基于稀疏自动编码器的发动机机载模型建模方法研究

发布时间：2018-04-18 15:31

  本文选题：机载发动机自适应模型 + 智能发动机控制　； 参考：《推进技术》2017年06期

【摘要】：为解决分段线性化机载模型精度不足的问题,提出并设计了基于稀疏自动编码器的大包线、具有10输入11输出的发动机机载自适应模型,该模型由稳态、动态两部分组合而成。首先基于一种新的相似准则进行建模所需样本数据的压缩,在保留主要信息的同时,大大降低了数据量及采样时间。用BP算法对简化后的样本数据进行了机载模型稳态部分的建模。针对机载模型动态部分所需样本数据量巨大、BP算法难以训练的问题,建立了基于稀疏自动编码器的动态机载模型。引入准稳态判断逻辑,在动态过程使用稀疏自动编码器的动态机载模型,在稳态过程使用基于BP算法的稳态机载模型。仿真结果表明,所建立的发动机机载模型具有优良的动稳态精度,且实时性好、存储量小,其中动态精度小于1%,稳态精度小于0.6%,一次模型计算时间不大于1ms,模型存储量不大于100kB。
[Abstract]:In order to solve the problem of insufficient accuracy of piecewise linearized airborne model, a large envelope model based on sparse automatic encoder with 10 inputs and 11 outputs is proposed and designed. The model is composed of steady and dynamic parts.Firstly, the compression of the sample data needed for modeling based on a new similarity criterion is carried out, which saves the main information and greatly reduces the data volume and sampling time.BP algorithm is used to model the steady-state part of the airborne model.The dynamic airborne model based on sparse automatic encoder is established to solve the problem that the sample data required in the dynamic part of the airborne model is huge and the BP algorithm is difficult to train.The quasi-steady-state judgment logic is introduced, the dynamic airborne model of sparse automatic encoder is used in the dynamic process, and the steady-state airborne model based on BP algorithm is used in the steady-state process.The simulation results show that the established airborne engine model has good dynamic and steady accuracy, good real-time performance and low storage capacity, in which the dynamic accuracy is less than 1, the steady-state accuracy is less than 0.6, the calculation time of the model is less than 1msand the storage capacity of the model is less than 100kB.
【作者单位】： 南京航空航天大学能源与动力学院江苏省航空动力系统重点实验室;中国航空发动机集团公司航空动力控制系统研究所;
【基金】：国家自然科学基金(51576096) 江苏省“青蓝工程” “333”人才工程
【分类号】：V233.7
，

本文编号：1768971