Integrated Cost Prediction for Distribution Network Engineering Based on the Novel Multi-objective Coati Algorithm and Laguerre Polynomials

Abstract

To achieve precise and stable cost prediction for distribution network engineering, a novel multi-objective coati optimization algorithm (MOCOA) is proposed, and based on this algorithm, a new multi-objective integrated prediction model based on Laguerre polynomials (LNN-Adaboost) is developed. Firstly, utilizing the nonlinear approximation properties of Laguerre orthogonal polynomials, a Laguerre neural network is constructed. Secondly, targeting prediction accuracy and stability, the Laguerre neural network is optimized using the multi-objective coati algorithm. Finally, ensemble learning Adaboost is introduced to correct the prediction errors of the model, achieving automatic allocation and recombination of error weights. By comparing with three mainstream multi-objective optimization algorithms on eight test problems, the effectiveness of the proposed algorithm is verified. Taking the cost data of the Ningxia distribution network engineering as the research object, the proposed model MOCOA-LNN-Adaboost is compared with several mainstream models. The results demonstrate that the proposed prediction model exhibits higher prediction accuracy and better prediction stability. Compared to MOCOA-ELM-Adaboost, the proposed prediction model shows an increase of 9.64% in R2, 2.65% in IA, and a decrease of 53.62% in RMSE, and 61.87% in SDEX..