Embedded clonal squirrel evolutionary programming for resilience improvement in power system

The increasing demand for reliable electricity supply, coupled with the growing frequency of high-impact low-probability (HILP) events such as hurricanes, natural disasters, and critical equipment failures, has underscored the urgent need for grid hardening and resilience improvement in modern power systems. Severe disruptions are not adequately addressed by conventional reliability measures, as widespread outages and significant degradation in system performance are often caused as a result. This thesis presents the Embedded Clonal Squirrel Evolutionary Programming (ECSEP) for Resilience Improvement in Power Systems. To address these limitations, this study pursues three primary objectives. The first objective is to develop a novel ECSEP algorithm for grid hardening in power systems. The second objective is to formulate a Multi-Objective ECSEP (MOECSEP) for simultaneous loss minimisation and voltage security enhancement in power system planning. The third objective is to develop a Composite Survivability Strategy (CSS) for grid hardening by integrating intelligent distributed generation (DG) installation and load shedding through coordinated optimisation using ECSEP. The proposed ECSEP algorithm is developed by embedding cloning operators and elements of the Squirrel Search Algorithm (SSA) into the Evolutionary Programming (EP) framework, thereby enhancing exploration capability, convergence speed, and solution accuracy. ECSEP is first applied to single-objective optimisation, focusing on minimising transmission losses and improving voltage stability through optimal DG placement and intelligent load shedding under both normal and outage conditions. Subsequently, MOECSEP is formulated using the weighted-sum principle to simultaneously address the conflicting objectives of system loss reduction and voltage stability enhancement. The CSS framework is then developed by combining DG installation and load shedding as a coordinated grid-hardening scheme governed by 22 decision parameters, enabling the simultaneous determination of optimal DG locations, sizes, load shedding locations, and their corresponding power reduction levels. Validation was conducted on the IEEE 30-Bus and IEEE 57-Bus Reliability Test Systems under two hurricane scenarios representing progressively severe disruption conditions. Comparative studies against benchmark algorithms, namely Evolutionary Programming (EP), Clonal Search Optimizer (CSO), and Squirrel Search Algorithm (SSA), consistently demonstrated the superiority of ECSEP, MOECSEP, and CSS across all test cases. The proposed techniques achieved significant reductions in transmission losses, improved voltage stability indices (FVSI), and enhanced overall system resilience compared to pre-optimisation baselines and competing methods. In conclusion, the ECSEP-based framework presents a robust, computationally efficient, and practically relevant solution for grid hardening and resilience planning in power systems subjected to HILP events. Future work is recommended to extend the framework to larger network benchmarks such as the IEEE 118-Bus and IEEE 300-Bus systems, incorporate energy storage systems alongside DG and load shedding, and integrate stochastic modelling to account for uncertainties in renewable generation and load profiles. Further extensions to real-time optimisation, cyber-physical resilience, and advanced multi-criteria decision-making approaches are also encouraged to broaden the practical applicability of the developed techniques.