The maintenance of road infrastructure is a classic social challenge, especially in the context of a decreasing maintenance budget and the advent of autonomous vehicle traffic. Road markings need an accurate replacement strategy to guarantee that the markings remain perceptible. The retroreflective luminance of markings is currently dynamically quantifiable only by using a retroreflectometer such as the Ecodyn from MLPC. The main objective of this research is to construct a performance-based approach for retroreflective marking replacement adapted to a given road network. This approach involves three main tasks: localize the strategic area based on past inspections, determine an adapted decay model for a given area, and evaluate the economic impact of replacing markings. This paper focuses on the first task. We apply the Agglomerative Hierarchical Clustering (AHC) method to a given dataset to obtain a suitable markings line segmentation. Markings whose retroreflective luminance exhibits similar evolution over time are interpreted to belong to a specific area of the road network. When no follow-up replacement has occurred, a replacement detector deduces the date at which markings were laid from the clusters. The broken center line of the French National Road 4 illustrates the proposed approach; the road is divided into 5 clusters and 34 lifecycles. A study of markings laid in 2008 and replaced in 2012 shows important variations in the decay of the retroreflective luminance as identified by the clustering approach. Even for a single road, an optimal replacement strategy for retroreflective road markings is necessary and is composed of several local maintenance strategies.

In the different structure of coordinated circuits, microwave transfer stations in broadcast communications, oil pipe frameworks, vacuum frameworks in quickening agents, PC ring systems, and hand-off stations in space, the consecutive k-out-of-n: F system is useful. We consider n segments that are arranged in a model of successive k-out-of-n: F systems. Without lifetime information about the whole structure, it is attractive to utilize prior belief or experience on its segments in the Bayesian examination of the system. The exact Bayesian reliability formulas, mean time to system failure, and confidence interval are obtained for the proposed system.

Manufacturing industries are facing challenging situations to produce efficient products for unexpected market responsiveness. Environmental policies and legislation pressures without any compromise in the quality of the product are other challenges linked to building a market-responsive product at a lower cost. Earlier manufacturers were only concerned with fast delivery of products, but now, because of new policies, they are made to follow up work, repair/maintenance, and disposal. Hence, to meet the above demands optimally under established machinery and setup, researchers are developing many design strategies such as DFA, DFMA, DFE, DFD, DFR, and DFQ. The overwhelming problem intensifies the efforts in design and assembly variations, including selecting the material, machine, tool, method, market management, and more. In this article, an attempt is made to solve the problems that are encountered at the product development stage to meet the guidelines of the existing strategies. The obtained results are examined and discussed with suitable illustrations.

Recently, friction stir processing (FSP) has emerged as a pioneering approach for the manufacture of composites with enhanced mechanical and tribological properties. The present study examines the impact of process parameters such as tool rotation speed and FSP pass number on the AZ61A/TiC magnesium metal composite for responses such as hardness and wear resistance. To minimize number of experiments, the design of experiments (DOE) was configured according to the response surface methodology (RSM) using central composite design (CCD). Analysis of variance (ANOVA) was conducted to develop a mathematical and empirical model for studying the relationship between tool rotation and number of passes for responses such as microhardness and wear resistance. Microhardness was checked on the Vickers microhardness testing machine, and tribological behavior was examined on the pin-on-disc tribotester. Wear tracks were analyzed via scanning electron microscopy (SEM). The responses were predicted using validated mathematical model, and contour plots were generated to study the interaction and influence of process parameters. Finally, the findings suggested that both selected parameters are significant and largely influence the responses.

Network function virtualization can provide the required functionality at run-time on demand by its virtualization technologies. Network functions are implemented in software using shared hardware, such as X86-based servers, and service function chaining allows a flow forwarding along a chain of network functions. Such architecture can improve the network flexibility and reduce the energy consumption, which can be achieved without impacting the acceptance ratio of requests. However, reducing energy consumption means activating as few nodes as possible, which leads to the unbalanced load and the low acceptance ratio. Therefore, to achieve a trade-off between the energy consumption and acceptance ratio, in this paper, we propose a new scheme called the node evaluation method. It is based on multiple types of status characteristics, which can evaluate each candidate node for virtual network function placement with multiple types of node status characteristics as input. Finally, the simulations results demonstrate that the proposed method can obtain a better trade-off between energy and acceptance ratio than the existing approaches.

Deep learning is a branch of machine learning. It uses neural networks as a bridge to represent a large number of data. It is also one of the research directions of artificial intelligence. At present, it is widely used in computer vision, speech recognition, audio recognition, fault diagnosis, and other fields, and it has achieved good results. In view of deep learning in modern complex mechanical equipment, fault diagnosis, and health, health management plays an important role. In this paper, the convolution neural network method for equipment fault diagnosis is summarized based on the structural characteristics of modern large-scale mechanical equipment and the advantages of deep learning. With the help of hardware in the loop simulation simulator, starting from the process of fault mechanism modeling, fault simulation, data processing, and so on and compared with other neural networks in deep learning, the experiment shows that the method has high accuracy, which is of great significance for improving the efficiency of equipment fault diagnosis.

To address the dynamic decision making problem, in this paper, we propose a probabilistic decision method based on bimodal implicit information quantum model. By means of constructing the quantum dynamic decision making system, we analyze the superposition nature of states and describe the time-varying evolution and state projection response. Secondly, on account of the bounded rationality hypothesis, the negatively related Hamilton operator is designed. After introducing subjective implicit information of decision-makers as the characteristic index of projection operator, we revise the state response matrixes. Then, the implicit information quantum weighted model is established for a bimodal decision system. Finally, according to synthesizing time-varying evolution and wave packet projection collapsing, an example analysis is given to illustrate the effectiveness of the proposed approach in categorization decision making application.

With the rapid development of intelligent driving and self-driving, how to quickly identify traffic signs in traffic scenes image is an urgent problem that needs to be solved. The existing object detection method can be divided into two categories: the one-staged method, which has a fast detection speed, and the two-stage method, which has higher detection accuracy. How to quickly and accurately detect targets in traffic scenes images is a current research focus. In this paper, an effective detection operator for the region of interest of traffic signs that utilizes the color, shape, and layout characteristics of traffic signs was proposed. It can accurately extract the region of interest in the traffic scene image for detection stage. The existing two-stage network was also fine-tuned to improve the accuracy of traffic sign detection. On the basis of the existing public data set, 13,000 images were collected and annotated to expand the training and test data. These data were used to verify the method proposed in this article. Experiments demonstrated that the proposed method has been improved in detection speed and detection accuracy.

To construct a representative driving cycle for light-duty vehicle, a new method has been proposed based on the PCA and t-SNE algorithm and K-means clustering algorithm in this paper. This method not only greatly reduces the computational pressure, but also retains more useful nonlinear characteristic information. It shows that the speed-acceleration frequency distribution of the driving cycle is basically consistent with the sample data with a relative error within ±4%, and the average relative error of the overall characteristic parameters is about 3.21%. These indicate the effectiveness and accuracy of our new method.

Traffic sign recognition systems are crucial for autonomous vehicles. They assist autonomous driving systems by collecting road-related information, such as speed limits, stop signs, etc., that are necessary for safe driving. However, as evidenced by recent autonomous vehicle crashes and recognition system failure-related studies, there are serious concerns about the inadequacies of the traffic sign recognition systems and their used techniques. In response to the industrial needs and to help practitioners improve the reliability and safety of the traffic sign recognition systems, this paper discusses the general architectural outline of traffic sign recognition systems and the challenges that must be overcome, in order for traffic sign recognition systems to be safe and reliable. An in-depth discussion of various solutions is given to provide practitioners valuable insight into the improvement of traffic sign recognition systems.

Network traffic monitoring and analysis are useful and crucial because they allow networks to be better understood, more efficiently used, and more accurately evaluated. Heterogeneous networks are widely used because they enable public networks with improved coverage, capacity, and communication. This paper proposes a system design using both sFlow and NetStream technologies for traffic performance analysis in heterogeneous networks. This sFlow+NetStream design should enable the system to obtain network traffic data more comprehensively and accurately (compared with a system that uses only one of the two technologies) and to largely reduce the pressure on the monitoring and analysis system (compared with a system that uses traditional port mirroring technology). To verify the design objects, we built a test environment with an H3C router and an eNSP simulator and conducted experimental tests in it. The test results proved the aforementioned system design to be valid and feasible.

The collaborative creation model generates tremendous application prospects. In order to improve efficiency and reliability of collaborative creation, we propose a blockchain mapping method for the collaborative creation application system. We use blockchain to store tree structural content of collaborative creation. This makes it easy for creations to trace sources and assign rewards to individual contributors in an impartial and trustworthy manner.

Reliability test and finite element analysis have been conducted to study the reliability of PBGA (Plastic Ball Grid Array Package) under random vibration load. Firstly, random vibration tests with various levels of excitation were conducted on the PBGA PCB (Printed Circuit Board) samples; the resistances of the components are monitored on real time, and a two-parameter Weibull distribution model was used to analyze the test results. Thus, the failure time was acquired. Secondly, finite element analysis was conducted on the 3D model of the test samples in order to ensure the model accuracy. Frequency and displacement were measured to compare the analysis results, and the vibration fatigue life was calculated based on Steinberg model. Lastly, by comparing the test results and the analysis results, Steinberg model is modified to ensure the prediction error of vibration fatigue model of PBGA package is less than 70%, which proves the random vibration fatigue model has good precision in engineering.

In the cloud environment, the resource utilization and overall efficiency of traditional task allocation and scheduling algorithms are very low. In order to improve the overall efficiency of Map/Reduce processing task allocation and improve the resource utilization efficiency of nodes, a pruning algorithm based on task processing time is proposed. In the pruning algorithm, the processing time of each task in each node is used for quantitative modeling, the task and processing time measurement matrix is established, and the rank pruning method is used to reduce the size of the assigned tasks. Only by assigning (N-1) tasks can the allocation problem of N nodes processing N tasks be solved, and the optimal solution of the task can be obtained. This article uses MATLAB to simulate the cloud environment and compares the pruning algorithm with traditional algorithms (including FIFO scheduling and capacity scheduling algorithms). Simulation results show that the pruning algorithm can significantly improve the overall efficiency of task scheduling under a large amount of data testing, and make full use of the computing power of nodes to improve the efficiency of Map/Reduce scheduling.

It is difficult to implement rapid and current fault diagnosis for the lithium battery because of its strong coupling and uncertainty. Moreover, the problem of getting lithium battery fault data will cause a lack of training samples, thus resulting in the unsatisfactory performance of the diagnosis model. Furthermore, once the parameters of the normal fault diagnosis model are confirmed at first, they cannot be readjusted. This will lead to poor generation ability of the model, which is not suitable for the complicated and various lithium battery working conditions. To resolve these above problems, a diagnosis method that is combined with Least Square Support Vector Machine (LSSVM) and optimized by Ant Lion Algorithm (ALO) and Online learning based on sample centre distance, is presented. ALO simulates the process of ant lion capturing ants to implement parameter optimization, which can improve model performance. Online learning based on sample centre distance will remove the samples that have little influence on model training. In this way, ALO will avoid local optimal solutions, and the parameters of the model can be updated according to the new data, thus improving the adaptability of the model to actual working conditions. To verify the possibility and effectiveness of the proposed method, experiments of lithium battery under various fault states are taken. Fault feature data are achieved from these experiments and are used for training, testing and comparing with diagnosis models. The comparative results reveal that the model presented above can guarantee the speed of model parameter updating, showing better generation ability. It's more suitable to implement rapid and current fault diagnosis for the lithium battery.

The ADS1299 chip of the OpenBCI development board is used as the core design system to collect non-invasive EEG signals and transmit the data to the upper computer via the WiFi module for display to realize portable monitoring of EEG. It can perform signal preprocessing and other operations on the upper computer as required. It uses the graphical editing language LabVIEW to develop according to the "producer-consumer" model. The producer part is mainly responsible for data analysis and reception, and it judges whether the data is lost. The consumer part is mainly responsible for data preprocessing, waveform display, recording, etc. The designed monitoring device is mainly realized by the EEG cap, OpenBCI development board, WiFi module, battery and other components. Using this monitoring device, it is possible to further study brain-controlled electronic entertainment products. Through this interactive behavior, you can achieve more precise control of your brain and enhance the immersive experience.

In order to reduce the calculation burden of the signal number estimation and improve the accuracy against the background of small snapshots, an idea for counting signal numbers based on a support vector machine is provided. First, the features of the signal and noise are extracted by the orthogonality between the noise vector and the array manifold. Then, a classifier based on a support vector machine is designed. Finally, the optimal structure of the classifier and the corresponding coefficients are trained by theoretical analysis and training data. The proposed algorithm performs well in both Gaussian white noise and colored noise. The validity and feasibility of the proposed theory are verified by the simulations.

Different from the traditional privacy protection of relational data, in this paper we focus on the protection of graph data in social networks. The problem of identity disclosure on graph data publication in social networks has caused an increase in attention and many existing methods of protecting graph data are based on the properties of the vertices. These methods can resist invasions such as degree attacks or neighbor attacks. Once an adversary knows the structural information of the graph, the probability of a vertex being recognized will be greatly increased. Considering the community structure of the graph, we propose a k-community anonymity model, in which the probability of an adversary identifying a vertex is no more than 1/k. We conduct our experiments in real social network datasets and compare it with the traditional k-degree anonymity model. The results show that the new protection scheme has better anonymous performance on resisting the structural attacks and a greater impact on the community structure in the graph.

It is an important issue in the management of scenic passengers to regulate passenger flow during peak hours, especially in emergency situations. Based on passenger flow, routing environment and control measures of a given scenic spot, we built a multi-agent model of the scenic spot. It is a simulation system that mainly consists of two multi-agent methods, i.e., the cellular automata model and ripple-spreading algorithm, and reflects the emergency evacuation of passenger flow in the scenic spot under emergent events. This paper uses a famous Chinese scenic spot, the Summer Palace in Beijing as a research case, and conducts an empirical research. The result shows that the reported simulation system can help to effectively assess the evacuation effect of purpose-designed regulation measures according to passenger distribution and road network characteristics. Thus, this simulation system can provide effective decision-making support to scenic spot managers.