设置隶属度函数改善测量结果的判别策略

Study on the improvement of the discrimination tactic for test resultby setting membership function

针对传统的测量值判别策略中,由于误差量的影响造成误判率增加的问题,提出了一种设置隶属函数的判别策略。该策略通过设置适当的隶属函数,获得测量值属于各类别的隶属度,在不增加测量次数的情况下,可大大降低误判率。考虑了二元判别问题,从检测概率的角度,分析了这种策略的可行性。并结合实际问题给出了应用该策略的具体方法,给出了例证,仿真结果证明了该策略的有效性。

To solve the problem of high mistake rate caused by margin of error in classic discrimination tactic of test result, a discrimination tactic using membership function is introduced. In the tactic, by setting proper membership function in the process of test result discrimination, membership can be calculated to describe the level of test result under one classification, thus mistake rate is reduced without additional tests. By dividing test result into two classes, the feasibility of the tactic is analyzed in terms of detection probability. And the method using the tactic is given. Theoretical calculation and computer simulation are also fulfilled. At last, an example is given, which shows that the tactic is effective.