Assess transferability to other wastewater treatment plants

Assess whether the multivariate LSTM wastewater-quality forecasting workflow, including data preprocessing and 24-hour-ahead forecasting, transfers effectively to wastewater treatment plants beyond the Seine aval plant.

Background

The paper develops and evaluates a multivariate LSTM model for forecasting pH, temperature, conductivity, and total suspended solids at the Seine aval wastewater treatment plant. The workflow incorporates high-frequency sensor measurements, precipitation data, upstream-plant information, preprocessing, and comparisons with persistence and SARIMA benchmarks.

The authors suggest that the workflow could be applied to other wastewater treatment plants when comparable high-frequency, multivariate, and sufficiently long historical datasets are available. However, the empirical study is limited to the Seine aval case, and the effectiveness and generalizability of the approach at other plants have not yet been established. Thus, transferability across plants remains an unresolved question.