Antenna performance optimisation using evolutionary algorithms

This thesis investigates the novel idea of using evolutionary algorithms to optimise control and design aspects of active array antenna systems. Active arrays differ from most mechanically scanned antennas in that they offer the ability to control the shape of their radiation pattern. As active arrays consist of a multiplicity of transmit and receive modules (TRMs), the task of optimally controlling them in order to generate a desired radiation pattern becomes difficult. The control problem is especially true of conformal (non-planar) array antennas that require additional phase control to achieve good radiation pattern performance. This thesis describes a number of significant advances in the optimisation of array antenna performance. Firstly a genetic algorithm (GA) is shown to be effective at optimising both planar and conformal antenna performance. A number of examples are used to illustrate and promote the basic optimisation concept. Secondly, in this thesis the techniques are advanced to apply multiobjective evolutionary optimisation algorithms to array performance optimisation. It is shown that Evolutionary Algorithms allow users to simultaneously optimise many aspects of array performance without the need to fine-tune a large number of weights. The multiple-objective analysis methods shown demonstrate the advantages to be gained by holding knowledge of the Pareto optimal solution set. Thirdly, this thesis examines the problems of optimising the design of large (many element) array antennas. Larger arrays are often divided into smaller sub-arrays for manufacturing reasons and to promote formation of difference beam patterns for monopulse operation. In the past, the partitioning has largely been left to trial-and-error or simple randomisation techniques. This thesis describes a new and novel approach for optimally subdividing both planar and conformal array antennas as well as improving gain patterns in a single optimisation process. This approach contains a new method of partitioning array antennas, inspired from a biological process and is also presented and optimised using evolutionary algorithms. Additionally, the technique can be applied to any size or shape of array antenna, with the processing load dependent on the number of subarrays, rather than the number of elements. Finally, the success of these new techniques is demonstrated by presenting a range of performance optimised examples of planar and conformal array antenna installations including examples of optimally evolved subarray partitions