An Optimization Strategy Of NSGA-Ⅱ In D-optimal Multi-objective Drug Formulation Researches

In industrial and agricultural production, medicine manufacturing, foodprocessing industries, there are a large number of mixture problem. When the amountof a finished product fixed, in order to make the mixture optimal, we need to adoptreasonable mixture experimental design and optimization methods. In the field ofdrug research,there are more common asks:1) The amount of each component in thedrug must be within the scope of the security, which limit the minimum andmaximum dosage of each composition;2) The drug formulation must be better inmulti-index. It contains both the upper and lower bound mixture problem, and themulti-objective optimization problem.Aiming at the upper and lower bound mixture problem, D-optimal mixturedesign is the most recommended. Under the condition of limited factor space, it canmaximal use the space information and arrange the test, and can improve theestimation precision of the model parameters, and greatly reduce the testing time. Inview of the multi-objective optimization mixture problem, the contour mapsuperposition method include a great deal of subjectivity. The Non-dominated SortingGenetic AlgorithmⅡ(NSGA-Ⅱ), as a kind of multi-objective genetic algorithm, intheory can parallel optimize the multiple target, and seek Pareto inferior solution set.But as a result of the mixture constraints, the existing program cannot solve theproblem directly.This paper, on the basis of studying the principle and the methods of D-optimummixture experimental design, will discuss the NSGA-Ⅱ with the strategy of mixtureconstraints into target application result between the upper and lower bounds mixtureproblems and the multi-objective optimization problems. This topic main researchcontents include:Section1The overview of the upper and lower bounds mixture problem. This part introduces the concept of mixture experiment, the principle and the method ofD-optimal mixture design, and the establishment and evaluation of regression model.Section2The overview of Non-dominated Sorting Genetic Algorithm Ⅱ.Introduces the concept and principle of genetic algorithm and the NSGA-Ⅱ,expounds the strategy of mixture constraints into target.Section3The test effect assessment of NSGA-Ⅱ with the strategy of mixtureconstraints into target. Choose blending test functions to test the program ofNSGA-Ⅱ, evaluate the feasibility of the strategy of mixture constraints into target.Test results show that the strategy of mixture constraints into target can expand theoriginal program applied to mixture constraint multi-objective optimization problem.The results are in good precision and accuracy.Section4The exploratory research on multi-objective formulationoptimization. Aiming at the experimental data from the literature, use the NSGA-Ⅱwith the strategy of mixture constraints into target to optimize the formulation ofMicroemulsion-Based Gel of Terbinafine (two goals) and the formulation ofSNEDDS of Lacidipine (three goals), and compare with the original results.In Terbinafine microemulsion gel formulation, an optimum formulation wasfound by optimizing strategy with particle size and solubility of17.04nm and44.32mg/ml at Oil, Smix and Water values of6.4%,53.5%and40.1%respectively.Compared with the contour map superposition method from the literature,microemulsion particle size reduced6.06%and solubility increased1.04%.In Lacidipine SNEDDS formulation, an optimum formulation was found byoptimizing strategy with absorbance, droplet size and15min cumulative releaseamount of0.657,21.896nm,100.619%at Oil, Surfactant and Co-surfactant values of30.92%,43.42%and25.66%respectively. Compared with the contour mapsuperposition method from the literature, dilution absorbance decreases12.4%,microemulsion particle size decreases1.7%,15min cumulative release increased by0.2%, target optimization effect is better.To sum up, the strategy of mixture constraints into target can make the original procedure of the NSGA-Ⅱ solve D-optimum mixture design and multi-objectiveoptimization problem, and the method is feasible, the effect is ideal. The method canbe used in other multi-objective optimization mixture design or be carried out andapplied in practical production.