Remaining-Amount Prediction Switching for Empty-Day Accuracy and Speed
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing remaining amount prediction systems for consumable products in electronic devices, such as toner in image forming apparatuses, face inaccuracies in predicting the empty day due to the influence of earlier or later consumable product amounts on the prediction, leading to potential misalignment with the actual empty day.
Innovation Solution
A system that switches prediction methods based on the amount of consumable product remaining, using linear approximation when the amount is above a threshold and a machine learning model when it drops below that threshold, thereby improving prediction accuracy and reducing processing time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a machine learning model is used to predict the amount of remaining consumable product, then prediction accuracy is improved, but processing time increases
Solution Approach 1:
The system dynamically switches between linear approximation and machine learning model based on the remaining amount of consumable product. When the remaining amount is large, linear approximation is used for fast prediction. When the remaining amount becomes small, the system transitions to using the machine learning model to maintain high prediction accuracy, thus resolving the contradiction between speed and accuracy.
Solution Approach 2:
The system changes the prediction method based on the parameter of remaining consumable product amount. By setting a threshold amount, the system switches from linear approximation to machine learning model when the remaining amount drops below the threshold, optimizing both processing time and accuracy based on the current state.
2Loss of time
If linear approximation is used to predict the empty day, then processing time is reduced, but prediction accuracy deteriorates
Solution Approach 1:
The system dynamically selects the prediction method based on the remaining consumable product amount. For large remaining amounts, linear approximation provides sufficient accuracy with fast processing. When the remaining amount decreases below a threshold, the system switches to the machine learning model to maintain accuracy, thus resolving the trade-off between speed and precision.
3Device complexity
If a single prediction method is used throughout the entire prediction period, then device complexity is reduced, but prediction accuracy deteriorates
Solution Approach 1:
The system employs a dynamic prediction approach where the method (linear approximation or machine learning) is selected based on the remaining consumable product amount. This dynamic switching increases operational complexity but maintains manageable system architecture while significantly improving prediction accuracy across different stages of consumable depletion.
Solution Approach 2:
The system changes the prediction approach based on the parameter of remaining consumable amount. By using a threshold-based switching mechanism, the system adapts its behavior based on current conditions, improving accuracy without requiring an overly complex unified model.
Data Source
AI summary
A remaining amount prediction system includes a storage device and a controller. The storage device stores actual measurement value data indicating an amount of actually remaining consumable product used in an electronic device. The controller includes a processor and functions as a remaining amount predictor by the processor executing a remaining amount prediction program. The remaining amount predictor predicts an empty day when the amount of remaining consumable product will become equal to or less than a specific amount on the basis of the amount of actually remaining consumable product. The remaining amount predictor switches a prediction method to be used to predict the empty day from a linear approximation using method using linear approximation to a machine learning model using method using a machine learning model in a process in which the amount of actually remaining consumable product decreases.