Parameter Estimation in Linear Rate Process: Modified Maximum Likelihood Estimation Using Particle Swarm Optimization, Least Square Estimation, and Simulation Methods

The current study presents a novel approach for parameter estimation in LRP based on a linear intensity parameter model of the Non-Homogeneous Poisson Process (NHPP). The Modified Maximum Likelihood Estimation- Particle Swarm Optimization (MMLE-PSO) method enhances prediction accuracy and computational efficiency by integrating the PSO algorithm. Compared to Least Square Estimation (LSE) and PSO, MMLE-PSO achieves superior parameter estimation, reducing errors by 61.2% and 92.6%, respectively. Additionally, it accelerates computational performance by 97% over Maximum Likelihood Estimation (MLE) and PSO due to its faster convergence. The method's effectiveness in event pattern modeling is demonstrated using outage data from the Mosul Dam power facility. In statistical evaluation, MMLE-PSO attains the lowest RMSE value 0.0253, outperforming LSE 0.0652 and PSO 0.3429. With its enhanced estimation precision and operational efficiency, MMLE-PSO proves to be a reliable tool for reliability engineering applications