Research On Spectral Wavalength Selection Problem Based On Ant Colony Algorithm

Nowadays, with the rapid development of economy and the rising of people’s living standard, Consumers began to pay more and more attention to the problems of food’s components. The content of food is closely related to the food’s safe and Nutrient and health. People pay more and more attention to the food safety hazard which was caused of environmental pollution, excessive pesticide residues, additives abuse, improper storage and other factors. The detection of food component plays an increasingly important role in ensuring food’s safety. The traditional food detection technology has disadvantages, such as long period, high degree of specialization and higher cost, as a result, it can not detect chemical component content of food in quick and in time. The technique of NIR-PLS is a new food detection technology which was used widely. It uses a mechanism that different substances in the near infrared spectrum at various wavelengths have different absorption rates by establishing a model between the spectral data and food ingredients to distinguish the chemical composition of the material. The NIR-PLS don’t add Reagents to food, destroy the food samples, pollute the environment and can achieve rapid measurement and online measurement, it is a green technology. However, the near-infrared spectra have phenomena, such as absorption overlap seriously, weak signal, serious interference etc. Therefore, it must have operation to select wavelength to eliminate irrelevant or nonlinear variable, get a calibration models of strong forecast capacity and good robustness.At present, in the near-infrared spectroscopy model combined with quantitative calibration, the mainly wavelength method is no information variable elimination method, genetic algorithms and so on. But they have some disadvantages. So, we need to find a new method to select wavelength, apply it to wavelength selection in the near-infrared spectrum.Ant colony algorithm is a new intelligent optimization algorithm produced by simulating the foraging behavior of ants, which was firstly proposed by Italian scholars M.Dorigo in the early1990s, and it can successfully solve the traveling salesman problem. Ant algorithm has characteristic of strong robustness parallelism, distributed computer system and easy to implement, etc. It has been widely used in the field of combinatorial optimization, network routing, function optimization, data mining, robot path planning, performance optimization for wireless sensor networks and other fields. Many data show that the ant colony algorithm has achieved remarkable results in a series of complex and difficult optimization problems, and the superiority of solving complex optimization problems in discrete optimization problems. Due to the near infrared spectrum wavelength selection problem is also a kind of combinatorial optimization problem. So the ant colony algorithm applied in the study will be subject to the near-infrared spectral wavelength selection, exploring its properties in the wavelength selection.The whole subject research process is as follows:1) firstly introduced the basic theory of ant colony algorithm, analyzed ant colony algorithm’s principle of positive feedback and ant colony find the shortest path, and then specifically described the basic model of ant colony algorithm, given the specific implementation steps and algorithms ant colony algorithm flowchart. Finally introduced the characteristics and application status of ant colony algorithm;2)Detailed analyze the performance effects by set various parameters α,β,p, m, Q, for specific TSP problem by matlab simulation experiment, gives reasonable ranges for these parameters;3)Introduces several improved ant colony algorithm, describe their improvement measures, analyze their performance by matlab simulation;4)Expounded the principle of near infrared spectral quantitative analysis, gives the specific implementation steps, introduce the advantage of this technology, and detailed introduce near-infrared spectroscopy quantitative analysis modeling tools—partial least squares algorithm;5)Apply ant colony algorithm to near-infrared spectrum wavelength selection, gives the feasibility and the concrete implementation steps of wavelength selection, made a comparative analysis with the experimental results and the model of wavelength with ant colony algorithm and all established model, the experiment result shown that using ant colony algorithm to select wavelength is better than the other two models. In another word, ant colony algorithm is a feasible algorithm, it can be applied to other problems in later study and exploration.