Intelligent Analysis On Chromatogram Of Traditional Chinese Medicine

Traditional Chinese medicine (TCM) is our national gem and cultural crystal. It has did many indelible contributions on medicine,health care and propagation for thousand years. At present it is facing a new trend which needs to make for international market and realize modernization of TCM. The quality control and management are the key of modernization of TCM.Analysis of TCM is the important and specifically demanded field of pharmaceutical analysis in China. But TCM always has very complex chemical composition, and its pharmadynamical function often acts in the way of the integral regulation of multicomponent,multitarget and multichannels. Most of the pharmadynamical components of TCM are unknown until now, so the quality control of TCM becomes a worldwide difficulty. Information technology of TCM apply mathematics,artificial intelligence and information knowledge, uses computers as its tool, whose aim is to reveal internal information law and research the methods of application. In this paper, high performance liquid chromatograph is applied to obtain experiment samples, and information technology is applied to improve traditional analytical method, which drives the process of modernization of TCM. The main content and academic contribution of the dissertation are described as follows:1. The application of wavelet transform in chromatogram data preprocessing is researched. Aimed at big size,noise interference and baseline drift of chromatogram datas, wavelet transform is applied to realize data compression,noise reduction and baseline correction. The experiments show that the effect of data preprocessing is very good. Wavelet transform will become an important tool of data preprocessing.2. A new model-resolution of overlapping chromatographic peaks of unknown components number is presented. The singularity detection principle of wavelet transform was used to extract characteristic points which included the information of chromatographic peak shape. According to extractive characteristic points, the number of hidden layer nodes and initial values of the parameters were estimated. Radial basis function (RBF) neural network was employed to fit overlapped chromatographic signals. The analytical results were from network parameters after training network with gradient descent method. Experimental results indicate that the proposed method has good performance with fast training speed,high accuracy and reliability, and can resolve overlapped peaks which has unexpected number of components.3. Due to the fixed regular of competing-layer structure, features of primary datasets can't be reflected by rule and line. Aimed at solving this problem, a novel competing-layer structure adjustable SOM algorithm (CSA-SOM) is proposed. This algorithm can adaptively adjust the positions of competing-layer neurons based on position of primary datasets. As a result, the neurons in mapping space can keep the features, such as the distance,the angle and the distribution, of primary datasets. The CSA-SOM algorithm is successfully applied in pattern recognition of red-spotted stonecrop samples. Both theory analysis and Experimental results indicates that CSA-SOM is an effective algorithm which can map dataset's inherent feature quickly and accurately. It resolves the conventional SOM's problem that mapped dataset's structure in competing-layer is distorted. The results indicate that CSA-SOM can effectively be applied in habitat differentiation of TCM fingerprint.