Based on the analysis of autonomous underwater vehicle ( auv ) sensor failure , diagnosis methods corresponding to three kinds of sensor failure are presented , that is , through that no sensor output for a long time to diagnose the fault of sensor signal gathered keeping unchanged , linear smoothing to solve the vibration of sensor signal , wavelet transform to inspect the broken sensor signal 摘要在分析水下机器人传感器故障形式的基础上，对传感器可能出现的三种故障形式分别给出了相应的诊断方法，即通过传感器长时间采集不到数据来诊断传感器信息保持不变的故障，线性平滑滤波解决传感器输出振荡，小波变换检测传感器信息的突变。
The core of floating threshold value system is the data sampling technology . we will introduce the key chip - tlc5540 and tlc5615 and utilize mcs51 to control the whole system in the article . it examines the original signal from the hgcdte sensor and realizes the control of floating threshold value 浮动阈值系统的核心是数据采集技术，文中介绍了核心芯片tlc5540和tlc5615 ，利用mcs51单片机控制整个系统，对hgcdte传感器输出的原始信号进行检测，实现了智能浮动阈值控制；同时阐述了性能先进的核心芯片? ? ad603及其应用于agc放大器的特殊性。
The method is as follows : when the video signals from cmos image sensor pass through a high - pass analog filter , the microprocessor samples and processes them and then judge whether the lens is in focus . if the lens is not well focused , the microprocessor will control the lens to be in focus by moving the stepping motor 这一系统采用的方法是通过对cmos图像传感器输出的模拟视频信号进行滤波，提取出代表图像清晰程度的高频成分，选用单片机组成硬件电路，对信号进行处理并分析判别聚焦位置、控制马达转动镜头至对焦清楚。
The signal handling and analyzing unit dispose of the sensor ' s output signal by amplifying , plasticisering , a / d transforming and measuring , then displays it digitally . at the same time , power speech alarm system and speech alarm system , speech alarm system will give alarm , and control the related equipment if it is necessary 信号处理单元对传感器输出的信号进行放大、整形、 a / d转换和测量等处理，并进行数字显示，同时接通语音报警系统电路，语音报警系统进行相应的预警和报警，在必要时对相关设备进行自动控制。
At the beginning of this thesis , the principle of fiber fabry - perot strain sensor is presented briefly , and the influence on the results of the peak - to - peak algorithm , brought by the spectrum distribution of light source , wavelength quantification or the noise in the output of the fabry - perot sensor is investigated . the limitation of the peak - to - peak algorithm is pointed out 论文首先阐述了光纤法珀应变传感器的测量原理，分析了实际光源光谱的非均匀分布、波长量化和传感器输出信号中的噪声等对波长域的条纹峰值解调算法的应变测量结果的影响，深入分析了条纹峰值解调算法的局限性。
The specific route is as follows : firstly , the output signals from the multipath sensor are nornalized , i . e . , all sensor signals ( including the analog data ) are transformed into multipath square wave pulse signals to form multipath parallel condition codes . these codes are treated as input signals for dac to obtain a series of dispersed analog signals for output as input signals for the vco . finally , a high frequency modulation signal is conducted at the vco ' s output 具体技术路线如下：先将由多路传感器输出的信号进行归一化处理，即将所有传感器信号(包括模拟量)转换成多路方波脉冲信号，以形成多位并行信号的状态码，将其作为dac的数据输入信号，从而得到一系列离散的模拟信号输出，作为vco的输入信号，最终在vco输出端形成高频的调制信号。
And then , the error matrixes of location and kinemics of the end point , and the on - line error compensation method are given based on robot ' s dynamics . finally based on puma robot , three simulation examples are given respectively ; the first is about the location error and on - line location error compensation , the second is about the kinetic error and on - line kinetic error compensation , the third is about location and kinetic errors causing by robot ' s dynamics and the on - line error compensations . the simulation results show that : a ) location error of the end point based on elastic deformation of the sensor will be about millimeter ' s degree under the permitting load , b ) the on - line error compensation methods given are available 第三章首先概括了目前机器人连杆惯性参数识别的四种方法，总结这些方法的优、缺点；指出这些方法存在的问题是：或者需将机器人解体，不能在线进行参数识别，或者不能给出机器人连杆独立的惯性参数值，只能获得惯性参数的组合值，而这些方法的共同问题是：不能考虑机器人连杆的关节特性；本章提出了一种基于腕力传感器的机器人末端连杆惯性参数在线识别方法，给出了该方法的理论计算和推导；研究提出了以腕力传感器输出为前提的、基于newton - euler动力学的机器人动力学正向、逆向递推公式；针对机器人负载参数辨识必须在线、实时的特点提出了基于腕力传感器的负载参数在线识别方法，给出了负载参数识别的步骤。
By studying the characteristics of the eddy displacement sensor , it adopts the control method which include using the fitting line of sensor ' s output characteristic curve in certain block as control characteristic curve , and stipulating door limit district and a certain of backward overshoot , the stabilization of welding torch is gained under the certain track precision 摘要通过对涡流式位移传感器的特性研究，采用以传感器输出特性曲线在一定区间内的拟合直线作为控制特性曲线、设定规定门限区和一定反向调节量的控制方法，得到了在保证一定的跟踪精度条件下的焊枪稳定。
If we concatenate the recognition results of each sensors and consider it as a joint feature vector of target , then decision level fusion for target recognition can be considered as a conventional pattern recognition problem . this dissertation uses multilayer perceptron to classify the joint feature vector 将各传感器输出的识别结果连接起来作为目标的一个联合特征矢量，则决策层融合目标识别可看作是一个常规的模式识别问题，本文采用多层感知器网络对这个联合特征矢量进行分类，实现决策层融合目标识别。
