Study On Sustainable Evaluation Of Agricultural Insurance Based On AHP-MGABP-SI

The regional unbalance issues,have always existed during the process of China's agricultural insurance business operation.In addition,the different standards of the assessment level,as well as the level of constraints,cannot fully assess the agricultural insurance developmental level.The use of multi-generation genetic algorithm was proposed in this thesis to optimize BP neural network under the selective integration environment,the analytic hierarchy process was introduced as well to guide the mechanism of the neural network and further assess the sustainable development of China's regional agricultural insurance.The main content of this thesis is as follows:(1)This thesis proposes a multi-generational genetic algorithm based on the traditional genetic algorithm.It has two chromosome exchange processes for each crossover operation during the running of the cross-operation of the genetic algorithm,consequently,each generation crossover process can produce two pairs of offspring,thus,the number of offspring doubled,and further accelerating the genetic algorithm's fitness and convergence speed.Then according to the multi-generation genetic algorithm to optimize the initial weights and thresholds of BP neural network(MGABP model).(2)By introducing the analytic hierarchy process as the guiding mechanism of the neural network,to obtain the a priori guidance sample through the expert knowledge of the analytic hierarchy process,and then using the BP neural network which optimized by the multi-generation genetic algorithm to reproduce the expert's knowledge and experience and thereby establish a new evaluation model(AHP-MGABP evaluation model).This model achieves complementarity between the two models,which not only reduces the subjective randomness of BP neural network learning,but also enhances the noise immunity of the analytic hierarchy process.(3)A selective integration scheme is proposed,which uses the information gain and the gain rate to select the agricultural insurance evaluation index set that is most relevant to the evaluation target.The bagging sampling method is used to realize the diversity of ensemble learning and the accuracy of ensemble learning is ensured by selecting the optimal index set.In the selective integration environment,AHP-MGABP is used as a base learner to learn on independent samples,and an integrated learner is established through an integrated solution(AHP-MGABP-SI evaluation model).(4)The AHP-MGABP-SI evaluation model which established by the above improved algorithm was applied to the evaluation of sustainable development of regional agricultural insurance in China.Each base learner was independently evaluated and trained on the annual insurance data,and the comprehensive evaluation over the years was integrated at the integrated end.A scientific and practical evaluation program was provided for the sustainable development of agricultural insurance in China,and further promotes the sustainable and healthy development of agricultural insurance in China.