Sensitivity Analysis of the Shipboard Integrated Power System

Abstract

Stochastic sensitivity analysis is a valuable tool in ranking inputs and in investigating the degree of interaction of its components. In this paper, we present stochastic simulation results for a shipboard integrated power system and study its sensitivity. Specifically, we apply sensitivity analysis to two high‐fidelity models of shipboard subsystems, investigating open‐ and closed‐loop control of the propulsion system. The results show that different inputs are most important for the open‐ and closed‐loop control, with sensitivities that change dramatically in time as they reflect the transition from the fast electrical scales to slower mechanical scales. We also demonstrate how sensitivity analysis can be used to establish the robustness of the AC drive.

Sensitivity analysis and low-dimensional stochastic modeling of shipboard Integrated Power Systems

“…Our gradient-based sensitivity analysis technique is not restricted on the grid structure, and as a result the sensitivity index converges faster for the same computational cost [8]. All i x are assumed to fluctuate within a specified range, b a, .…”

“…According to [5], the constant-slip current control is composed of an inner loop with a current feedback from IM and an outer loop with IM's shaft angular velocity feedback for controlling motor torque and speed, respectively. The detailed derivation of all machine equations can be found in [5], [8] and [9]. The model of AC power distribution and propulsion systems is composed of 30 state variables (see Table II) and 31 parameters (see Table III).…”

“…As a result, sensitivity analysis that can identify the influential and interactive parameters from a large number of parameters is needed for the IPS designer to improve performances of the integrated system and to prevent a cascaded failure. The sensitivity analysis, based on the One-At-a-Time stochastic variation [7], [8], has been shown to be able to identify and prioritize the important parameters of an AC subsection in the entire IPS [4]. In addition, to further accelerate large-scale simulation or even guide experimental studies, a reduction of parametric space can be accomplished by fixing the less important parameters at their nominal values.…”

“…The direction of in the gradient computation is random such that the OAT perturbation is equally distributed in the parametric space. Using the Monte Carlo (MC) or Quasi Monte Carlo (QMC) sampling method [8] IV. RESULTS In IPS simulation, we assume that the generator initially operates at its rated speed near steady-state conditions.…”

“…SENSITIVITY ANALYSIS Global sensitivity analysis [11], [14] is an effective tool in identifying the most significant parameters of the system, particularly in large-scale systems with many parameters, i.e., the electric ship IPS [8], [9]. First, we …”