Design of a weapon assignment subsystem within a ground-based air defence environment

Lotter, Daniel Petrus (2017-12)

Thesis (PhD)--Stellenbosch University, 2017.


ENGLISH SUMMARY : A number of assets on the ground typically require protection from aerial threats in a military ground-based air defence environment. The problem of defending these assets is twofold: Incoming aircraft rst have to be identied and classied as friendly or hostile, and the level of threat posed to defended assets by each hostile aircraft has to be assessed, after which available ground-based weapon systems secondly have to be assigned to engage aerial threats with a view to scare them away or to neutralise them. The latter problem is known in the military operations research literature as the weapon assignment problem. A re control ocer is responsible for solving both these sub-problems in real time, usually under very stressful conditions. The officer therefore typically employs a computerised threat evaluation and weapon assignment decision support system to aid him in this task. An architecture is put forward in this dissertation for the weapon assignment part of such a decision support system. The proposed architecture contains two subsystems, namely an engagement quantisation subsystem and a weapon assignment subsystem. The purpose of the engagement quantisation subsystem is to quantify single shot hit probabilities achievable by weapon systems in conjunction with other information within the format required by the weapon assignment subsystem. The working of the various components of the engagement quantisation subsystem is illustrated by means of a series of small numerical examples. The weapon assignment subsystem forms the heart of the proposed architecture and a weapon assignment model classication is proposed for use in this subsystem. This classication consists of four classes of weapon assignment models ranging in different levels of complexity. The classes are single-objective static weapon assignment models, multi-objective static weapon assignment models, single-objective dynamic weapon assignment models and multi-objective dynamic weapon assignment models. A model prototype is proposed for default inclusion in each of the aforementioned weapon assignment model classes. The working of each of these models is illustrated by solving it in the context of a hypothetical, but realistic, ground-based air defence environment. A conventional genetic algorithm is used to solve the single-objective static weapon assignment model prototype, while an extension of this algorithm, a nondominated sorting genetic algorithm (specially designed for solving multi-objective optimisation problems) is used to solve the multi-objective static weapon assignment model prototype. The method of simulated annealing is used to solve the single-objective dynamic weapon assignment model prototype, while a variant of the aforementioned nondominated sorting genetic algorithm is used to solve the multi-objective dynamic weapon assignment model prototype. The results returned by the algorithms are discussed and validated by means of three methods, including a subjective face validation, a random benchmark validation and a validation consultation with two independent military experts. It is found that the results are plausible in terms of realism and practical executability. The models also outperform solutions put forward by the military experts when asked to solve the models by hand in the context of the same ground-based air defence scenario.

AFRIKAANSE OPSOMMING : Raadpleeg teks vir opsomming.

Please refer to this item in SUNScholar by using the following persistent URL:
This item appears in the following collections: