当前位置: 首页>博士论文>资源详情
极端环境下多维力/触觉传感器解耦及信号补偿研究
中文摘要

近年,智能机器人技术受到了学术界和工业界的广泛关注,已成为未来机器人领域发展的主要趋势之一。传感器作为智能机器人感知外界环境的重要媒介,是智能机器人系统中不可缺少的重要部件。其中,多维力触觉传感器不仅可以实现传统的刚性力/力矩检测,也可以实现柔性触觉检测,具有十分重要的研究意义。尤其,在太空等极端环境中,如何解决多维力触觉传感器的冗余操作、大温差下的精密操作以及柔性操作等问题,是航空航天等极端环境中机器人研究的关键问题之一。 本文首先根据传感器制造材料的柔性及刚性特点分别介绍了三维柔性触觉传感器和六维力/力矩传感器(以下用六维力传感器表示)的研究现状,同时也总结了智能算法在多维力触觉传感器的发展现状。然后针对多维力触觉传感器各桥路之间的维间耦合、柔性传感器柔软性、传感器稳定性、温漂等一系列问题,综合运用材料力学、弹性力学、人工智能、传感器技术等学科理论,对传感器容错处理方法、温度补偿以及柔性传感器结构设计、受力分析、解耦算法等方面内容进行了深入的研究。本论文主要研究内容总结如下: 1.提出利用多维刚性力传感器维间耦合信息开展容错解耦算法研究。 基于双E型膜片式六维力传感器的结构,分析了各维桥路之间的耦合关系,以斜率形式定量表示出各维桥路之间的耦合关系的强弱。提出一种基于线性方法的可以舍弃故障桥路的容错解耦方法,以及利用非线性BP神经网络对故障桥路进行信号补偿的容错解耦算法。仿真实验结果表明,利用该算法在指定桥路故障状态下该传感器能够满足测量要求; 2.开展对多维刚性力传感器在大温差环境下的温度补偿方法研究。 基于双E型膜片式六维力传感器应变片布局方式,分析了温度变化对六维力传感器测量精度的影响。在不改变多维力传感器惠斯通全桥电路及采用应变片自补偿法的前提下,通过构建自适应模糊神经网络对传感器进行温度补偿。实验结果表明,该方法明显提高了传感器的稳定性,传感器最大的温度漂移相对误差不超过0.1%。 3.以N型柔性触觉传感器为基础开展解耦研究。 对由导电橡胶制备的N型微结构敏感单元进行了受力分析,通过有限元仿真实验获得基于该种微结构传感器的标定数据,提出了一种基于RBF神经网络的解耦算法,仿真结果证明了以上工作的有效性; 4.具有容错能力的多维柔性触觉传感单元结构设计与其解耦研究。 针对传感器的柔性、冗余性和温度补偿等问题,采用柔性压阻膜片作为传感器敏感元件。设计了一种正四面体结构的传感单元可将三维力分解到该结构对应的各个侧面。对该结构的传感单元进行受力分析,利用有限元分析验证了结构的合理性及容错性,通过设计实验获取标定数据,以BP神经网络对传感器进行非线性解耦研究,实验结果表明该传感器实现了量程范围内的三维力测量。 以上研究成果解决了多维力触觉传感器面临的解耦精度低、柔性传感器不柔软、及稳定性一致性差等问题,提高了多维力触觉传感器系统的稳定性、灵敏度及可靠性。 关键词:多维力触觉传感器,三维柔性触觉传感器,六维力传感器,神经网络,解耦,容错,温度补偿,

英文摘要

Recently, research on intelligence robot technology has attracted wide attentions from the academic circles and the industrial circles. It has become one of major trends of robot development in future. As an important medium for intelligence robot to perceive external environment, sensor occupy the core position in intelligence robot system. Among them, the multi-dimensional force tactile sensor not only can realize the traditional rigid force/torque detection, but also can realize flexible tactile detection, which has very important research significance. Especially, in the extreme environment such as space, how to solve the problem of redundant operation of multidimensional force tactile sensor, precision operation under large temperature difference and flexible operation is one of the key issues in the research of robots in the extreme environment of aerospace. Research status on the multidimensional force tactile sensor was introduced in this paper. Multidimensional force tactile sensors can be divided into three-dimensional flexible tactile sensor and six-dimensional force/torque sensor (hereinafter referred as the six-dimensional force sensor) according flexibility and rigidity of the texture materials. At the same time, we also summarized the development status of intelligent algorithms in multi-dimensional force tactile sensors. This paper is to address a series of problems in application of multidimensional force tactile sensors, such as strong coupling among different dimensions of the sensor bridge circuit, poor flexibility of flexible sensor and poor robustness of rigid sensors. Therefore, structural design, stress analysis, decoupling, fault-tolerance processing and temperature compensation of sensors were explored by combining mechanics of materials, elastic mechanics, artificial intelligence and sensor technology. It lays a foundation for designing a multidimensional tactile sensor with soft stability and excellent fault tolerance. Main research contents are organized as follows. 1.The fault-tolerance decoupling was performed by using coupling information among dimensions of the six-dimensional force sensor. The structure of six-dimensional force sensor with double E-membrane was introduced. The coupling relationship among bridge circuits in the six-dimensional force sensor was analyzed in detail. Coupling strengths among bridge circuits were expressed by slope quantification. The fault-tolerance decoupling by abandoning the fault bridge circuits was proposed based on the linear method. Besides, fault bridge circuit signals of the sensor were compensated and fault-tolerance decoupling was performed by using the nonlinear BP neural network algorithm. Experimental results demonstrated that the coupling intensity can influence the fault-tolerance decoupling capability directly. 2.Research on temperature compensation method for multi-dimensional force sensor under large temperature difference environment The influence of temperature variation on the measurement accuracy of six-dimensonal force sensor was analyzed based on the double E diaphragm type six-dimensional force sensor strain gauge layout method. The adaptive fuzzy neural network is used to compensate the temperature of the sensor by constructing adaptive fuzzy neural network without changing the multidimensional force sensor of Wheatstone full bridge circuit and using the strain gauge self compensation method. The experimental results show that the stability of the sensor is obviously improved and the relative error of the maximum temperature drift of the sensor is not more than 0.1%. 3.Decoupling study was based on N-type flexible sensors The force analysis of N-type microstructure sensitive unit prepared by conductive rubber is carried out. The calibration data based on this kind of microstructure sensor was obtained through finite element simulation experiments. A decoupling algorithm based on RBF neural network was proposed. The simulation results prove the effectiveness of the above work. 4.Research on structure design and decoupling of multi-dimensional flexible tactile sensing unit with fault tolerance. For the flexibility, redundancy and temperature compensation of the sensor, a flexible piezoresistive film was used as a sensor sensor. A sensing element with a regular tetrahedral structure was designed to decompose three-dimensional forces to the corresponding sides of the structure. The force analysis of the sensor unit of the structure is carried out, the use of finite element analysis to verify the structure of the rationality and fault tolerance, through the design of experimental calibration data obtained by BP neural network nonlinear decoupling sensor, the experimental results show that The sensor realizes three-dimensional force measurement within the measuring range. The above research results have solved the problems such as the low measuring accuracy of multi-dimensional force tactile sensors, the inflexibility of flexible sensors, and the poor consistency of stability, which has improved the stability, sensitivity and reliability of multi-dimensional force tactile sensor systems. Keyword: Multi-dimensional force/tactile sensor, Three-dimensional flexible tactile sensor, Six-dimensional force sensor, Neural network, Decoupling, Fault-tolerant, Temperature compensation

作者相关
主题相关
看过该书的人还在看哪些书