Ville Mattila

Improving Maintenance Decision Making in the Finnish Air Force Through Simulation

We used discrete-event simulation to model the maintenance of fighter aircraft and improve maintenance-related decision making within the Finnish Air Force. We implemented the simulation model as a stand-alone tool that maintenance designers could use independently. The model has helped the designers to study the impact of maintenance resources, policies, and operating conditions on aircraft availability. It has also enabled the Finnish Air Force to advance the operational capability of its aircraft fleet. We designed the model to simulate both normal and conflict operating conditions. The main challenge of the project was the scarcity and confidentiality of data about the fighter aircraft, their maintenance, and various operational scenarios, especially during conflict situations.

Ranking and selection for multiple performance measures using incomplete preference information

This paper presents two new procedures for ranking and selection (R&S) problems where the best system designs are selected from a set of competing ones based on multiple performance measures evaluated through stochastic simulation. In the procedures, the performance measures are aggregated with a multi-attribute utility function, and incomplete preference information regarding the weights that reflect the relative importance of the measures is taken into account. A set of feasible weights is determined according to preference statements that are linear constraints on the weights given by a decision-maker. Non-dominated designs are selected using two dominance relations referred to as pairwise and absolute dominance based on estimates for the expected utilities of the designs over the feasible weights. The procedures allocate a limited number of simulation replications among the designs such that the probabilities of correctly selecting the pairwise and absolutely non-dominated designs are maximized. The new procedures offer ease of eliciting the weights compared with existing R&S procedures that aggregate the performance measures using unique weights. Moreover, computational advantages are provided over existing procedures that identify non-dominated designs based on the expected values of the performance measures. The new procedures allow to obtain a smaller number of non-dominated designs. They also identify these designs correctly with a higher probability or require a smaller number of replications for correct selection. Finally, the new procedures allocate a larger number of replications to the non-dominated designs that are therefore evaluated with greater accuracy. These computational advantages are illustrated through several numerical experiments.

Maintenance scheduling of a fleet of fighter aircraft through multi-objective simulation-optimization

The maintenance scheduling problem of a fleet of fighter aircraft is considered through multi-objective simulation-optimization (MOSO). In the problem, a maintenance schedule consisting of target starting times of the maintenance activities of the aircraft is determined. The objectives are to minimize average deviation between the target and actual starting times of the activities and to maximize average aircraft availability. The objectives depend on the maintenance schedule through complex interactions due to a policy in which the need for maintenance is based on the flight hours of the aircraft cumulated during flight missions. In addition, the durations of the flight missions, maintenance activities, and failure repairs are uncertain. Therefore, an MOSO approach is applied to the problem. The approach includes a discrete-event simulation model and a state-of-the art multi-objective simulated annealing algorithm for determining non-dominated schedules. Moreover, a multi-attribute value (MAV) function is used for supporting a maintenance decision-maker (DM) in selecting the preferred non-dominated schedule for implementation. The MAV function captures incomplete information on the values of the objectives as well as on the DM’s preference statements regarding the weights of the objectives. The approach is implemented as an MOSO tool whereby the DM can consider the complex interactions and uncertainties of the problem which have not been addressed in the existing literature on maintenance scheduling. The approach and the tool are illustrated with a set of test problems as well as a real-life example problem.

Scheduling fighter aircraft maintenance with reinforcement learning

This paper presents two problem formulations for scheduling the maintenance of a fighter aircraft fleet under conflict operating conditions. In the first formulation, the average availability of aircraft is maximized by choosing when to start the maintenance of each aircraft. In the second formulation, the availability of aircraft is preserved above a specific target level by choosing to either perform or not perform each maintenance activity. Both formulations are cast as semi-Markov decision problems (SMDPs) that are solved using reinforcement learning (RL) techniques. As the solution, maintenance policies dependent on the states of the aircraft are obtained. Numerical experiments imply that RL is a viable approach for considering conflict time maintenance policies. The obtained solutions provide knowledge of efficient maintenance decisions and the level of readiness that can be maintained by the fleet.

Flight time allocation for a fleet of aircraft through reinforcement learning

Fighter aircraft are typically maintained periodically on the basis of cumulated usage hours. In a fleet of aircraft, the timing of the maintenance therefore depends on the allocation of flight time. A fleet with limited maintenance resources is faced with a design problem in assigning the aircraft to flight missions so that the overall amount of maintenance needs will not exceed the maintenance capacity. We consider the assignment of aircraft to flight missions as a Markov Decision Problem over a finite time horizon. The average availability of aircraft is taken as the optimization criterion. An efficient assignment policy is solved using a Reinforcement Learning technique called Q-learning. We compare the performance of the Q-learning algorithm to a set of heuristic assignment rules using problem instances that involve varying number of aircraft and types of periodic maintenance. Moreover, we consider the possibilities of practical implementation of the produced solutions.

Optimizing locations of decoys for protecting surface-based radar against anti-radiation missile with multi-objective ranking and selection

This paper considers the decoy location problem, i.e., the problem of determining optimal locations for decoys that protect a surface-based radar against an anti-radiation missile. The objectives of the problem are to simultaneously maximize distances between the missiles detonation point and the radar as well as the decoys. The problem is solved using a stochastic simulation model providing the distances as well as a ranking and selection procedure called MOCBA-p. In the procedure, location combinations are evaluated through a multi-attribute utility function with incomplete preference information regarding weights related to the objectives. In addition, multi-objective computing budget allocation is used for allocating simulation replications such that the best combinations are selected correctly with high confidence. Numerical experiments presented in the paper illustrate the suitability of MOCBA-p for solving the decoy location problem. It provides computational advantages over an alternative procedure while also enabling ease of determining the weights.