Improved Kidney Algorithm And Its Application In Neural Network

The kidney-inspired algorithm(KA)is a stochastic optimization algorithm that simulates the working principle and process of human kidneys.It has the ability to perform partial fine search for better individuals and global search for poor individuals.It has been successfully applied in the fields of rainfall prediction and power system optimization.However,compared with other group intelligent algorithms,the research on the theoretical and practical applications of kidney algorithms is still lacking.There are still related many problems that need further research.In order to further improve the convergence speed of the algorithm and the accuracy of the algorithm,the standard kidney algorithm is improved from two aspects: algorithm update formula and introduction of cooperation operator.Then the improved kidney algorithm is used to optimize the weight and threshold of BP neural network.And the optimized BP neural network model is applied to the prediction of grain yield.The main work of the article is as follows:(1)Firstly,a kidney algorithm based on scaling factor and cooperation operator is proposed.Based on the kidney algorithm,the scaling factor is added to the reabsorption update formula.By adjusting the scaling factor,the algorithm is prevented from falling into local optimum.Then,after the algorithm completes one iteration,the cooperative operator operation is performed,and the information sharing of the better individual among the populations is used to increase the diversity of the population and enhance the optimization ability of the algorithm.Eight typical test functions are selected and compared with the basic kidney algorithm.The test results show that the new algorithm is superior to the improved kidney algorithm in convergence speed and convergence precision.(2)According to the characteristics of the kidney algorithm in solving the complex problem optimization process and the basic principle of BP neural network algorithm,the improved kidney algorithm is used to replace the gradient descent method in BP neural network to train the neural network weight and threshold,and a kind of neural network is established.BP neural network prediction model based on improved kidney algorithm.(3)The new prediction model is applied to the grain yield prediction problem.The simulation results show that the prediction accuracy is higher than BP neural network and kidney algorithm optimization BP neural network model.