本文選題：四旋翼無(wú)人機(jī)　切入點(diǎn)：管道巡線系統(tǒng)　出處：《昆明理工大學(xué)》2017年碩士論文　論文類型：學(xué)位論文

【摘要】：長(zhǎng)距離漿體管道輸送技術(shù),已經(jīng)成功解決偏遠(yuǎn)山區(qū)礦產(chǎn)資源輸送問(wèn)題。漿體管道大部分鋪設(shè)在山地、丘陵地帶,極端天氣一旦發(fā)生極易對(duì)管道造成危害,產(chǎn)生巨大的經(jīng)濟(jì)損失。管道巡線人員沿.山路對(duì)管線進(jìn)行巡視,無(wú)法及時(shí)發(fā)現(xiàn)并確定事故管線地點(diǎn),導(dǎo)致故障無(wú)法及時(shí)得到處理。因此,采取一種既能快速查找定位礦漿管道事故地點(diǎn),又能節(jié)省人力物力的方式具有重要的意義。針對(duì)管道所處山地位置環(huán)境復(fù)雜,巡線人員不易到達(dá)現(xiàn)場(chǎng),造成管道巡線維護(hù)成本較大問(wèn)題,本文設(shè)計(jì)了一種四旋翼無(wú)人機(jī)的輔助巡線系統(tǒng),用于礦漿管道運(yùn)行狀態(tài)的監(jiān)測(cè)。巡線人員可以通過(guò)輔助巡線機(jī)器人,在無(wú)法到達(dá)管道位置的情況下,通過(guò)無(wú)人機(jī)上所攜帶的圖像處理設(shè)備和圖傳設(shè)備自動(dòng)地對(duì)管道進(jìn)行跟蹤巡視,觀察礦漿管道運(yùn)行狀態(tài)。本文的主要工作內(nèi)容和研究成果如下:(1)建立基于深度卷積神經(jīng)網(wǎng)絡(luò)(CNN)的管道分類模型。采集了實(shí)驗(yàn)室條件下的模擬管道數(shù)據(jù)作為數(shù)據(jù)集,對(duì)神經(jīng)網(wǎng)絡(luò)管道分類模型進(jìn)行訓(xùn)練。通過(guò)在測(cè)試集上對(duì)模型進(jìn)行驗(yàn)證,該管道識(shí)別分類模型可以識(shí)別復(fù)雜條件下的管道,結(jié)果顯示其分類正確率基本滿足要求。(2)設(shè)計(jì)了基于STM32F405單片機(jī)的飛行控制系統(tǒng)。首先對(duì)四旋翼飛行器進(jìn)行數(shù)學(xué)建模,分析其模型特性,確定飛行器的控制策略。其次,對(duì)各個(gè)傳感器,包括加速度計(jì)、陀螺儀和磁力計(jì)等進(jìn)行擬合校正和數(shù)據(jù)濾波,計(jì)算得到較為準(zhǔn)確的姿態(tài)信息。最后實(shí)現(xiàn)了基于串級(jí)PID算法的飛行控制器。(3)實(shí)現(xiàn)了管道分類模型軟件、飛控系統(tǒng)軟件和飛控系統(tǒng)硬件。使用Python腳本語(yǔ)言和TensorFlow深度學(xué)習(xí)框架實(shí)現(xiàn)了模型的軟件部分。使用C語(yǔ)言完成了飛行控制系統(tǒng)的軟件部分、傳感器數(shù)據(jù)的處理與校準(zhǔn)程序、姿態(tài)控制程序和位置控制程序,并最終在STM32單片機(jī)上運(yùn)行。(4)將機(jī)載電腦和飛控系統(tǒng)搭載到四旋翼飛行器上,在實(shí)驗(yàn)室模擬管道下進(jìn)行模擬礦漿管道巡線飛行,飛行器可按模擬管道路線自主飛行,達(dá)到了礦漿管道巡線要求,實(shí)現(xiàn)了對(duì)礦漿管道運(yùn)行狀態(tài)的監(jiān)測(cè)。
[Abstract]:Long distance slurry pipeline transportation technology has successfully solved the problem of mineral resources transportation in remote mountainous areas. Most of the slurry pipelines are laid in mountainous and hilly areas. If extreme weather occurs, it is easy to cause harm to pipelines. Great economic losses are caused. Pipeline patrol personnel patrol the pipeline along the. Mountain road, unable to find and determine the location of the accident pipeline in time, resulting in failure can not be dealt with in time. Therefore, It is of great significance to adopt a way that can locate the accident site of the slurry pipeline quickly and save manpower and material resources. In view of the complex environment of the mountain location of the pipeline, it is difficult for the line patrol personnel to get to the site. In this paper, a kind of auxiliary line patrol system for four-rotor UAV is designed to monitor the running state of slurry pipeline. When the pipeline position cannot be reached, the pipeline is automatically tracked and patrolled through the image processing and transmission equipment carried on the UAV. The main contents and research results of this paper are as follows: 1) the classification model of pipelines based on deep convolution neural network (CNN) is established. The simulated pipeline data under laboratory conditions are collected as data sets. The neural network pipeline classification model is trained. By verifying the model on the test set, the pipeline identification and classification model can identify the pipeline under complex conditions. The results show that the classification accuracy basically meets the requirements. (2) the flight control system based on STM32F405 microcontroller is designed. Firstly, the mathematical model of the four-rotor aircraft is established, and its model characteristics are analyzed, and the control strategy of the aircraft is determined. Fitting and correcting various sensors, including accelerometers, gyroscopes and magnetometers, and data filtering, Finally, a flight controller based on cascade PID algorithm. The software of the flight control system and the hardware of the flight control system are realized by using Python script language and TensorFlow depth learning framework. The software part of the flight control system and the processing and calibration program of the sensor data are completed by using C language. Attitude control program and position control program, and finally run on the STM32 microcontroller. 4) the airborne computer and flight control system are mounted on the four-rotor aircraft, and simulated slurry pipeline inspection flight is carried out under the laboratory simulated pipeline. The aircraft can fly independently according to the simulated pipeline route, which can meet the requirements of slurry pipeline inspection, and realize the monitoring of slurry pipeline running state.
【學(xué)位授予單位】：昆明理工大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類號(hào)】：TD561

【參考文獻(xiàn)】

相關(guān)期刊論文 前9條

1 夏紀(jì)冬;劉子龍;張文文;;基于四元數(shù)解算和模糊PID控制的四軸旋翼機(jī)[J];電子科技;2017年01期

2 熊風(fēng)光;李希;韓燮;;基于整體最小二乘的橢圓擬合方法[J];微電子學(xué)與計(jì)算機(jī);2017年01期

3 龍?jiān)坡?陳洋;滕雄;;四旋翼飛行器姿態(tài)解算與濾波[J];計(jì)算機(jī)測(cè)量與控制;2016年10期

4 張合新;王強(qiáng);張騰飛;孟飛;劉峗;;激光主動(dòng)成像圖像邊緣檢測(cè)算法研究[J];激光與紅外;2016年06期

5 陳孟元;謝義建;陳躍東;;基于四元數(shù)改進(jìn)型互補(bǔ)濾波的MEMS姿態(tài)解算[J];電子測(cè)量與儀器學(xué)報(bào);2015年09期

6 李亞楠;趙耀;林春雨;白慧慧;劉美琴;;基于圖像分割的金字塔Lucas-Kanade光流法提取深度信息[J];鐵道學(xué)報(bào);2015年01期

7 孫茂林;;長(zhǎng)輸管道的安全問(wèn)題[J];化工管理;2014年18期

8 鄒偉生;袁海燕;羅紹卓;;長(zhǎng)距離管道輸送發(fā)展現(xiàn)狀及在礦山的應(yīng)用前景[J];金屬材料與冶金工程;2009年01期

9 張榮輝;賈宏光;陳濤;張躍;;基于四元數(shù)法的捷聯(lián)式慣性導(dǎo)航系統(tǒng)的姿態(tài)解算[J];光學(xué)精密工程;2008年10期

相關(guān)博士學(xué)位論文 前1條

1 戴金波;基于視覺(jué)信息的圖像特征提取算法研究[D];吉林大學(xué);2013年

相關(guān)碩士學(xué)位論文 前2條

1 胡畔;基于單一二維照片的三維模型重建研究[D];上海交通大學(xué);2015年

2 姜成平;一種四旋翼無(wú)人機(jī)控制系統(tǒng)的設(shè)計(jì)與實(shí)現(xiàn)研究[D];哈爾濱工業(yè)大學(xué);2014年

，

本文編號(hào)：1577584