群智能搜索在基础油性能预测模型中的优化效能

润滑油基础油组成成分的变化对其性能有重要影响，本文中针对复合基础油体系中KN4010、PAO40和PriEco 3000三种成分含量的不同配比，对润滑油运动黏度、黏度指数和旋转氧弹性能指标变化的影响，建立基于最小二乘支持向量机(LSSVM)基础性能预测模型，并进行对比分析及筛选；选用经典的粒子群优化算法(PSO)、蜻蜓算法(DA)和鲸鱼优化算法(WOA)等仿生群智能搜索策略构建混合模型，对优选的预测模型进行参数优化. 测试结果表明：机器学习技术对油液性能具有良好的预测能力，并且LSSVM基础模型可以在小样本条件下得到相对较好的预测结果，而WOA-LSSVM能够显著降低模型的预测误差；并且通过测试和留一交叉验证法分析，WOA-LSSVM的预测结果明显优于其余模型，具有良好的预测精度和泛化能力.

Changes in the composition of lubricant base oils exert a significant influence on their performance. In this paper, a basic performance prediction model based on least squares support vector machine (LSSVM) is established to explore the effects of different mixing ratios of three components (KN4010, PAO40 and PriEco 3000) in the composite base oil system on the kinematic viscosity, viscosity index and rotating bomb performance indexes of lubricating oil, followed by comparative analysis and model screening. Classic bionic swarm intelligence search strategies including particle swarm optimization (PSO), dragonfly algorithm (DA) and whale optimization algorithm (WOA) are adopted to construct hybrid models for parameter optimization of the selected optimal prediction models. The test results demonstrate that machine learning technologies exhibit excellent predictive capability for lubricating oil performance. The basic LSSVM model can obtain relatively good prediction results under small-sample conditions, while WOA-LSSVM can significantly reduce the prediction error of the model. Moreover, through test and leave-one-out cross-validation analysis, the prediction results of WOA-LSSVM are remarkably superior to those of other models, featuring good prediction accuracy and generalization ability.

搜集汇总

背景与挑战

背景概述

该数据集聚焦于润滑油基础油性能预测，通过分析KN4010、PAO40和PriEco 3000三种成分的不同配比对运动黏度、黏度指数和旋转氧弹性能的影响，建立最小二乘支持向量机（LSSVM）预测模型。数据集采用群智能搜索策略（如粒子群优化、蜻蜓算法和鲸鱼优化算法）进行参数优化，测试结果表明鲸鱼优化算法-LSSVM模型能显著降低预测误差，并展现出良好的预测精度和泛化能力。