Search

High performance analysis and processing technology of load data is an important basic of situation awareness technology in the context of power Internet of things. Under this background, a high performance deep learning load classification method for the massive class imbalanced load data is presented. Firstly, the mogrifier long short-term memory (Mogrifier-LSTM) network is presented as the basic classification algorithm to solve the deficiency of information interaction between input layer and hidden layer in traditional LSTM. In addition, attention mechanism and bidirectional feature fusion mechanism are introduced to further improve the ability of learning the association characteristics of time series data. Secondly, the Spark distributed computing framework is employed to realize the parallel processing of massive load data, which can improve the efficiency of classification task. And then, ensemble learning is introduced to improve the classification accuracy of the parallel processing. At last, the ADASYN method is employed to balance the load data, which can solve the potential class imbalance issue. The experiments show that the presented load classification method has advantages in terms of classification accuracy and efficiency.

Due to the rapid development of solar cells based on organic metal halide perovskite as light-absorber, the power conversion efficiency (PCE) of perovskite solar cells (PSCs) has increased from 3.8% to 25.5%, which exceeds that of silicon ones. In the meantime, the stability of the device has also made breakthrough progress. However, the degradation of perovskite is the primary reason which largely limits large-scale application. PSCs quickly chemically react with the water/oxygen in the air and decompose when exposed to ambient condition. It is urgently necessary to develop appropriate encapsulation techniques to isolate perovskite from external stimuli including moisture, oxygen and UV light to effectively extend lifetime. In this paper, the influences and physical mechanisms of the stability of perovskite are summarized. Then, the main encapsulation methods and technical approaches to improve and enhance the stability of perovskite are also reviewed, including traditional rigid-substrate-based encapsulation, thin film encapsulation, and advanced encapsulation strategies. Finally, it overviews the challenges and further development strategies faced by PSCs encapsulation techniques in practical applications.

Smart audio devices are gated by an always-on lightweight keyword spotting program to reduce power consumption. It is however challenging to design models that have both high accuracy and low latency for accurate and fast responsiveness. Many efforts have been made to develop end-to-end neural networks, in which depthwise separable convolutions, temporal convolutions, and LSTMs are adopted as building units. Nonetheless, these networks designed with human expertise may not achieve an optimal trade-off in an expansive search space. In this paper, we propose to leverage recent advances in differentiable neural architecture search to discover more efficient networks. Our searched model attains 97.2% top-1 accuracy on Google Speech Command Dataset v1 with only nearly 100K parameters., Comment: ICASSP 2021

Cell-Free massive multiple-input multiple-output (MIMO), which is expected to become one of the core technologies of the B5G and 6G, has attracted more and more researchers' attention. This paper considers a Cell-Free Massive MIMO system with multiple antennas at access points (AP) and a single antenna at users. At the same time, this system applies AP selection technology. To reduce pilot contamination, this paper proposes a new pilot assignment algorithm based on weighted counting, which uses the user's prior geographic information to maximize the pilot reuse weighted distance, and also considers pilot power control in the pilot training phase and AP power control in the downlink phase to verify system performance. Simulation results show that the algorithm can significantly improve system performance.

As an important information asset, the combination of energy and big data is the future development direction. The construction of energy big data center is a part of the national energy security strategy and also a guarantee for the construction of digital economy. In this paper, we put forward the construction idea and theoretical system of provincial energy big data center from the two aspects of overall architecture and implementation path respectively, which provides a replicable reference case for the actual construction of various regions. The contribution of this paper are two folds. First, the overall framework of the provincial energy big data center has been systematically studied, so as to make it serve the government, enterprises and the public. Second, a set of reproducible and scalable paths for construction and implementation of the center for the initial stage has been proposed, in order to offer references to construct an "open-ended, integrated, creative, safe, reliable and normative" provincial energy big data center.

Due to the development of the SIN (Space Information Network) and its characteristics which lead to scenarios that could not be encountered in terrestrial network, large-scale emulation and experimental verification are urgently needed. Various kinds of network simulation methods exist for SIN but are generally lack of flexibility (e.g. real network emulation) or accuracy (e.g. OPNET, NS and etc.). In previous works, we had presented a laboratory scale platform, based on SDN (Software Defined Networking) and virtualization technologies, to conduct credible research and testing on SIN. In this paper, we describe a distributed and extensible architecture on the basis of the original platform and use Docker to reduce the need of computing resources. Then, we construct a Spark-based network analysis system which supports capturing network traffic and analyzing network performance for different emulation scenarios among our platform. Experiments demonstrate that the platform performs well for the emulation of SIN and the analysis system can analyze network performance efficiently.

Particle swarm optimization (PSO) attracts much attention due to its ability in solving complex practical engineering problems effectively. To further improve its performance, a heterogeneous particle swarm optimizer (HPSO) is proposed in this work. Five widely used benchmark functions are selected to test its efficiency. Furthermore, five state-of-the-art improved PSO variants are selected for a comparisons purpose. The results demonstrate that HPSO is better than the other five algorithms. A logistics problem in aircraft manufacturing is then studied and solved. The results show HPSO's superiority over its tested PSO variants.

With superior flexibility, mobility, ease of use and integration, vehicle-mounted command and scheduling system has irreplaceable advantages in emergency and disaster relief, military command, urban management, border patrol and other fields. The current system login authentication and scheduling of each dispatching terminal control depend on dialogue with the information interaction and communication between the scheduling server, if the scheduling server software and hardware failures or offline, the entire vehicle scheduling command system will lose key scheduling command center is paralyzed, and all vehicle terminal equipment will not be able to certification and control scheduling. Aiming at this problem, this paper proposes a dynamic password authentication server without center node type method, when the vehicle scheduling command system scheduling server cannot be included in the communication network, can be either car terminal equipment as scheduling server starts the server program, quickly and play the function of the center node, a certified instructions parsing, dynamic instruction generation and dynamic matching certification process for each terminal equipment, such as identity authentication and scheduling control, effective guarantee the stability of the vehicle scheduling command system operation.

Giant potential value lies in the application of energy big data on energy resource distribution and energy structure transformation. On the basis of the investigation of the national energy information system and big data platform for energy enterprises, development goals are specified, the overall structural design of the "three platforms and two systems" and the scheme of data management are proposed. Aiming at government, enterprises, and the public, design and blueprint of the multi-dimensional application scenario system are presented, and the function scheme of typical application scenarios serving the government and the public are discussed.

A Bi-swarm Particle Swarm Optimizer with novel neighborhood topology strategy (BPSO-NT) is proposed in this paper. The increase of its population diversity helps to improve its global search ability. The strategy of updating neighborhood topology that plays a vital role in the particle swarm optimization algorithm (PSO) is studied by leveraging link prediction techniques. Different learning strategies are utilized to update the velocity of individuals in the two swarms of BPSO-NT. From comparison results with the state-of-the-art PSO variants on ten benchmark functions, the superiority of the proposed algorithm is demonstrated. Furthermore, BPSO-NT is applied to the intermodal transportation planning, and statistical results show that BPSO-NT outperforms other PSO variants in this practical optimization problem.

Human-following robots can bring a lot of convenience to our lives. The function of human-following can be decomposed into three processes: human detection, tracking and following. Considering cost and adaptability. In this paper, a low-cost laser scanner is used to achieve it. And an improved method of threshold-based clustering is proposed, which can effectively solve the problem of excessive clustering when it is far away from the laser scanner. Three types of features are extracted to recognize the legs from other objects by the random forest classifier. The position in the middle of the legs is used to indicate the position of the target person, which will be tracked by the robot using a Kalman filter algorithm and nearest neighbor algorithm. After inputting the coordinates of the followed target into the controller, the linear velocity and angular velocity of the following can be obtained. Experiments show that the robot can follow the target person stably.

With the ever-increasing sensor types and complexity in the field of activity recognition, proper multi-sensor configuration system is essential to balance the recognition performance improvement and increased computational complexity caused by the use of multiple homogeneous or heterogeneous sensors. The multi-sensor deployment problem is normally transformed to multiple ensemble classifier pruning problem, while the competence of various ensemble pruning approaches for a particular subject are generally different. In this paper, a dynamic ensemble pruning selection model using meta-learning (META-DEPS) is proposed to recommend the most competent ensemble pruning algorithm for a given test sample. We utilize statistics features of processed sensor dataset to form a representative meta-feature set. Then meta-features extracted from the training dataset are used to train the instance-based meta-learner with newly designed ranking rules. Finally, an ensemble pruning algorithm is selected based on the trained meta-learner and realize the prediction of test instance in the generalization phase. The system conducts empirical studies on real-world activity recognition dataset and the results show that the proposed meta-learning based framework improves the accuracy of activity recognition when compared against using conventional ensemble pruning.

Delay-/disruption-tolerant networking (DTN) is developed for reliable data delivery in challenging communication environments such as the harsh deep-space environment where the traditional Internet protocols (mainly TCP/IP) typically do not work effectively. Because the energy resources available on space spacecrafts are generally very limited, energy consumption is an important performance issue for space communications. Energy resource consumption is even critical for communication systems in deep-space flight missions, and it must be evaluated with respect to the transmission efficiency. Bundle protocol (BP) was developed as the main protocol of DTN for reliable data transmission. Some studies have been done in evaluating the transmission performance, mainly throughput/goodput, on the use of the “reliable BP” for file delivery in deep-space communication scenarios. However, a very limited number of studies have been done in investigating the energy resource consumption issue of BP. In this paper, analytical modeling and the numerical evaluation results are presented to study this issue. The analysis is done by considering both the traditional “reactive” retransmission method and “proactive” retransmission method for reliable transmission service of BP. It is found that the “proactive” retransmission method of BP transmits a much larger amount of data than the traditional transmission method for reliable delivery of a file, leading to extra energy consumption.

Delay-/disruption-tolerant networking (DTN) is developed for reliable data delivery in challenging communication environments characterized by a very long propagation delay and/or lengthy, random link disruption and highly lossy data links. Space-based networks and deep-space communication scenarios are typical of this type of communication environment. Bundle protocol (BP) and Licklider transmission protocol (LTP) were developed as the main protocols of DTN for reliable data transmission in space. Some studies have been done in evaluating the transmission performance, mainly throughput/goodput, on the use of BP/LTP for reliable file delivery in deep-space communication scenarios. The National Aeronautics and Space Administration (NASA) has also demonstrated DTN and implemented the use of its BP protocol on the International Space Station (ISS) for data delivery to the earth ground station. However, a little work has been done in analyzing the performance of BP/LTP for reliable data/file delivery from the ISS to the ground stations, especially in presence of link disruption. In this paper, analytical modeling and the numerical evaluation results are presented to study the effect of link disruptions on the use of BP/LTP for reliable file delivery over a relay-based communication architecture between the ISS and the ground stations.

The traditional Siamese network is often used to solve the similarity classification problem with a small amount of data and large number of categories. This method can determine the similarities and differences between different objects, but only limited to two classifications: Same or different. This paper proposes a similarity classification method based on a new Siamese architecture ---- Multi-State Siamese (MSS) network, which is used to distinguish multi-class similar scenes. And it is applied to indoor visual navigation to assist the agent to decide how to move from the current observation scene to the target scene. In the MSS network, a new loss function model is established to minimize the distance between the same category during training. By classifying indoor visual scenes, the agent is guided to realize navigation function. The experiments verify the convergence of the model and achieve good accuracy in recognition.

Photovoltaic (PV) power generation has broad prospects as one of the renewable generations, and its penetration in the power grid keeps increasing. However, PV system does not have the advantages of large inertia and output impedance as synchronous generators, and cannot realize active frequency regulation in real time. Some scholars have proposed a droop control algorithm based on the self-droop characteristics of synchronous generators, which makes the PV system have the primary frequency regulation capability, but it still lacks inertia and damping links in the control process. To solve this problem, in this paper, a frequency control strategy for PV system using virtual synchronous generator (VSG) control is proposed. Based on droop control, the swing equation is added to improve the dynamic characteristics of PV system during frequency control. Simulation results verified that the proposed algorithm can achieve the same primary frequency regulation capability as the droop control, and also effectively improve the dynamic characteristics of frequency change.

Quick UDP Internet connection(QUIC) protocol and bottleneck bandwidth and round-trip propagation time (BBR) based congestion control algorithm are two major contributions out of Google’s persistent attempts to make the Internet faster. QUIC is an user level reliable protocol running on top of UDP to enable evolution of transport mechanisms, while BBR is designed for better TCP congestion control. Although in two different directions, they also converge with BBR support in QUIC. Many researches have proved the performance of QUIC and BBR in different network scenarios, but the evaluation in satellite Internet with long propagation delays and relatively high error rates is still missing.In this paper, we have a preliminary evaluation of the performance of QUIC with BBR via GEO satellite Internet access on dedicated network emulation testbeds. The results and our analysis confirms that for satellite Internet QUIC with the new BBR congestion control has improvements compared with the classic CUBIC.

The Internet of things (IoT) is embedded into the industrial scene and combined with various devices, facilities and materials to form Smart Objects (SOs). The SOs could be thought of as a heterogeneous swarm in a smart factory under Industrial IOT environment. The process of production is the coordinated operation of various production resources, but dynamic events often interfere with production activities. To solve this problem, an swarm collaboration architecture is proposed. Specifically, it considers the influence of human factors with Human-in-the-Loop on workshop intelligent equipment interaction, and the multi-level dynamic scheduling method of swarm collaboration under the framework is discussed.

This paper introduces a design scheme of a Mine Network camera based on Open Network Video Interface Forum (ONVIF). The device adopts Cortex-A8 processor as the central control unit, adopts infrared laser as auxiliary illumination unit, adopts CMOS Charge-coupled Device (MCCD) image sensor as acquisition unit and adopts power supply adaptation scheme of intrinsically safe circuit technology to obtain video collection function in all black environment. It compresses, stores, and transmits captured video images. It overcomes the shortcomings of unclear collection of dark environment in the existing products and serious red storms. It supports Transmission Control Protocol/Internet Protocol (TCP/IP), User Datagram Protocol (UDP) and ONVIF protocols, and realizes seamless connection with common clients on the market. After testing, the visual distance of the dark environment is 50 meters. Through the network, the monitoring information can be obtained in real time on the client supporting ONVIF, and the high-definition video collection can be realized in the dark environment under the mine.

Existing routing protocols of wireless sensor networks (WSNs) attempted to optimize the energy efficiency and the routing reliability from the perspective of the network itself and failed to take into consideration the environmental impact from outside, causing them cannot make prompt reaction to the dynamic changes of the environments (e.g., wildfire). Thus, in these routing protocols the routing survivability under harsh environments is questionable. To tackle this issue, in this paper by referencing the concept of potential field, we design an environment-cognitive multipath routing protocol (ECMRP) in order to provide sustainable message forwarding service under harsh environments. In ECMRP, routing decisions are made according to a mixed potential field in terms of depth, residual energy and environment. The basic idea of this approach is to instruct data packets to select routes with the tradeoff among latency, energy conservation and routing survivability. As the environmental field is constructed and updated using the sensing capability of WSN itself, constructed routes can avoid crossing through the danger zones to keep the paths safe. The experimental results show that ECMRP can obtain significant improvements in packet delivery ratio and network lifetime under harsh conditions.

Aiming at the issues of information sensing, transmission, analysis, feedback control and information exchange in metropolitan area networks, a new wireless mesh transmission network based on network coding technology is designed in this paper. The network nodes store, forward and process the received data with transmission characteristics of the wireless channel, and optimize inter-message client communication. By analyzing the connection mode of relay nodes and the multi-source information coding, a multicast relay network coding algorithm and a multi-source random network coding algorithm are proposed. The test results indicate that the bandwidth of 54Mbps can reach 49.3Mbps after 3-hop transmission, and can still reach 10.43Mbps through 24-hop, the packet loss rate of 220GB within 10s is as low as 0.19%, which solves the problem of multi-hop transmission bandwidth utilization and effective transmission of massive data.

In this paper, DC power and voltage characteristics in MMC-HVDC grid are analyzed. Coupling between DC voltage and power in MMC grid under power imbalance is proposed based on MMC converter response, and time-varying curve of DC voltage is analyzed. A simplified model of four-terminal MMC-HVDC grid is built, and simulations of DC voltage change under power imbalance is done. Analyses and simulation results show that, under fixed control response; a positive proportional relationship exists between power variation and DC voltage variation. Time-varying curve of DC voltage can be calculated by a first order differential equation.

An innovative Big Data enabled Intelligent Immune System (12S) has been developed to address manufacturing dynamics over life cycles to achieve sustainable manufacturing. The 12S is a novel inter-disciplinary integrated system of the artificial immune mechanism, the Wireless Sensor Network (WSN)-based Cyber Physical System (CPS), Artificial Neural Networks (ANNs) and sustainable manufacturing scheduling optimization algorithm. Abnormal energy consumption patterns of manufactured components from monitored Big Data are identified using ANNs. An intelligent immune mechanism is devised to adapt to the pattern/condition changes of machine tool systems and process dynamics. A rescheduling algorithm is triggered if abnormal manufacturing conditions are detected thereby achieving adaptive multiobjective optimization of energy consumption and manufacturing performance. Computer Numerical Controlled (CNC) machining processes have been used for system validation via industrial deployment into multiple machine lines in manufacturing factories. Around 30% in energy saving and over 50% in productivity improvement have been achieved by adopting 12S.

Cyber-Physical System (CPS) is a combination of closely integrated physical processes, network and computation. It has been widely used in manufacturing because of its advantages in connecting cyber world and physical world. Computer numerical control (CNC) machine tools play a vital role in modern manufacturing, and their machining accuracy is greatly affected by thermal errors. In this paper, a thermal error compensator based on CPS for CNC machine tools is designed and developed. According to the requirements of a CPS, the compensator is developed on an embedded system, which can speedily collect data from sensors, precisely predict thermal errors, and effectively communicate with CNC system as well as cloud service platform. The compensator is applied in a heavy-duty CNC machine tool and the result shows that it produces intended effect in different machining conditions.

Falls are a critical public health issue that requires continuous monitoring, especially for the elders. This paper proposed a method based on a tri-axial gyroscope for fall events recognition. A tri-axial gyroscope is placed at the user's waist to collect tri-axial angular velocity information. In order to facilitate data processing and extract features, real-time data are divided into a set of consecutive and partially overlapping windows. Three time-domain features that reflect the differences between the falls and other movements in our daily lives are extracted from these consecutive data windows. Then, each of these windows is classified as representing either a fall or a non-fall event by using a trained machine learning classifier. Decision Tree is chosen as the classifier because of its low algorithm complexity and easy implementation on embedded systems. Experimental results have shown that our proposed method can effectively differentiate the fall events from other human daily activities in spite of their high similarity in some cases, with the Accuracy of 99.52%, Precision of 0.993, Recall of 0.995 and F-measure of 0.994.

The guidance system is an important component of AGV. In order to obtain the guidance system with low cost, high flexibility and precision, a combinatorial inertial guidance method is proposed in this paper. The inertial guidance system is composed of the encoder and the inertial measurement unit (IMU). And, on this basis, the magnetic sensor (MS) is added to correct the cumulative error. Moreover, because the IMU and MS can't identify the current location of AGV, this paper adopts the passive RFID tag to calibrate the important location of AGV to assist the AGV positioning. In addition, in order to obtain more accurate control parameters, the kalman filter (KF) is used for data processing in the system. Finally, the experiments demonstrate the effectiveness and accuracy of the proposed system.

By combining the polarized modulation and the spatial modulation technique, dual-polarized spatial modulation (DPSM) has been recently proposed for MIMO transmission to improve the spectral efficiency and reduce the space occupancy. The performance of DPSM over correlated Rayleigh and Rice fading channels has been studied, showing that DPSM technique outperforms uni-polarized SM technique. For a general investigation, in this paper, we study the performance of DPSM scheme over Nakagami-m fading channel, which is a more general model than Rayleigh and Rice channel. A tight upper bound of average bit error rate (ABER) for DPSM scheme over Nakagami-m fading channel is derived, based on the ABER framework. In addition, we present detailed analysis of performance for different shape parameter m in the channel model. Finally, we validate the accuracy of the theoretical derivation through the results of the numerical simulations.