioning method has a faster convergence speed and better positioning accuracy of 2.48 meters.

The existing approaches usually treat the recognition of pornographic images as a binary or multi-class classification task, which is realized by extracting a variety of features. However, these approaches ignore the problem of infinite kinds in the negative samples and lose sight of the impact of uncertainty in classification tasks, resulting in inadequate negative data sets and incorrect recognition of samples that are not in the training set. In order to address this challenge, this paper proposes a method named Deep One-Class with Attention for Pornography (DOCAPorn) that recognizes the pornographic images through the one-class classification model based on neural networks and introduces the visual attention mechanism to enhance the performance of recognition. Moreover, since the existing approaches based on deep convolutional neural networks (CNNs) require a fixed-size (e.g., $224\times224$ ) input image, the geometric distortion caused by image scaling is not considerd in the existing approaches, which reduces the pornographic image recognizition accuracy. In order to solve this issue, this paper proposes the Scale Constraint Pooling (SCP) that converts the inputs of different dimensions into outputs of the same dimension. In addition, all the existing approaches ignore the adversarial attacks in the field of pornographic image recognition. In order to deal with this problem, the paper proposes the Preprocessing for Compressing and Reconstructing (PreCR), a pre-processing approach that reduces the subtle perturbation through compressing the images and then reconstructs the purified image for recognition. The proposed approach is verified by conducting comparative experiments using custom datasets. The experimental results showed that we achieved an accuracy of 98.419% on our dataset. In addition, the proposed approach yielded a recognition accuracy of 95.632% on the NPDI dataset. Furthermore, the obtained results demonstrate that the proposed approach not only effectively recognizes pornographic images but also effectively defends the adversarial attacks.

The information of earth structure is curial for designing the grounding system and evaluating the transmission line parameters. In order to obtain the earth structure using the measured data, an efficient algorithm for the parameter estimation of a horizontally multilayered soil by using a novel integration method is proposed. In the proposed algorithm, an adaptive iteration process will terminate till the predefined error reaches the requirement, where the earth parameters, such as the layer depth, layer resistivity, and layer number, are adjusted accordingly with the help of the Levenberg-Marquardt algorithm. In each iteration, the key is to evaluate the apparent resistivity and its partial derivatives with respect to soil parameters, which are expressed in the form of general Sommerfeld integral (GSI). In this paper, the GSI is divided into two parts. The first part with an infinite upper limit is integrated along a novel complex integration path instead of the conventional real axis path, which makes the evaluation of the GSI highly efficient due to the fast damping property of the transformed kernel. The second part with the finite integral interval is calculated using the cubic spline technique to have a semi-analytic expression, which can avoid evaluating the oscillatory kernel. In comparison with the results in the published literatures, numerical examples show that the proposed soil parameters estimation approach can reach good accuracy, and is robust and efficient. The reason is simple as the GSIs are calculated by using the well-designed and sophisticated approach. With extensive numerical experiments, the proposed algorithm is believed to be a capable and applicable candidate in engineering to invert the information of the earth structure.

With the rapid advancement of the online social network, social media like Twitter has been increasingly critical to real life and become the prime objective of spammers. Twitter spam detection refers to a complex task for the involvement of a range of characteristics, and spam and non-spam have caused unbalanced data distribution in Twitter. To solve the mentioned problems, Twitter spam characteristics are analyzed as the user attribute, content, activity and relationship in this study, and a novel spam detection algorithm is designed based on regularized extreme learning machine, called the Improved Incremental Fuzzy-kernel-regularized Extreme Learning Machine (I2FELM), which is used to detect the Twitter spam accurately. As revealed from the experience validation results, the proposed I2FELM can efficiently identify the balanced and unbalanced dataset. Moreover, with few characteristics taken, the I2FELM can more effectively detect spam, which proves the effectiveness of the algorithm.

Online social networks (OSNs) are generally susceptible to Sybil attack, which causes a series of cybersecurity problems and privacy violations. Malicious attackers can create massive Sybils and further utilize those fake identities to launch various Sybil attacks. Therefore, Sybil detection in OSNs has become an urgent security research problem for both academia and industries. The existing content-based methods to detect Sybils base on the combination of manual-design features and machine learning algorithms, which requires lots of professional experiences and human effort. These methods divide the Sybil detection problem into two piece-wise sub-problems, which prevents us from getting the optimal solution. In this work, we propose a novel content-based method to detect Sybils. The proposed method is an end-to-end classification model that extracts features directly from the input data, and then output the classification results in a unified framework. The proposed method includes three main parts: first, the self-normalizing convolutional neural network (CNN) is adopted to extract lower features from the multi-dimensional input data; second, the bidirectional self-normalizing LSTM network (bi-SN-LSTM) is developed to extract higher features from the compressed feature map sequence; third, the dense layer and softmax classifier are stacked to output the classification results. Unlike the traditional bidirectional long short-term memory network (bi-LSTM), the proposed bi-SN-LSTM network utilizes SELU as the activation function of its recurrent step, which provides unbounded changes to the state value. Through the case study of the real-world dataset, the comparison experiments demonstrate that our method significantly outperforms other state-of-the-art methods.

Total focus method (TFM) is an ultrasonic phased array imaging method and has a larger dynamic focusing range and higher spatial resolution than traditional imaging methods. But TFM needs to process a large amount of data, resulting in a long imaging time, which greatly limits its application and development in the industrial field. The sparse method can improve the computing efficiency by reducing the amount of data, but the resolution and intensity of sparse image are very low. This paper proposes a virtual source sparse TFM (VSSTFM) method based on phase coherent weighting (PCW) to improve the spatial resolution and intensity of the sparse TFM imaging. The layout of virtual sources in the virtual source array is designed and optimized by genetic algorithm. According to the phase distribution characteristics of the ultrasonic testing signal, a phase coherence factor is constructed to weight the VSSTFM image to further improve the image quality. Experimental results show that compared with conventional TFM, the proposed method can improve the imaging efficiency by 66.56 % while providing significantly higher image quality than TFM imaging.

Motivated by the importance of fault-tolerant control for robot movement, a fault-tolerant motion planning scheme is proposed to eliminate the joint velocity jump of the redundant manipulator at the time of failure, and enhance the motion stability of the manipulator fault-tolerant operation. The joint velocity jump is eliminated by replacing the degradation scheme at the moment of joint failure, and a neural dynamic method is introduced to eliminate the position error of the end-effector of the manipulator, and the analysis method based on pseudoinverse and inverse-free is used to plan the motion of the manipulator in real time. The simulation results of the redundant manipulator based on the planar four-link indicate that the fault-tolerant scheme can not only realize the manipulator fault-tolerant control without joint velocity jump, but also effectively guarantee the manipulator operation precision.

With the improvement of hardware capability of mobile devices, mobile devices are more and more widely used. Mobile screen video recording is one of the important applications. However, the generated video files will occupy plenty of memory resources, which are limited in mobile devices. Fortunately, a well designed video coding method could effectively relieve such pressure. Therefore, aimed at the mobile screen video, this paper proposes a down-sampling based rate control algorithm to improve the compression efficiency. Firstly, the source video is down sampled by a factor of 4. Then, the down sampled video is encoded twice and the coding information is stored. Finally, based on the stored coding information, the real encoding process is optimized at bit allocation and bit control. Experimental results show that compared with the default rate control method in high efficiency video coding test model, the proposed method could obviously improve rate distortion performance and bit control accuracy.

Due to the complex and time-varying network environments, traditional methods are difficult to extract accurate features of intrusion behavior from the high-dimensional data samples and process the high-volume of these data efficiently. Even worse, the network intrusion samples are submerged into a large number of normal data packets, which leads to insufficient samples for model training; therefore it is accompanied by high false detection rates. To address the challenge of unbalanced positive and negative learning samples, we propose using deep convolutional generative adversarial networks (DCGAN), which allows features to be extracted directly from the rawdata, and then generates new training-sets by learning from the rawdata. Given the fact that the attack samples are usually intra-dependent time sequence data, we apply long short-term memory (LSTM) to automatically learn the features of network intrusion behaviors. However, it is hard to parallelize the learning/training of the LSTM network, since the LSTM algorithm depends on the result of the previous moment. To remove such dependency and enable intrusion detection in real time, we propose a simple recurrent unit based (SRU)-based model. The proposed model was verified by extensive experiments on the benchmark datasets KDD'99 and NSL-KDD, which effectively identifies normal and abnormal network activities. It achieves 99.73% accuracy on the KDD'99 dataset and 99.62% on the NSL-KDD dataset.

Screen content video (SCV) is generated by computers, including animation, texts and graphics. SCV contains continuous static frames and many scene changes, making SCV different from conventional camera captured video (CCV) in terms of temporal characteristic. Therefore, conventional quantization parameter (QP) cascading method may not be efficient for SCV. In this paper, a distortion propagation based QP cascading method is proposed for SCV. The special temporal characteristic of SCV is considered and the distortion propagation of every coding tree unit (CTU) is measured. Based on the CTU level distortion propagation, the improved QP cascading method is designed. Experimental results show that compared with other methods, the proposed could achieve better rate distortion (RD) performance and less encoding time.

This paper presents an efficient entry guidance law for no-fly zone avoidance, which is based on the recently developed linear pseudospectral model predictive control. First, a simple mapping relation of the position of the no-fly zone between the inertial frame (IF) and the auxiliary geographical frame (AGI) is derived in a geometric manner. Second, a model predictive method is used to judge whether the constraint of the no-fly zone is activated in AGI. Then, additional bank reversal is performed to shape the entry trajectory at the right time so as to avoid the no-fly zone. This method is very easy to be implemented onboard and does not increase the additional computational burden. The nominal and Monte Carlo simulations are conducted to show that the proposed method consistently offers very great, stable, and robust performances.

In this paper, a new two-stage optimization framework is proposed to determine the optimal-mix integration of dispatchable Distributed Generation (DG), in power distribution networks, in order to maximize various techno-economic and social benefits simultaneously. The proposed framework incorporates some of the newly introduced regulatory policies to facilitate low carbon networks. A modified Taguchi Method (TM), in combination with a node priority list, is proposed to solve the problem in a minimum number of experiments. Nevertheless, the standard TM is computationally fast but has some inherent tendencies of local trapping and usually converges to suboptimal solutions. Therefore, two modifications are suggested. A roulette wheel selection criterion is applied on priority list to select the most promising DG nodes and then modified TM determines the optimal DG sizes at these nodes. The proposed approach is implemented on two standard test distribution systems of 33 and 118 buses. To validate the suggested improvements, various algorithm performance parameters such as convergence characteristic, best and worst fitness values, and standard deviation are compared with existing variants of TM, and improved genetic algorithm. The comparison shows that the suggested corrections significantly improve the robustness and global searching ability of TM, even compared to meta-heuristic methods.

Negative selection algorithm (NSA) is an important method for generating detectors in artificial immune systems. Traditional NSAs randomly generate detectors in the whole feature space. However, with increasing dimensions, data samples aggregate in some specific subspaces, not uniformly distributed in the whole space. The detectors randomly generated by traditional NSAs cannot exactly fall into these specific subspaces, which results in a low coverage of detectors and a poor performance in a high-dimensional space. To overcome this defect, an improved real NSA based on subspace density seeking (SDS-RNSA) is proposed in this paper. In an SDS-RNSA, a subspace density seeking algorithm is adopted to procure the dense subspace regions of samples. Then, detectors are generated in each subspace region to cover up nonself-region efficiently and improve the performance of the algorithm. During the process of detector generation, the redundancy of candidate detectors is calculated, and the redundant is eliminated to minimize the time expense of the algorithm. Experimental results demonstrate that, compared with the classic NSAs, the SDS-RNSA can significantly improve the detection rate with an approximative false alarm rate and a smaller time expense. At the best case, the detection rate of the SDS-RNSA is increased by 14.7%, while the time expense is decreased by 78.1%.