摘要
围绕机电设备状态监测的需求,设计了基于视听信息融合的状态监测系统。分别利用图像可听化技术、数据归一化和主成分分析得到了较低维数的同质特征数据。以BP神经网络为融合模型,对所得特征数据进行识别与融合,进而得到对机电设备运行状态的决策输出。实验结果表明,在外界噪声环境下,融合视听信息的状态监测系统能维持较高的正确识别率,同时以神经网络为融合模型保证了系统的稳定性和鲁棒性。
Focused on the demand of mechatronical device status monitoring,a surveiliance system based on audio-visual information fusion was designed.The characteristic data with the same type and low dimensionality was obtained through the methods of image auralization,data normalization and principal component analysis.Based on BP neural network model,the charateristic data was recognised and fused,and the decisions of monitering of electromechanical device were obtain.The experimental results show that the correct recognition rate of monitoring based on audio-visual fusion is maintained at a high level under noisy environment,meanwhile the stability and the robustness of the system was guaranteed by the neural network model.
出处
《仪表技术与传感器》
CSCD
北大核心
2015年第9期56-59,共4页
Instrument Technique and Sensor
基金
国家自然科学基金资助项目(51175145)
关键词
状态监测
信息融合
图像可听化
主成分分析
神经网络
statu monitoring
information fusion
image auralization
principal component analysis
neural network
