RESUMO
The projection algorithm used in mixture analysis to determine whether there is unknown disturbance existing in grey system can not accurately identify different samples and similar samples at the same time when it is used in the identification of drugs, because of the insufficient criteria. In the present study, one of its criteria for whether the size of measurement error of testing sample is at a limited level is improved for whether the size and distribution of measurement error is equal and similar between testing sample and standard sample. By testing 6 kinds of normal drugs (including BAYER Aspirin Enteric-coated Tablets, TYLENOL Acetaminophen Sustained Release Tablets, BAYER Compound Paracetamol Tablets(II), HUAZHONG Compound Vitamin C, HUAZHONG Vitamin B and MADINGLIN Demperidone Tablets) and 3 kinds of similar drugs of aspirin (including BAYER Aspirin Enteric-coated Tablets, Shanghai SINE Aspirin Enteric-coated Tablets and Bamyl Aspirin Effervescent Tablets), it was found that the un-improved projection algorithm directly used in discrimination of drugs shows poor performance with many problems existing, however, the improved projection algorithm can discriminate different drugs and similar drugs with accuracy up to 100%. The improved projection algorithm can be a universal, accurate and reliable automated pharmaceutical identification algorithm and can provide a reference for the study on identification of substance.