Values of Intelligent Alarm System Under Photoelectric Sensor Networks

Abstract

This work was aimed to construct the intelligent alarm system with multiple photoelectric sensors as the core in this study. The system is first designed the circuit with microprocessor as the core, and then, there was a principle analysis of photoelectric measurement in the height, speed, and temperature to design a network mode of photoelectric sensor, circuits to control security doors and manage password, substation, and the monitoring center. The fusion approach based on deep learning is designed for the data collected by security alarm system. The 1-dimensional (1-D) representation of 2-dimensional (2-D) data is also designed according to the most of key information represented by the eigenvalue set of singular value decomposition of data matrix. The original 1-D signal sequence and the characteristics after 1-D were for data fusion, which is applied to identify, thus improving the accuracy of the alarm system and reducing its labor cost. During the experiment, the data fusion method proposed in this study is compared with naive bayes (NB) method and the weighted majority voting (WMV) method. The random data sets are generated with the help of a Gaussian function. The extreme learning machine (ELM) neural network classifier and k-nearest neighbors (KNN) classifier are carried in the alarm system designed in this study, respectively. The simulation analysis shows that WMV can obtain better performance of information classification compared with NB and data fusion methods, so the accuracy of classification is improved obviously. Besides, the fusion results accuracy of WMV is greatly higher than the other two.