Research On Intelligent Thermal Control Technology Of Space Telescopes Based On Machine Learning

As one of the basic support systems of space telescope,the main task of thermal control system is to ensure that the space telescope provides a reliable and suitable thermal environment for all space mission units,such as electromechanical equipment,key components,mirror groups and scientific payloads,throughout the mission cycle.Traditional thermal control strategies and measures are very practical and applicable for telescopes running on fixed orbits because of their high reliability and simple control.However,with the continuous introduction of various new space loads and the upgrading of telescope detection tasks,the sensitivity of space telescopes to temperature is getting higher and higher,and the temperature control requirements are getting higher and higher.At the same time,when the space telescope needs to complete complex space tasks such as orbital transfer and rapid maneuver,the thermal control system must be able to adjust the thermal control strategy independently according to the current environmental changes and thermal control requirements in time.These requirements pose a huge challenge to the traditional thermal control technology:on the one hand,it is required to improve the efficiency of traditional thermal analysis modeling and calculation,and realize multi-condition adaptive modeling;on the one hand,the thermal design optimization process is required to get rid of the shackles of over-reliance on the experience of thermal engineers,improve optimization efficiency and achieve automation and intelligence of the optimization process;on the other hand,it is required to realize the temperature prediction of space optical load,realize its pre-control temperature mechanism,“eliminate peak and fill valley”,and reduce the fluctuation of load temperature;on the other hand,it is required that the active thermal control system can adaptively adjust the system parameters online according to the environment and system changes under unsupervised conditions to achieve intelligent autonomous operation.Therefore,this paper summarizes the development status of space telescope and its thermal control technology.Based on the application advantages of artificial intelligence technologies such as machine learning in agent model,temperature prediction,multi-objective optimization and multi-software integrated simulation,an intelligent thermal control technology of space telescope based on machine learning is proposed,and the problems of temperature prediction,thermal design parameter optimization and multi-software integrated simulation are discussed.The key technologies of thermal analysis agent modeling of space telescope based on machine learning,multi-objective optimization of thermal design parameters based on machine learning,on-orbit temperature prediction of space telescope based on machine learning,and multi-software integrated simulation system based on DOS command are studied.Traditional thermal control technology and thermal design scheme are the foundation of intelligent thermal control.A good intelligent optimization thermal control scheme is based on a solid traditional thermal control design.A solid traditional thermal design scheme can also help us better understand and discover the intersection of intelligent algorithm and spacecraft thermal control.Therefore,based on the thermal design work of an atmospheric index detector space telescope（DQM）,this paper first elaborates the process and measures of thermal design,as well as the points that can be improved in this process and combined with machine learning and other algorithms.Based on this,a series of work such as intelligent thermal control described in the remainder is carried out.Then,the thermal design parameter optimization scheme of space telescope based on surrogate model is studied.The space telescope is developing towards deep space exploration and maneuvering orbit transfer,and its thermal environment is changing and complex.The temperature of the telescope directly affects its imaging quality.Reasonable thermal design is the basis for ensuring the stable operation of the telescope.The optimization of thermal design parameters of space telescopes is mainly through parameter traversal and repeated attempts.There are problems such as heavy dependence on engineer experience,large computational workload,time-consuming and difficult to achieve global optimization.In the third chapter,a surrogate model based on improved back propagation neural network（called GAALBP）is proposed,and the design method of model parameters is optimized by genetic algorithm（GA）,referred to as SMPO.The surrogate model of the atmospheric index detector（called DQM）is established by using GAALBP,and compared with the surrogate model established by traditional back propagation neural network and radial basis neural network.The results show that the regression rate of the surrogate model based on GAALBP reaches 99.99%,the MSE error is less than 2×10^-6,and the maximum absolute error is less than 4×10^-3.GA is used to optimize the thermal design parameters of the surrogate model.The optimization results are verified by finite element simulation.Compared with the design results of manual thermal design parameters,the maximum temperature of CMOS is reduced by 5.33℃,the minimum temperature is reduced by 0.39℃,the temperature fluctuation is reduced by 4 times,and the parameter optimization calculation time is reduced by nearly 10 times.In addition,SMPO shows versatility and can be used to provide better selection guidance parameters and optimization for various complex engineering applications.Then,a thermal design optimization strategy of heat source layout based on multi-software integration is proposed,and its feasibility and effectiveness are verified by simulation comparison.The Muti-chip-module（MCM）is a multi-chip module that integrates several LSI/VLSI/ASIC bare chips and other components on an interconnected substrate to achieve overall packaging.The bare chip is used as a heat source.Its position on the substrate is directly related to the thermal load distribution of the overall structure and the chip junction temperature,which in turn affects the performance and reliability of the component.Therefore,thermal layout optimization has become an important part of the thermal design of multi-chip modules.In this regard,this paper establishes a set of abstract multi-chip module model agent model,and proposes an intelligent optimization system based on multi-software integration to realize heat source layout design.MATLAB,PYTHON,NX/SST and other software are called by WINDOWS DOS command,and the multi-objective optimization algorithm（non-dominated sorting genetic algorithm NSGA-II）is nested to realize the intelligent optimization of MCM component heat source layout.The simulation shows that this method can effectively improve the speed,accuracy and intelligence of multi-chip module heat layout optimization,and verify the effectiveness and practicability of the intelligent optimization layout system proposed in this paper.Finally,the temperature prediction of space telescope based on PROPHET algorithm is studied.Temperature prediction can predict the future temperature change trend of the telescope,and realize pre-judgment and early intervention control.The temperature prediction of the telescope flight is extremely important for realizing the’peak shaving and valley filling’of the spacecraft temperature fluctuation and its own thermal management.At the same time,the telescope orbit temperature prediction has great practical guiding significance in thermal system fault analysis and diagnosis,guiding thermal design and realizing spacecraft thermal system agent.Temperature prediction is the basis of on-orbit temperature prediction and intelligent control of space telescope.In this paper,the temperature of space telescope is predicted based on PROPHET algorithm,and the temperature prediction model is established.Firstly,the thermal analysis model of space telescope is established and the dynamic characteristics of the system are analyzed.Then,the prediction model is established based on the PROPHET algorithm,and then the input and output variables of the PROPHET prediction model and the data samples of the training set and the verification set are given.Finally,the temperature values of the two adjacent orbits of the key temperature control points of the space telescope are used as the input of the PROPHET prediction model,and the temperature fluctuation of the next orbit is the output of the model,and the prediction results are given.The simulation results show that the temperature of the space telescope can be effectively predicted.This is the first time that PROPHET has been applied to the temperature prediction of space telescope.The telescope temperature data has obvious time series properties and the temperature fluctuates periodically,which makes the PROPHET algorithm perform well in the on-orbit temperature prediction of spacecraft.