A method based on big data mining to classify and eliminate flight parameter outliers

Abstract

The invention provides a big data mining-based method for deleting flight parameter outliers in a classified manner. In a flight process of an aircraft, flight parameters are recorded in a flight recorder; and data contains illogical outliers, so that postmortem analysis requirements can be met by deleting the outliers. The flight data is regularly decomposed by adopting a data segmentation method; related data points can be quickly matched by utilizing a characteristic of curve fitting, and related statistical information is obtained; the statistical information of different flight parameters are combined into a variance set; flight parameter segments containing the outliers can be quickly identified by utilizing a radial basis neural network classification characteristic; then, for ill data segments, the positions of the outliers are quickly identified according to characteristics of maximum and minimum values and a mean value; and the subsequent process is performed in the same way, so that outlier points of all the data segments of the different flight parameters are obtained. The method adopts a big data mining thought, is different from a general method for mining the outliers by multi-point filtering, and has relatively high speed and accuracy especially for massive data containing the outlier points.

Description

technical field [0001] The invention provides a method for classifying and eliminating outliers of flight parameters based on big data mining, which belongs to the field of data processing in flight control systems and is mainly used in the outlier processing of flight data of unmanned aerial vehicles. Background technique [0002] In the UAV flight test, it is necessary to measure various flight parameters in the UAV flight and save them in the data recorder for later analysis. However, due to equipment or signal problems during the measurement, the recorded data contains outliers. If such outliers are not eliminated, it will have a great impact on the post-event analysis of aircraft performance. Therefore, it is necessary to take certain methods before data analysis. Eliminate outliers to ensure data integrity and reliability. [0003] In the existing outlier removal methods, multi-point smoothing filtering is generally used to calculate the prior mean value, and then the...

AI Technical Summary

Problems solved by technology

However, the above methods all have the disadvantages of large amount of calculation, long processing time or relatively complicated calculation method. When the flight data is large, it will take more time

Images