加权距离判别分析法在岩体质量等级分类中的应用

APPLICATION OF WEIGHTED MAHALANOBIS DISTANCE DISCRIMINANT ANALYSIS METHOD TO CLASSIFICATION OF ROCK MASS QUALITY

距离判别分析法是一种按多指标性能参数进行分类的方法,已被应用于工程岩体质量的等级分类中。但传统的距离判别法对各个指标的重要性同等对待,夸大一些微小变化指标的作用。因此,采用加权马氏距离对样本中各指标进行加权处理以区分其重要性。首先对工程岩体质量指标运用主成分分析思想,计算出一组新的指标参数和与之对应的反应原始指标信息的贡献率,以贡献率组成对角阵作为权值矩阵;然后把数据标准化处理,计算加权马氏距离判别岩体质量等级。将该方法应用于工程实践,并与人工神经网络法、距离判别法进行比较。研究结果表明,加权距离判别分析法具有更高的判别能力,是一种准确有效的岩体质量分级方法。

Distance discriminant analysis method is one of classification methods according to multi-index performance parameters;and it has been applied to classification of engineering rock mass quality. However,the traditional Mahalanobis distance discriminant method treats with the importance of all parameters equally,and exaggerates the role of parameters which changes a little. The weighted Mahalanobis distance is used in discriminant analysis method to distinguish the importance of each parameter to improve the classification of rock mass quality. In the concrete application,firstly based on the principal component analysis scheme,a new group of parameters and their corresponding percent contributions of the parameters of rock mass quality are calculated;and the weighted matrix is regarded as the diagonal matrix of the contributions rates. Setting data to standardization, then the weighted Mahalanobis distance can be calculated to classify rock mass quality. The new method is used in practical project and compared with the artificial neural network method and Mahalanobis distance discriminant analysis method. The results show that the weighted Mahalanobis distance method has a higher discriminant ability;and prediction results are in better agreement with the fact. It is an exact and effective classification method of rock mass quality.