基于灰色关联机理组合模型的压缩机电功率预测

Prediction of compressor electric power based on gray relational mechanism combination model

为了精确获得压缩机电功率,通过变频滚动转子制冷系统实验台,根据灰色关联(GRM(1,m))预测模型样本需求量少、预测精度高与机理(Mechanism)模型能够反映系统本质特性的特点,利用Matlab语言编程建立GRM(1,m)-Mechanism组合预测模型。通过3种模型对压缩机电功率进行预测,结果表明:GRM(1,m)-Mechanism组合模型的预测精确性和适用性更好,其最大相对误差、平均相对误差分别为4.05%,1.71%;比机理模型分别降低了1.29%,1.09%;比灰色关联(GRM(1,m))预测模型降低了1.02%,2.19%。最后,通过压缩机变转速试验验证组合模型预测平均相对误差在1.9%以内,进一步证明GRM(1,m)-Mechanism组合模型的精确性和适用性。

In order to accurately obtain the compressor electric power,the GRM(1,m)-Mechanism combination prediction model was established using Matlab language through the experimental bench for the variable frequency rolling rotor refrigeration system according to the characteristics of the (GRM(1,m)) prediction model such as low sample demand for gray correlation,high prediction accuracy and its ability to reflect the system essential characteristics.The compressor electric power was predicted by these three models respectively.The results show that the GRM(1,m)-Mechanism combination prediction model has better prediction accuracy and applicability than two other models.Its maximum relative error and average relative error were 4.05% and 1.71% respectively,which were 1.29% and 1.09% lower than that of the Mechanism model,and 1.02% and 2.19% lower than that of the gray correlation(GRM(1,m)) prediction model.Finally,the average relative error of combination prediction model was verified to be within 1.9% by the compressor variable speed experiments,which further proved the accuracy and applicability of the GRM(1,m)-Mechanism model.