硅微谐振式加速度计是利用微机械振梁的谐振频率随施加在其上的拉压力而变化的机理，来测量微质量块所敏感的加速度。与传统加速度计相比，硅微谐振式加速度计工作在谐振状态，具有低成本、小型化与高精度的特点，符合高精度武器装备的苛刻需求。本文以高精度硅微谐振式加速度计为研究对象，为了实现优于10μg的零偏稳定性与10ppm的标度因数稳定性，从微机械结构、闭环噪声、驱动检测电路、温漂抑制等方面进行了研究，论文具体研究内容如下：对硅微谐振式加速度计的关键结构进行了理论分析，针对谐振梁的非线性振动、双侧谐振器耦合自锁进行了理论推导与仿真，并给出了整体闭环方案。对硅微谐振式加速度计的结构进行了设计与优化，在课题目标与国内加工工艺的条件下，以第一版结构Y1为例，分别从总体设计流程、关键结构设计与优化、低温漂结构的设计、耦合自锁优化等方面进行了分析与参数设计，实现了64Hz/g的标度因数及单梁?1.14Hz/℃的频率温度系数；根据Y1的实验效果，再次对标度因数与低温漂结构进行了改进，设计了第二版结构Y2，实现了96Hz/g的标度因数及单梁?0.57Hz/℃的频率温度系数，该结构大幅降低了对热膨胀系数不匹配引起的热应力的敏感度，谐振频率与温度具有良好的线性度。为了减小闭环噪声和回路延迟，对硅微谐振式加速度计进行了系统优化。首先对结构进行真空封装，降低了机械噪声，提高了谐振器的品质因数，并提出了一种基于非线性模型估算品质因数的方法，以此监测真空封装的可靠性；其次根据非线性理论推导出的闭环最优工作点，并结合闭环方案的改进，大幅减小了非线性振动，实现了约0.5μg的加速度测量噪声；最后对驱动检测电路进行了详细分析与优化，实现了48pm/ 的振幅检测噪声，并将闭环回路的延迟从10μs减小到0.22μs。通过理论分析、仿真与实验建立了硅微谐振式加速度计的温漂模型，分别从结构、非线性振动、闭环相位、封装粘片方面对其进行了优化。其中，Y1结构由于差分频率的温度系数较大，采用高精度测温电路进行温度补偿，实现了低于5μg的零偏稳定性；Y2结构采用了一种可隔离粘片应力的H形玻璃基底，使差分频率的温度系数降为5mHz/℃，在无温度补偿的情况下，实现了低于2μg的零偏稳定性。最后在室温（22±1℃）下对研制的硅微谐振式加速度计进行了性能测试。Y1结构实现了5ppm的标度因数稳定性、4.4μg的阈值、5.2μg的零偏稳定性；Y2结构实现了1.6μg的分辨率和低于3μg的零偏稳定性。

Micromechanical Silicon Resonant Accelerometer (MSRA) measures the detected acceleration of the micro proof mass, based on the principle of resonant frequency shift with the tension force and compression force on the micro vibrating beam. Compared with the traditional accelerometer, MSRA works in resonant state with characteristic of low lost, miniaturization and high precision which meet the demand of high precision weapons. The dissertation takes the high precision MSRA as research content. In order to achieve the performance of better than 10μg zero bias stability and 10ppm scale factor stability, the dissertation researches from the aspects of the micro mechanical structure, the closed-loop noise, the drive and sense circuit, the temperature drift suppression. The detailed research content is as follows:Analyze the theoretical model of the key structure of the MSRA. The theoretical derivation and simulation of nonlinear vibration and self-locking of coupled resonators are carried out, and the overall closed-loop scheme is given.Design and optimize the mechanical structure of the MSRA. At the condition of research target and domestic processing capacity, the first structure edition Y1 analyzes and designs the key structure parameters from the aspects of overall design process, key parameters optimization, low temperature drift and self-locking optimization. The edition Y1 realizes the scale factor of 64Hz/g, and its temperature drift coefficient of single beam frequency is ?1.14Hz/℃. According to the experimental results of Y1, the second structure edition Y2 improves the scale factor and low temperature drift again. The edition Y2 realizes the scale factor of 96Hz/g and the temperature drift coefficient of ?0.57Hz/℃. The edition Y2 is almost insensitive with the thermal stress caused by the expansion coefficient. The resonant frequency has a good linearity with temperature.In order to reduce the closed-loop noise and loop delay, the dissertation optimizes the system of MSRA. First, the vacuum encapsulation of the structure is used to reduce the mechanical noise and improve the quality factor of the resonator. Propose a method of estimating the quality factor based on the nonlinear model. The method can monitor the reliability of the vacuum. Second, deduce the optimal working point according to the nonlinear theory. Combined with the improvement of closed-loop scheme, the system reduces nonlinear vibration obviously and realizes the acceleration measurement noise less than 0.5μg. At last, analyze and optimize the drive and sense circuit. The circuit implements the amplitude detection noise of 48pm/ , and the reduced closed-loop delay from 10μs to 0.22μs.Establish the temperature drift model of the MSRA through theoretical analysis, simulation and experiment. Optimize the temperature drift model from the aspects of structure, nonlinear vibration, the closed-loop phase and package bonding. The edition Y1 adopts temperature compensation by high precision temperature measuring circuit. It realizes the zero bias stability of less than 5μg. The edition Y2 uses H-shaped glass substrate which can isolate the stress of package bonding. The temperature coefficient of differential frequency is reduced to 5mHz/°C. Without temperature compensation, the edition Y2 realizes the zero bias stability of less than 2μg.Test the performance of the MSRA at room temperature (22±1℃). The edition Y1 realizes the scale factor stability of 5ppm, the threshold of 4.4μg, and the zero bias stability of 5.2μg. And the edition Y2 realizes the resolution of 1.6μg and the zero bias stability of less than 3μg.