Objective  With the development of infrared guidance technology, the ground targets with obvious infrared characteristics are increasingly threatened on the battlefield. In order to improve the survivability of ground targets on the battlefield, it is necessary to master the infrared radiation characteristics of ground targets. The traditional methods for studying the infrared radiation characteristics of ground targets include theoretical modeling analysis method and field testing method. The theoretical modeling analysis method is faced with such problems as a huge amount of calculation, high cost of field testing, obtained limited data, and being unable to obtain the overall infrared radiation characteristics of ground targets at any time. For the ground target, if the thermocouple is arranged on the typical part of the surface of the ground target and the relationship between the typical part and the overall temperature distribution is established, the rapid prediction of the temperature field of the ground target can be achieved. Therefore, how to arrange thermocouple on the surface of ground target and establish the relationship between the data of finite measuring points and the whole temperature field of ground target has become an urgent problem to be solved. Therefore, proper orthogonal decomposition (POD) method is introduced to extract the characteristics of the ground target temperature field, and a reduced order model of the temperature field is established. On this basis, combined with QR (orthogonal right triangular) decomposition algorithm, the rapid prediction of the temperature field of the ground target is realized by using the data of finite measuring points.

  Methods  POD method is introduced into temperature field characteristics analysis of ground targets, and the specific implementation process is shown (Fig.1). Taking square cavity and model tank as the research objects, POD method was used to extract the temperature field characteristics of ground targets, and two kinds of ground target temperature field reduction models were established to predict the temperature distribution of ground targets at multiple moments and compare it with the real temperature field at the same moment (Fig.7, Fig.15). Based on the reduced order model of temperature field combined with QR decomposition algorithm, the position of the best sensor was determined, and the temperature field was predicted by using the measured data of the best sensor, and compared with the real temperature field (Fig.9-10, Fig.16-17). Finally, the reliability of the method is verified by error calculation and analysis.

  Results and Discussions   POD method can extract the main characteristics of the ground target temperature field well. On this basis, the reduced order model is established, and the predicted temperature distribution is basically consistent with the real ground target temperature distribution. After combining POD and QR algorithm, the best sensor measurement data obtained by QR decomposition algorithm is used to predict the temperature field of two ground targets. The predicted temperature distribution trend of the two ground targets is basically the same as the real temperature distribution trend. Based on the POD temperature field reduction model, the calculation time is further reduced and the calculation efficiency is improved. The average absolute error of the temperature field of the square cavity with heat source is less than 1.5 K. The average absolute error of the temperature field of the model tank target is less than 2.5 K. This indicates the accuracy of the method.

  Conclusions  It can be seen that POD can better extract the characteristics of the temperature field of the ground target, so as to establish the reduced order model of the temperature field of the ground target. After combining with the QR algorithm, the temperature field can be quickly and accurately predicted through the calculated best sensor measurement data, greatly reducing the calculation time. If conditions permit, the number of training sets and sensors can be increased to improve the accuracy, which provides a new method and idea for the rapid prediction of transient temperature field of ground targets by using finite measuring point data.