Nitrous oxide (N₂O) emissions from wastewater treatment plants (WWTPs) exhibit significant seasonal variability, making accurate predictions with conventional biokinetic models difficult due to complex and poorly understood biochemical processes. This study addresses these challenges by exploring data-driven alternatives, using long short-term memory (LSTM) based encoder-decoder models as basis. The models were developed for future integration into a model predictive control framework, aiming to reduce N₂O emissions by forecasting these over varying prediction horizons. The models were trained on 12 months and tested on 3 months of data from a full-scale WWTP in Amsterdam West, the Netherlands. The dataset encompasses seasonal peaks in N₂O emissions typical for winter and spring months. The best performing model, featuring a 256-256 LSTM architecture, achieved the highest accuracy with test R² values up to 0.98 across prediction horizons spanning 0.5 to 6.0 hours ahead. Feature importance analysis identified past N₂O emissions, influent flowrate, NH₄⁺, NO_x, and dissolved oxygen (DO) in the aerobic tank as most significant inputs. The observed decreasing influence of historical N₂O emissions over extended prediction horizons highlights the importance and significance of process variables for the model's performance. The model's ability to accurately forecast short-term N₂O emissions and capture immediate trends highlights its potential for operational use in controlling emissions in WWTPs. Further research incorporating diverse datasets and biochemical process inputs related to microbial activities in the N₂O production pathways could improve the model's accuracy for longer forecasting horizons. These findings advocate for hybridising deep learning models with biokinetic and mechanistic insights to enhance prediction accuracy and interpretability.

中文翻译：

---

使用基于 LSTM 的深度学习模型预测大型污水处理厂的一氧化二氮排放


废水处理厂 （WWTP） 排放的一氧化二氮 （N₂O） 表现出显著的季节性变化，由于生化过程复杂且知之甚少，因此很难使用传统的生物动力学模型进行准确预测。本研究通过使用基于长短期记忆 （LSTM） 的编码器-解码器模型为基础，探索数据驱动的替代方案来应对这些挑战。这些模型是为了将来集成到模型预测控制框架中而开发的，旨在通过在不同的预测范围内预测 N₂O 排放来减少这些排放。这些模型进行了 12 个月的训练，并使用来自荷兰阿姆斯特丹西部全尺寸污水处理厂的 3 个月数据进行了测试。该数据集包括冬季和春季典型的 N₂O 排放的季节性峰值。性能最佳的模型采用 256-256 LSTM 架构，在提前 0.5 到 6.0 小时的预测范围内，测试 R² 值高达 0.98，实现了最高的准确性。特征重要性分析确定好氧罐中过去的 N₂O 排放、进水流速、NH₄⁺、NO_x 和溶解氧 （DO） 是最重要的输入。观察到的历史 N₂O 排放对扩展预测范围的递减影响突出了过程变量对模型性能的重要性和显著性。该模型能够准确预测短期 N₂O 排放并捕捉即时趋势，这凸显了其在控制污水处理厂排放方面的运营应用潜力。 进一步研究结合与 N₂O 生产途径中的微生物活动相关的不同数据集和生化过程输入，可以提高模型的准确性，从而获得更长的预测范围。这些发现主张将深度学习模型与生物动力学和机理见解相结合，以提高预测的准确性和可解释性。