无人直升机传感器故障诊断与容错控制方法研究
发布时间：2019-04-09 12:49
【摘要】：当前,无人机发展如火如荼,无人直升机由于能够垂直起降、能够低速飞行、定点悬停的优势应用更加广泛,发展势头更加迅猛。国际国内各大科研院机构争相研发新型无人直升机,投入到民用或者军用。但是由于无人直升机平台的具有强干扰、强振动等恶劣的工作环境。其飞行控制器对于传感器的要求较高。一旦发生传感器故障,会导致无人直升机的误控制甚至失控坠机。所以就出现了针对无人直升机飞行控制系统的传感器故障诊断与容错控制。无人直升机飞行控制系统的传感器故障诊断与容错控制有很多种方法,目前,最常用的就是传感器多余度设计,一旦有传感器故障就用余度传感器代替故障传感器,实现容错控制。由于无人直升机的成本和载重有限,这样多余度故障诊断与容错控制系统就使其增加了成本,降低了任务能力。所以本论文提出了一种新的基于解析模型的故障诊断与容错控制方法。该方法只需要一套传感器就可以实现,大大降低了无人直升机的成本。主要的研究过程如下:首先针对某型无人直升机建立数学模型,为后面仿真验证该故障诊断与容错控制方法做准备。又分析了无人直升机IMU模块传感器的工作原理,分析并建立了IMU模块传感器的常见故障模型。进一步提出无人直升机飞行控制系统传感器故障诊断与容错控制方法:利用所建立的无人直升机模型,设计全维主观测器和针对每个输出的降维观测器,实现对无人直升机模型的解耦输出。求出全维主观测器与无人直升机模型输出的主残差,以及各个降维观测器解耦输出与无人直升机模型输出的次残差。先应用序贯概率比准则判断该主残差,确定系统是否发生故障。一旦确定发生故障,就对比次残差,此时会发现只有发生故障传感器所对应的那个降维观测器的输出次残差变化非常明显,其他次残差基本为零。这样就确定了故障的传感器,此时立即隔离故障传感器,用其他传感器输出的加权值(重构信号)来代替故障传感器的输出值。然后将重构信号反馈给无人直升机飞行控制器。无人直升机飞行控制算法采用滑模变结构控制算法,这样会针对每一种传感器故障模型切换到相应的滑模面进行控制,实现更好的容错控制。最后利用Matlab的GUI环境建立了一个简单的仿真系统,来完成该算法的仿真验证。仿真时,分别假设各姿态角传感器发生了故障,以此验证该方法的有效性和实用性。仿真结果表明该传感器故障诊断与容错控制方法可行,且具有比较高的实时性。
[Abstract]:At present, the development of UAV is in full swing, unmanned helicopter can take off and land vertically, can fly at low speed, the advantages of fixed-point hovering are more widely used, and the momentum of development is more rapid. Research institutes at home and abroad are scrambling to develop new unmanned helicopters for civilian or military use. However, due to the unmanned helicopter platform has strong interference, strong vibration and other harsh working environment. Its flight controller has higher requirements for sensors. In the event of sensor failure, the unmanned helicopter will be miscontrolled or even out of control. Therefore, sensor fault diagnosis and fault tolerant control for unmanned helicopter flight control system appear. There are many methods of sensor fault diagnosis and fault-tolerant control in unmanned helicopter flight control system. At present, the most commonly used method is the redundancy design of sensors. If there is a sensor fault, the redundant sensor is used to replace the fault sensor. The fault-tolerant control is realized. Because the cost and load of unmanned helicopter are limited, the redundant fault diagnosis and fault tolerant control system increases the cost and reduces the mission capability. Therefore, a new fault diagnosis and fault tolerant control method based on analytic model is proposed in this paper. This method can be realized only by a set of sensors, which greatly reduces the cost of unmanned helicopter. The main research process is as follows: firstly, a mathematical model for a certain unmanned helicopter is established to prepare for the later simulation verification of the fault diagnosis and fault-tolerant control method. The working principle of IMU module sensor of unmanned helicopter is analyzed and the common fault model of IMU module sensor is established. Furthermore, a sensor fault diagnosis and fault tolerant control method for unmanned helicopter flight control system is proposed. Based on the model of unmanned helicopter, a full-dimensional master observer and a reduced-order observer for each output are designed. The decoupling output of unmanned helicopter model is realized. The main residuals of the output of the full-dimensional master observer and the unmanned helicopter model, and the sub-residual of each reduced-order observer decoupling output and the unmanned helicopter model output are calculated. First, the sequential probability ratio criterion is used to judge the principal residual to determine whether the system is faulty or not. Once the fault is determined, the sub-residuals are compared, and it is found that only the output sub-residuals of the reduced-order observer corresponding to the fault sensor are very obvious, and the other sub-residuals are basically zero. In this way, the fault sensor is determined, the fault sensor is isolated immediately, and the output value of the fault sensor is replaced by the weighted value (reconstruction signal) of the output of the other sensors. Then the reconstructed signal is fed back to the unmanned helicopter flight controller. The flight control algorithm of unmanned helicopter adopts sliding mode variable structure control algorithm, which can control each sensor failure model to the corresponding sliding mode surface and realize better fault tolerant control. Finally, a simple simulation system based on the GUI environment of Matlab is established to complete the simulation verification of the algorithm. In the simulation, each attitude angle sensor is assumed to fail, so as to verify the effectiveness and practicability of the proposed method. The simulation results show that the sensor fault diagnosis and fault-tolerant control method is feasible and has high real-time performance.
【学位授予单位】：南昌航空大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：V279
，
本文编号：2455199
[Abstract]:At present, the development of UAV is in full swing, unmanned helicopter can take off and land vertically, can fly at low speed, the advantages of fixed-point hovering are more widely used, and the momentum of development is more rapid. Research institutes at home and abroad are scrambling to develop new unmanned helicopters for civilian or military use. However, due to the unmanned helicopter platform has strong interference, strong vibration and other harsh working environment. Its flight controller has higher requirements for sensors. In the event of sensor failure, the unmanned helicopter will be miscontrolled or even out of control. Therefore, sensor fault diagnosis and fault tolerant control for unmanned helicopter flight control system appear. There are many methods of sensor fault diagnosis and fault-tolerant control in unmanned helicopter flight control system. At present, the most commonly used method is the redundancy design of sensors. If there is a sensor fault, the redundant sensor is used to replace the fault sensor. The fault-tolerant control is realized. Because the cost and load of unmanned helicopter are limited, the redundant fault diagnosis and fault tolerant control system increases the cost and reduces the mission capability. Therefore, a new fault diagnosis and fault tolerant control method based on analytic model is proposed in this paper. This method can be realized only by a set of sensors, which greatly reduces the cost of unmanned helicopter. The main research process is as follows: firstly, a mathematical model for a certain unmanned helicopter is established to prepare for the later simulation verification of the fault diagnosis and fault-tolerant control method. The working principle of IMU module sensor of unmanned helicopter is analyzed and the common fault model of IMU module sensor is established. Furthermore, a sensor fault diagnosis and fault tolerant control method for unmanned helicopter flight control system is proposed. Based on the model of unmanned helicopter, a full-dimensional master observer and a reduced-order observer for each output are designed. The decoupling output of unmanned helicopter model is realized. The main residuals of the output of the full-dimensional master observer and the unmanned helicopter model, and the sub-residual of each reduced-order observer decoupling output and the unmanned helicopter model output are calculated. First, the sequential probability ratio criterion is used to judge the principal residual to determine whether the system is faulty or not. Once the fault is determined, the sub-residuals are compared, and it is found that only the output sub-residuals of the reduced-order observer corresponding to the fault sensor are very obvious, and the other sub-residuals are basically zero. In this way, the fault sensor is determined, the fault sensor is isolated immediately, and the output value of the fault sensor is replaced by the weighted value (reconstruction signal) of the output of the other sensors. Then the reconstructed signal is fed back to the unmanned helicopter flight controller. The flight control algorithm of unmanned helicopter adopts sliding mode variable structure control algorithm, which can control each sensor failure model to the corresponding sliding mode surface and realize better fault tolerant control. Finally, a simple simulation system based on the GUI environment of Matlab is established to complete the simulation verification of the algorithm. In the simulation, each attitude angle sensor is assumed to fail, so as to verify the effectiveness and practicability of the proposed method. The simulation results show that the sensor fault diagnosis and fault-tolerant control method is feasible and has high real-time performance.
【学位授予单位】：南昌航空大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：V279
，
本文编号：2455199
本文链接：https://www.wllwen.com/kejilunwen/hangkongsky/2455199.html
