An Enquiry Into Linear Conflict Models

This monograph generalises, and extends, the classic dynamic models in conflict analysis (Lanchester 1916, Richardson 1919, Boulding 1962). Restrictions on parameters are relaxed to account for alliances and for peacekeeping. Incrementalist as well as stochastic versions of the model are reviewed. These extensions allow for a rich variety of patterns of dynamic conflict. Using Monte Carlo techniques as well as time series analyses based on GDELT data (for the Ethiopian-Eritreian war, 1998–2000), we also assess the empirical usefulness of the model. It turns out that linear dynamic models capture selected phases of the conflict quite well, offering a potential taxonomy for conflict dynamics. We also discuss a method for introducing a modicum of (bounded) rationality into models from this tradition.

Neural Computing Enhanced Parameter Estimation for Multi-Input and Multi-Output Total Non-Linear Dynamic Models

In this paper, a gradient descent algorithm is proposed for the parameter estimation of multi-input and multi-output (MIMO) total non-linear dynamic models. Firstly, the MIMO total non-linear model is mapped to a non-completely connected feedforward neural network, that is, the parameters of the total non-linear model are mapped to the connection weights of the neural network. Then, based on the minimization of network error, a weight-updating algorithm, that is, an estimation algorithm of model parameters, is proposed with the convergence conditions of a non-completely connected feedforward network. In further determining the variables of the model set, a method of model structure detection is proposed for selecting a group of important items from the whole variable candidate set. In order to verify the usefulness of the parameter identification process, we provide a virtual bench test example for the numerical analysis and user-friendly instructions for potential applications.

Linear Grey-Box Modeling of Gas Turbine Engines for MIMO Control Design

High-fidelity models are an increasingly indispensable tool for evaluating high-performance control and optimization algorithms of gas turbine (GT) engines. However, GT high-fidelity models (HFM’s) are often too complex for synthesizing model-based control and optimization algorithms, which are much more amenable to low-order linear dynamic models. Obtaining models suitable for control synthesis can be arduous and costly. White-box (first-principles) methods may produce models that lose fidelity in off-nominal conditions, while black-box (data-driven) methods yield model parameters without physical significance, preventing generalization of the model across product configurations. This paper presents a grey-box method for obtaining low-order linear dynamic models for a 5.5 MW GT engine that retain fidelity at off-nominal conditions and generalize to multiple product configurations. The approach exploits the structure of an available HFM by grouping its constituent component-level models according to their suitability to white-box or black-box modeling. Specifically, the rotor model is linearized, yielding a first-order linear model with adjustable physical parameters, e.g. inertia. A second-order open-loop linear model is obtained from remaining component-level models via system identification using closed-loop data generated by the HFM. The linear models are combined to form a third-order open-loop linear GT engine model, which retains fidelity with the HFM in transient validation experiments, and multiple rotor inertia values.