中文摘要：

为了提高检测率，采用DS证据融合技术融合多个ELM，能够提高整个检测系统的精确性。但是传统的DS技术处理冲突信息源时并不理想。因此，通过引入证据之间的冲突强度，将信息源划分成可接受冲突和不可接受冲哭，给出了新的证据理论（improvedDempster-Shafer，I-DS),同时针对ELM随机产生隐层神经元对算法性能造成影响的缺陷作出了改进。通过实验表明，结合I-DS和改进的ELM能够更高速、更有效地判别入侵行为。

英文摘要：

In order to im prove the detection ra te , DS evidence fusion technology and E L M could im prove ove rall detection system. H o w e ve r, DS technology was not id e a l when the tra d itio n a l sources was c o n flic t. T h e re fo re , the te xt gave the new e v idencetheory (im p ro v e d D em pster-S h a fe r,I -D S ) by d iv id in g the source in to acceptable and unacceptable c o n flic t ju d g in g fromin te n sity c o n flic t o f evide nces, at same tim e ,it made an im provem ent about the disadvantages. Because the disadvantages affectedthe a lgo rithm perform ance de ficie ncies fo r the random ly p roducin g hidd en neurons. E xperim ents show that com binedw ith the I-D S and im proved E L M can be a faster and more effective id e n tific a tio n in tru s io n .