A Support Vector Machine Algorithm for PIR Special Processor

• Published in: 2020 IEEE International Conference on Computational Electromagnetics (ICCEM) pp. 279 - 280
• Main Authors: Yu, Cao Ling, Gang Han, Zhi, Xiao, Wei Hua, Song Tong, Mei
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.08.2020
• Subjects: Classification algorithms; data acquisition; fast Fourier transform; Intellectual property; mathematical statistics; Microcontrollers; Motion detection; PIR sensor; Real-time systems; Support vector machines; SVM classification; Timing
• DOI: 10.1109/ICCEM47450.2020.9219356

With the continuous improvement of people's safety awareness, infrared products as human motion detection technology have been widely used in the field of security. In order to better apply infrared products to life, improving the performance of infrared products and reducing the cost of products has become the main goal. According to the signal collected by Pyroelectric infrared (PIR) sensor, this paper establishes a database model. According to the data collected, Kalman filter is used to preprocess the data. The validity of the data after preprocessing is judged by the algorithm. The experimental results show that the accuracy of the model can reach 97% by using a support vector machine (SVM) algorithm incorporated with Fast Fourier Transform (FFT). According to the above algorithm flow, a real-time intellectual property (IP) core is designed by using hardware description language, after establishing the data processing algorithm. The interface design, timing design and function design of the IP core are designed. The IP core can be connected to the microcontroller unit (MCU) as an independent peripheral to form a PIR special processor, which can detect the distance of 15 m in real time.