In brief

QuestionCan an electrochemical sensor plus machine learning turn COD into a 10-second online measurement?
ResultThe reported system performs online COD detection in about 10 s with 96% accuracy and was deployed in an industrial setting.
Why it mattersIt is an end-to-end example in which electrode design, electrochemical response, and model inference are built around a process-monitoring task.
Machine learning-driven COD real-time monitoring electrochemical analysis system based on nitrogen-etched boron-doped diamond electrode
Ruiyu Rong et al. · Chemical Engineering Journal · Published 1 May 2026
DOI: 10.1016/j.cej.2026.175515

The electrochemical problem

Chemical oxygen demand is widely used as an aggregate measure of oxidizable material in water, but conventional COD analysis is not naturally suited to fast, continuous process control. An electrochemical method could be faster, provided that complex sample responses can be related reliably to COD.

The study combines an engineered diamond electrode with machine learning to convert electrochemical measurements into a rapid online estimate.

What the researchers did

The authors fabricated a nitrogen-etched boron-doped diamond electrode and evaluated its electrocatalytic response to organic compounds. Machine learning was then incorporated into the electrochemical analysis system to map the measured response to COD.

The complete system was designed for online operation rather than only benchtop proof-of-concept testing, and it was subsequently evaluated in an industrial environment.

What they found

The paper reports online COD detection in approximately 10 seconds with 96% accuracy. The highlights also state that the system was successfully deployed and validated in an industrial setting.

The value of the result is the integration: the electrode provides a stable electrochemical signal, while the model handles the multicomponent relationship between that signal and the operational COD quantity.

Why this matters

This is closer to process analytics than to a conventional single-analyte electrochemical sensor. COD is itself a composite quantity, so a data-driven mapping from electrochemical oxidation response to the reference metric is a natural approach.

Rapid online inference could enable feedback control in treatment systems that would be difficult with slow laboratory COD measurements.

One important caution

COD is an operational aggregate metric, not a unique molecular composition. A model calibrated on one family of industrial wastewaters may encounter a distribution shift when the mixture of organics changes substantially.

Therefore, the reported 96% accuracy should be understood in the context of the study dataset and deployment, not as a universal accuracy for every wastewater stream.

Primary source

Ruiyu Rong et al. Machine learning-driven COD real-time monitoring electrochemical analysis system based on nitrogen-etched boron-doped diamond electrode. Chemical Engineering Journal (2026). DOI: 10.1016/j.cej.2026.175515.

Research-note facts and quantitative values were checked against the publisher record linked above. The “Why this matters” and “One important caution” sections are EchemAI editorial interpretation.