The evaluation of engine performance and emission of biodiesel produced from African pear seed oil (APO) and its prediction using multi‐input‐multi‐output (MIMO) artificial neural network (ANN) technique was studied. A standard diesel test bed was used to carry out the evaluation using petrol‐diesel, biodiesel, and its blends, operating at various engine revolutions per minute. Sensitivity analysis used the connection weight method. The performance results showed that petro‐diesel blended with a small amount of APO methyl ester (B20) improved the engine efficiency and emissions better than only petrol‐diesel. The engine performance was predicted well, with the MIMO model having better correlation coefficients of 0.99779, 0.99993, and 0.99774 on training, validation, and testing, respectively, than the Multiple input Single Output (MISO) model. The desired outputs compared between their measured and simulated values exhibited a very low mean square error of 0.0000014488. The blend had the highest relative impact (62.08%) on the output variables. The MIMO model has therefore shown the potential to adopt two input variables equally in predicting the engine performance of biodiesel rather than only complex variables. © 2021 Society of Chemical Industry and John Wiley & Sons, Ltd

中文翻译：

---

通过多输入多输出人工神经网络（ANN）和敏感性分析评估非洲梨籽油（APO）生物柴油的发​​动机性能和排放及其预测

研究了非洲梨籽油（APO）生产的生物柴油的发​​动机性能和排放评估以及使用多输入多输出（MIMO）人工神经网络（ANN）技术进行的预测。使用标准柴油测试台使用汽油-柴油，生物柴油及其混合物进行评估，并以每分钟不同的发动机转速运行。灵敏度分析使用连接权重方法。性能结果表明，掺有少量APO甲酯（B20）的汽油柴油比仅汽油柴油更好地改善了发动机效率和排放。与多输入单输出（MISO）模型相比，MIMO模型在训练，验证和测试上的相关系数分别为0.99779、0.99993和0.99774，具有良好的发动机性能预测能力。在其测量值和模拟值之间进行比较的所需输出显示出非常低的均方误差0.0000014488。共混物对输出变量的影响最大（62.08％）。因此，MIMO模型显示了在预测生物柴油的发​​动机性能方面平均采用两个输入变量的潜力，而不仅仅是复杂变量。©2021化学工业协会和John Wiley＆Sons，Ltd