Search

Deadlocks are a highly undesirable situation in flexible manufacturing systems (FMSs). This article presents a direct and novel method to detect such markings by constructing the resource flow graph of a Petri net that models an FMS and to recover such markings by adding a set of recovery transitions. First, an algorithm is developed to build a new kind of directed graph called the resource flow graph of a Petri net. Resource flow graphs can well represent the competition for shared resources by different processes. Second, based on the resource flow graph, loop graphs can be found. Furthermore, partial deadlock markings can be easily detected due to their relationship with loop graphs. Then, we propose an algorithm to design a set of recovery transitions for loop graphs that are enabled at partial deadlock markings. The proposed approach can detect partial deadlock markings without generating a complete reachability graph of a Petri net and the resulting net is deadlock-free with all reachable markings by adding the obtained recovery transitions. Finally, some widely used examples are provided to demonstrate the proposed approach.

With integer linear programming problems (ILPPs) being formulated and solved, the existing approaches design optimal Petri-net supervisors via nonpure net structures, including self-loops and data inhibitor arcs. Nonpure net structures are powerful for control of Petri-net-modeled discrete-event systems. However, in the existing work, the formulated ILPPs contain a large number of constraints, which is computationally inefficient. In this article, we propose approaches that formulate ILPPs with fewer constraints such that the computational efficiency is significantly improved. To do so, in formulating ILPPs for optimal Petri-net controllers by using self-loops and data inhibitor arcs, we remove the reachability conditions for legal markings. By doing so, an obtained solution may result in some legal markings unreachable. To solve this problem, a novel technique is developed to design an optimal controller by modifying the initial marking and structure of the obtained supervisor. It is shown that, by the reduced ILPPs, one can find the same feasible solutions as that obtained by the existing work. Finally, the proposed approaches are demonstrated by examples.

The hypotactic structural relation between clauses plays an important role in improving the discourse coherence of document-level translation. However, the standard neural machine translation (NMT) models do not explicitly model the hypotactic relationship between clauses, which usually leads to structurally incorrect translations of long and complex sentences. This problem is particularly noticeable on Chinese-to-English translation task of complex sentences due to the grammatical form distinction between English and Chinese. English is rich in grammatical form (e.g. verb morphological changes and subordinating conjunctions) while Chinese is poor in grammatical form. These linguistic phenomena make it a challenge for NMT to learn the hypotactic structure knowledge from Chinese as well as the structure alignment between Chinese and English. To address these issues, we propose to model the hypotactic structure for Chinese-to-English complex sentence translation by introducing hypotactic structure knowledge. Specifically, we annotate and build a hypotactic structure aligned parallel corpus that provides rich hypotactic structure knowledge for NMT. Moreover, we further propose a structure-infused neural framework to combine the hypotactic structure knowledge with the NMT model through two integrating strategies. In particular, we introduce a specific structure-aware loss to encourage the NMT model to better learn the structure knowledge. Experimental results on WMT17, WMT18 and WMT19 Chinese-to-English translation tasks demonstrate the effectiveness of the proposed methods.

Flying insects are remarkably agile and robust. As they fly through cluttered natural environments, they can demonstrate aggressive acrobatic maneuvers such as backflip, rapid escape, and in-flight collision recovery. Current state-of-the-art subgram microaerial-vehicles (MAVs) are predominately powered by rigid actuators such as piezoelectric ceramics, but they have low fracture strength (120 MPa) and failure strain (0.3%). Although these existing systems can achieve a high lift-to-weight ratio, they have not demonstrated insect-like maneuvers such as somersault or rapid collision recovery. In this article, we present a 665 mg aerial robot that is powered by novel dielectric elastomer actuators (DEA). The new DEA achieves high power density (1.2 kW/kg) and relatively high transduction efficiency (37%). We further incorporate this soft actuator into an aerial robot to demonstrate novel flight capabilities. This insect-scale aerial robot has a large lift-to-weight ratio (>2.2:1) and it achieves an ascending speed of 70 cm/s. In addition to demonstrating controlled hovering flight, it can recover from an in-flight collision and perform a somersault within 0.16 s. This work demonstrates that soft aerial robots can achieve insect-like flight capabilities absent in rigid-powered MAVs, thus showing the potential of a new class of hybrid soft-rigid robots.

Decision fusion is an effective way to resolve the conflict of diagnosis results. Aiming at the problem that Dempster-Shafer (DS) theory deals with the high conflict of evidence and produces wrong results, a decision fusion algorithm for fault diagnosis based on closeness and DS theory is proposed. Firstly, the relevant concepts of DS theory are introduced, and the normal distribution membership function is used as the evidence closeness. Secondly, the harmonic average is introduced, and the weight of each evidence is established according to the product of closeness of each evidence and its harmonic average. Thirdly, the weight of conflicting evidence is regularized, and the final decision fusion result is obtained by using the Dempster’s rule. Lastly, the simulation and application examples are designed. Simulation and application results show that the method can effectively reduce the impact of diagnostic information conflicts and improve the accuracy of decision fusion. What’s more, the method considers the overall average distribution of evidence in the identification framework, it can reduce evidence conflicts while preserving important diagnostic information.

Millimeter-sized electrostatic film actuators, inspired by the efficient spatial arrangement of insect muscles, achieve a muscle-like power density (61 W kg−1) and enable robotic applications in which agility is needed in confined spaces. Like biological muscles, these actuators incorporate a hierarchical structure, in this case building from electrodes to arrays to laminates, and are composed primarily of flexible materials. So comprised, these actuators can be designed for a wide range of manipulation and locomotion tasks, similar to natural muscle, while being robust and compact. A typical actuator can achieve 85 mN of force with a 15 mm stroke, with a size of [Formula: see text] mm3 and mass of 92 mg. Two millimeter-sized robots, an ultra-thin earthworm-inspired robot and an intestinal-muscle-inspired endoscopic tool for tissue resection, demonstrate the utility of these actuators. The earthworm robot undertakes inspection tasks: the navigation of a 5 mm channel and a 19 mm square tube while carrying an on-board camera. The surgical tool, which conforms to the surface of the distal end of an endoscope, similar to the thin, smooth muscle that covers the intestine, completes tissue cutting and penetrating tasks. Beyond these devices, we anticipate widespread use of these actuators in soft robots, medical robots, wearable robots, and miniature autonomous systems.

In an integrated radar and communication system, the filter output of each pulse of the integrated radar and communication waveform usually has high sidelobe level (SLL) and signal-to-noise ratio (SNR) loss as well as serious range sidelobe modulation (RSM). To solve these problems, a joint weighted optimization (JWO) method is proposed to design the filters. In this method, the filters for different integrated radar and communication waveforms are designed by jointly minimizing the RSM, SNR loss, and SLL according to different practical demands. In terms of the RSM, SNR loss, and SLL, the performance of the designed filters is tunable by reasonably changing the weight factors. Finally, several numerical results are presented to verify the effectiveness of the proposed method.

The goal of sentence matching is to determine the semantic relation between two sentences, which is the basis of many downstream tasks in natural language processing, such as question answering and information retrieval. Recent studies using attention mechanism to align the elements of two sentences have shown promising results in capturing semantic similarity/relevance. Most existing methods mainly focus on the design of multi-layer attention network, however, some critical issues have not been dealt with well: 1) the higher attention layer is easily affected by error propagation because it relies on the alignment results of preceding attentions; 2) models have the risk of losing low-layer semantic features with the increase of network depth; and 3) the approach of capturing global matching information brings about large computing complexity for model training. To this end, we propose a Deep Bi-Directional Interaction Network (DBDIN) to solve these issues, which captures semantic relatedness from two directions and each direction employs multiple attention-based interaction units. To be specific, the attention of each interaction unit will repeatedly focus on the original sentence representation of another one for semantic alignment, which alleviates the error propagation problem by attending to a fixed semantic representation. Then we design deep fusion to aggregate and propagate attention information from low layers to high layers, which effectively retains low-layer semantic features for subsequential interactions. Moreover, we introduce a self-attention mechanism at last to enhance global matching information with smaller model complexity. We conduct experiments on natural language inference and paraphrase identification tasks with three benchmark datasets SNLI, SciTail and Quora. Experimental results demonstrate that our proposed method can achieve significant improvements over baseline systems without using any external knowledge. Additionally, we conduct interpretable study to disclose how our deep interaction network with attention can benefit sentence matching, which provides a reference for future model design. Ablation studies and visualization analyses further verify that our model can better capture interactive information between two sentences, and the proposed components are indeed able to help modeling semantic relation more precisely.

Nowadays, cluster tools tend to concurrently process multiple types of wafers with similar recipes in order to improve their utilization and flexibility in semiconductor manufacturing. Different wafer types may have different wafer flow patterns, resulting in that cluster tools are deadlock-prone. It is challenging to develop a general method to solve the deadlock problem of cluster tools without restriction on the wafer types. This work aims at solving such a challenging problem for single-arm cluster tools. To do so, a general Petri net model is developed for single-arm cluster tools. Given the wafer flow patterns of all wafer types to be processed in a single-arm cluster tool, such a Petri net model can be easily obtained by defining the relationship between places and transitions. Then, a control method by using self-loops is presented to prevent the model from deadlocks during the evolutions from the initial state to the final state. Furthermore, such a control method is proved to be optimal. Illustrative examples are given to verify the proposed method at last.

State space techniques are one of the main approaches deployed for the analysis of concurrent systems. However, state space construction is stalled by a common phenomenon called the state explosion problem which makes it a tough task or even impossible when the state space computation demands prohibitive cost (time and memory). We limit general resource allocation systems (RASs) to a certain class whose state space can be hierarchically constructed, yet it comprises various enough types of real-world discrete event systems, such as automated manufacturing systems. This paper focuses on a class of RASs modeled with Petri nets (PNs), where, through pure algebraic operations, a novel method to compute the state spaces is proposed, which is motivated by the superposition property. Given a PN model of a system and a target resource configuration, we first propose a special initial marking called the initial basis marking and compute the corresponding reachability graph. Then, we increase the capacity of the resource places in an incremental way and generate the reachability graphs by taking advantage of the PN structure and the previously computed reachability graph until the capacity function of resources reaches the target resource configuration. A complete enumeration of reachable states can be obtained by a recursive scheme. Experimental studies also demonstrate the efficiency of the proposed approach in terms of computational cost and its high-potential to cope with the state-explosion problem.

In the intelligent transportation system, it is an efficient way for the intelligent vehicle to use the integrated radar and communications system (IRCS) to obtain the range and velocity estimations of other vehicles and simultaneously communicate with other facilities, such as vehicles and base stations. In the IRCS, the transmitted waveform is the orthogonal frequency division multiplexing (OFDM) integrated radar and communications waveform that contains communications information. Due to the existence of communications information, the traditional range and velocity estimation methods in radar can not be directly utilized in the IRCS. Moreover, to improve the resolution of range and velocity estimations, the signal bandwidth and coherent processing interval (CPI) are usually required to be increased, which will result in the increase of system cost and the reduction of update rate. To solve these problems, a auto-paired super-resolution range and velocity estimation method is proposed by using the OFDM integrated radar and communications waveform. First, the communications information in the received signals is compensated. Then, the frequency smoothing is performed to reduce the correlation between the echoes reflected by different targets. Finally, the auto-paired super-resolution range and velocity estimations are obtained by exploiting the translational invariance structure of the underlying signal model in both frequency and pulse domains. Furthermore, Cramer-Rao bounds (CRBs) on range and velocity estimations with the OFDM integrated radar and communications waveform are derived. Several numerical examples are presented to verify the effectiveness of the proposed method.

In recent years, with the explosion of information on the Internet, there has been a large amount of data produced, and analyzing these data is useful and has been widely employed in real world applications. Since data labeling is costly, lots of research has focused on how to efficiently label data through semi-supervised learning. Among the methods, graph and hypergraph based label propagation algorithms have been a widely used method. However, traditional hypergraph learning methods may suffer from their high computational cost. In this paper, we propose a Hypergraph Label Propagation Network (HLPN) which combines hypergraph-based label propagation and deep neural networks in order to optimize the feature embedding for optimal hypergraph learning through an end-to-end architecture. The proposed method is more effective and also efficient for data labeling compared with traditional hypergraph learning methods. We verify the effectiveness of our proposed HLPN method on a real-world microblog dataset gathered from Sina Weibo. Experiments demonstrate that the proposed method can significantly outperform the state-of-the-art methods and alternative approaches.

Petri nets are an important and popular tool to model and analyze deadlocks in flexible manufacturing systems. The state space of a Petri net model can be divided into two disjoint parts: a live-zone and a dead-zone. Reachability graph analysis plays an important role in the modeling and control of Petri nets. Most existing studies have to fully enumerate the reachable markings of a Petri net to obtain the first-met bad markings (FBMs), which exacerbates the computational overheads. In this paper, a computationally efficient method to find dead markings in Petri nets is presented. We first introduce an algorithm to find dead markings by solving an integer linear programming problem. Then, the set of markings in the dead-zone is calculated, including the set of dead markings and the set of bad markings. Then we can find all the FBMs. By using a vector covering approach, the minimal covered set of FBMs is computed. The proposed approach can obtain the dead markings and FBMs by searching only a part of a reachability graph. Finally, examples are provided to demonstrate the proposed method.

As effective tools for time series analyzing, a variety of information entropies have been widely applied in engineering, economic, biomedicine and other fields. In this paper, we define a new distance between finite sequence based on inversion and derive a new entropy, Fuzzy Permutation Entropy(FPE). A comparison of recognition performance for WGN, 1/f noise, periodical, or chaotic sequence and sine waves shows that FPE is valid on distinguishing deterministic signals from stochastic signals. Further contrast studying versus FE and PE shows that FPE is more sensitive to the alteration of the complexity of time series and more effective on separating different signals. Moreover, FPE is used to explore the distinction of the complexity of various traffic flows via the time headways which is simulated from the improved brake light rule model. We hope it is conducive to design Intelligent Transportation Management System.

In this paper, we present a novel cooperative scheme of joint optimal resource allocation, such that the overall performance of the coexisting radar-communications (CRC) system can be improved. In our proposed scheme, target detection and multiuser communication are performed by radar and communication subsystems at the same time, as well as a control center, which is responsible for joint resource management. We aim to minimize the ISLR for target detection and maximize the sum-rate for communications simultaneously by jointly optimizing the band assignment and transmit power allocation. Since the resulting optimization problem involving two performance metrics and a binary constraint is a multiobjective nonconvex problem, a two-tier iterative decomposition (TT-ID) approach is devised to obtain the globally optimal solution. However, compared with the conventional radar signals, the autocorrelation function of the devised radar signal may still have relatively high sidelobes. In particular, when the data transmission becomes the primary purpose of the CRC system, the sidelobe performance gets worse. As a consequence, some weak targets are most likely overshadowed by the adjacent strong targets through the matched filtering at the radar receiver. To address this, a spectral estimation algorithm based on the Bayes Cauchy–Gaussian (Bayes–CG) model is employed to further reduce the range sidelobes of the matched filter output at the radar receiver according to the prior distribution of the desired autocorrelation. Finally, several numerical results are provided to show the merits of the proposed method.

With task-oriented dialogue systems being widely applied in everyday life, slot filling, the essential component of task-oriented dialogue systems, is required to be quickly adapted to new domains that contain domain-specific slots with few or no training data. Previous methods for slot filling usually adopt sequence labeling framework, which, however, often has limited ability when dealing with the domain-specific slots. In this paper, we take a new perspective on cross-domain slot filling by framing it as a machine reading comprehension (MRC) problem. Our approach firstly transforms slot names into well-designed queries, which contain rich informative prior knowledge and are very helpful for the detection of domain-specific slots. In addition, we utilize the large-scale MRC dataset for pre-training, which further alleviates the data scarcity problem. Experimental results on SNIPS and ATIS datasets show that our approach consistently outperforms the existing state-of-the-art methods by a large margin.

Stylized neural machine translation (NMT) aims to translate sentences of one style into sentences of another style, which is essential for the application of machine translation in a real-world scenario. However, a major challenge in this task is the scarcity of high-quality parallel data which is stylized paired. To address this problem, we propose an iterative dual knowledge transfer framework that utilizes informal training data of machine translation and formality style transfer data to create large-scale stylized paired data, for the training of stylized machine translation model. Specifically, we perform bidirectional knowledge transfer between translation model and text style transfer model iteratively through knowledge distillation. Then, we further propose a data-refinement module to process the noisy synthetic parallel data generated during knowledge transfer. Experiment results demonstrate the effectiveness of our method, achieving an improvement over the existing best model by 5 BLEU points on MTFC dataset. Meanwhile, extensive analyses illustrate our method can also improve the accuracy of formality style transfer.

Aiming at the dimensional disaster and over-fitting problems in data analysis, this paper proposes a feature selection method using hybrid integration of difference models and random forests (Integrate-RF), firstly, Integrate-RF use CART, CHAID, SVM, BN, NN, K-Means, Kohonen to evaluate the importance of features, and then, for the above seven sorts, Integrate-RF use the arithmetic average method to calculate the importance of the features; secondly, Integrate-RF select the most important features from the remaining features into features subset, and use random forest classification to get the corresponding out-of-bag(OOB) data classification error rate; finally, the optimal features subset can be selected based on the OOB data classification error rate. Experiments show that feature selection methods proposed in this paper effectively reduces the data dimension, selects features better and more adaptable.

Learning to order events at discourse-level is a crucial text understanding task. Despite many efforts for this task, the current state-of-the-art methods rely heavily on manually designed features, which are costly to produce and are often specific to tasks/domains/datasets. In this paper, we propose a new graph perspective on the task, which does not require complex feature engineering but can assimilate global features and learn inter-dependencies effectively. Specifically, in our approach, each document is considered as a temporal graph, in which the nodes and edges represent events and event-event relations respectively. In this sense, the temporal ordering task corresponds to constructing edges for an empty graph. To train our model, we design a graph mask pre-training mechanism, which can learn inter-dependencies of temporal relations by learning to recover a masked edge following graph topology. In the testing stage, we design an certain-first strategy based on model uncertainty, which can decide the prediction orders and reduce the risk of error propagation. The experimental results demonstrate that our approach outperforms previous methods consistently and can meanwhile maintain good global consistency.

Scheduled sampling is an effective method to alleviate the exposure bias problem of neural machine translation. It simulates the inference scene by randomly replacing ground-truth target input tokens with predicted ones during training. Despite its success, its critical schedule strategies are merely based on training steps, ignoring the real-time model competence, which limits its potential performance and convergence speed. To address this issue, we propose confidence-aware scheduled sampling. Specifically, we quantify real-time model competence by the confidence of model predictions, based on which we design fine-grained schedule strategies. In this way, the model is exactly exposed to predicted tokens for high-confidence positions and still ground-truth tokens for low-confidence positions. Moreover, we observe vanilla scheduled sampling suffers from degenerating into the original teacher forcing mode since most predicted tokens are the same as ground-truth tokens. Therefore, under the above confidence-aware strategy, we further expose more noisy tokens (e.g., wordy and incorrect word order) instead of predicted ones for high-confidence token positions. We evaluate our approach on the Transformer and conduct experiments on large-scale WMT 2014 English-German, WMT 2014 English-French, and WMT 2019 Chinese-English. Results show that our approach significantly outperforms the Transformer and vanilla scheduled sampling on both translation quality and convergence speed., Findings of ACL-2021, code at https://github.com/Adaxry/conf_aware_ss4nmt

This article addresses the forbidden state problem in discrete-event systems (DESs) modeled with Petri nets. Given a control specification, we first decide the sets of forbidden and admissible markings. Then, the minimal mask set of first-met forbidden markings (FFMs) and the minimal root set of admissible markings are computed by marking mask that is implemented using a class of special places in a plant, called competitive places. Marking mask can effectively filtrate the markings to be processed such that the two obtained sets are in general much smaller than the sets of originally specified forbidden and admissible markings, respectively. Monitors computed by place invariants are used to forbid the forbidden markings. It is shown that a maximally permissive (optimal) supervisor can be computed if it exists. Integer linear programming is used to optimize the structure of a supervisor. The minimal mask set of FFMs and root set of admissible markings efficiently reduce the computational overhead because of much fewer constraints and variables in the formulated programming problem. The developed methodology is illustrated by parameterized examples.

In this work, we proposed a long-term tracking strategy to deal with the occlusion, out-of-plane rotation, and the confusing non-target object. Our tracking system is composed of two parts, the CA-CF tracker, an efficient correlation method for short-term tracking, and the SVM-based re-detector, which prevents the CA tracker from degradation. When the tracker works with confidence, the CA-CF module ensures an accurate tracking result and the SVM updates accordingly. When the response maps fluctuate heavily, the SVM switches to work as a re-detector and the tracker will be initialized. We also introduced to adopt both the maximum response criterion and the APCE criterion to judge the performance of the tracker in time. By evaluating our algorithm on the OTB benchmark datasets, we proposed to analyze the result affected by the parameters of our CA-CF-SVM strategy. The experimental results show that our method has a significant improvement than the state-of-the-art methods for the long-term tracking both in accuracy and robustness.

With the development of the meteorological IoT (Internet of Things) and meteorological sensing network, the collected multi-source meteorological data have the characteristics of large amount of information, multidimensional and high accuracy. Cloud computing technology has been applied to the storage and service of meteorological big data. Although the constant evolution of big data storage technology is improving the storage and access of meteorological data, storage and service efficiency is still far from meeting multi-source big data requirements. Traditional methods have been used for the storage and service of meteorological data, and a number of problems still persist, such as a lack of unified storage structure, poor scalability, and poor service performance. In this study, an efficient storage and service method for multidimensional meteorological data is designed based on NoSQL big data storage technology and the multidimensional characteristics of meteorological data. In the process of data storage, multidimensional block compression technology and data structures are applied to store and transmit meteorological data. In service, heterogeneous NoSQL common components are designed to improve the heterogeneity of the NoSQL database. The results show that the proposed method has good storage transmission efficiency and versatility, and can effectively improve the efficiency of meteorological data storage and service in meteorological applications.

Hyperspectral imaging has great achievements in agriculture, astronomy, surveillance, and so on. However, the inherent low spatial resolution of hyperspectral imaging, unfortunately, limits its more widespread applications. Recently, hyperspectral image (HSI) super resolution addresses this problem by fusing a low spatial resolution HSI (LR-HSI) with a high spatial resolution multispectral image (HR-MSI), but most of these methods did not consider real-time restoration of high spatial resolution HSI. In this study, the authors propose a fast HSI super-resolution method which fills this blank. Specifically, they model the hyperspectral super resolution as a linear regression problem according to the fact that the imaging process is a linear transform and the inverse of this transform can be approximately estimated, as the spectra of a typical scene lie in a very low-dimensional space. To further exploit the low-dimensional nature of the spectra, they divide the HR-MSI and LR-HSI into several patches and learn the inverse transform patch-by-patch. Experiments on several public datasets show that their method approximates state-of-the-art methods in accuracy, but is several orders of magnitude faster than all of them. Furthermore, they provide an efficient C language implementation of their methods, which can meet the real-time request.

Along with the rapid development of information technology, the information technology education trend is emerged. The enhancement of quality becomes the cross-century direction of education reform. To grasp the trend of time and integrate information multimedia into traditional teaching courses for enhancing teaching quality are the primary issues for professional teachers. With an experimental design model, Total 100 students of Hubei University of Automotive Technology, as the research objects. The class is an 8-week experimental instruction every Thursday (32 hours in total, 16 hours in class, another 16 hours are for out-of-class activities). The research results reveal that information multimedia instruction shows higher self-efficacy than general instruction and achieves the significance, information multimedia instruction reveals higher learning effect than general instruction and reaches the significance, and self-efficacy appears notably positive effects on learning effect. According to the results to propose suggestions, it is expected, through the integration of information multimedia, to induce and maintain students’ interests in the curricula and even the willingness to make efforts towards the learning goal for better learning effect.

Based on the hypothesis that all vehicles on roads are floating vehicles equipped with vehicular terminals and upload their traffic information to the traffic control center in real time, many traffic flow guidance algorithms are proposed in order to improve the road capacity in Intelligent Transportation System. Based on this complete information scenario, the traffic control center can obtain all traffic information on roads and compute the guidance metrics according to the traffic flow guidance algorithm. The vehicle therefore can select the route with the guidance. However, due to the cost and technology limits, in short time, some vehicles cannot be equipped with vehicular terminals, which are non-floating vehicles and cannot upload their traffic information to the traffic control center. In this incomplete information scenario, the guidance effect will be affected by non-floating vehicles. In this paper, the method of estimating non-floating vehicles’ driving information according to floating vehicles’ information is introduced. With the estimation method, a new traffic flow guidance algorithm, Estimated Weighted Vehicle Density Feedback Strategy based on Weighted Vehicle Density Feedback Strategy (WVDFS) is proposed. In the simulation, the guidance effect of the new algorithm is performed based on the two-route scenario, which is the simplified model of traffic network. NaSch model is also used as the vehicle mobility model, which can simulate the vehicle motion. The simulation results show that non-floating vehicles have great negative influence on the performance of WVDFS, and the applicability of our algorithm is validated to have better performance in incomplete information scenarios. Our algorithm can provide solutions for current traffic flow guidance with incomplete information scenarios.

Based on the brake light (BL) model, Knospe et al. proposed a symmetric two-lane cellular automaton (BL-STCA) model, which could reproduce various empirically observed two-lane phenomena. In real traffic, the effect of brake light on the lane-changing behavior cannot be ignored. Therefore, BL-STCA model is interesting. However, there are two problems with BL-STCA model, too strong exchange of vehicles between lanes and unreasonable phenomenon in some special scenarios, such as a broken-down vehicle parked on one lane due to traffic accident. In order to solve the problems, we introduce the dynamic lane-changing probability and proposed an improved BL-STCA model with modification of lane-changing rules. The simulation results show as below. (1) Our new model effectively solves the above two problems of BL-STCA model. In addition, the lane-changing frequency is consistent with the real traffic data, which means the validity of our new model. (2) Compared with single-lane scenario, the lane-changing behaviors in two-lane scenario can effectively suppress the emergence of wide moving jam. (3) From the microcosmic level, the lane-changing behaviors can well explain the moving blank phenomenon within wide moving jam.

This paper focuses on solving deadlock problems in flexible manufacturing systems modeled with Petri nets by adding a set of recovery transitions. Different from the traditional deadlock control policies that add control places to a net model, this paper employs transitions to recover all the deadlock markings to be legal ones. A live net system can be obtained with all reachable markings. First, according to reachability graph analysis, a set of recovery transitions is obtained for each deadlock marking to be recovered. Second, we present a vector intersection approach to compute a recovery transition to recover multiple deadlock markings. Then, an iterative method is developed to find a set of recovery transitions to recover all deadlock markings. At each iteration step, a recovery transition is found to recover more than one deadlock markings. This iterative method cannot obtain the minimal number of recovery transitions in theory, but it can usually lead to a minimal one supported by extensive experimental studies. Finally, several widely used examples are provided to demonstrate the presented approach. The experimental results show that the reported deadlock recovery technique is effective and efficient.

Petri nets are an important and popular tool to model and analyze deadlocks in automated manufacturing systems. The state space of a Petri net model can be divided into two disjoint parts: a live-zone and a dead-zone. A first-met bad marking (FBM) is a marking in the dead-zone, representing the very first entry from the live-zone to the dead-zone, and the calculation of FBMs to a large extent contributes to the complexity of designing optimal liveness-enforcing supervisors. Most existing studies have to fully enumerate the reachable markings of a Petri net model to obtain the FBMs, which exacerbates the computational overheads. This paper first explores a variation mechanism of calculating FBMs with respect to the resource capacity in a class of S3PR (Systems of Simple Sequential Processes with Resources) from the structural analysis perspective, which contains a ξ-resource. More generally, for the class of S3PR with an η-resource as defined in this paper, the FBMs can be calculated in an algebraic way by a customized structural analysis technique without enumerating all the reachable markings. Finally, the variation mechanism of calculating FBMs is revealed for these considered classes of Petri net models. Examples are given to demonstrate the proposed method.

It is difficult to find the hidden danger or defect of substation equipment accurately and quickly when the substation equipment is inspected manually. In this paper, an electric inspection robot equipped with ultraviolet camera was studied. The dual-light image detection module of visible and ultraviolet camera was built on the robot's head. The effect of air pressure on the results of ultraviolet imaging was studied experimentally. The results show that with the increase of air pressure, the spot area decreases obviously, and the corona discharge weakens. Robot patrol inspection equipped with ultraviolet camera is helpful to realize non-contact live detection, which is of great significance to improve the intelligent detection level of substation.

There are many book recommendation systems, but there are few recommendation systems for Chinese elementary school students. Children's cognitive models are very different from adults, the personalized and accurate recommendation algorithm for them needs to be recommended based on the knowledge structure of children at different ages. In the paper, a new framework model of classified books for Chinese elementary school students was proposed. First of all, a words library and a character library are extracted from six grades textbooks, which named as B_W and B_C respectively; secondly, four machine learning methods are used for prediction, through the indicator of ACC, Recall and F1-Scores, iRF has the best performance in four methods, the max ACC of iRF is 95.3%, which is 10 percentage points better than NB method.

Node.js is one of the most popular frameworks for building web applications. As software systems mature, the cost of running their entire regression test suite can become significant. Selective Regression Testing (SRT) is a technique that executes only a subset of tests the regression test suite can detect software failures more efficiently. Previous SRT studies mainly focused on standard desktop applications. Node.js applications are considered hard to perform test reduction because of Node's asynchronous, event-driven programming model and because JavaScript is a dynamic programming language. In this paper, we present NodeSRT, a Selective Regression Testing framework for Node.js applications. By performing static and dynamic analysis, NodeSRT identifies the relationship between changed methods and tests, then reduces the regression test suite to only tests that are affected by the change to improve the execution time of the regression test suite. To evaluate our selection technique, we applied NodeSRT to two open-source projects: Uppy and Simorgh, then compared our approach with the retest-all strategy and current industry-standard SRT technique: Jest OnlyChange. The results demonstrate that NodeSRT correctly selects affected tests based on changes and is 250% faster, 450% more precise than the Jest OnlyChange.

Neural chat translation aims to translate bilingual conversational text, which has a broad application in international exchanges and cooperation. Despite the impressive performance of sentence-level and context-aware Neural Machine Translation (NMT), there still remain challenges to translate bilingual conversational text due to its inherent characteristics such as role preference, dialogue coherence, and translation consistency. In this paper, we aim to promote the translation quality of conversational text by modeling the above properties. Specifically, we design three latent variational modules to learn the distributions of bilingual conversational characteristics. Through sampling from these learned distributions, the latent variables, tailored for role preference, dialogue coherence, and translation consistency, are incorporated into the NMT model for better translation. We evaluate our approach on the benchmark dataset BConTrasT (English-German) and a self-collected bilingual dialogue corpus, named BMELD (English-Chinese). Extensive experiments show that our approach notably boosts the performance over strong baselines by a large margin and significantly surpasses some state-of-the-art context-aware NMT models in terms of BLEU and TER. Additionally, we make the BMELD dataset publicly available for the research community., Comment: Accepted as a long paper at ACL 2021. Code and data are available at https://github.com/XL2248/CPCC

Zero-resource named entity recognition (NER) severely suffers from data scarcity in a specific domain or language. Most studies on zero-resource NER transfer knowledge from various data by fine-tuning on different auxiliary tasks. However, how to properly select training data and fine-tuning tasks is still an open problem. In this paper, we tackle the problem by transferring knowledge from three aspects, i.e., domain, language and task, and strengthening connections among them. Specifically, we propose four practical guidelines to guide knowledge transfer and task fine-tuning. Based on these guidelines, we design a target-oriented fine-tuning (TOF) framework to exploit various data from three aspects in a unified training manner. Experimental results on six benchmarks show that our method yields consistent improvements over baselines in both cross-domain and cross-lingual scenarios. Particularly, we achieve new state-of-the-art performance on five benchmarks., Comment: 9 pages, ACL21 Findings

The predominant approach of visual question answering (VQA) relies on encoding the image and question with a “black box" neural encoder and decoding a single token into answers such as “yes” or “no”. Despite this approach’s strong quantitative results, it struggles to come up with human-readable forms of justification for the prediction process. To address this insufficiency, we propose LRRA [Look, Read, Reasoning,Answer], a transparent neural-symbolic framework for visual question answering that solves the complicated problem in the real world step-by-step like humans and provides human-readable form of justification at each step. Specifically, LRRA learns to first convert an image into a scene graph and parse a question into multiple reasoning instructions. It then executes the reasoning instructions one at a time by traversing the scene graph using a recurrent neural-symbolic execution module. Finally, it generates answers to the given questions and makes corresponding marks on the image. Furthermore, we believe that the relations between objects in the question is of great significance for obtaining the correct answer, so we create a perturbed GQA test set by removing linguistic cues (attributes and relations) in the questions to analyze which part of the question contributes more to the answer. Our experiments on the GQA dataset show that LRRA is significantly better than the existing representative model (57.12% vs. 56.39%). Our experiments on the perturbed GQA test set show that the relations between objects is more important for answering complicated questions than the attributes of objects.

Entity Linking (EL) refers to the task of linking entity mentions in the text to the correct entities in the Knowledge Base (KB) in which entity embeddings play a vital and challenging role because of the subtle differences between entities. However, existing pre-trained entity embeddings only learn the underlying semantic information in texts, yet the fine-grained entity type information is ignored, which causes the type of the linked entity is incompatible with the mention context. In order to solve this problem, we propose to encode fine-grained type information into entity embeddings. We firstly pre-train word vectors to inject type information by embedding words and fine-grained entity types into the same vector space. Then we retrain entity embeddings with word vectors containing fine-grained type information. By applying our entity embeddings to two existing EL models, our method respectively achieves 0.82\(\%\) and 0.42\(\%\) improvement on average F1 score of the test sets. Meanwhile, our method is model-irrelevant, which means it can help other EL models.

This paper presents the systems developed by Beijing Jiaotong University and Toshiba (China) Co., Ltd. for the CCMT 2021 quality estimation (QE) and automatic-post editing (APE) task. For QE task, we mainly rely on multiple pretrained language models, and propose a multi-phase pre-finetuning scheme, to adapt the pretrained models to the target domain and task. The pre-finetuning scheme consists of language-adaptative finetuning, domain-adaptative finetuning and task-adaptative finetuning. For APE task, we use BERT-initialized Transformer as the backbone model, and create different groups of synthetic data by different data augmentation methods, i.e. forward translation, round-trip translation and multi-source denoising autoencoder. Multi-model ensemble is adopted in both tasks. Experiment results on the development set show high accuracy on both QE and APE tasks, demonstrating the effectiveness of our proposed methods.

Syntactically controlled paraphrase generation can produce diverse paraphrases by exposing syntactic control, where both semantic preservation and syntactic variations are two important factors. Previous works mainly focus on using fine-grained syntactic structures (e.g., full parse tree) as syntactic control. While these methods can achieve excellent syntactic controllability, leads to failing to preserve the semantics of the input sentence. The main reason is that it is difficult to retrieve perfectly compatible syntactic structures with the input sentences. In this paper, we explore coarse-grained syntactic structures to trade-off semantic preservation and syntactic variations. Furthermore, to improve semantic preservation and syntactic controllability, we propose a Syntax Attention-Guided Paraphrase (SAGP) model that can correctly select syntactic information according to the current state for surface realization. Experiment results show that SAGP outperforms the previous state-of-the-art method under the same setting. Additionally, we validate that using coarse-grained structures can generate more semantically reasonable text without affecting the syntactic controllability.

Deep neural networks have achieved state-of-the-art performances on named entity recognition (NER) with sufficient training data, while they perform poorly in low-resource scenarios due to data scarcity. To solve this problem, we propose a novel data augmentation method based on pre-trained language model (PLM) and curriculum learning strategy. Concretely, we use the PLM to generate diverse training instances through predicting different masked words and design a task-specific curriculum learning strategy to alleviate the influence of noises. We evaluate the effectiveness of our approach on three datasets: CoNLL-2003, OntoNotes5.0, and MaScip, of which the first two are simulated low-resource scenarios, and the last one is a real low-resource dataset in material science domain. Experimental results show that our method consistently outperform the baseline model. Specifically, our method achieves an absolute improvement of 3.46% \(F_1\) score on the 1% CoNLL-2003, 2.58% on the 1% OntoNotes5.0, and 0.99% on the full of MaScip.

Machine translation quality estimation (QE) aims to evaluate the result of translation without reference. Existing approaches require large amounts of training data or model-related features, leading to impractical applications in real world. In this work, we propose a contrastive learning framework to train QE model with limited parallel data. Concretely, we use denoising autoencoder to create negative samples based on sentence reconstruction. Then the QE model is trained to distinguish the golden pair from the negative samples in a contrastive manner. To this end, we propose two contrastive learning architectures, namely Contrastive Classification and Contrastive Ranking. Experiments on four language pairs of MLQE dataset show that our method achieves strong results in both zero-shot and supervised settings. To the best of our knowledge, this is the first trial of contrastive learning on QE.

This paper presents the systems developed by Beijing Jiaotong University for the CCMT2021 evaluation tasks. We joined four translation tasks of Chinese-English, English-Chinese, Uyghur-Chinese, Tibetan-Chinese. In all directions, we build our system based on transformer architecture and Dynamic-Conv. Additionally, we apply Byte Pair Encoding (BPE) to all translation tasks to resolve the out-of-vocabulary (OOV) problem. We also adopt some techniques that have been proven effective recently in academia, such as data augmentation, finetuning, model ensemble and reranking. Experiments show that our machine translation systems achieved high accuracy on all directions.

Insulator is one of the most widely used components in the power grid. Identifing and locating insulator accurately and efficiently is the basis of intelligent evaluation of insulator defects based on UV imaging. The traditional insulator identification and location is mostly done by the back-end server, which has poor real-time performance. Based on this, this paper proposes an insulator intelligent recognition method based on machine vision and edge calculation, builds an edge calculation platform and loads the labeled insulator data set into SSD model for training. The accuracy of recognition is 91.46% and the model file is transplanted to the edge device. At the same time, the edge equipment is equipped with a visual sensor, which can identify the real-time video stream of the insulator. It has a good real-time performance and provides a new scheme for the fault detection of the external insulation equipment.

Insulator is one of the most widely used electrical equipment in power system, and its insulation state is related to the safe and stable operation of power system. In order to realize the intelligent diagnosis of the insulation state of the field running insulators, an intelligent diagnosis framework based on cloud computing and deep learning algorithm is designed; the browser / server development mode manages the UV images and videos of insulator discharge, and develops a set of image and video management data According to the database system, the database management system of UV imaging samples is constructed, which integrates the function of insulator visible light recognition and insulator discharge severity assessment based on UV image. The developed software is applied to the inspection of transmission lines and substations, and the application result is good.

We designed a virtual reality system which can assist training of vascular interventional surgery. The hardware including electromagnetic navigation, performs real-time guide wire position tracking and display. Through k-dimensional tree data structure we store and calculate the distance between guide wire tip and vascular centerline for security warning, avoiding guide wire tip penetration through vascular wall. The neural network model aims to predict the next operation of guide wire status in real time. The operation information will be displayed by virtual reality devices.

We propose a system for automatic household design, which is based on Generative Adversarial Networks (GAN) to construct a multi-layer network structure system. Based on the given home building structure information and in accordance with the real rules, specific furniture objects can be reasonably placed in the corresponding home building space. First, we constructed the household design dataset, which includes a building structure diagram with constraint information as input, and a real home layout based on the structure diagram design. including part of the design and the complete design. The multi-layer GAN structure can generate the household design results with structural relations according to the design steps. The system introduces the attention network, which can generate the structural details in the image by focusing on the constraint conditions of the input image. Conditional regression is designed at the back of the system, so that the generated results can have diverse characteristics. Our system can quickly generate a diversified real home layout, which has good practical value.

Providing an intelligent campus guiding application is a good example of smart campus. Our smartphone APP focused on monitor the drivers' illegal behaviors, such as illegal parking for safety and management issues. For illegal parking detection, the motion mode recognition algorithm is focused on to detect the walking mode after drivers park their cars. When distinguishing low-speed driving mode and walking mode, the accuracy rate of Decision Tree algorithm is 67.03%, and that of our algorithm is 89.91%. When the walking mode is detected, the car location will be computed and if the location is not within the parking lot, the illegal parking occurs and the illegal parking alert will be sent to the APP. The application will play an important role in smart campus to improve the traffic safety and traffic management in campus.

Motors have a wide range of applications in various aspects such as automotive, medical, industrial production, etc. Commonly used motors can generally be divided into DC motors and AC motors.Permanent Magnet Synchronous Motor (PMSM) is a type of AC motors with strong starting ability, high peak efficiency and high reliability, and with greater application value. This paper mainly studies the control technology of PMSM, and designs a set of motor controller and its protection system based on FPGA. First, based on the development and simulation platform of FPGA, the important algorithms of the motor controller and protection system are simulated and analysed. Then the system is built in the actual environment. Through actual testing, the speed control of the motor can be accurately achieved and various fault protection and instructions of the motor controller can be realized. The system is based on the development platform of FPGA, which with fast running speed, high flexibility, short development cycle, high resource utilization rate, and strong portability.

V2X (Vehicle-to-X) network is an important support to automated vehicle. However, for automated vehicle safety, if the security of V2X network cannot be guaranteed, automated vehicles may not obtain required correct information, which is a fatal problem. In V2X network, black hole and gray hole attacks are critical threats of network availability and may prevent legitimate users from receiving packets. In this review, the definition, classification and impact of the attacks on V2X network are presented, and the detection and prevention technologies are classified and analyzed, respectively. Finally, three open issues and future trends are proposed, which are considering specific characters of V2X network, paying attention to artificial intelligent technologies in detection and utilizing edge computing architecture as the local trusted environment in detection and prevention.