Bio-inspired Intelligent Algorithms And Their Applications In Parameter Prediction For Renewable Power Systems

Due to the increasing problems of rapid fossil consumption and environmental pollution,renewable energies have been paid more attention.As two key new energy types,wind power and solar power have been developing dramatically whereas more problems during the development process should be studied and addressed.Solar power(named PV)and wind power have explicit characteristics of parameter uncertainty and model randomness,which lead to variable power output and difficult configuration of hybrid energies.To address these problems,this thesis blends bio-inspired algorithms with support vector machines(SVM)for parameter tuning,and develops prediction models for PV output,wind speed,and parameter configuration of hybrid energies.The main contributions of this work are as follows:(1)An improved Backward Bat Algorithm(BBA)based SVM called BBA-SVM is presented.Then,the genetic mechanism is adopted to build up the Genetic-BBA,i.e.,G-BBA,who adopts variable steps of BAT with adjusted searching direction.By using the backward searching mode,the local optimization can be avoided so that the effective global tuning of SVM can be ensured.(2)For illustration,the PV system is adopted to validate the proposed BBA-SVM and G-BBA-SVM power prediction methods.The software of SAM is used for generating the data sets of PV power,illumination,and temperature for system modeling.Comparative studies are carried out for various bio-inspired SVM approaches.(3)The BBA-SVM is applied to the wind power prediction,with adopting the Bladed software to simulation the wind speed input.The wind power regression prediction model is developed with simulations and data analysis.(4)The system parameter configuration is studied based on the BBA-SVM load prediction.After comparing different power prices and efficiencies for various resources,the reduced analysis model is developed for power generation and delivery.