A Parameter-Estimation Method Using the Ensemble Kalman Filter for Flow and Thermal Simulation in an Engine Compartment

Abstract

The feasibility of the parameter estimation on the basis of the ensemble Kalman filter (EnKF) for a practical simulation involving model errors was investigated. The three-dimensional flow and thermal simulations for the engine compartment of a test excavator were simulated, and several unknown temperatures used for boundary conditions were estimated with the method. The estimation method was validated in two steps. First, the estimation method was tested with the influence of the model errors removed by virtually creating true values with a simulation. These results showed that the proposed parameter-estimation method can successfully estimate surface temperatures. They also suggested that the appropriate ensemble size can be evaluated from the number of unknown parameters. Second, the estimation method was tested under a practical condition including model errors by using actual measurement data. Model errors were statistically estimated using prior obtained error data concerning other design configurations, and they were added to the observation error in the EnKF. These results showed that taking model errors into account in the EnKF provides more-accurate parameter-estimation results. Moreover, the uncertainty of an estimated parameter can be evaluated with the standard deviation of its distribution.