Cigarette smoke colour spectrum baseline correction MPLS (mathematically proven learning system) method

The invention relates to a cigarette smoke colour spectrum baseline correction MPLS (mathematically proven learning system) method and belongs to the technical field of tobacco chemistry and application. The method provided by the invention comprises the following steps of: introducing peak recognition and penalized least squares method principle based on a mathematical morphology in a stechiometry to carry out cigarette smoke colour spectrum information baseline correction, namely, utilizing an opening operation of the mathematical morphology to find out peak position in a cigarette smoke colour spectrum signal and then recording the lowest point of adjacent peaks, utilizing the lowest point as a weight penalized least square to fit out a background, carrying out manual baseline correction process through artificial intelligence expert, and carrying out automation. The method provided by the invention has the beneficial effects that a principal component analysis result of the cigarette smoke after background correction is greatly improved, and the validity of the algorithm is verified relative to the condition before the algorithm is carried out, the MPLS method is a useful tool of cigarette smoke GC/oa-TOF-MS colour spectrum baseline correction, the method develops a new application field for the tobacco chemistry and application field, and a new method and concept are provided for cigarette smoke research.