基于AFSA优化的支持向量机柴油机性能预测模型

电控高压共轨技术的采用使得船用柴油机性能及排放具有了更大的优化空间，但同时柴油机控制参数增多使得柴油机性能的预测变得更为复杂。为了建立精确的柴油机性能预测模型，利用回归支持向量机，通过对实验数据的学习以获得预测模型。支持向量机的预测精度会因其参数的选择出现一定的差异，所以需要对支持向量机参数选择进行研究和优化。以某型船用高速大功率电控高压共轨柴油机为研究对象，建立支持向量机预测模型，分析其预测性能受参数选择的影响，并利用人工鱼群算法对其进行优化。结果表明，基于人工鱼群算法优化的回归支持向量机能够建立精度较高的柴油机性能预测模型，且人工鱼群算法具有很好的寻优性能。

Diesel engine performance prediction model research using AFSA

The introduction of common rail direct injection system(CRDI) has extended the optimization space of marine diesel engine, meanwhile, the prediction of engine performance became more complex due to the increase of control parameters. In order to develop an accurate model for the engine performance, support vector regression(SVR) is employed for the modeling based on engine test data. The accuracy of SVR will be varied due to its parameters, so the research and optimization of these parameters are necessary. Based on a CRDI-assisted marine diesel engine, the SVR predictive model is developed; the parameters of SVR are investigated and optimized using artificial fish school algorithm(AFSA). The results indicate that the SVR predictive model optimized by AFSA has the ability to develop perfect engine performance predictive models. In addition, AFSA has an excellent performance of optimization.