论文部分内容阅读
应用计算机模式识别技术进行抽油机井示功图识别的关键，是根据示功图的形状特征，选择和提取能够最有效地反映示功图形状信息的特征不变量，从而实现快速准确地识别示功图的目的。本文引入“矩”这一特征不变量、提取了能够详细描述示功图形状的１２个特征，并在分类器设计中采用“以形状判别为主，以参数判别为辅”的分类决策规则，通过编程更好地提高了示功图模式识别的准确性。实际应用证明，识别准确率达到９０％以上。
The key to identify the dynamometer of pumping well with computer pattern recognition is to select and extract the feature invariants which can most effectively reflect the shape information of dynamometer according to the shape features of dynamometer to realize the fast and accurate identification The purpose of the work diagram. In this paper, we introduce the feature invariant of “Moments”, extract 12 features that can describe the shape of dynamometer in detail, and use the classification decision rule of “discernibility based on shape discernibility and supplemented by parameter discernment” By programming, the accuracy of pattern recognition is improved. Practical application shows that the recognition accuracy rate of more than 90%.