Method, apparatus and computer readable medium for anomaly detection of substance

Abstract

In anomaly detection of a substance, a first set of chemical fingerprints is obtained. Each fingerprint in the first set is indicative of a plurality of physicochemical characteristics of a normal sample of the substance. The first set is converted into clusters of data points in a multi-dimensional principal component analysis (PCA) map. Each dimension of the map is based on a principal component (PC) corresponding to one of the physicochemical characteristics. The contour pattern of the cluster of data points is constructed as a predictive model to identify a new sample having a chemical fingerprint outside of the contour pattern as an anomaly. The predictive model is optimized using the second set of chemical fingerprints. Each chemical fingerprint in the second set is indicative of a plurality of physicochemical characteristics of a test sample including a normal test sample and an abnormal sample of the substance. 在物质的异常检测中，获得第一组化学指纹。第一组中的每个指纹指示物质的正常样品的多个物理化学特性。第一组在多维主成分分析(PCA)图中被转换为数据点群集。该图的每个维度均基于对应于物理化学特性中的一个的主成分(PC)。将数据点群集的轮廓图案构建为预测模型，以将具有在轮廓图案之外的化学指纹的新样品识别为异常。使用第二组化学指纹优化预测模型。第二组中的每个化学指纹均指示包括正常检验样品和物质的异常样品的检验样品的多个物理化学特性。