SINS_WSN组合定位下采煤机精确位姿感知理论及技术研究.pdf
博士学位论文 SINS/WSN 组合定位下采煤机精确位姿感 知理论及技术研究 Research on Accurate Position and Pose Perception Theory and Technology for Shearer with SINS/WSN Integrated Localization 作 者杨海 导 师李威 教授 中国矿业大学 二○一六年十二月 万方数据 中图分类号 TD67 学校代码 10290 UDC 622.2 密 级 公开 中国矿业大学 博士学位论文 SINS/WSN 组合定位下采煤机精确位姿感知理论 及技术研究 Research on Accurate Position and Pose Perception Theory and Technology for Shearer with SINS/WSN Integrated Localization 作 者 杨海 导 师 李威 申请学位 工学博士 培养单位 机电工程学院 学科专业 机械电子工程 研究方向 综采装备位姿检测 答辩委员会主席 龚烈航 评 阅 人 二○一六年十二月 万方数据 学位论文使用授权声明学位论文使用授权声明 本人完全了解中国矿业大学有关保留、使用学位论文的规定,同意本人所撰 写的学位论文的使用授权按照学校的管理规定处理 作为申请学位的条件之一, 学位论文著作权拥有者须授权所在学校拥有学位 论文的部分使用权,即①学校档案馆和图书馆有权保留学位论文的纸质版和电 子版,可以使用影印、缩印或扫描等复制手段保存和汇编学位论文;②为教学和 科研目的,学校档案馆和图书馆可以将公开的学位论文作为资料在档案馆、图书 馆等场所或在校园网上供校内师生阅读、浏览。另外,根据有关法规,同意中国 国家图书馆保存研究生学位论文。 (保密的学位论文在解密后适用本授权书) 。 作者签名 导师签名 年 月 日 年 月 日 万方数据 论文审阅认定书论文审阅认定书 研究生 杨海 在规定的学习年限内,按照研究生培养方案的 要求,完成了研究生课程的学习,成绩合格;在我的指导下完成本学 位论文,经审阅, 论文中的观点、 数据、 表述和结构为我所认同, 论文 撰写格式符合学校的相关规定, 同意将本论文作为学位申请论文送专 家评审。 导师签字 年 月 日 万方数据 致谢致谢 时光荏苒,五年的硕博生涯即将画上一个圆满的句号。回首走过的岁月,求 学期间,身边的良师益友给予了很多的帮助,在此表示感激之情。 首先, 衷心感谢我的导师李威教授, 2011 年,本人本着对机电自动化的浓厚 兴趣以及导师严谨治学精神的向往,有幸加入了导师的研究课题组。从本科的毕 业设计到硕士阶段的工程历练, 再到博士阶段的论文选题等研究过程始终得到了 导师的悉心指导和帮助。李老师那渊博的知识、独到的眼光、睿智的见解、敏锐 的洞察力以及富有启发性的分析使我受益匪浅。期间,李老师为我提供了优越而 舒适的学习、 实验、 工作环境, 在学习和生活中更是给予了我无私的帮助与关心, 无限的信任与支持。在此,向导师表示深深的谢意,感谢您多年来对我的关心和 帮助。 然后,感谢母校中国矿业大学这九年来对我的培养。感谢王禹桥副教授、范 孟豹副教授以及杨雪峰副教授, 三位老师在学习和生活上给予我很多无私的关怀 和帮助。三位老师渊博的学识、活跃的学术思想在很多关键地方对都我起到了很 好的引导作用,产生了深远的影响。感谢张嗣鸣老师多年来在生活和学习上对我 的关心和照顾。 其次, 感谢同课题组的樊启高博士、 罗成名博士、 应葆华硕士、 司卓印硕士、 张金尧硕士以及王超硕士, 正是由于你们的陪伴使我在课题的研究过程中不再孤 单前行,在遇到困惑的时候每次和你们讨论都能够有意想不到的收获。 感谢实验室的许少毅博士、刘玉飞博士、魏华贤博士、路恩博士、鞠锦勇博 士、杨盼盼博士、王茗硕士、须晓峰硕士、杨康硕士以及实验室的兄弟姐妹们, 感谢你们多年来的陪伴,是你们让实验室充满活力与欢声笑语,让我感觉到咱们 这个集体充满了家一样的温暖。 感谢国家留学基金委对我的资助, 使我能够有一年的时间来到德国进行学术 交流学习。感谢德国弗莱堡大学的 Leonhard M. Reindl 教授,让我有机会进入 您的研究团队学习历练。特别感谢弗莱堡大学的张睿博士,在德国的这一年里, 从生活上、学习上和工作上你都给予了我无微不至的关心和照顾,同时你严谨的 科研作风、扎实的知识储备以及活跃的科研思路都对我产生了深远的影响。同时 感谢同课题组的Fabian Hflinger博士、 Johannes Wendeberg博士、 Joan Bordoy 博士、Dominik Schott 博士、Joachim Hoppe 硕士以及 Sebastian 硕士等,正因 为有你们在科研和生活中的陪伴, 使我在德国的异乡求学之旅变得特别的丰富多 彩。 感谢我的父母,离家异乡求学十载,对家里面的照顾甚少,聚少离多,你们 对我的要求甚少,但是自始至终给予我无私的关怀、支持和理解,我将永远怀着 万方数据 一颗感恩的心并用我的一切去报答你们。特别感谢我的妻子,在这几年最艰苦的 博士求学生涯中,是你无微不至的照顾和不离不弃的陪伴,始终激励着我奋发前 行,给了我无穷的动力。谨向我的父母、岳父母、妻子以及家人致以最崇高的敬 意和最诚挚的谢意 感谢所有在学习和生活上给予我帮助的人们。 感谢论文中引用文献的作者所 提供的理论和方法。 最后,感谢各位专家和学者在百忙之中审阅我的论文,并提出宝贵的指导意 见和建议。在此,谨向各位专家和学者表示深深的感谢 万方数据 I 摘摘 要要 随着近年来国家能源战略的发展实施, 使得煤炭开采逐步从高产量向绿色开 采、安全高效开采发展。煤矿井下综采工作面的自动化和智能化是实现矿井无人 化、安全高效开采的关键步骤,也是发展“数字矿山”,提高矿井机电装备信息化 和自动化水平的重要组成部分。本文在国家高技术研究发展计划项目的资助下, 以综采工作面“三机”装备采煤机、刮板输送机和液压支架为研究对象,以捷 联惯性导航系统和无线传感器网络定位系统为基础, 以实现井下综采“三机”自动 化为目标,开展采煤机复杂环境下的精确位姿感知技术研究。通过建立采煤机动 力学模型和“三机”运动学模型, 在此基础上研究采煤机复杂振动环境下的捷联惯 导误差补偿策略及多约束定位解算策略, 构建采煤机 SINS/WSN 组合定位模型, 在此基础上分析异类传感器数据传输特性以及井下工作面无线传感器网络定位 影响因素,提出组合定位系统数据异步融合方法以及 WSN 不同失效情况下的紧 耦合容错组合定位方法来实现采煤机复杂环境下的定位定姿。 主要研究工作包括 1)采煤机动力学模型下捷联惯导解算误差补偿策略研究。建立了采煤机的 多刚体动力学模型,并进行数值计算下的采煤机机身振动形式求解。分析了动力 学模型下的采煤机机身振动形式,设计了针对采煤机机身的等效二阶振动模型。 利用等效的机身角振动和线振动特性, 构建了基于四元数多子样算法的捷联惯导 圆锥误差和划船误差补偿模型。 2)采煤机运动学约束下的捷联惯导测量误差校正研究。分析了捷联惯导加 速度计和陀螺仪的测量偏差特性, 提出了基于中值滤波的采煤机静止状态检测方 法,利用采煤机运动前期的静止状态,构建了基于最小二乘法的加速度计和陀螺 仪静态测量偏差校正模型。利用采煤机的行驶速度特性,建立了基于模糊推理机 制的加速度动态测量偏差实时校正模型。 研究了采煤机受刮板机的运动学约束特 性,针对采煤机工作时停车速度为零的特点,构建了基于采煤机运动学辅助下的 捷联惯导零速校正模型。 3)时间异步传输下 SINS/WSN 组合定位策略研究。建立了无线传感器网络 节点部署与定位解算模型,分析了组合定位系统中 SINS 和 WSN 之间的时间配 准误差以及传统同步融合算法下引入的异步数据定位误差, 构建了基于融合周期 的批处理数据伪量测模型,研究了融合周期中 WSN 量测数据采样特性,设计了 基于批处理异步融合与单惯导状态更新自动切换的定位模型, 实现了量测数据时 间异步下基于无迹卡尔曼滤波的组合定位系统融合解算。 4)WSN 失效下采煤机多模型容错组合定位策略研究。研究了 WSN 在定位 过程中粗大定位误差和定位数据丢失的情况,建立了基于 WSN 定位失效情况下 万方数据 II 决策树容错判断策略, 构建了基于卡尔曼滤波的容错组合定位模型。 研究了 WSN 定位过程中出现的部分锚节点测距失效导致无法解算出准确采煤机位置信息的 情况, 利用剩余的准确测距信息建立了基于测距误差校正下的紧耦合模糊自适应 组合定位模型。分析了以上两种容错定位模型在不同 WSN 失效状态下的各自定 位优势,建立了基于卡尔曼滤波残差检测的多模型自动切换定位策略,实现了针 对 WSN 不同失效情况下的自适应容错组合定位。 本论文有图 72 幅,表 5 个,参考文献 154 篇。 关键词关键词综采工作面;采煤机定位定姿;捷联惯导;无线传感器网络;组合定位 万方数据 III AbstractAbstract With the implementation and development of national energy strategy, the coal mining has been made to advances to the green, safety and high efficient mining gradually from high-volume production. Automation and intelligence for fully mechanized coal mining face is one of the key technologies to realize unmanned operation, safety and high efficient mining, which is an important component in improvement of inationization and automation standards for mining machinery. Taking fully mechanized coal mining “three machines” i.e. Shearer, Scraper conveyor and Hydraulic support, as the object, this dissertation researches the high-accuracy position and pose perception technology supported by the National High Technology Research and Development Program of China. On the basis of the Strap-down Inertial Navigation System SINS and Wireless Sensor Network WSN, implementing the automation of “three machines” in the coal mine is the aim of our researches. Through building the dynamic model of shearer and kinematic model of three-machine, SINS error compensation and multi-constraint localization solution have been researched. Moreover, the SINS/WSN integrated positioning system for shearer is built. The data transmission characteristics of different sensors are analyzed. Meanwhile the influence factor of WSN is studied based on the environment of a coal mine. Data asynchronous fusion of integrated positioning system and the fault- tolerant tightly-coupled integrated positioning are proposed based on different failure modes of WSN. Then the position and pose perception of shearer can be achieved with a complex environment of coal mine. The main research works includes 1 Researching on solution error compensation strategy of SINS based on the dynamic model of shearer. After building the multi-rigid-body dynamics model of shearer, the vibration of shearer fuselage is calculated with numerical simulation algorithm. Through analyzing the vibration characteristic of shearer fuselage with a dynamic model, the equivalent second-order vibration model for shearer fuselage is designed. By means of the angular vibration and linear vibration of shearer fuselage, the coning error and sculling error compensation models of SINS are established based on the quaternion multi-sample orientation solution algorithm. 2 Researching on the measured error calibration strategy for SINS based on the kinematic constraint of shearer. After analyzing the measurement biases of accelerometer and gyroscope in SINS, the stationary state detection of Shearer 万方数据 IV is proposed based on the median filter algorithm. Measurement bias calibration models for accelerometer and gyroscope are built with Least Square algorithm at the stationary state before shearer motion. Moreover, according to the velocity characteristic of shearer, the measured acceleration correction model of SINS is established based on the Fuzzy Inference System FIS. Finally, the kinematic constraint characteristic of shearer from the Scraper conveyor has been analyzed. According to the zero-velocity condition for the stop of shearer, the SINS zero velocity updated model aided by kinematic of shearer is obtained. 3 Researching on the integrated positioning strategy of SINS/WSN for the asynchronous data. Firstly, the node deployment model and localization solution model of Wireless Sensor Network are built. Then the time alignment error between SINS and WSN is analyzed and the positioning error of traditional synchronous fusion algorithm for integrated positioning system is studied with asynchronous data. Moreover, the batch processing-based pseudo measurement equation is built with fixed fusion period. According to the measured data of WSN at every fusion period, the auto-changed asynchronous fusion positioning model is designed based on the batch processing . Finally, the improved asynchronous fusion positioning model for the non- alignment data is obtained based on the Unscented Kalman Filter. 4 Researching on the multi-model fault-tolerant integrated positioning strategy of shearer for the failure of WSN. After analyzing the conditions of gross position error and data missed for WSN in the positioning process, the fault-tolerant judgement model with decision tree algorithm is built based on the failure conditions of WSN. Moreover, the fault-tolerant integrated positioning model of SINS/WSN system can be established based on the Kalman Filter algorithm. If some anchor nodes cannot measure the accurate distance between anchor node and mobile node, the WSN will not calculate the accurate position of shearer. Aimed at above condition, the fuzzy adaptive tightly- coupled integration positioning model is built using remaining accurate measured distances based on the calibration for the measured distance error of WSN. Finally through analyzing the advantage of two proposed fault-tolerant positioning models for different failure modes of WSN, the automatic switching multi-model positioning model is established based on the residual detection of Kalman Filter. Then the adaptive fault-tolerant integrated positioning strategy has been achieved for different failure states of WSN. The dissertation has 72 figures, 5 tables and 154 references. 万方数据 V Keywords fully mechanized coal mining face; position and pose perception of shearer; Strap-down Inertial Navigation System; Wireless Seneor Networks; integrated localization 万方数据 VI Extended AbstractExtended Abstract Coal is the important basic energy and raw materials in our country. Meanwhile the energy structure based on coal wont change in the short term. With the development of the national economy, the demand of coal resources is increasing and the coal mine safety accident is becoming more frequent. With the implementation and development of a national energy strategy, coal mining has been made to advances to the green, safety and high efficient mining gradually from high-volume production. The safe efficient development and utilization technology of coal resources is a hotspot issue that has been researched widely by domestic and overseas scholars. One of the most effective solution way is the mechanization and automation exploitation for the mining equipment. Then the unmanned exploitation for the fully mechanized coal mining face can be achieved. Moreover, the ination sensation technology for coal mining face is one of the key technologies to realize three-machine automation, which is an important component in improvement of inationization and automation standards for mining machinery. Taking fully mechanized coal mining as the object, this dissertation researches the high-accuracy position and pose perception technology supported by the National High Technology Research and Development Program of China. On the basis of the Strap- down Inertial Navigation System SINS and Wireless Sensor Network WSN, implementing the automation of “three machines” in the coal mine is the aim of our researches. Through building the dynamic model of shearer and kinematic model of three-machine, SINS error compensation and multi-constraint localization solution have been researched. Moreover, the SINS/WSN integrated positioning system for shearer is built. The data transmission characteristics of different sensors are analyzed, meanwhile the influence factor of WSN is studied based on the environment of coal mine. The data asynchronous fusion of integrated positioning system and the fault-tolerant tightly-coupled integrated positioning are proposed based on different failure modes of WSN. Then the position and pose perception of shearer can be achieved with complex environment of coal mine. The main research work includes 1 Researching on solution error compensation strategy of SINS based on the dynamic model of shearer. After building the multi-rigid-body dynamics model of shearer, the vibration of shearer fuselage is calculated with numerical simulation 万方数据 VII algorithm. Through analyzing the vibration characteristic of shearer fuselage with dynamic model, the equivalent second-order vibration model for shearer fuselage is designed. By means of the angular vibration and linear vibration of shearer fuselage, the coning error and sculling error compensation models of SINS are established based on the quaternion multi-sample orientation solution algorithm. The error compensation models of the shearer are analyzed in simulation under multi-constraint conditions. Simulation results indicate that the coning and sculling errors are composed of different periodical components. The coning errors of the two-sample and non-compensation algorithms diverge. Compared with the non-compensation algorithm, the coning and sculling errors with the four-sample algorithm decrease by 89.4 and 68.7, respectively. Considering the different coal and rock traits such as coal, rock parting and gangue, the coning and sculling errors are 3.934 10-5 deg and 0.112 m/s, respectively, when the shearer cut the coal. When cutting the gangue, the coning and sculling errors increase to 5.987 10-4 deg and 0.246 m/s, respectively, owing to the shock and vibration of the shearer drums. In addition, because of the different vibration frequencies of the shearer, the decreased positioning error with the four-sample algorithm compared with the three-sample algorithm is different, and the decreased percentage can be used to make a reference for error compensation . 2 Researching on the measured error calibration strategy for SINS based on the kinematic constraint of shearer. After analyzing the measurement biases of accelerometer and gyroscope in SINS, the stationary state detection of Shearer is proposed based on the median filter algorithm. The measurement bias calibration models of accelerometer and gyroscope are built with Least Square algorithm at the stationary state before shearer motion. Moreover, according to the velocity characteristic of shearer, the measured acceleration correction model of SINS is established based on the Fuzzy Inference System FIS. Finally, the kinematic constraint characteristic of shearer from the Scraper conveyor has been analyzed. According to the zero-velocity condition for the stop of shearer, the SINS zero velocity updated model aided by kinematic of shearer is obtained. The experimental results show that the positioning parameter of SINS has been corrected with kinematic constraint of shearer, then the positioning trajectory of SINS can track the heading angle of reference trajectory and surpasses the pure SINS with 51.2 improvement. On the basis of the motion constraint for shear