融合音频和视频的配电室智能监控技术研究

Research on intelligent monitoring technology of distribution room based on audio and video

结合智能监控系统的多功能检测特征和实时跟踪优势,研究设计出融合音频和视频的配电室智能监控技术,即通过构建高斯混合模型(GMM),研究了特征参数与混合阶数对于异常检测的识别率影响,同时基于Mel频率倒谱系数(MFCC)评估方法,分析了MFCC系数对于时间复杂度的影响。研究表明,混合阶数为32时,时间复杂度最低,且平均识别率达到85%以上。融合音频和视频的配电室智能监控技术,既实现了音频与视频多功能异常检测,又摆脱了传统视频异常检测的单向性与盲区性,继而提升了对于配电室安全检测的综合判断。

Combined with the multi-functional detection features and real-time tracking advantages of intelligent monitoring system,this research designs an intelligent monitoring technology of distribution room integrating audio and video,that is,by constructing Gaussian mixture model(GMM),the influence of characteristic parameters and mixing order on the recognition rate of anomaly detection is studied,and based on Mel frequency cepstrum coefficient(MFCC)evaluation method,the influence of MFCC coefficient on time complexity is analyzed.The results show that when the mixing order is 32,the time complexity is the lowest,and the average recognition rate is more than 85%.The intelligent monitoring technology of power distribution room integrated with audio and video not only realizes the multi-functional anomaly detection of audio and video,but also gets rid of the unidirectionality and blind area of traditional video anomaly detection,therefore improves the comprehensive judgment of power distribution room safety detection.