摘要
在个体行为数据聚类的双重混合高斯模型算法假设的基础上,提出一种打火机微量气体泄漏动态检测的双重算法。算法通过动态空间扫描和浓度空间补偿的方法,完成了泄漏打火机的甄别并消除了打火机泄漏气体对相邻对象检测准确度的影响,解决了具有多峰值特征的单支打火机微量气体泄漏程度判别的问题。试验证明在工厂环境下使用该算法,漏检率优于0.06%、误检率优于0.18%,有效地提升了检测效率,能够为同类小型密封容器密封性能的检测提供借鉴。
This paper proposed a dual algorithm for dynamic detection of lighter’s trace gas leakage,mainly based on novel clustering Gaussian model algorithm for individual behavior data.By means of dynamic space scanning and spatial compensation concentration,it realized trace gas leakage scanning and discrimination.Interference caused by the contiguous leakage one will be eliminated evidently as well,with which it is easier and more reliably to work on signal trace leakage lighter with multiple-peak.Experimental results show that the missing detected accuracy is less than 0.06%and the miss ratio is less than 0.18% when in factories,resulting in high detection efficiency.This algorithm can be applied to trace leakage detection for other small pressure vessels.
出处
《仪表技术与传感器》
CSCD
北大核心
2012年第6期66-68,共3页
Instrument Technique and Sensor
基金
国家质检总局公益性行业科研专项(201110045-2)