Permeable blocks rainwater tank (RW tank) source heat pump system (RWHP system) can contribute to the development of an eco-friendly energy system. Therefore, this study evaluated the temperature behavior of the RW tank through the testbed operation performance analysis of RWHP system. Additionally, in order to develop a model that predicts the temperature of the RW tank, four models (Bagging, Boosting, Long short-term memory, Gated recurrent unit) were reviewed based on measured testbed data. As a result, it was confirmed that the temperature of the RW tank is affected by factors such as ambient temperature, heat pump outlet fluid temperature, the heat exchange amount of the heat pump, RW tank water level, and system operating mode. Among the four predictive models that predict the temperature of the RW tank using these variables as input values, the bagging model showed the highest predictive performance. The bagging model showed very good predictive performance, with a MAE of 0.38 °C, an R2 of 0.9940 in training and a MAE of 1.06 °C, R2 of 0.9672 in testing. This study demonstrated the feasibility of predicting RW tank temperatures using real-time field data. The proposed RW tank temperature prediction model is expected to be incorporated into energy simulation tools such as TRNSYS or LCEM, contributing to performance prediction and improvement studies of RWHP systems. This approach allows for the quantitative assessment of the development value of RWHP systems and will be used to review optimal design strategies and operating manuals for RWHP systems.

中文翻译：

---

地下透水块体长期运行雨水池源热泵系统温度行为分析及预测模型


透水块雨水箱 （RW 水箱） 源热泵系统 （RWHP 系统） 有助于开发环保能源系统。因此，本研究通过 RWHP 系统的试验台运行性能分析，评价了 RW 罐的温度行为。此外，为了开发预测 RW 水箱温度的模型，根据测量的测试台数据审查了四种模型（Bagging、Boosting、Long short-term memory、Gated Recurrent Unit）。结果证实，RW 槽的温度受环境温度、热泵出口流体温度、热泵热交换量、RW 槽水位、系统运行模式等因素的影响。在使用这些变量作为输入值预测 RW 罐温度的四个预测模型中，装袋模型显示出最高的预测性能。装袋模型显示出非常好的预测性能，MAE 为 0.38 °C，训练中的 R2 为 0.9940，测试中的 MAE 为 1.06 °C，R2 为 0.9672。这项研究证明了使用实时现场数据预测 RW 水箱温度的可行性。拟议的 RW 水箱温度预测模型有望被纳入 TRNSYS 或 LCEM 等能源模拟工具中，有助于 RWHP 系统的性能预测和改进研究。这种方法允许对 RWHP 系统的开发价值进行定量评估，并将用于审查 RWHP 系统的最佳设计策略和操作手册。