Research On Improved LM-BP Neural Network Temperature Control Algorithm For ASE Source

Fiber optic gyro becomes more and more popular in inertial navigation system because of it has many unique advantages.With the development of the times and the inertial navigation system's requirements,High-precision fiber optic gyroscope has become the focus of research institutes and universities.Fiber optic gyro light source has also been widely attention because of it is an important part of fiber optic gyroscope.High-precision fiber optic gyroscope has strict requirements in the input light source,Light source wavelength instability directly affects the working accuracy of fiber optic gyroscope,temperature is the key variable that affects the output wavelength of the light source.Therefore,it is necessary to improve the precision of the fiber optic gyroscope so that it can be applied to higher precision,first of all we need to use the light source to study,and light source temperature control system is its most important.The main research contents of this paper are to improve the LM algorithm to improve the back propagation algorithm used by the traditional BP neural network.This paper uses it to form a new light source temperature control algorithm instead of the PID control algorithm which is often used in the fiber optic gyro light source temperature control system to improve the output light source stability.First the basic principle of fiber optic gyroscope's ASE source is analyzed,the influence of light source performance on the accuracy of fiber optic gyroscope and the effect of temperature on the performance of light source are studied.Secondly,the basic model and principle of neural network are briefly summarized.BP neural network as an example,the basic model and back propagation algorithm of BP neural network are deduced in detail.Thirdly,the rationality and superiority of BP neural network and PID control are expounded,and the main steps and working process of BP neural network and PID control are summarized.Finally,this paper analyzes the shortcomings of reverse propagation algorithm of traditional BP neural network,improves the adaptive parameter factor and momentum factor of LM algorithm to improve the back propagation algorithm of BP neural network,and form the imprones LM-BP neural network temperature control algorithm which is better used in the fiber optic gyroscope light source temperature control system,we have to make its MATLAB simulation experimental analysis and made into finished product.The results show that the improned LM-BP neural network temperature control algorithm has better dynamic,fast and stable compared with the PID algorithm based on BP neural network,and its light source temperature performance satisfies the requirement of the fiber optic gyroscope to the light source,and it has a high reference and reference significance on the high precision fiber optic gyroscope research.