Design And Implementation Of Nonlinear Function Extreme Value Calculation System Based On SAGA And BP Networks

In the study of the optimization model of modern engineering problems,many practical problems are finally abstracted into optimization problems such as "maximum output","lowest cost" and "best efficiency".In the final analysis,these problems are all problems of seeking extreme values.Therefore,how to find the optimal extremum quickly and accurately is an important issue.With the rapid development of science and technology and the continuous improvement of production tools,the process of obtaining the optimal extreme value becomes more and more complicated,among which there is a class of non-linear problems.The function of this problem is very complex or even difficult to express with mathematical functions.Known conditions have only some discrete input and output data.It is difficult to obtain the extreme values of the problem based on these data alone,and the problem often has multiple local extreme values,so it is difficult to find the optimal extreme value of the problem and the conditions for obtaining the extreme value.Aiming at this kind of problem,this paper proposed an extremum calculation algorithm for dual optimization of BP network with sinusoidal adaptive genetic algorithm,and used a mixed programming technology of VC ++ and MATLAB to develop a non-linear function extreme value obtaining system.The research content and innovation of the paper mainly include the following aspects:1.Analyzed the research status and optimization theory of nonlinear function extremum calculation problem,and explored the structure and characteristics of BP neural network and genetic algorithm.2.A sine adaptive genetic algorithm(SAGA)double optimization BP neural network algorithm is given to obtain the optimal extreme value,which uses the BP neural network's fitting prediction ability and genetic algorithm's global optimization ability.The algorithm is centered on the BP neural network training and fitting problem function.According to the problems of genetic algorithm in search speed and convergence,the sine adaptive cross rate and mutation rate calculation method is introduced into the genetic algorithm,and SAGA is obtained.Uses the SAGA to optimize the initial weight threshold of the BP neural network and the finaloptimal extreme value two aspects.In order to test the performance of the algorithm model,the traditional GABP method,GA algorithm double optimization BP neural network method and SAGA algorithm double optimization BP neural network method were simulated and compared.Experimental results show that the proposed algorithm effectively improves the accuracy of the nonlinear function extreme value.3.Using a mixed programming technology of VC ++ and MATLAB,a nonlinear function extremum calculation system is developed.In the system,a graphical interface was implemented with VC ++,and the extremum calculation algorithm was realized by calling the interface function of neural network and genetic algorithm in MATLAB engine.The user only needs to set the corresponding parameters in the system interface to obtain the optimal extreme value of the system and the point where the extreme value is obtained.