Search

With the obvious improvement in people’s quality of life, the use of natural gas is becoming increasingly common. Household natural gas has hidden safety hazards while being hygienic and convenient. There are many gas explosion accidents every year, and life and property are seriously damaged. When the natural gas body leaks a little and the methane concentration in the air is extremely low, the methane sensor may fail to report, and the signal is weak or the early warning signal cannot be processed in time, which brings hidden dangers to the accident of gas explosion. In this paper, an optimized array of antenna arrays is designed based on the ZigBee network. Use multiple MQ-4 sensor arrays to detect the methane concentration in the kitchen, and send the weak signals collected by the four methane sensors to the ZigBee router in the same direction through the four antennas of the ZigBee module. The router timely and accurately transmits to the ZigBee coordinator and uses ZigBee network technology to upload the alarm information to the cell monitoring center through wireless. The simulation test results show that the diamond antenna array can quickly detect the change of methane concentration, and has high sensitivity and reliability.

Named entity recognition (NER) is a common task in Natural Language Processing (NLP). To this end, we propose a novel approach based on pre-training model to complete the sequence labeling tasks by learning the large-scale real-world data from Brazilian legal documents. Especially, combining iterated dilated convolution[1] (IDCNN) and Bi-LSTM, we develop the scalable sequence labeling model named Sequence Tagging Model (STM) and extensive experiments validate the effectiveness of STM for NER tasks. Furthermore, compared with the IDCNN-CRF model, the experimental results show that the STM is better and the F1 score is 93.23%, which provides an important basis for NER tasks.

Driving behavior is one of the most important reasons that affect traffic safety and energy saving. In view of the management problems existing in traffic safety, it is necessary to strengthen the driver’s safety and energy conservation awareness. To this end, based on the driving status information collected by the road transportation industry in the vehicle networking system, we propose a data-driven approach to dig deeper into driving behaviors that affect driving safety and energy efficiency. Among them, we used the quantitative description of idle speed preheating, overspeed, fatigue driving and other series of indicators, and combined with the AHP (Analytic Hierarchy Process) to construct a comprehensive driving safety evaluation index system. Assisting the Ministry of Transportation in managing roads more efficiently and creating a harmonious driving environment via our evaluation model.

Social change has a direct influence on the family loss. To this end, we propose a novel approach to model the probability of family loss by learning the large-scale real-world survey data from China Family Panel Studies (CFPS). Especially, combining individual economic, health, identity, family profile and other information, we develop the scalable tree boosting model named Family Loss Prediction Model (FLPM) to model the family loss. Extensive experiments on real-world survey data validate the effectiveness of FLPM for measuring the probability of family loss in China. Furthermore, compared with the XGBoost and GBDT models, the experimental results show that the FLPM model is better and the precision is 98.48%, which provides an important basis for this study.

This paper presents an integrated delta-sigma digital current sensor. With high sensitivity and good linearity, GMR sensor is a good choice for current measurement application. In this design, a GMR sensor which is configured as a Wheatstone bridge of four spin-valve sensing elements is integrated with CMOS signal processing circuit together, and the output signal is converted to digital domain by an on-chip 10-bit delta-sigma ADC. The whole circuit is designed and simulated using CSMC standard 0.5 μm process.