Modeling And Simulation Analysis Of Machine Taste Based On Nerve Conduction Machanism

At present,artificial intelligence is a hot technology in the world.As an important part of artificial intelligence,intelligent perception is an important source of information acquisition and interaction with the outside world.Machine perception includes machine vision,machine olfaction and machine taste.Its main purpose is to simulate,extend and expand human perception or cognitive ability by using machine.Neural perception model based on physiological basis is one of the effective means to realize machine perception.Therefore,it is of great practical significance for the development of machine taste perception to establish taste nerve conduction model based on the physiological basis of taste and to explore the neural system's action mode of processing external stimuli based on neural conduction mechanism.The physiological perception model based on neural conduction mechanism is an effective means to study the physiological activities of biological nervous system.In order to reveal the mechanism of taste nerve conduction and explore the response mode of taste system to external stimuli,a taste nerve conduction model with chaotic network characteristics was established based on the physiological anatomy of taste nerve.By combing the taste nerve conduction pathway and abstracting the topological structure of neural network,each node of neural conduction was expressed digitally,and the second-order differential equation was used to represent the neural dynamic characteristics of each node in the taste model,and the digital expression of the taste nerve conduction mechanism from the taste receptor to the taste area of the cerebral cortex was completed.At the same time,EEG signals of brain response to different intensity of taste stimulation were collected.On the one hand,the response law of brain response to external stimulation was explored through taste EEG experiment.On the other hand,the bionic performance and rationality of the model were verified through comparative analysis with the response results of taste nerve conduction model.The fourth-order Runge-Kutta method was used to decompose the model numerically to complete the simulation analysis of the model.The results showed that the model was bionic to a certain extent by comparing the response of brain waves to taste stimulation and the 1/f characteristics of response output in taste EEG experiment.The gradient descent method and the back propagation learning rule in the artificial neural network were introduced to optimize the model parameters.The effect of the model parameters optimization was verified by comparing the changes of 1/f characteristics before and after the optimization.Through the electronic tongue beer experiment,the 30s voltage of the response curve of the sensor was used as the electrical signal generated by simulating the taste perception of human body.The taste signals of different taste quality are input into the model,and the classification accuracy of different taste signals of the model was analyzed to verify the pattern recognition ability of the model.The results showed that the gustatory model could respond to different external stimuli in time,and it could return to the state before the stimulus after the external stimulus was removed,and the output waveform presented random and disordered oscillation state;the output response of the gustatory model was similar to that of the gustatory EEG in response law to a great extent,and the power spectrum of the gustatory model presented power law with the increase of frequency The drop characteristics were consistent with the 1/f characteristics of EEG signals and EEG signals in actual physiological experiments.After parameter optimization,the power spectrum of the model decreased more obviously according to the power law.Under the condition of 3 fold cross validation,the classification accuracy of the taste model for three brands of beer reached 58.3%.It could be concluded that the taste model has good perception ability and self-recovery ability when responding to different external stimuli,has the same characteristics as the EEG signal in the actual physiological experiment,has certain pattern recognition ability for different taste stimuli,and showed a certain Bionics in function and performance,and the 1/f characteristic of the model became more and more after the parameter optimization of the taste model Precise.