摘要
描述了概率文法的基本特性及其在铁路路轨探伤中的应用．文章首先给出了概率文法的定义，描述了它的树型结构，并为了得到非歧义的产生规则序列，对概率文法进行了修改，然后利用概率文法构造了模式识别器．介绍了应用于路轨探伤的识别程序的实现过程，并结合实例讨论了利用概率文法在提高识别器的可靠性。
In pattern recognition, the probabilities of which all character strings in Language L(G) occur, are different. Thus the probabilities of using the production rules in grammar G, or say, their importance is also different. If the probability of using the production forms is taken as a measure of their importance, it is possible to increase the correctness of pattern recognition, and to decrease the mistakes, losses of decision and misclassifications. This conclusion is implemented in "pattern recognition and monitor system of rail anomalies"designed by us. This system has passed the provincial appraisal. This paper describes the features of probabilistic grammar and its application in rail flaws tests. The paper gives the definition of probabilistic grammar first, and then explains the tree structure of probabilistic grammar. In order to get non-equivocal production rules sequence, it is necessary to improve the probabilistic grammar. Thus paragraph 3 describes this improvement process. Constructing pattern recognizers by means of probabilistic grammar are discussed in paragraph 4. Finally, the paper introduces a real application of probabilistic grammar in railroad track flaws test.
出处
《南京大学学报（自然科学版）》
CAS
CSCD
1997年第4期495-502,共8页
Journal of Nanjing University（Natural Science）
基金
江苏省自然科学基金
关键词
概率文法
模式识别器
铁路
路轨探伤
probabilistic grammar
pattern recognizer
anomaly pattern
