Water Infrastructure Setpoint Scheduling Using Predictive Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional control systems for water resource infrastructures are cumbersome, expensive, and computationally inefficient, requiring significant resources and being difficult to implement in real-time due to their complexity and reliance on physically-based numerical models that are not easily integrable with SCADA systems.
Innovation Solution
A control mechanism scheduler that receives operating and disturbance data to generate simulations and schedules for actuable infrastructure components, using pattern recognition algorithms and machine learning to automatically adjust control mechanisms and achieve predetermined objectives such as improved reliability and reduced costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional physically-based numerical models are used to model water resource infrastructures, then the model can capture system characteristics, but the model becomes computationally inefficient and difficult to implement in real-time
Solution Approach 1:
The patent replaces traditional physically-based numerical models with machine learning-based predictive models. The system uses historical data and sensor inputs to train models that predict system behavior without requiring complex physical equations, thereby substituting mechanical/mathematical modeling with data-driven approaches that are computationally more efficient
Solution Approach 2:
The patent creates simplified digital representations of the water resource infrastructure using machine learning models that copy the essential behavior patterns from historical data. These models replicate system characteristics without requiring the full complexity of physical models, enabling real-time prediction and control
2Reliability
If traditional physically-based numerical models are used to model water resource infrastructures, then the model can capture system characteristics, but the model design and calibration becomes cumbersome and expensive
Solution Approach 1:
The machine learning models automatically learn system characteristics from historical data without requiring manual calibration or tuning by experts. The models self-adjust their parameters during the training process, eliminating the cumbersome manual calibration process required by traditional physical models
Solution Approach 2:
The patent replaces the manual model design and calibration process with automated machine learning model training. Instead of requiring experts to manually adjust physical parameters, the system automatically learns from data, significantly reducing development time and cost
3Reliability
If traditional physically-based numerical models are used, then system characteristics can be modeled, but the models require a large number of parameters that are not directly measurable
Solution Approach 1:
The patent uses machine learning models that learn system behavior patterns directly from measurable sensor data without requiring unmeasurable physical parameters. The models copy the essential input-output relationships observed in historical data, eliminating the need for difficult-to-measure parameters
4Reliability
If traditional physically-based numerical models are used, then system characteristics can be captured, but the models require significant staff resources to maintain
Solution Approach 1:
The machine learning models are designed to automatically update and adapt to changing system conditions using ongoing sensor data. The models self-maintain their accuracy without requiring significant manual intervention or expert staff resources, unlike traditional models that require continuous calibration and validation
5Reliability
If traditional physically-based numerical models are used, then system characteristics can be modeled, but the models are not available for integration with SCADA systems
Solution Approach 1:
The patent replaces complex physical models with machine learning models that have standardized interfaces compatible with SCADA systems. The data-driven approach allows for easier integration with existing control systems through standard data exchange protocols, unlike traditional models that require custom integration efforts
Data Source
AI summary
A system and method control a water resource infrastructure (WRI). The WRI has infrastructure components that are actuatable to cause a change to the WRI. A monitoring system has sensors that collect operating data that describes a state of the infrastructure components. A disturbance data provider provides disturbance data that may be expected to have an impact on operational parameters of the infrastructure components. A control mechanism scheduler receives the disturbance data and the operating data, trains to generate a schedule of setpoints for a control system in accordance with approaching a predetermined objective, and retrieves and outputs the schedule of setpoints in response to receiving real-time operational data. A control system receives the schedule of setpoints controls the infrastructure components based thereon.