Sensor-Trained Peripheral Machine Control for Factory Energy Demand
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing industrial control systems for manufacturing facilities require specialized knowledge and are difficult to update, leading to inefficient energy usage and resource management due to the need for custom-made solutions that are challenging to maintain with changes in manufacturing processes or equipment.
Innovation Solution
A machine learning-based control system that uses sensor data to train a model for real-time control of peripheral machines, optimizing energy consumption while ensuring energy demand is met, without requiring specialized knowledge of the manufacturing facility or process, and can be updated automatically.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If custom-made control solutions are implemented for each factory, then the system can be tailored to specific manufacturing needs, but the system becomes difficult to update and maintain when manufacturing processes or equipment change
Solution Approach 1:
The control system automatically adapts to changes in the manufacturing facility by using machine learning algorithms that learn from sensor data and adjust control parameters without requiring expert intervention. The system performs self-updating and self-optimization, eliminating the need for manual reconfiguration when processes or equipment change.
Solution Approach 2:
The system transitions from static custom-made configurations to dynamic adaptive control. Control parameters are continuously adjusted based on real-time sensor data and learned patterns, allowing the system to automatically adapt to changing manufacturing conditions without manual intervention.
2Loss of energy
If expert designers manually plan and program control solutions, then the system can be optimized for energy efficiency, but the process requires specialized knowledge and is time-consuming
Solution Approach 1:
The patent replaces manual expert planning and programming with automated machine learning algorithms. The system uses sensor data to automatically learn optimal control strategies for energy efficiency, substituting human expert knowledge with computational intelligence that continuously optimizes energy usage without requiring specialized human intervention.
Solution Approach 2:
The system implements continuous feedback loops where sensor data from the manufacturing facility is constantly monitored, analyzed, and used to adjust control parameters. This closed-loop approach automatically optimizes energy efficiency by learning from actual system performance and making real-time adjustments without expert intervention.
3Ease of manufacture
If traditional control systems are used, then implementation is straightforward, but the systems cannot automatically adapt to changes in the manufacturing facility
Solution Approach 1:
The system performs preliminary learning during a training period where sensor data is collected and used to train machine learning models before actual control begins. This preliminary action prepares the system to automatically adapt to the specific manufacturing facility characteristics without requiring manual configuration, combining implementation simplicity with adaptive capability.
Data Source
AI summary
A method of applying feedback control on peripheral units supplying energy to manufacturing units in a manufacturing facility uses a trained model, and includes:mapping supply relations between peripheral units and manufacturing units into a schema;establishing communications with sensors monitoring peripheral unit metrics indicative of energy transfer from the peripheral units to the manufacturing units;training the model based on the schema and training data gathered by communication with the sensors during a training period, the trained model predicting energy usage by the peripheral machines with a specified degree of accuracy; andduring a control period following the training period, gathering further data from the sensors and minimizing energy usage by the peripheral units while supplying the total energy demanded by the manufacturing units by controlling at least one of the peripheral units based on an outcome of inputting the further data into the trained model.