基于Gentle Adaboost的气密性检测系统

差压法气密性检测易受外部因素与预设参数影响。针对问题基于集成学习建立气密性检测系统，包含传感器终端数据采集系统、人机交互界面，并用最小二乘法对传感器进行线性拟合，利用Gentle Adaboost算法寻找每轮迭代中最佳弱分类器并更新下一轮样本权重，通过集成数轮迭代中最佳弱分类器组成强分类器，对被测物的气密性能进行判断。实验结果表明：所提系统在气密性检测中的准确度、精确度与召回率皆优于传统方法与单一分类模型，准确度达到99.8%，能有效克服外部因素对检测结果的影响，提高了差压法气密性检测的准确性与稳定性。

Air-Leakage Detection System Based on Gentle Adaboost

The differential pressure air-leakage detection is easily affected by external factors and preset parameters.Aiming at the problem,an air-leakage detection system based on ensemble learning was established,which included the sensor terminal data acquisition system,human-computer interface,and linear fitting of the sensor with the least square method.The Gentle Adaboost algorithm was used to find the best weak classifier in each iteration and update the sample weight in the next round.A strong classifier was formed by integrating the best weak classifier in several rounds of iterations to judge the sealing performance of the tested object.The experimental results show that the accuracy,precision and recall of the proposed system in air-leakage detection are superior to traditional methods and single classification model,and the accuracy is 99.8%.It can effectively overcome the influence of external factors on the test results,and improve the accuracy and stability of differential pressure air-leakage detection.