System for energy management in a water treatment plant

A system for energy management in a water treatment plant, the system comprising:

• at least one processor;
• at least one non-transitory computer-readable medium coupled to the processor, the computer-readable medium storing:
• a low-level energy optimization machine-learned model (LLEOMLM) comprising processor-executable instructions that when executed by the processor, cause the system to perform operations, the operations comprising:
• obtaining LLEOMLM input data from an industrial control system associated with the water treatment plant, the LLEOMLM input data comprising plant operating data for the water treatment plant;
• obtaining LLEOMLM constraints data for the water treatment plant from the industrial control system;
• and
• processing the LLEOMLM input data and LLEOMLM constraints data to provide as output, a current maintenance schedule configured to minimize operating cost for the water treatment plant during a LLEOMLM optimization period, the current maintenance schedule comprising one or more maintenance process performance times and a corresponding predicted operating cost value calculated for each of the maintenance process performance times;
• and
• a clean-in-place (CIP) system configured to execute a CIP maintenance process on equipment of the water treatment plant responsive to the output of the LLEOMLM for maintaining the water treatment plant.

A computer-implemented method for maintaining a water treatment plant, the method comprising:

• receiving input data from an industrial control system associated with the water treatment plant, the input data comprising plant operating data for the water treatment plant;
• receiving constraints data for the water treatment plant from the industrial control system, the constraints data indicating at least one or more optimization periods, an energy consumption objective, and a cost objective;
• determining whether to execute at least one of a high-level energy optimization machine-learned model (HLEOMLM) and a low-level energy optimization machine-learned model (LLEOMLM) based on at least one of the input data and constraints data;
• executing at least one of the HLEOMLM and the LLEOMLM in response to the determining, wherein executing the HLEOMLM comprises processing the input data and constraints data to provide as output, a long-term maintenance schedule configured to minimize energy consumption for the water treatment plant, the long-term maintenance schedule comprising one or more maintenance process performance times and a corresponding predicted energy consumption value calculated for each of the maintenance process performance times, and wherein executing the LLEOMLM comprises processing the input data and constraints data to provide as output, a current maintenance schedule configured to minimize operating cost for the water treatment plant, the current maintenance schedule comprising one or more maintenance process performance times and a corresponding predicted operating cost value calculated for each of the maintenance process performance times;
• and
• executing at a clean-in-place (CIP) system, a CIP maintenance process on equipment of the water treatment plant responsive to the output of at least one of the HLEOMLM and the LLEOMLM for maintaining the water treatment plant.