Search

With the development of advanced metering infrastructure (AMI), large amounts of electricity consumption data can be collected for electricity theft detection. However, the imbalance of electricity consumption data is violent, which makes the training of detection model challenging. In this case, this paper proposes an electricity theft detection method based on ensemble learning and prototype learning, which has great performance on imbalanced dataset and abnormal data with different abnormal level. In this paper, convolutional neural network (CNN) and long short-term memory (LSTM) are employed to obtain abstract feature from electricity consumption data. After calculating the means of the abstract feature, the prototype per class is obtained, which is used to predict the labels of unknown samples. In the meanwhile, through training the network by different balanced subsets of training set, the prototype is representative. Compared with some mainstream methods including CNN, random forest (RF) and so on, the proposed method has been proved to effectively deal with the electricity theft detection when abnormal data only account for 2.5% and 1.25% of normal data. The results show that the proposed method outperforms other state-of-the-art methods.

This paper proposes a robust and computationally efficient control method for damping ultra-low frequency oscillations (ULFOs) in hydropower-dominated systems. Unlike the existing robust optimization based control formulation that can only deal with a limited number of operating conditions, the proposed method reformulates the control problem into a bi-level robust parameter optimization model. This allows us to consider a wide range of system operating conditions. To speed up the bi-level optimization process, the deep deterministic policy gradient (DDPG) based deep reinforcement learning algorithm is developed to train an intelligent agent. This agent can provide very fast lower-level decision variables for the upper-level model, significantly enhancing its computational efficiency. Simulation results demonstrate that the proposed method can achieve much better damping control performance than other alternatives with slightly degraded dynamic response performance of the governor under various types of operating conditions.

The socioeconomic growth of China and local urban vitality have both been significantly impacted by the outbreak of COVID-19. It is of great significance to scientifically evaluate the changes of urban vitality in the Short Period of Time before and after the Outbreak of the Epidemic. Taking Zhengzhou City as an example, this study selected the epidemic period in May 2022, based on traffic big data and traffic simulation data, and applied the AHP-entropy weight method, gray relational TOPSIS method, time series method, spatial analysis method, visual analysis method, gravity model, standard deviation ellipse, multiple linear regression model and other methods. The daily characteristics, stage characteristics and temporal and spatial patterns of urban vitality affected by COVID-19 in Zhengzhou were revealed at multiple scales. The results are as follows: The change of urban vitality in the short term of the epidemic outbreak is basically in line with the change model of urban vitality. Erqi District has the greatest impact on the change of urban vitality, which is the center of gravity of urban vitality change. Among the means of transportation, pedestrians, nonmotor vehicles, taxis, private cars, and buses all had a positive correlation effect on the change of urban vitality in Zhengzhou, with nonmotor vehicles contributing the most. Trucks had a negative correlation with the change of urban vitality. The results of this study indicated that traffic data had good application prospects in the real-time evaluation and analysis of major health emergencies.

The multi-directional flow of energy in a multi-microgrid (MMG) system and different dispatching needs of multiple energy sources in time and location hinder the optimal operation coordination between microgrids. We propose an approach to centrally train all the agents to achieve coordinated control through an individual attention mechanism with a deep dense neural network for reinforcement learning. The attention mechanism and novel deep dense neural network allow each agent to attend to the specific information that is most relevant to its reward. When training is complete, the proposed approach can construct decisions to manage multiple energy sources within the MMG system in a fully decentralized manner. Using only local information, the proposed approach can coordinate multiple internal energy allocations within individual microgrids and external multilateral multi-energy interactions among interconnected microgrids to enhance the operational economy and voltage stability. Comparative results demonstrate that the cost achieved by the proposed approach is at most 71.1% lower than that obtained by other multi-agent deep reinforcement learning approaches.

Image segmentation is an essential part of remote sensing preprocessing. This paper proposes a deep learning-based multiscale pyramid sieve and analysis module to solve the problems associated with multiscale object coexistence in the complex semantic environments of multispectral remote sensing images. The image information distribution features, which change with scales in different convolutional feature map pyramids, were investigated, and the relationship between informative scales and the variation function curve was examined. Then, the proposed sieve and analysis module was designed according to the results. The proposed module is described as follows. (1) First, a feature map pyramid is built via a convolution calculation. (2) Sieve informative feature layers from the feature map pyramid with variation analysis. (3) Fuse informative feature layers. (4) The final feature representation is input into subsequent calculations for image segmentation. Experimental results demonstrate that, compared to the control group, the precision of the experimental group improved by 3.57%-5.89%, and the intraclass conformity improved by 2.74%-5.58%. In addition, the intraclass chaos decreased by 1.5%-7.9%. The experimental results demonstrate that the proposed multiscale pyramid sieve and analysis module can separate informative feature layers and improve the hierarchical segmentation accuracy of remote sensing multispectral images.

Accurate and timely fault diagnosis is of great significance for the safe operation and power supply reliability of distribution systems. However, traditional intelligent methods limit the use of the physical structures and data information of power networks. To this end, this study proposes a fault diagnostic model for distribution systems based on deep graph learning. This model considers the physical structure of the power network as a significant constraint during model training, which endows the model with stronger information perception to resist abnormal data input and unknown application conditions. In addition, a special spatiotemporal convolutional block is utilized to enhance the waveform feature extraction ability. This enables the proposed fault diagnostic model to be more effective in dealing with both fault waveform changes and the spatial effects of faults. In addition, a multi-task learning framework is constructed for fault location and fault type analysis, which improves the performance and generalization ability of the model. The IEEE 33-bus and IEEE 37-bus test systems are modeled to verify the effectiveness of the proposed fault diagnostic model. Finally, different fault conditions, topological changes, and interference factors are considered to evaluate the anti-interference and generalization performance of the proposed model. Experimental results demonstrate that the proposed model outperforms other state-of-the-art methods.

The existing approach for evaluating the statistical characteristics of speckle intensity distribution is applicable only to completely decorrelated laser beams. In the current study, a theoretical method based on a coherent matrix is developed for evaluating the first-order speckle statistics of partially correlated beams. A rigorous analytical formula for the calculation of speckle contrast is obtained based on the proposed coherent matrix method. Subsequently, an approximate analysis model of the speckle intensity distribution in a laser spot illuminated by partially correlated beams is proposed. The simulation results show that the accuracy of the approximate analysis model increases with the number of beams and is acceptable for practical application, even for a small number of beams. Applications of the proposed method are presented for analyzing two different optical schemes for speckle suppression. It is shown that the developed theoretical method is straightforward and effective for the analysis of light intensity statistics of laser spots illuminated by partially correlated beams. Therefore, the developed method can be easily generalized to imaging systems with partially correlated laser beams for evaluating the speckle intensity statistics.

Individual tree detection from airborne laser scanning (ALS) point clouds is the basis for forestry inventory and further applications. In the past decade, many methods have been developed to localize tree instances in ALS point clouds. These methods rely on empirical rules and field measurements that may change from plot to plot. Besides, most existing methods cannot consider multiple clues (e.g., shape priors and neighboring trees) under the same framework, which makes them not flexible and extensible. In this letter, we devise a new point-based and model-driven framework named “crown guess and selection”. This framework first generates crown candidates automatically, and then the qualities of candidates and their neighboring information are both considered. Finally, expected crowns are selected from candidates simultaneously. The proposed framework is tested and evaluated in a benchmark dataset. We also compare the new framework with several existing methods, and it turns out that the proposed framework outperforms others in terms of model flexibility and detection accuracy.

A time-variable time-of-use electricity price can be used to reduce the charging costs for electric vehicle (EV) owners. Considering the uncertainty of price fluctuation and the randomness of EV owner's commuting behavior, we propose a deep reinforcement learning based method for the minimization of individual EV charging cost. The charging problem is first formulated as a Markov decision process (MDP), which has unknown transition probability. A modified long short-term memory (LSTM) neural network is used as the representation layer to extract temporal features from the electricity price signal. The deep deterministic policy gradient (DDPG) algorithm, which has continuous action spaces, is used to solve the MDP. The proposed method can automatically adjust the charging strategy according to electricity price to reduce the charging cost of the EV owner. Several other methods to solve the charging problem are also implemented and quantitatively compared with the proposed method which can reduce the charging cost up to 70.2% compared with other benchmark methods.

This study proposes a deep reinforcement learning (DRL) based approach to analyze the optimal power flow (OPF) of distribution networks (DNs) embedded with renewable energy and storage devices. First, the OPF of the DN is formulated as a stochastic nonlinear programming problem. Then, the multi-period nonlinear programming decision problem is formulated as a Markov decision process (MDP), which is composed of multiple single-time-step sub-problems. Subsequently, the state-of-the-art DRL algorithm, i.e., proximal policy optimization (PPO), is used to solve the MDP sequentially considering the impact on the future. Neural networks are used to extract operation knowledge from historical data offline and provide online decisions according to the real-time state of the DN. The proposed approach fully exploits the historical data and reduces the influence of the prediction error on the optimization results. The proposed real-time control strategy can provide more flexible decisions and achieve better performance than the pre-determined ones. Comparative results demonstrate the effectiveness of the proposed approach.

Deep learning has become a mainstream method in marine data processing field. However, the raw marine data, characterized by fluctuations, outliers and noise, is a serious obstacle to its performance improvement. To address this problem, a hybrid deep computing model is developed for marine sensor data prediction, combining smoothing and deep belief echo state network (DBEN). The proposed structure adopts a two-stage data processing mode to deal with the complex characteristics of marine time series. In the preprocessing stage, four smoothing methods are considered for fluctuation reduction and outlier handling, so that purer data conducive to achieving a higher prediction accuracy can be gained. In the prediction stage, DBEN equipped with efficient feature learning serves as a nonlinear approximator. The effectiveness of the constructed hybrid model is tested on real-world marine time series of different sources and traits, and its superiority in prediction accuracy is demonstrated by readout comparisons and statistical measures.

Seawater quality prediction has a tremendous potential of enabling future smart ocean. However, this time-sensitive application puts forward a strict delay requirement, thus easily leading to overwhelmed networks. Edge computing is emerging as an effective means of solving network overload, due to its edge-based distributed processing. Therefore, we develop a hybrid multivariate prediction model for seawater quality assessment in an edge computing environment, considering the combination of principal component analysis (PCA) and relevance vector machine (RVM). The PCA method is employed for dimension reduction of ten seawater quality factors in advance. Six principal components are extracted from multiple features, used as input variables of the subsequent predictor. Finally, a RVM is developed to predict the future trends of dissolved oxygen and pH, measuring seawater quality. Experimental results on the real-world ocean sensor data show that our PCA-RVM based multivariate prediction model outperforms single RVM, SVM and its extended version in prediction accuracy and efficiency, meanwhile statistical testings confirm this finding.

A 3-D printed antenna subsystem, which proves its effectiveness in cost- & weight-reduction and reliability improvement over conventional CNC fabrication, is proposed in this paper. The proposed structure which integrates an orthogonal-mode transducer (OMT) and a corrugated horn is fully compatible to fabrication using high temperature resistant resin Acrylonitrile Butadiene Styrene (ABS) with 3-D printing technology, as well as coating using copper with electroless metallization method. Detailed design procedures are provided and a prototype operating at X-band is designed, fabricated and measured. Over the whole operating frequency range of 11.7 GHz~12.7 GHz, measured reflection coefficients are better than -22 dB and -12 dB for different polarization directions. A very good isolation better than 60 dB is achieved between two input ports. The cross polarization level is at least 30 dB lower than the co-polarization. Moreover, the effectiveness of the fabricated antenna is also verified in a polarimetric radiometer system.

With the growing integration of distributed energy resources (DERs), flexible loads, and other emerging technologies, there are increasing complexities and uncertainties for modern power and energy systems. This brings great challenges to the operation and control. Besides, with the deployment of advanced sensor and smart meters, a large number of data are generated, which brings opportunities for novel data-driven methods to deal with complicated operation and control issues. Among them, reinforcement learning (RL) is one of the most widely promoted methods for control and optimization problems. This paper provides a comprehensive literature review of RL in terms of basic ideas, various types of algorithms, and their applications in power and energy systems. The challenges and further works are also discussed.

Fingerprint liveness detection has gradually been regarded as a primary countermeasure for protecting the fingerprint recognition systems from spoof presentation attacks. The convolutional neural networks (CNNs) have shown impressive performance and great potential in advancing the state-of-the-art of fingerprint liveness detection. However, most existing CNNs-based fingerprint liveness methods have a few shortcomings: 1) the CNN structure used on natural images does not achieve good performance on fingerprint liveness detection, which neglects the inevitable differences between natural images and fingerprint images; or 2) a relative shallow architecture (typically several layers) has not paid attention to the capability of deep network for spoof fingerprint detection. Motivated by the compelling classification accuracy and desirable convergence behaviors of the deep residual network, this paper proposes a new CNN-based fingerprint liveness detection framework to discriminate between live fingerprints and fake ones. The proposed framework is a lightweight yet powerful network structure, called Slim-ResCNN, which consists of the stack of series of improved residual blocks. The improved residual blocks are specifically designed for fingerprint liveness detection without overfitting and less processing time. The proposed approach significantly improves the performance of fingerprint liveness detection on LivDet2013 and LivDet2015 datasets. Additionally, the Slim-ResCNN wins the first prize in the Fingerprint Liveness Detection Competition 2017, with an overall accuracy of 95.25%.

Spoof surface plasmon polaritons (SSPPs) fed open-loop elements without a metallic ground plane are investigated for the SSPPs-based antenna design. By employing a straight SSPPs section close to the open end of the loop as the feed, the one-wavelength and one-and-a-half-wavelength modes of the open-loop are excited effectively, which can generate radiation fields with orthogonal polarizations. Based on the open-loop arrays operating in different modes, two Ka-band SSPPs-fed leaky-wave antennas with frequency-scanning in the H- or E- plane are then designed, fabricated and measured. Wide impedance bandwidths, stable gain of around 13 dBi and radiation efficiency of higher than 70% are confirmed experimentally. The studies in this letter provide a new means to adjust the polarization features of the SSPPs-fed antennas. Benefitted from the compact single-layered geometry that is easy to integrate with circuits and good radiation performance, the proposed SSPPs-fed antenna arrays would be an attractive candidate for future millimeter-wave wireless applications.

As a frequent natural disaster, red tide has attracted more and more attentions. In fact, red tide results from the joint actions of multiple complex marine environmental factors. Unfortunately, there is no work on the interaction analysis between these factors. To inaugurate a systematic research of this area, a novel machine learning based framework is developed for marine environmental series analysis. It combines cross recurrence plot (CRP), cross recurrence quantification analysis (CRQA) and statistical analysis. This framework provides a general way to transform two marine series into a high-dimensional space. CRP is used to visualize internal dynamics, while the influence of factors is quantitatively analyzed through CRQA. Finally, the representative factors in each field are statistically determined by boxplot. This is the first analysis framework attempting to reveal the similarity in intrinsic dynamics of marine factors. Experimental results show that the framework is competent to perform the visualization of marine time series. Besides, the results also demonstrate the degree of similarity between different marine factors through quantitative analysis.

In this paper, a modified artificial bee colony (MABC) algorithm is proposed to estimate the unknown fractional-order nonlinear systems. First, the parameter estimation problem can be mathematically transformed into a multi-dimensional continuous function optimization problem, which is solved via the MABC algorithm. The proposed MABC algorithm well combines the original artificial bee colony algorithm (ABC) and the Nelder-Mead simplex method (NMS) in a very simple way, which takes full advantage of the exploration ability of the ABC algorithm and the exploitation ability of the NMS. And in the meanwhile, the proposed algorithm improves the speed of convergence. To evaluate the effectiveness of the proposed MABC algorithm, numerical simulations contain two typical uncertain fractional-order nonlinear systems. The results show that compared with the other population-based algorithm and ABC-variants, the proposed MABC for solving the problem of parameter estimation get the faster convergence speed and higher calculation accuracy.

A novel adaptive radial basis function neural network H-infinity control strategy with robust feedback compensator using linear matrix inequality (LMI) approach is proposed for micro electro mechanical systems vibratory gyroscopes involving parametric uncertainties and external disturbances. The proposed system is comprised of a neural network controller, which is designed to mimic an equivalent control law aimed at relaxing the requirement of exact mathematical model and a robust feedback controller, which is derived to eliminate the effect of modeling error and external disturbances. Based on the Lyapunov stability theorem, it is shown that H-infinity tracking performance of the gyroscope system can be achieved, all variables of the closed-loop system are bounded, and the effect due to external disturbances on the tracking error can be attenuated effectively. Numerical simulations are investigated to demonstrate that the satisfactory tracking performance and strong robustness against external disturbances can be obtained using the proposed adaptive neural H-infinity control strategy with robust feedback compensator by LMI technique.

This paper extends a conventional adaptive fuzzy sliding mode controller with integer order to fractional ones for a three-phase active power filter (APF) in order to track instruction current quickly and precisely and offset the harmonic current in the electrical grid. A fractional sliding surface is utilized to make the system work on the designed sliding surface stably. A fuzzy system with fractional adaptive laws is employed to precisely approximate the unknown parameters in the APF system. Simulations are given to demonstrate the excellent performance of the proposed controller. Moreover, the superiority of the proposed controller over the conventional adaptive fuzzy sliding mode controller is also illustrated by showing the smaller total harmonic distortion, which is a significant index to evaluate the current quality in the smart grid.

A novel adaptive method for the micro-electro mechanical systems (MEMSs) gyroscope based on a dynamic surface control and combined with the approach of adaptive fuzzy and sliding mode control (SMC) was shown in this paper. The first-order filter was introduced to the conventional adaptive backstepping technique in the dynamic surface control, which not only maintains the advantage of original backstepping technique, but also reduces the number of parameters and avoids the problem of parameter expansion. The adaptive fuzzy logic system is employed to approximate the dynamic characteristics of the gyroscope; the SMC is a kind of compensation for the error of the fuzzy approximation, which reduced the chattering phenomenon in adaptive control significantly. The proposed control scheme has improved the dynamic characteristics of the gyroscope, reducing the chattering of inputs and improving the timeliness and effectiveness of tracking. Simulation results indicate that the proposed control scheme has superior performance compared with conventional SMC.