CAJ | 학술논문

针对现役装备技术状态评估多依赖于手工拆卸的现状,提出一种基于AdaBoost-SVM模式识别算法的在线技术状态评估方法。利用人工后坐在线检测设备对炮闩装置技术状态参数进行检测,在对检测数据进行相关性分析特征提取的基础上,引入支持向量机模式识别方法,建立炮闩装置技术状态评估模型。通过将评估模型与Ada-Boost算法相结合,每次迭代都根据测试精度对分类错误的样本点和各分量分类器的权重重新赋值,在下一次迭代中形成新的分量分类器以优化分类结果,最终将各分量分类器依其权重综合完成评估。实例分析结果验证了评估模型的正确性和有效性。
침대현역장비기술상태평고다의뢰우수공탁사적현상,제출일충기우AdaBoost-SVM모식식별산법적재선기술상태평고방법。이용인공후좌재선검측설비대포산장치기술상태삼수진행검측,재대검측수거진행상관성분석특정제취적기출상,인입지지향량궤모식식별방법,건립포산장치기술상태평고모형。통과장평고모형여Ada-Boost산법상결합,매차질대도근거측시정도대분류착오적양본점화각분량분류기적권중중신부치,재하일차질대중형성신적분량분류기이우화분류결과,최종장각분량분류기의기권중종합완성평고。실례분석결과험증료평고모형적정학성화유효성。
A kind of on-line technical condition assessment method based on AdaBoost-SVM pattern recognition is introduced aiming at the actualities that the technical condition assessment of active arming lies mainly on the manual disassembly.The breech mechanism technical condition parameters are tested by artificial recoiling on-line test equipment.The support vector machine is introduced to build the breech mechanism technical condition assessment model based on the correlativity analysis character extraction of test data.Each time of iteration will reevaluate weight of the wrong swatch and every heft class organs by combining AdaBoost algorithm with the assessment model,the new heft class organ would be engendered in the next iteration to optimize the result.At last,all heft class organs are combined with each other to fulfill the assessment according to their weights.The correctness and validity of assessment model are proved by instance analysis.