Research On The Key Points Of Life-cycle Control And Analysis Of Civil Aircraft Maintenance Costs

With science and technology developing further and further, the performance and complexity of equipments are increasing more and more. The maintenance cost of a civil aircraft has grown up to two thirds of the price and 10%-20% of the direct operating cost, which has became the most important part of life cycle cost. So, both aircraft manufactures and airlines have paid much attention on it. China is enhancing the force of R&D of civil aircraft these years. Under the circumstances and in order to improve aircraft competence in the market, form full system and life cycle, this dissertation mainly researched how to implement maintenance cost control and analysis in our country. The basic concept, main contents and implementary processes of maintenance cost control and analysis were included in it, and after settling some difficult problems, a comparatively integrated methods system was established,a decision support software for maintenance cost control and analysis was also developed.Because of the development of civil aircraft in our country correspondingly behindhand, we have no baseline aircraft and only have limited maintenance cost data in the stage of aircraft development. This brought comparatively difficulties to forecast maintenance cost of new aircraft in the early stage. In order to resolve this difficulty, a concept that the maintenance cost could be forecasted through the comparison between new aircraft and similar aircrafts was firstly introduced, Case-Based Reasoning (CBR) theory was secondly applied, and CBR-DMC method was lastly proposed. Four steps were included in CBR-DMC method: retrieve, reuse, revise and retain, and it could satisfy the real need of engineering.For the simultaneous coherence between maintenance cost target and design requirements and maintenance requirements, the evaluation of maintenance cost target in the early stage of aircraft development is very complicated. Existing evaluating methods waste time and money and can not give out a proper revising strategy when the target is not reasonable. A two-stage Data Envelopment Analysis (DEA) approach with fuzzy inputs and fuzzy outputs was researched. In the application of this approach, future maintenance of new aircraft and its similar aircrafts were regarded as decision-making units, direct manhour costs, direct material costs and indirect maintenance costs were regarded as fuzzy inputs, and dispatch reliability, available seat miles and reciprocal loss costs were regarded as fuzzy outputs, after two-stage analysis, the efficiency of maintenance cost target could be evaluated and the inefficient target could be revised.As similar components are not easily to be found and experiential cost data of different shop maintenances are difficlutily to be gathered, predicting the maintenance cost in the stage of aircraft design is very hard. According the parameters that could be obtained in the design stage, component type, unit price, mean shop processing time and mean time between unscheduled removals, a combined prediction method for the maintenance cost of components named IA-PSO was brought forward. Multiple single models was based in this method, Particle Swarm Optimization algorithm (PSO) improved by Immunity Algorithm (IA) was used to set weights for every single model which attended combing, and different kinds of information form different single models could be integrated by it.It is a comparatively difficult problem for maintenance cost control in the stage of aircraft operation that how to estimate the engine shop visit cost in the condition of small samples. For this problem, a Support Vector Machine (SVM) ensemble method was proposed which based the technical condition and maintenance level of engine shop visit.In this method, the classical and variable precision rough set theory was used to generate several component SVMs by disturbing the samples, the grey relational coefficient theory was used to calculate the probability of being chosen, and simple average was used to ensemble the good ones after setting a threshold of probability. On the base of upward problems being resolved, the working targets, working scopes, working contents and working processes of life-cycle maintenance cost control and analysis were proposed, a comparatively integrated methods system was established, and a decision support software, CAMCAS, was also developed. The theories and methods of this dissertation got good effects when they were applied in AVIC I Commercial Aircraft Co., Ltd and West South Subsidiary Company, Air China. The research of this dissertation could help the aircraft companies of our country to grasp the main idea of life-cycle maintenance cost control and analysis, and it provided several convenient and practical methods and tools also.