Satellite component layout temperature field prediction method based on uncertainty

Abstract

The invention discloses a satellite component layout temperature field prediction method based on uncertainty. The method comprises the following steps: S1, constructing a deep neural network model; s2, generating a training data set and a pool data set; s3, training a deep neural network model by using the training data set; s4, repeatedly performing temperature field prediction on each pool of data for multiple times by using the model, and calculating the variance of multiple prediction results; s5, setting the pool data in a descending order according to the variance, selecting a preset number of pool data located in the front order, taking the selected pool data and the corresponding temperature field as new training data to be added into the training data set, and deleting the selected pool data from the pool data set; s6, continuing to train the model; s7, judging whether the prediction precision of the model meets a preset requirement or not; and if not, repeatedly performing temperature field prediction on the data of each pool for multiple times by using the model, calculating the variance of multiple prediction results, and returning to the step S5. According to the invention, the high-precision agent model can be obtained through a small amount of training data, and the calculation cost is reduced.

Description

technical field [0001] The invention relates to the technical field of satellite layout design, in particular to an uncertainty-based temperature field prediction method for satellite component layout. Background technique [0002] Satellite technology plays an irreplaceable and important role in communication, remote sensing, navigation, military reconnaissance and other fields, and is a hot research topic in the current industrial field. In order to meet the overall performance of the satellite, when determining the layout of various components inside the satellite, it is usually necessary to consider whether the temperature field under the current component layout meets the design requirements, such as whether the maximum temperature is too high, whether the temperature of a specific location is too high, etc. Therefore, how to obtain the temperature field under the component layout is a problem that needs to be solved when designing satellites. As for how to obtain the ...

AI Technical Summary

Problems solved by technology

[0003] However, the simulation calculation method based on finite element analysis determines that the calculation efficiency of a single simulation is not too high, and determining the final satellite component layout is an iterative optimization process. During the iterative optimization process, it is necessary to analyze the satellite component layout multiple times The temperature field greatly reduces the optimization efficiency, and for the calculation of the temperature field of the complex component layout, the calculation cost and calculation time will gradually increase with the complexity

Although the prediction method based on the deep neural network proxy model can realize the rapid prediction of the temperature field of the satellite component layout, thereby significantly improving the efficiency of satellite component layout optimization, but in order to ensure that the obtained deep neural network proxy model has sufficient prediction accuracy, it is necessary to A large amount of training data is used to train the deep neural network model. Due to the high cost of satellite development, it is impossible to obtain a large amount of experimental data sets to train the deep neural network model as in traditional fields (such as image recognition). The acquisition of each training data is Consumes more computing resources and computing time

Moreover, the existing prediction method based on the proxy model of the deep neural network can only give a positive temperature field prediction result after the satellite component layout is given, and cannot provide a data to evaluate the credibility of the current prediction result

However, in practical engineering applications, data may be affected by various uncertain factors such as noise, neural network model parameters and structure selectivity, etc., which makes the constructed deep neural network model have certain uncertainties, and this uncertainty will have a great impact on the training process and prediction results of the deep neural network model

Images