Research On The Measurement Model Of Body Skin Temperature Field Based On Improved GA-BP Neural Network

Skin temperature can reflect the health of human body.Obtaining the human skin temperature distribution,that is,the human skin temperature field,which has a great significance in clinical medicine,HVAC and aerospace.Because of the high complexity of the heat transfer process in the human body,it is difficult to accurately describe the mathematical problems.In this paper we use data-driven method to build human skin temperature field instead of finite element method.The method based on data-driven model building is based on the process data collected.It has many advantages,such as do not need to know the process mechanism and algorithm universality,so it is widely used in the modeling and optimization with complicated mechanism processes.In this paper,we use BP neural network and genetic algorithm in combination with optical temperature measurement system to measure body surface temperature field successfully.The influences of different topological structures on temperature measurement system and different networks reliability under topological structures are also discussed.Main work of this paper:1.According to the human body temperature characteristics and the existing human body structural parameters,the temperature field simulation model of the human thoracic cavity based on the finite element analysis method is established by ANSYS software.By establishing the temperature field simulation model of human thoracic cavity,the finite element method's problems in existing human skin temperature field is analyzed.2.A human skin temperature measurement system based on Fiber Bragg Grating(FBG)temperature sensor is set up to measure body surface temperature.According to the characteristics of human skin temperature field such as high data dimension,nonlinearity and dynamic characteristics,a human skin temperature field measurement model is established based on BP neural network.3.After analyzing the defects of BP neural network,the feasibility of combining BP neural network and genetic algorithm is discussed.The existing problems of genetic algorithm are also discussed and the improvement measures are given.The measurement model based on BP neural network which is optimized by improved genetic algorithm is established.Comparing the model output value and experimental data,it is found that the model has a high accuracy which can control the error in ± 0.40 ?..4.The influence of the topological structure of FBG sensor network on human skin temperature measurement system is analyzed.A FBG sensor network with self-healing function is proposed.The reliability of the proposed network topology and four basic network topologies are analyzed theoretically and verified by experiments.