LSTM-Based Wastewater Treatment Plants Operation Strategies for Effluent Quality Improvement

Industrial environments are characterised by the non-lineal and highly complex processes they perform. Different control strategies are considered to assure that these processes are correctly performed. Nevertheless, these strategies are sensible to noise-corrupted and delayed measurements. For that reason, denoising techniques and delay correction methodologies should be considered but, most of these techniques require a complex design and optimisation process as a function of the scenario where they are applied. To alleviate this, a complete data-based approach devoted to denoising and correcting the delay of measurements is proposed here with a two-fold objective: simplify the solution design process and achieve its decoupling from the considered control strategy as well as from the scenario. Here it corresponds to a Wastewater Treatment Plant (WWTP). However, the proposed solution can be adopted at any industrial environment since neither an optimization nor a design focused on the scenario is required, only pairs of input and output data. Results show that a minimum Root Mean Squared Error (RMSE) improvement of a 63.87% is achieved when the new proposed data-based denoising approach is considered. In addition, the whole system performance show that similar and even better results are obtained when compared to scenario-optimised methodologies.

show abstract

“…They correspond to the

and the

, both measured in (mg/L). They are two of the variables showing the highest mutual information with respect to the

concentration [ 43 , 56 , 57 ].…”

Section: Data-based Enhanced Control Strategymentioning

confidence: 99%

Industrial Control under Non-Ideal Measurements: Data-Based Signal Processing as an Alternative to Controller Retuning

Pisa

Morell

Vilanova

et al. 2021

Sensors

Self Cite

View full text Add to dashboard Cite

show abstract

“…22,27 Another limitation in the literature derives from ANNs trained solely with simulated data based on the activated sludge models (ASMs) with simple process layouts. 25,26 Moreover, simulating the long-term dynamic treatment performance of WRRFs is challenging because of the lack of detailed knowledge of influent wastewater characteristics and how rainfall events would translate to plant influent and operational changes (e.g., sludge wastage, airflow, recycle rate, etc.) over the long-term.…”

Section: Introductionmentioning

confidence: 99%

Evaluating Long-Term Treatment Performance and Cost of Nutrient Removal at Water Resource Recovery Facilities under Stochastic Influent Characteristics Using Artificial Neural Networks as Surrogates for Plantwide Modeling

Emaminejad

Aguiar

et al. 2021

ACS EST Eng.

View full text Add to dashboard Cite

Integrated watershed modeling is needed to couple water resource recovery facilities (WRRFs) with agricultural management for holistic watershed nutrient management. Surrogate modeling can facilitate model coupling. This study applies artificial neural networks (ANNs) as surrogate models for WRRF models to efficiently evaluate the long-term treatment performance and cost under influent fluctuations. Specifically, we first developed five WRRFs, including activated sludge, activated sludge with chemical precipitation (ASCP), enhanced biological phosphorus removal (EBPR), EBPR with acetate addition (EBPR-A), and EBPR with struvite recovery (EBPR-S), in a high-fidelity simulation program (GPS-X). The five WRRFs were based on an existing plant that treats combined domestic and industrial wastewater. The ANNs have satisfactory performance in capturing nonlinear biological behaviors for all five WRRFs, even though the prediction performance (R-square) slightly decreases as the model complexity increases. We advanced ANNs application in WRRF models by simulating long-term (10-yr) performance with monthly influent fluctuations using ANNs trained by simulation data from steady-state models and evaluated their performance on Phosphorus (P) and Nitrogen (N) removal. EBPR-S shows the most resilience, while EBPR is more sensitive to influent characteristics impacted by stormwater inflow. When comparing life cycle costs of N and P removal for each layout over the 10-yr simulation period, EPBR-S is the most cost-effective alternative, highlighting both the operational and cost benefits of side-stream P recovery. By capturing both nonlinear behaviors of biological treatment and operating costs with computationally lean ANNs, this study provides a paradigm for integrating complex WRRF models within integrated watershed modeling frameworks.

show abstract

“…Recently, industrial scenarios are the study-case for soft-sensing based on LSTM. In [ 40 , 41 ], authors proposed soft sensors based on LSTM for quality prediction in wastewater treatment plants. Moreover, LSTM-based soft sensors to estimate key quality variables in the fermentation process and debutanizer column [ 42 ].…”

Section: Introductionmentioning

confidence: 99%

Industrial Semi-Supervised Dynamic Soft-Sensor Modeling Approach Based on Deep Relevant Representation Learning

Lima

Araújo

2021

Sensors

View full text Add to dashboard Cite

Soft sensors based on deep learning have been growing in industrial process applications, inferring hard-to-measure but crucial quality-related variables. However, applications may present strong non-linearity, dynamicity, and a lack of labeled data. To deal with the above-cited problems, the extraction of relevant features is becoming a field of interest in soft-sensing. A novel deep representative learning soft-sensor modeling approach is proposed based on stacked autoencoder (SAE), mutual information (MI), and long-short term memory (LSTM). SAE is trained layer by layer with MI evaluation performed between extracted features and targeted output to evaluate the relevance of learned representation in each layer. This approach highlights relevant information and eliminates irrelevant information from the current layer. Thus, deep output-related representative features are retrieved. In the supervised fine-tuning stage, an LSTM is coupled to the tail of the SAE to address system inherent dynamic behavior. Also, a k-fold cross-validation ensemble strategy is applied to enhance the soft-sensor reliability. Two real-world industrial non-linear processes are employed to evaluate the proposed method performance. The obtained results show improved prediction performance in comparison to other traditional and state-of-art methods. Compared to the other methods, the proposed model can generate more than 38.6% and 39.4% improvement of RMSE for the two analyzed industrial cases.

show abstract

LSTM-Based Wastewater Treatment Plants Operation Strategies for Effluent Quality Improvement

Cited by 40 publications

References 22 publications

Industrial Control under Non-Ideal Measurements: Data-Based Signal Processing as an Alternative to Controller Retuning

Industrial Control under Non-Ideal Measurements: Data-Based Signal Processing as an Alternative to Controller Retuning

Evaluating Long-Term Treatment Performance and Cost of Nutrient Removal at Water Resource Recovery Facilities under Stochastic Influent Characteristics Using Artificial Neural Networks as Surrogates for Plantwide Modeling

Industrial Semi-Supervised Dynamic Soft-Sensor Modeling Approach Based on Deep Relevant Representation Learning

Contact Info

Product

Resources

About