The carbon emission and maintenance-cost guided optimization of aero-engine clearance schedule

Aero-engine clearance is an important way to reduce carbon emission, save fuel consumption, and decrease maintenance costs. To solve the problem of insufficient clearance caused by the current regular cleaning schedule, this paper proposes an aero-engine clearance optimization method. This method has two innovations: a reinforcement optimized clearance strategy and a mixed transfer process neural network. The mixed transfer process neural network (MTPNN) is composed of two process neural networks and a multi-layer perceptron, predicting exhaust gas temperature margin (EGTM) and fuel flow after engine cleaning. The MTPNN is training by the real data from the monitoring engine of the airline. The reinforcement optimized clearance strategy is a customized reinforcement learning framework for optimizing the aero-engine cleaning schedule. It uses the data provided by MTPNN to evaluate the carbon emissions and company benefits after engine cleaning, thereby maximizing the economic benefits and the performance of the engine. According to the simulation, the proposed method is capable of increasing the airline’s profit by $36,386.5 per engine and reducing carbon emissions by 135.83 tons during 2750 flight cycles.