基于改进注意力机制CNN-ATT的区域性ZTD预测模型

基于天顶对流层延迟（ZTD）的强时空特征,提出了一种融合卷积神经网络的改进注意力机制（CNN-ATT）的多站点ZTD组合预测模型。该模型首次将多源数据（包括日解算精度、年积日(DOY) 和三维坐标）综合运用于ZTD预测任务。通过对南宁市的5个参考站（CORS）和14个国际GNSS服务（IGS）站点共1 501个年积日的观测数据进行研究,选取传统BP模型、GPT2w模型和ATT模型作为基线模型进行实验对比分析。研究结果显示,在预测精度方面,改进的CNN-ATT模型与BP模型相比其均方误差（MSE）和平均绝对误差（MAE）分别减少了5.5 mm和 4.4 mm,预测精度分别提高了41.4%和67.8%;与ATT模型相比,CNN-ATT模型的预测MSE和MAE也分别减少了4.6 mm和2.1 mm,预测精度分别提升了36.2%和50.0%。在定位精度方面,改进的CNN-ATT模型的精度表现优于SAAS,GPT2w,BP以及ATT模型。并且与传统SAAS对流层模型相比,CNN-ATT模型在N,E,U 3个方向的精度提升高达18.2%,12.6%和31.0%。此外,研究还发现CNN-ATT模型在长预测时间步长中的精度表现更为稳定,更适合多测站预测任务,并且其精密单点定位（PPP）收敛速度更快。     

Robust fault-tolerant attitude control for satellite with multiple uncertainties and actuator faults

In this paper, the satellite attitude control system subject to parametric perturbations, external disturbances, time-varying input delays, actuator faults and saturation is studied. In order to make the controller architecture simple and practical, the closed-loop system is transformed into a disturbance-free nominal system and an equivalent disturbance firstly. The equivalent disturbance represents all above uncertainties and actuator failures of the original system. Then a robust controller is proposed in a simple composition consisting of a nominal controller and a robust compensator. The nominal controller is designed for the transformed nominal system. The robust compensator is developed from a second-order filter to restrict the influence of the equivalent disturbance. Stability analysis indicates that both attitude tracking errors and compensator states can converge into the given neighborhood of the origin in finite time. To verify the effectiveness of the proposed control law, numerical simulations are carried out in different cases. Presented results demonstrate that the high-precision attitude tracking control can be achieved by the proposed fault-tolerant control law. Furthermore, multiple system performances including the control accuracy and energy consumption index are fully discussed under a series of compensator parameters.     

卫星精密测定轨中的对流层延迟估算系统

基于V isual Basic编程,设计了卫星精密测定轨中的实时气象采集与对流层延迟估算系统。利用实时采集ZQZ-A型自动气象站的温度、相对湿度、大气压等气象要素,采用Saastamoinen模型进行估算天顶方向对流层静力延迟和湿延迟,同时采用地基双波段微波辐射计测量大气水汽总量和云液态水总量,进而估算更高精度的湿延迟,在其有效时与模型计算的静力延迟相加,得到天顶方向总延迟;无效时采用模型计算的总延迟。最后将其投影到卫星方向得到总延迟。实际应用表明,设计的对流层延迟估算系统人机界面友好,具有易用性,能实时、稳定地为卫星精密测定轨提供服务。     

不确定性因素下航班延误波及建模与分析

针对机场终端区日益严重的航班延误,提出了一种基于有色－时间Petri网的不确定性因素下航班延误波及分析模型。模型综合考虑航班运行尾流、跑道运行模式、跑道占用时间等运行安全间隔约束,同时根据连续航班间的信息更新,设计进离场航班序列动态调整流程,并由计算模型可得出航班延误总架次及时间。最后,依据历史延误数据分析得出不确定性因素延误的概率分布。选取上海浦东机场为背景对模型进行仿真验证,仿真结果表明,所构建模型可用于战略阶段的航班延误波及程度与过程的评估。     

Carbon monoxide spatial gradients over source regions as observed by SCIAMACHY: A case study for the United Kingdom

Carbon monoxide (CO) is an important air pollutant whose emissions and atmospheric concentrations need to be monitored. The measurements of the SCIAMACHY instrument on ENVISAT are sensitive to CO concentration changes at all atmospheric altitude levels including the boundary layer. The SCIAMACHY CO measurements therefore contain information on CO emissions. Until now no studies have been published where the SCIAMACHY CO measurements have been used to quantify CO emissions by applying, for example, inverse modelling approaches. Here we report about a step in this direction. We have analysed three years of CO columns to investigate if spatial gradients resulting from United Kingdom (UK) CO emissions can be observed from space. The UK is an interesting target area because the UK is a relatively well isolated CO source region. On the other hand, the UK is not the easiest target as its emissions are only moderate and because the surrounding water has low reflectivity in the 2.3 μm spectral region used for CO retrieval. We determined horizontal CO gradients from seasonally and yearly averaged CO during 2003–2005 over the UK taking into account daily wind fields. We show that the measured CO longitudinal (downwind) gradients have the expected order of magnitude. The estimated 2σ error of the gradients depends on time period and applied filtering criteria (e.g., land only, cloud free) and is typically 10–20% of the total column. The gradients are barely statistically significant within the 2σ error margin. This is mainly because of the relatively high noise of the SCIAMACHY CO measurements in combination with a quite low number of measurements (∼100) mainly due to cloud cover.     

滞后不确定组合大系统的鲁棒控制

本文讨论了具有时滞互联不确定组合大系统的镇定问题。这种不确定性是时变的,它可作为不确定参数及输入摄动存在于每个子系统中,也可能存在于互联项中。不确定性在假设可能知道它的边界的情况下,利用本文构造的线性分散状态反馈控制律,系统是实际可稳定的,即系统可镇定。     

A new weighted mean temperature model in China

The Global Positioning System (GPS) has been applied in meteorology to monitor the change of Precipitable Water Vapor (PWV) in atmosphere, transformed from Zenith Wet Delay (ZWD). A key factor in converting the ZWD into the PWV is the weighted mean temperature ($T_m$), which has a direct impact on the accuracy of the transformation. A number of Bevis-type models, like $T_m-T_s$ and $T_m-T_s\text{,}{P}_s$ type models, have been developed by statistics approaches, and are not able to clearly depict the relationship between $T_m$ and the surface temperature, $T_s$. A new model for $T_m$, called weighted mean temperature norm model (abbreviated as norm model), is derived as a function of $T_s$, the lapse rate of temperature, δ, the tropopause height, $h_{\mathit{trop}}$, and the radiosonde station height, $h_s$. It is found that $T_m$ is better related to $T_s$ through an intermediate temperature. The small effects of lapse rate can be ignored and the tropopause height be obtained from an empirical model. Then the norm model is reduced to a simplified form, which causes fewer loss of accuracy and needs two inputs, $T_s$ and $h_s$. In site-specific fittings, the norm model performs much better, with RMS values reduced averagely by 0.45 K and the Mean of Absolute Differences (MAD) values by 0.2 K. The norm model is also found more appropriate than the linear models to fit $T_m$ in a large area, not only with the RMS value reduced from 4.3 K to 3.80 K, correlation coefficient $R^2$ increased from 0.84 to 0.88, and MAD decreased from 3.24 K to 2.90 K, but also with the distribution of simplified model values to be more reasonable. The RMS and MAD values of the differences between reference and computed PWVs are reduced by on average 16.3% and 14.27%, respectively, when using the new norm models instead of the linear model.     

深空干涉测量对流层经典模型改进

对流层延迟修正误差是深空干涉测量的重要误差源之一。通过修正对流层经典天顶延迟模型和Niell映射函数NMF（Niell Mapping Function）构建了一种高精度区域对流层延迟模型。首先，结合我国深空网喀什深空站对流层延迟实测数据，对Saastamoinen模型的适用性进行分析，通过线性最小二乘拟合修正天顶干延迟参数，模型精度相对改善29.6%；然后，针对NMF低仰角情况下映射偏差较大的问题，构建偏差函数模型，显著改善了低仰角下的映射性能，经实测数据验证：仰角在0°~30°时，对流层延迟模型偏差相对改善约30%。改进后的对流层区域模型估计精度高，可为我国深空干涉测量对流层延迟修正提供参考。     

Locally generalised multi-agent reinforcement learning for demand and capacity balancing with customised neural networks

Reinforcement Learning (RL) techniques are being studied to solve the Demand and Capacity Balancing (DCB) problems to fully exploit their computational performance. A locally generalised Multi-Agent Reinforcement Learning (MARL) for real-world DCB problems is proposed. The proposed method can deploy trained agents directly to unseen scenarios in a specific Air Traffic Flow Management (ATFM) region to quickly obtain a satisfactory solution. In this method, agents of all flights in a scenario form a multi-agent decision-making system based on partial observation. The trained agent with the customised neural network can be deployed directly on the corresponding flight, allowing it to solve the DCB problem jointly. A cooperation coefficient is introduced in the reward function, which is used to adjust the agent’s cooperation preference in a multi-agent system, thereby controlling the distribution of flight delay time allocation. A multi-iteration mechanism is designed for the DCB decision-making framework to deal with problems arising from non-stationarity in MARL and to ensure that all hotspots are eliminated. Experiments based on large-scale high-complexity real-world scenarios are conducted to verify the effectiveness and efficiency of the method. From a statistical point of view, it is proven that the proposed method is generalised within the scope of the flights and sectors of interest, and its optimisation performance outperforms the standard computer-assisted slot allocation and state-of-the-art RL-based DCB methods. The sensitivity analysis preliminarily reveals the effect of the cooperation coefficient on delay time allocation.