针对液体火箭发动机(LRE)启动过程中推力变化大、仿真精度低的问题,用部分试车数据建立仿真模型,分析建模数据的变化特征,构造出非线性搜索函数;并用未建模的试车数据对模型进行验证。结果表明:克隆选择原理建模方法仿真精度高,不需要训练;与BP算法相比,总误差更小,而且对于启动过程中推力变化较大的采样点,仿真精度也有较大的提高。

For the problem of low accuracy simulation caused by quickly and quantitatively alterative force in start-up process of liquid rocket engine (LRE), the clone selection mechanism of artificial immune system was introduced. Constructing the simulation model with parts of the test data and analyzing the characters of the data, a non-linear searching function was established firstly. Then the model was validated with the test data that were not used in model constructing and parameters were optimized. The results show that the model of clone selection mechanism is a high precision and does not need training simulation model. Compared with BP algorithm, the precision of clone selection mechanism model is improved in the period of biggish force alteration during the start-up process and the total error is lower.