Scholarlink

<p>To address the issue of discrepancies in main engine power estimation caused by differences in the main dimensions of inland vessels during emission calculations, a prediction method based on clustering analysis and particle swarm-grey wolf optimization (PSO-GWO) is proposed. This method accurately fits various inland vessel main dimension data, detects the correlation between vessel dimensions and main engine power, and applies clustering analysis with different numbers of centers. The clustering results are then incorporated into an artificial neural network optimized by PSO-GWO to train multiple models for predicting and evaluating the main engine power of inland vessels. The most accurate model is selected as the final prediction model to enhance the overall prediction capability. Data analysis shows that the model performs best when the number of clusters is four, achieving a coefficient of determination (R<sup>2</sup>) of 0.96104 and a mean squared error (MSE) of 40,399.43. Compared to the non-clustered model (R<sup>2</sup> = 0.92479, MSE = 75,288.06), the error is significantly reduced, indicating the effectiveness of this method in predicting inland vessel power. This method provides reliable data support for emission calculations of inland vessels and has practical applications in optimizing shipping energy efficiency and environmental management.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Research on Inland Ship Main Engine Power Prediction Based on Clustering and Particle Swarm-Grey Wolf Optimization

  • 期刊论文

  • 发表日期：

    2025年1月6日

查看全文
  • Liang Tong,
  • Shijie Sun,
  • Xiankun Tan,
  • Chunyu Zhang,
  • Dan Liu,
  • Lumeng Wang

摘要

To address the issue of discrepancies in main engine power estimation caused by differences in the main dimensions of inland vessels during emission calculations, a prediction method based on clustering analysis and particle swarm-grey wolf optimization (PSO-GWO) is proposed. This method accurately fits various inland vessel main dimension data, detects the correlation between vessel dimensions and main engine power, and applies clustering analysis with different numbers of centers. The clustering results are then incorporated into an artificial neural network optimized by PSO-GWO to train multiple models for predicting and evaluating the main engine power of inland vessels. The most accurate model is selected as the final prediction model to enhance the overall prediction capability. Data analysis shows that the model performs best when the number of clusters is four, achieving a coefficient of determination (R2) of 0.96104 and a mean squared error (MSE) of 40,399.43. Compared to the non-clustered model (R2 = 0.92479, MSE = 75,288.06), the error is significantly reduced, indicating the effectiveness of this method in predicting inland vessel power. This method provides reliable data support for emission calculations of inland vessels and has practical applications in optimizing shipping energy efficiency and environmental management.