Prediction Model Generation Device for Power Demand Forecasting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing prediction models for power demand are not effective across varying conditions such as date, time, and weather, leading to fluctuating prediction performance.
Innovation Solution
A prediction model generation device that generates multiple prediction models based on classification of data into classes, using a learning dataset and evaluation dataset to calculate parameters and evaluate performance, with the option to adjust weights for ensemble models, thereby improving prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single prediction model is used for power demand forecasting, then the model structure is simple and easy to implement, but the prediction performance fluctuates greatly depending on date, time, and weather conditions
Solution Approach 1:
The patent divides the single prediction model into multiple submodels, each specialized for specific conditions (date, time, weather). The class generation unit creates distinct classes based on these conditions, and separate prediction models are generated for each class, allowing each model to be optimized for its specific condition rather than trying to handle all conditions with one general model
Solution Approach 2:
The system dynamically selects and switches between different prediction models based on current conditions. The class generation unit determines which class the current situation belongs to, and the corresponding prediction model is selected from the plurality of generated models, making the system adaptable to changing conditions rather than static
2Reliability
If multiple prediction models are generated for different conditions, then the prediction performance becomes more stable across varying conditions, but the device complexity and computational requirements increase
Solution Approach 1:
Each prediction model is specialized for specific local conditions (particular date types, time periods, weather conditions) rather than being a general-purpose model. The class generation unit identifies specific condition combinations, and prediction models are tailored to each class, giving each model local expertise for its designated condition range
Solution Approach 2:
While each prediction model is specialized, the overall system achieves universality by having multiple models that collectively cover all possible conditions. The class generation unit manages the selection among these models, making the system as a whole capable of handling diverse conditions through a unified framework
Data Source
AI summary
A prediction model generation device has a first storage unit that stores a plurality of explanatory variables, a second storage unit that stores a plurality of objective variables, an input unit that inputs instruction information on classification, a class generation unit that generates a plurality of classes based on the instruction information, and a prediction model calculation unit that calculates a plurality of prediction models corresponding to the plurality of classes. The prediction model calculation unit has a learning data set extraction unit that extracts a learning data set corresponding to each of the plurality of classes from among the plurality of explanatory variables and the plurality of objective variables.