规模数据支持的飞机油耗监控方法研究

发布时间：2018-01-26 19:20

本文关键词： 航空运输飞机油耗神经网络 RELAX算法参数估计　出处：《中国民航大学》2015年硕士论文　论文类型：学位论文

【摘要】：高效的燃油利用对我国提倡建设“资源节约型”和“环境友好型”社会具有重要意义。在民航业领域,提高燃油利用效率也是每个航空公司在飞机运行管理中的重要目标之一。但由于航线运行复杂化、多样化的特点,航空公司的飞机燃油效率的评价很难通过监控飞机的单个或某几个指标来衡量。需要针对飞行运输过程配置多点数据监控。这样使监控数据十分庞杂,影响评价指标的确定。本文以航空公司机队大量运行数据为基础,侧重于从飞机油耗监控评价体系方法中的监控指标确定、油耗性能模型设计、基准线与边界线设计等方面,建立合理的监控方法体系,为我国航空公司飞行机队的油耗监控提出适用性方法,并进行方法学的构建探索。首先,分析课题的背景及意义,提出在数据资源急速扩张的时代,航空公司需要充分利用飞行运行数据开展各方面的研究,其中飞行油耗类数据分析是重要的一个环节。重点综述国内外航空企业在油耗监控方法及其评价方法的应用与研究,分析若干航空公司的油耗监控系统的特点,并对航空公司油耗及其影响因素进行理论介绍。其次,提出一套基于规模数据的飞机油耗监控方法理论,细化分析过程,构建需求分析框架,借鉴CDM方法学进行油耗监控方法学的构建尝试,通过指标选取原则优选几类指标。构造一种基于飞行距离的指数衰减型油耗性能模型,该类模型可分类不同油耗模式并界定性能与距离的参数。针对当前学科领域对于飞机油耗模型所采取的研究方法措施进行分类总结,结合油耗评价指标方法分析各类模型适用性,选择适合航空公司油耗评价监控的方法。再次,确定油耗性能模型的监控评价指标基准线与边界线方法,使用人工神经网络、信号分离理论与最小二乘三类方法进行标准确定。重点介绍了基于已建立的油耗模型的信号分离RELAX算法与非一致性检测广义内积综合的一种交叉迭代优化计算方法,该方法能够在强“噪声”下有效估计基准线,并合理确定油耗模型的输入和输出数据。最后对比三类方法的优缺点,提出针对不同油耗模型的方法优选,方法验证正确,结果有一定参考性。最后使用三类方法对基于距离的油耗性能评价模型进行实例验证与单因素模型构建验证。根据某航空公司B737航班550架次航班数据与A330飞机的8430架次航班数据进行归类建模分析,分析研究算法的噪声性能、拟合性能。综合三类方法的优缺点进行总体模型估计和分类模型估计,为航空公司油耗监控方法研究提出有效方法。通过单因素分析和数据对比补充完善并验证了油耗监控方法体系。总结各类方法的优缺点,提出对工作的总结与展望。
[Abstract]:Efficient fuel utilization is of great significance to the construction of "resource-saving" and "environment-friendly" society in our country, which is in the field of civil aviation. Improving the efficiency of fuel utilization is also one of the important objectives of each airline in the aircraft operation management, but because of the complexity of the route operation, the characteristics of diversification. The evaluation of aircraft fuel efficiency of airlines is difficult to be measured by a single or several indicators of monitoring aircraft. It is necessary to configure multi-point data monitoring for the flight transportation process, which makes the monitoring data very complex. Based on a large number of airline fleet operation data, this paper focuses on the determination of monitoring indicators and the design of fuel consumption performance model from the methods of aircraft fuel consumption monitoring and evaluation system. Base line and boundary line design, the establishment of a reasonable monitoring method system for the aviation fleet of China's fuel consumption monitoring method, and to explore the construction of methodology. Analysis of the background and significance of the subject, proposed that in the era of rapid expansion of data resources, airlines need to make full use of flight data to carry out various aspects of research. Among them, the analysis of fuel consumption data is an important link. The application and research of domestic and foreign aviation enterprises in fuel consumption monitoring methods and their evaluation methods are summarized, and the characteristics of fuel consumption monitoring systems of some airlines are analyzed. And the aviation fuel consumption and its influencing factors are introduced theoretically. Secondly, a set of aircraft fuel consumption monitoring method theory based on scale data is proposed to refine the analysis process and build a demand analysis framework. Using the CDM methodology for the construction of oil consumption monitoring methodology, selected several types of indicators through the principle of index selection to construct an exponential attenuation performance model based on flight distance. This kind of model can classify different fuel consumption models and define the parameters of performance and distance. Combined with the fuel consumption evaluation index method to analyze the applicability of all kinds of models, select a suitable fuel consumption evaluation monitoring method. Thirdly, determine the fuel consumption performance model monitoring evaluation index baseline and boundary line method. Using artificial neural networks. The theory of signal separation and three methods of least squares are used to determine the standard. This paper mainly introduces the RELAX algorithm of signal separation based on the established fuel consumption model and a cross-iterative optimization of the generalized inner product synthesis of non-consistency detection. The method of computing. This method can estimate the datum effectively under the strong "noise", and reasonably determine the input and output data of the fuel consumption model. Finally, by comparing the advantages and disadvantages of the three methods, the optimal selection of different fuel consumption models is put forward. Method to verify the correctness. Finally, three methods were used to verify the fuel consumption performance evaluation model based on distance and the single factor model was constructed. According to the data of 550 flights of a certain airline flight B737. The data of 8430 flights with the A330 were classified and modeled. This paper analyzes the noise performance and fitting performance of the algorithm, and combines the advantages and disadvantages of the three methods to estimate the overall model and the classification model. It provides an effective method for the study of fuel consumption monitoring methods of airlines. Through single factor analysis and data contrast, the oil consumption monitoring method system is perfected and verified. The advantages and disadvantages of various methods are summarized. Put forward the summary and prospect of the work.
【学位授予单位】：中国民航大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：V241.72;TP277

【参考文献】