Chang Gao
Chang Gao has a total of 14 co-authors. Here are the top co-authors by number of shared publications:
- Weiping Liu
SICHUAN SHUDAO NEW SYSTEM RAIL GROUP CO., LTD., China
4 shared publications - Fanqiang Lin
College of Mechanical and Electrical Engineering, Chengdu University of Technology, Chengdu 610059, China
3 shared publications - Wugang Lai
College of Mechanical and Electrical Engineering, Chengdu University of Technology, Chengdu 610059, China
3 shared publications - Hongyu Sun
Chengdu University of Technology, China
2 shared publications - Xvheng Zhao
Chengdu University of Technology, China
2 shared publications - Guoqian Yang
Chengdu University of Technology, China
1 shared publication
Latest Publications
- Open Access
- 28 November 2022
- Published by Association for Computing Machinery (ACM)
- p. 255-259
Design of Intelligent Classroom Energy Management System
In recent years, with the rapid development of society and technology, the size of teaching facilities in colleges and universities has continued to expand, and most of the energy management in teaching areas is still controlled by people, leading to problems such as low management efficiency and serious energy waste. To address these types of problems, an intelligent classroom management system is proposed in this paper, which is based on the efficient implementation of low-complexity convolutional neural networks in STM32H7 series microprocessors and the coordination of each sensing module. The system has the option to automatically detect changes in the classroom environment to control the switching of various energy sources, or to remotely control the switching of the energy system via Wi-Fi from a mobile phone terminal. After testing, the system can efficiently and automatically manage the classroom energy system.
- 28 November 2022
- Published by Association for Computing Machinery (ACM)
- p. 255-259
Open AccessDesign of Intelligent Classroom Energy Management System
In recent years, with the rapid development of society and technology, the size of teaching facilities in colleges and universities has continued to expand, and most of the energy management in teaching areas is still controlled by people, leading to problems such as low management efficiency and serious energy waste. To address these types of problems, an intelligent classroom management system is proposed in this paper, which is based on the efficient implementation of low-complexity convolutional neural networks in STM32H7 series microprocessors and the coordination of each sensing module. The system has the option to automatically detect changes in the classroom environment to control the switching of various energy sources, or to remotely control the switching of the energy system via Wi-Fi from a mobile phone terminal. After testing, the system can efficiently and automatically manage the classroom energy system.
- Open Access
- 28 November 2022
- Published by Association for Computing Machinery (ACM)
- p. 244-249
Design and Implementation of a Human Flow Visual Analysis System Based on RT-Thread System
The most effective measure to control and prevent COVID-19 is to maintain a safe social distance. The Chinese government has previously adopted a physical social distance of 1m as a standard for safe social distance in public places such as campuses and factories and has achieved good results. Social distance has also been widely recognized as an implementation program by the World Health Organization to minimize the spread of COVID-19 in public places. Therefore, for the social safety distance risk of pedestrians during the COVID-19 epidemic, this study proposes an RT-Thread-based social safety distance risk assessment under the COVID-19 epidemic. Firstly, the pedestrian is trained with fine-tuned YOLO-Fastest for target extraction, and the pedestrian key points are extracted to treat the continuous pedestrian motion as a continuous motion of prime points to achieve tracking processing of pedestrians; secondly, the trained weight files are deployed to the RT-Thread OS engineering project through STM32CubeMX AI and RT-AK toolkit to identify the face position and detect whether a mask is worn or not. The recognition rate is 30.71% for unmasked faces and 78.84% for masked faces on the ART-Pi with a maximum main frequency of 480Mhz. It has good performance in human flow calculation and spacing calculation by pixel position as a difference. This can provide ideas for handling safe social distancing of pedestrians in places such as campuses for epidemic prevention and control.
- 28 November 2022
- Published by Association for Computing Machinery (ACM)
- p. 244-249
Open AccessDesign and Implementation of a Human Flow Visual Analysis System Based on RT-Thread System
The most effective measure to control and prevent COVID-19 is to maintain a safe social distance. The Chinese government has previously adopted a physical social distance of 1m as a standard for safe social distance in public places such as campuses and factories and has achieved good results. Social distance has also been widely recognized as an implementation program by the World Health Organization to minimize the spread of COVID-19 in public places. Therefore, for the social safety distance risk of pedestrians during the COVID-19 epidemic, this study proposes an RT-Thread-based social safety distance risk assessment under the COVID-19 epidemic. Firstly, the pedestrian is trained with fine-tuned YOLO-Fastest for target extraction, and the pedestrian key points are extracted to treat the continuous pedestrian motion as a continuous motion of prime points to achieve tracking processing of pedestrians; secondly, the trained weight files are deployed to the RT-Thread OS engineering project through STM32CubeMX AI and RT-AK toolkit to identify the face position and detect whether a mask is worn or not. The recognition rate is 30.71% for unmasked faces and 78.84% for masked faces on the ART-Pi with a maximum main frequency of 480Mhz. It has good performance in human flow calculation and spacing calculation by pixel position as a difference. This can provide ideas for handling safe social distancing of pedestrians in places such as campuses for epidemic prevention and control.
- Open Access
- 28 November 2022
- Published by Association for Computing Machinery (ACM)
- p. 175-180
Pipeline Leak Audio Detection System based on Machine Learning
It is of great practical significance to study intelligent robots for crack detection of urban underground pipes instead of manual detection, which not only can save time and cost, but also can improve the accuracy rate of identification. The system consists of two parts: mobile platform and system detection platform. The mobile platform is obstacle-avoiding by ultrasonic module, segmenting the pipeline color by camera and issuing platform steering command. The system detection platform adopts Raspberry Pi as the core control system, and collects the pipeline leakage audio through the pickups, and the collected audio is passed through Gaussian mixture model-Hidden Markov model (GMM -HMM) and Deep neural networks- Hidden Markov model (DNN -HMM) respectively in Raspberry Pi model (DNN -HMM) for the purpose of detection. The accuracy and running speed of the two methods were tested separately through simulation experiments. GMM -HMM achieves 74.1% accuracy and 14 (frame/s) running speed, while DNN -HMM achieves 80.6% accuracy and 9 (frame/s) running speed, DNN -HMM has higher accuracy and running speed to meet the real-time requirements for pipeline leakage audio detection.
- 28 November 2022
- Published by Association for Computing Machinery (ACM)
- p. 175-180
Open AccessPipeline Leak Audio Detection System based on Machine Learning
It is of great practical significance to study intelligent robots for crack detection of urban underground pipes instead of manual detection, which not only can save time and cost, but also can improve the accuracy rate of identification. The system consists of two parts: mobile platform and system detection platform. The mobile platform is obstacle-avoiding by ultrasonic module, segmenting the pipeline color by camera and issuing platform steering command. The system detection platform adopts Raspberry Pi as the core control system, and collects the pipeline leakage audio through the pickups, and the collected audio is passed through Gaussian mixture model-Hidden Markov model (GMM -HMM) and Deep neural networks- Hidden Markov model (DNN -HMM) respectively in Raspberry Pi model (DNN -HMM) for the purpose of detection. The accuracy and running speed of the two methods were tested separately through simulation experiments. GMM -HMM achieves 74.1% accuracy and 14 (frame/s) running speed, while DNN -HMM achieves 80.6% accuracy and 9 (frame/s) running speed, DNN -HMM has higher accuracy and running speed to meet the real-time requirements for pipeline leakage audio detection.
- Open Access
- 28 November 2022
- Published by Association for Computing Machinery (ACM)
- p. 106-110
Emitter identification based on multi-class spectrograms in time-frequency domain
In the identification of emitter, there are some problems such as poor identification effect and incomplete feature expression of emitter by using single signal processing method to obtain spectrogram with feature of emitter. In order to solve these problems, a method of emitter identification based on multi-class spectrograms in time-frequency domain is proposed. Specifically, three time-frequency analyses, including short-time Fourier transform (STFT), smoothed pseudo Wigner-Ville distribution (SPWVD) and reduced interference distribution with Hanning kernel (RIDHK), are performed on signal data to obtain the time-frequency spectrograms with different feature expressions. Three convolutional neural networks (CNN) were used to extract the independent features from three classes of spectrograms, and the three features were stacked to obtain more complete features of emitter, to improve the recognition effect. Experimental results show that compared with the single feature recognition method, the proposed method improves the accuracy of emitter identification effectively.
- 28 November 2022
- Published by Association for Computing Machinery (ACM)
- p. 106-110
Open AccessEmitter identification based on multi-class spectrograms in time-frequency domain
In the identification of emitter, there are some problems such as poor identification effect and incomplete feature expression of emitter by using single signal processing method to obtain spectrogram with feature of emitter. In order to solve these problems, a method of emitter identification based on multi-class spectrograms in time-frequency domain is proposed. Specifically, three time-frequency analyses, including short-time Fourier transform (STFT), smoothed pseudo Wigner-Ville distribution (SPWVD) and reduced interference distribution with Hanning kernel (RIDHK), are performed on signal data to obtain the time-frequency spectrograms with different feature expressions. Three convolutional neural networks (CNN) were used to extract the independent features from three classes of spectrograms, and the three features were stacked to obtain more complete features of emitter, to improve the recognition effect. Experimental results show that compared with the single feature recognition method, the proposed method improves the accuracy of emitter identification effectively.