小型多旋翼无人机发动机气路故障残差参数偏差值挖掘

Mining of Gas Path Fault Residuals Parameter Deviation for Small Uav Rotorcraft Engine

随着发动机工作性能的不断提高,获取的数据信号种类和数量也随之增加,为判断发动机故障增加了较大的难度。提出基于改进支持向量机数据挖掘算法完成对发动机气路故障偏差值进行挖掘和分析。首先将发动机工作状态实时监测系统收集的数据信息输入到支持向量机训练集中,计算发动机气路参数偏差值并输出;然后将输出的数据作为初始样本进行分析后,判断此数据里包括的发动机故障信息属性,以此为基础,采用指数平滑算法对不属于异常点的突变数据进行平滑处理,输出结果即为小型多旋翼无人机发动机气路故障残差参数偏差值,完成气路故障残差参数偏差值的挖掘。实验证明利用改进后的支持向量机数据挖掘算法,可以准确地计算发动机偏差值数据,更加精确地反映发动机的气路故障,为后期数据平滑处理提供了数据支持。

With the continuous improvement of engine performance, the acquisition of data type and quantity also will increase, for the judgment, increased the difficulty of the larger engine failure.Data mining based on improved support vector machine(SVM) algorithm is proposed to complete the engine gas path fault deviation for mining and analysis.Engine working state real time monitoring system of the collected data information into the support vector machine training focus, deviation calculation engine gas path parameters and output;Then through analysis of the output data as the initial samples, determine the engine fault information included in the data attribute, on this basis, the exponential smoothing algorithm is adopted to do not belong to the abnormal point mutation data smoothing processing, the output is more than a small uav rotorcraft engine gas path fault residuals parameter deviation, gas path fault residuals parameters deviation of mining.Experiments prove that using the improved support vector machine(SVM) algorithm, data mining can accurately calculate the engine deviation data, more accurately reflect the engine gas path fault, late for data smoothing processing provides data support.