Food security has become a key issue worldwide in recent years. According to the Department for Environment Food and Rural Affair (DEFRA) UK, the key components of food security are food availability, global resource sustainability, access, food chain resilience, household food security, safety and confidence of public towards food system. Each of these components has its own indicators which need to be monitored. Only a few studies had been made towards analysing food security and most of these studies are based on conventional data analysis methods such as the use of statistical techniques. In handling food security datasets such as crops yield, production, economy growth, household behaviour and others, where most of the data is imprecise, non-linear and uncertain in nature, it is better to handle the data using intelligent system (IS) techniques such as fuzzy logic, neural networks, genetic algorithm and hybrid systems, rather than conventional techniques. Therefore this thesis focuses on the modelling of food security using IS techniques, and a newly developed hybrid intelligent technique called a 2-stage hybrid (TSH) model, which is capable of making accurate predictions. This technique is evaluated by considering three applications of food security research areas which relate to each of the indicators in the DEFRA key food security components. In addition, another food security model was developed, called a food security risk assessment model. This can be used in assessing the level of risk for food security. The TSH model is constructed by using two key techniques; the Genetic Algorithm (GA) module and the Artificial Neural Network (ANN) module, where these modules combine the global and local search, by optimizing the inputs of ANN in the first stage process and optimizing of weight and threshold of ANN, which is then used to remodel the ANN resulting in better prediction. In evaluating the performance of the TSH prediction model, a total of three datasets have been used, which relate to the food security area studied. These datasets involve the prediction of farm household output, prediction of cereal growth per capita as the food availability main indicators in food security component, and grain security assessment prediction. The TSH prediction model is benchmarked against five others techniques. Each of these five techniques uses an ANN as the prediction model. The models used are: Principal Component Analysis (PCA), Multi-layered Perceptron-Artificial Neural Network (MLP-ANN), feature selection (FS) of GA-ANN, Optimized Weight and Threshold (OWTNN) and Sensitive Genetic Neural Optimization (SGNO). Each of the application datasets considered is used to show the capability of the TSH model in making effective predictions, and shows that the general performance of the model is better than the other benchmarked techniques. The research in this thesis can be considered as a stepping-stone towards developing other tools in food security modelling, in order to aid the safety of food security.

published_or_final_version / Geography and Geology / Doctoral / Doctor of Philosophy

published_or_final_version / Geography and Geology / Doctoral / Doctor of Philosophy

published_or_final_version / Urban Planning / Master / Master of Science in Urban Planning

published_or_final_version / Urban Planning / Master / Master of Science in Urban Planning

published_or_final_version / Urban Planning / Master / Master of Science in Urban Planning

published_or_final_version / Urban Planning / Master / Master of Science in Urban Planning

published_or_final_version / Business Administration / Master / Master of Business Administration

published_or_final_version / Architecture / Master / Master of Philosophy

published_or_final_version / Urban Planning and Design / Master / Master of Urban Design