Abstract A group contribution method is developed for estimating the electrical conductivity of ionic liquids (ILs). Based on 1578 collected experimental data covering 57 ILs in a wide range of temperature (248.05–468.15 K) and electrical conductivity (0.0017–9.1670 S m−1), the parameters of each group (cation, anion, substituent) in the method are optimized. A deterministic algorithm-based on multivariate linear regression is used with 1121 data points covering 57 ILs used as training set and the reaming 457 data points covering 20 ILs are used for validation. The prediction results (from test set) expressed as an average absolute relative deviation (AARD %) of 6.8% between the experimental and predicted electrical conductivities of ILs illustrate the good predictive capability of this group contribution method, with a maximum relative deviation (RD %) of 26.6%. This group contribution-based method can be easily extended to new IL groups that are not involved in this study once the experimental data of ILs involving these new groups become available.

中文翻译：

---

一种基于群贡献的离子液体电导率预测方法

摘要 开发了一种用于估计离子液体 (ILs) 电导率的群贡献方法。基于 1578 条收集的 57 种离子液体在较宽的温度范围（248.05–468.15 K）和电导率（0.0017–9.1670 S m-1）下的实验数据，该方法中各基团（阳离子、阴离子、取代基）的参数优化。使用基于多元线性回归的确定性算法，将覆盖 57 个 IL 的 1121 个数据点用作训练集，并使用覆盖 20 个 IL 的 457 个数据点进行验证。预测结果（来自测试集）表示为 IL 的实验和预测电导率之间 6.8% 的平均绝对相对偏差 (AARD %)，说明了这种组贡献方法的良好预测能力，最大相对偏差 (RD %) 为 26.6%。一旦涉及这些新组的 IL 的实验数据可用，这种基于组贡献的方法可以很容易地扩展到未参与本研究的新 IL 组。