Study On The Risk Assessment Model Of Farmer’s Small Credit Loans With The Improved BP Neural Network

With the rapid development of the rural economy, rural credit loans demand gradually increased. Rural credit loans is mainly farmer’s small credit loans. Credit risk assessment is crucial in farmer’s small credit loans. It directly restricts the development of rural economy. Therefore,it is necessary to study the risk evaluation system of farmer’s small credit loans.This paper first introduces the concept of farmer’s credit evaluation and farmer’s small credit loans, then made a detailed analyzes on the research about the risk evaluation of farmer’s small credit loans. It shows that in this field domestic research is obviously behind the foreign.Because the artificial intelligence has great advantage in credit evaluation, it has been widely used in foreign countries. The BP neural network technology is an important technology in the artificial intelligence technology. It has many advantages, such as better learning and adaptive ability, inherent parallel computing and storage, using a stable nonlinear methods. Therefore, BP neural network has been applied in the study of small farmers credit risk assessment. However, the traditional BP neural network has the disadvantages, such as slow convergence,easy to fall into local minimum of traditional BP algorithm. In order to overcome these disadvantages, many kinds of improvements are proposed, such as adding momentum term,the adaptive learning rate,the LM algorithm, artificial immune, genetic algorithm,particle swarm optimization algorithm.This paper uses an emerging intelligent algorithm which is quantum particle swarm optimization algorithm (QPSO) to improve BP neural network model. Quantum particle swarm optimization algorithm (QPSO) has the advantages of less adjustable parameters, simple and easy to implement, and also has better convergence and global search ability. So it can overcome the shortcomings in convergence performance of BP neural network algorithm in a certain extent.By adding adaptive mutation to improve quantum particle swarm optimization (QPSO) and achieved good result. Although quantum particle swarm algorithm has a faster convergence speed in the early, but in late it may have been gathered to a certain point and formed local minima before reach the global optimum. Therefore, it uses adaptive mutation to overcome this drawback it.The improved quantum particle swarm optimization optimizes BP neural network weights and thresholds, then it was got the optimized BP network model. It can improve the convergence rate of BP neural network more efficiently and avoid falling into local minimum compared with the improved BP neural network model by the genetic algorithm etc.Then, the improved BP neural network model applied to the risk evaluation system of farmer’s small credit loans. In the simulation experiment, I randomly selected5data sets from the data, and then take the average value of the results of the5set of experiments, to compare with the traditional BP neural network model of experimental results. By the simulation experiments, it shows that the improved BP neural network algorithm can improve the accuracy of credit assessment and reducing the error rate.At last, the research prospect is presented from the aspects of small farmer’s credit risk assessment and improvement of BP neural network.