A Quantum-Enhanced PSO-LSSVM Architectural Framework for Predicting Residential Complex Nursery Temperatures in Baghdad

Abstract

Accurate temperature prediction is crucial for the success of  and urban nurseries buildings in residential complex, Baghdad, where severe seasonal temperature extremes can cause significant indoor environmental damage. These microclimates are highly vulnerable to the adverse effects of both acute frost and intense desert heat. Developing a reliable temperature prediction framework capable of forecasting microclimatic shifts several hours in advance is essential to minimizing economic losses. This paper introduces a novel temperature forecasting technique utilizing a Least Squares Support Vector Machine (LSSVM) model whose hyperparameters are dynamically tuned via an Improved Particle Swarm Optimization (IPSO) approach. The IPSO algorithm incorporates an adaptive mutation probability to prevent premature convergence and efficiently discover the optimal parameter boundaries of the LSSVM framework. To evaluate its efficacy, the proposed IPSO-LSSVM model was tested against conventional predictive models using a comprehensive structural dataset. The IPSO-LSSVM model demonstrated superior predictive performance compared to traditional Support Vector Machine (SVM) and Back Propagation Neural Network (BPNN) architectures in forecasting both maximum and minimum temperature boundaries. These results establish the framework as a highly effective tool for automated climate control engineering in urban nurseries.