Font Size: a A A

Research On Data Cleaning Method Based On Infrared Data

Posted on:2018-08-23Degree:MasterType:Thesis
Country:ChinaCandidate:Q Q LinFull Text:PDF
GTID:2428330551957972Subject:Software engineering
Abstract/Summary:PDF Full Text Request
As people deeply exploring the natural science,more and more mysteries and nature rules have been discovered and successfully applied to serve human beings' daily life.Infrared is one of typical cases.Since it was discovered,it has been widely used in all walks of life.An obvious issue is,all the objects whose temperatures above absolute zero can launch infrared so that the infrared launched from our remote control is easy to be disturbed,resulting in inaccuracy of the data received by infrared sensors,which leads to our remote control equipment often fail to work.In this paper,we propose a strategy to identify the dirty infrared data,and on the basis of the infrared data identifying means we recommend,we put forward the "key combination" scheme to overcome the shortcomings in the current intelligent remote control equipment.In order to recognize infrared data,this paper has done the following research work:(1)After researching and summarizing present methods of data cleaning,the paper reclassifies the data quality problems according to their causes and then proposes the method of cleaning the wrong data caused by the objective factors.(2)On the basis of understanding and mastering the working principle of infrared remote control equipment,the paper makes a simple infrared remote control system by utilizing ARDUNO.Taking the influence of distance and direction to infrared remote control equipment into account,the paper has collected experimental data under all kinds of situations.(3)Based on the characteristics of the collected infrared data and the purpose of the experiment,this paper proposes and designs the decision tree classification method of supervised learning to distinguish infrared data by studying the knowledge of machine learning and data mining.Meanwhile,after understanding the principles of C4.5,this paper proposes a strategy to cope with the continuous attributes based on boundary point theory and Fayyad theory.(4)Focus on the problems of existing intelligent infrared remote control equipment——it is easily to be interfered and the number of controlled infrared equipment is confined because of its limited storage.In this paper,based on the infrared data identifying method,we propose a"key-combination”solution to solve these two problems and verify its feasibility though relevant experiments.Furthermore,considering the security of transmitting the infrared data,we design an encryption/decryption model.
Keywords/Search Tags:infrared remote control, data identifying, machine learning, data mining
PDF Full Text Request
Related items