Stochastic simulators based optimization by Gaussian process metamodels - Application to maintenance investments planning issues

TL;DR

This paper develops a Gaussian process-based metamodel for stochastic simulators to efficiently estimate output distributions and optimize quantiles, demonstrated on maintenance investment planning.

Contribution

It introduces a novel quantile function metamodel using Gaussian processes and an adaptive design method (QFEI) for optimization with few simulator runs.

Findings

Effective approximation of stochastic simulator quantiles.

Successful application to maintenance investment optimization.

Reduced number of simulations needed for optimal solutions.

Abstract

This paper deals with the construction of a metamodel (i.e. a simplified mathematical model) for a stochastic computer code (also called stochastic numerical model or stochastic simulator), where stochastic means that the code maps the realization of a random variable. The goal is to get, for a given model input, the main information about the output probability distribution by using this metamodel and without running the computer code. In practical applications, such a metamodel enables one to have estimations of every possible random variable properties, such as the expectation, the probability of exceeding a threshold or any quantile. The present work is concentrated on the emulation of the quantile function of the stochastic simulator by interpolating well chosen basis function and metamodeling their coefficients (using the Gaussian process metamodel). This quantile function metamodel is then used to treat a simple optimization strategy maintenance problem using a stochastic code, in order to optimize the quantile of an economic indicator. Using the Gaussian process framework, an adaptive design method (called QFEI) is defined by extending in our case the well known EGO algorithm. This allows to obtain an "optimal" solution using a small number of simulator runs.