本文以实现移动小型智能化系统的实时自主定位为目标,针对激光里程计误差累计大,旋转估计不稳定,以及观测信息利用不充分等问题,提出一种LiDAR/IMU紧耦合的实时定位方法—Inertial-LOAM.数据预处理部分,对IMU数据预积分,降低优化变量维度,并为点云畸变校正提供参考.提出一种基于角度图像的快速点云分割方法,筛选结构性显著的点作为特征点,降低点云规模,保证激光里程计的效率;针对地图构建部分存在的地图匹配点搜索效率低和离散点云地图的不完整性问题,提出传感器中心的多尺度地图模型,利用环形容器保持地图点恒定,并结合多尺度格网保证地图模型中点的均匀分布.数据融合部分,提出LiDAR/IMU紧耦合的优化方法,将IMU和LiDAR构成的预积分因子、配准因子、闭环因子插入全局因子图中,采用基于贝叶斯树的因子图优化算法对变量节点进行增量式优化估计,实现数据融合.最后,采用实测数据评估Inertial-LOAM的性能并与LeGO-LOAM,LOAM和Cartographer对比.结果表明,Inertial-LOAM在不明显增加运算负担的前提下大幅降低连续配准误差造成的误差累计,具有良好的实时性;在结构性特征明显的室内环境,定位精度达厘米级,与对比方法持平;在开阔的室外环境,定位精度达分米级,而对比方法均存在不同程度的漂移.

超宽带UWB(Ultra Wide Band)技术是一种新兴的室内定位技术,具有功耗低、定位精度高、穿透能力强等特点,被广泛研究.利用到达时间差TDOA(Time Difference of Arrival)测量方法,通过UWB标签向多个基站发送信号得到标签到各个基站的时间戳差,利用Chan算法实时解算标签的运动位置.建立"当前"统计模型,设置阈值改进自适应卡尔曼滤波迭代过程,利用改进后的自适应卡尔曼滤波算法消除非视距误差和噪声对TDOA定位的影响,实现了更准确的TDOA室内定位.实验表明,基于自适应卡尔曼滤波的TDOA定位方法具有较高的精度.

同时定位与地图构建(SLAM)系统在机器人、自动驾驶和移动测量等领域有广泛的应用,但目前系统精度的测试方法尚不完善.针对SLAM系统精度测试的实际需求,本文提出了一种SLAM高精度测试场的建立方法.首先建立三维控制网,然后利用三维激光扫描技术对环境信息进行采集,最终构建出用于测试SLAM系统精度的高精度三维点云特征库.试验结果表明,建立的三维点云特征库包含丰富的特征信息,具有真实地理坐标,精度达到毫米级,可以较好地满足SLAM精度测试的需求.

A semi-fuzzy c-means algorithm based on revised Euclidean distance was proposed to improve the real-time capability and precision in fault pattern classified. Effectiveness of threshold parameters on clustering was investigated, and then program steps were given. The example of fault diagnosis in an airborne fire control system BIT was developed. The results show that the new algorithm can recognize fault pattern adaptively and precisely.

To improve the accuracy of air-to-surface multi-target attack,the priority decision model of air-to-surface multi-target attack under uncertainty is researched.Uncertainty factors,which affect the decision of air-to-surface multi-target attack,are carefully analyzed and an interval-based method is proposed to describe the uncertainty of attributes.A multi-attribute value tree is used to construct the multifactor cooperative priority decision model.The RICH(Rank inclusion in criteria hierarchies) method is applied to solve the model.Simulation results of 2 vs.4 air-to-surface multi-target attacks demonstrate that the cooperative priority decision model proposed here is effective and can give reasonable decision support for air-to-surface cooperative multi-target attack.

In order to correct evaluation of radar system performance,the preference programming is introduced to the radar system performance evaluation.The multi-level value tree model is used to establish model,and the algorithm of RICH is extended to the solution of the multi-level value tree.The weight algorithm based on multi-level value tree is presented.Finally,the method is verified with a simulation example for synthetic performance evaluation of radars with uncertain information.The simulation results show that the method is effective,the radar performance under uncertain information can be ranked by it.

According to the feature of airborne mobile platforms with multiple sensors,the significance of multi-sensor target assignment in fusion system of airborne mobile platforms was introduced.Then the shortages of current assignment algorithm were analyzed.A mathematical model of the sensor distribution was established with consideration of priority of target and cooperation of sensors.Then the model based on genetic particle swarm optimization algorithm was built up.Simulation was made and the result showed that this new target assignment method is effective,which has higher stability and faster convergence rate.

In this paper,according to the feature of multi-aircraft cooperative combat sensor platform the overall structural model of the multi-aircraft cooperative combat sensor platform composed of MAS(Multi-Agent System)is built up together with the structural model of the single agent based on the BDI(Belief,Desire,Intention)architecture.A study is conducted on sensor Agent definition,operation principle and process.An effective method is provided for the establishment of intelligent multi-sensor system.

针对单处理器嵌入式系统软硬件划分问题,采用带权有向无环图进行建模,并将之约简,进而转换为多约束条件的0/1背包问题求解.由于基本粒子群优化算法无法求解0/1背包问题,故将遗传算法中的交叉、变异思想引入粒子群优化算法,提出了求解离散组合优化问题的遗传粒子群优化（GPSO）算法,采用两点交叉算子和非均匀变异算子对粒子的位置和速度更新方法进行了重新定义.实验结果表明,采用文中算法能有效地解决软硬件划分问题,具有良好的全局搜索能力,其寻优能力和执行时间优于遗传算法和模拟退火算法.

Plug and play weapon system was analyzed and modeled based on AADL,and the modeling of its subsystem was described taking weapon management computer system as an example.Considering the characteristics of modeling based on AADL,the mission data exchange format and structure,periodic task scheduling algorithm,sporadic task scheduling algorithm and mixed task scheduling algorithm based on server were investigated.Plug and play weapon system was modeled and analyzed by using OSATE software.The result showed that this method can satisfy the modeling of the plug and play weapon system.

An optimum particle filter method based on artificial fish-swarm algorithm was proposed to deal with ballistic missile tracking problem.Swarm intelligence was introduced into particle filtering to solve its shortcomings of particle weight degradation and large requirement to particles.In this paper,the principle of particle filter and the problem existed were described.Then artificial fish-swarm algorithm was combined with particle filter to make the particle swarm move to the region of high likelihood,thus to solve the problem of particle weight degradation.Experimental results are given to demonstrate the feasibility and effectiveness of the proposed approach.

Numerical simulations for the aerodynamic characteristics of near space hypersonic projectile have been computed using the Spalart-Allmaras (SA) turbulence model and thermally perfect model, a time-dependent, finite-volume numerical method and AUSM~+ scheme are adopted to solve the three dimensional Navier-Stokes equation. Numerical solutions of hypersonic projectile flow-field with attack angle are obtained, and the numerical results can be used in the further study of the near space hypersonic projectile and missile.

Based on the status quo and development of the airborne storage management system(SMS),three kinds of simulation systems are designed in this paper: the control algorithm simulation,the hardware-in-the-loop simulation and the simulation of the standard interface weapons SMS,whose functions and top design schemes are expounded in detail.The constructions and uses of those simulation systems will promote the further research on SMS.