基于多值测试的诊断策略优化生成

Optimized diagnostic strategy generation with multivalued tests

研究了多值输出测试条件下的诊断策略优化生成问题,把该问题形式化为一个最优多值与或决策树的搜索问题,然后将已有的基于二值测试的优化算法(信息增量启发式算法)同多值逻辑相结合,提出了包括多值输出测试和非对称测试的故障诊断策略优化生成算法,最后应用测试案例验证了该算法的有效性。结果表明,本方法对二值或多值测试以及非对称测试均适用,可以获得诊断精度和测试费用的理想权衡。

The problem of optimized diagnostic strategy generation with multivalued tests was researched. The problem was formulated as an optimal multivalued AND/OR decision tree search problem. Traditional binary test-based algorism （i. e. information gain heuristic based algorithm） was combined with multivalued logic, and the optimized diagnostic strategy generation algorithm was proposed, which includes multivalued tests and asymmetrical tests. Finally, a test case was used to verify the presented method. Experimental results show that this method is suitable for binary tests, multivalued tests and asymmetrical tests, and can obtain an efficient tradeoff between diagnostic precision and test cost.