针对实物保护系统(Physical Protection System，PPSY)中单一传感器报警准确率较低的问题，提出了一种基于改进ID3的CAC-ID3(Confidence And Correlation-ID3)算法在多传感器实物保护系统中数据融合的新方法。与传统的单一传感器数据信息处理相比，多传感器数据融合能够更加准确、全面的得到被测对象的数据信息，有效地利用多传感器资源。CAC-ID3算法首先在ID3 的基础上引入属性置信度重新计算期望熵，解决属性和价值不对等的问题，克服多传感器数据分类时多值偏向的缺点，其值由经验和相关领域知识决定。然后通过引入属性间的相关度来调整信息增益值，提高分类精度。实验结果表明:基于CAC-ID3 的决策树算法的多传感器PPSY能有效提高报警准确率和可靠性，防止敌对分子入侵，提高传感器对PPSY的检测的效能，且该算法的分类精度高于ID3 算法。
Application of Optimized ID3 Algorithm in Multi-Sensor Physical Protection System
Aiming at the low alarm accuracy of a single sensor in the Physical Protection System(PPSY),a new data fusion method based on improved ID3 Confidence And Correlation-ID3(CAC-ID3)algorithm was proposed in the multi-sensor Physical Protection System.Compared with the traditional single sensor data processing,the multi-sensor data fusion system can obtain more accurate and comprehensive data information of the tested objects and make effective use of multi-sensor resources.CAC-ID3 algorithm firstly introduces attribute confidence to recalculate ex-pected entropy on the basis of ID3 to solve the problem of mismatch between attributes and values and overcome the shortcoming of multi-value bias in multi-sensor data classification,whose value is determined by experience and rele-vant domain knowledge.Then the information gain is adjusted by introducing the relevancy between attributes to im-prove the classification accuracy.The experimental results show that the multi-sensor PPSY based on CAC-ID3 deci-sion tree algorithm can effectively improve alarm accuracy and reliability,prevent the invasion of hostile molecules,and improve the sensor's detection efficiency of PPSY;The classification accuracy of this algorithm is higher than that of ID3 algorithm.
NETL
NSTL
万方数据
