Research On Modeling And Analysis Method Of Biological Systems Based On Hybrid Petri Nets

In order to study dynamic biological system,mathematical model need to be established and computer modeling and simulation should be carried out.Continuous Petri net is a system modeling and simulation tool,which can realize the visual modeling and simulation of ordinary differential equations and is often used to simulate biological systems.However,due to the complexity of biological systems and errors in data measurement,biological systems are uncertain,and continuous Petri nets cannot model and simulate biological systems that possess uncertainties.Fuzzy neural network is a kind of artificial intelligence technology,which can utilize expert knowledge to solve the modeling problems of nonlinear system without the help of accurate mathematical model.However,fuzzy rules need pre-setting by experts,which fails to acquire and adjust automatically.To solve the above problems,in this paper,based on the continuous Petri nets and fuzzy neural network,the author proposes a modeling and analysis method named hybrid Petri nets to model the uncertain biological systems.In the initialization phase of fuzzy neural network,the author use clustering analysis technology to optimize the fuzzy neural network modeling algorithm,which can reduce the reliance on expert knowledge.The main research contents of this paper are as follows:(1)Summarizing recent research on Petri nets and fuzzy neural network and introducing related technologies including Petri nets related theory,fuzzy system,fuzzy neural network and common clustering algorithms,etc.(2)To solve the problem of modeling uncertain biological systems,this paper proposes a new modeling method named hybrid Petri nets.Hybrid Petri nets is a biological system modeling method based on fuzzy neural network and continuous Petri net.Without the accurate mathematical model,this method can model and simulate the uncertain biological system.(3)Aiming at the problem that the initialization of fuzzy neural network depends heavily on expertise,this paper proposes a structure design method of fuzzy neural network based on FCM recursive clustering.This method is able to recursively divide input spaces driven by the fluctuation of the output sample,automatically acquire the fuzzy rules,determine the initial value of parameters,and reduce the dependence of the fuzzy neural network on expertise.(4)In order to verify both the approximation capability of fuzzy neural network and hybrid Petri net simulation ability,this paper demonstrates two biological system simulation experiments.The results of experimental show that,hybrid Petri nets can model biological systems that possess uncertainties and recursive structure design method based on FCM recursive clustering can reduce the dependence of the fuzzy neural network on expertise,as well as accelerate the algorithm convergence speed in training period.