Electricity Demand Prediction Using Individual Vehicle Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing electricity-demand prediction methods face challenges in accurately predicting electricity demand in specific areas due to limited record data, particularly when analyzing group behavior trends, as the number of probe cars may be insufficient to gather substantial data.
Innovation Solution
An electricity-demand prediction device and system that generates individual models based on vehicle probe data, incorporating factors like operating rate, presence proportion, and battery charging rate, and constructs lifestyle models to predict electricity demand with higher accuracy by reflecting user decision-making features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If group behavior trends in a specific area are statistically analyzed to construct models for reproducing features, then prediction accuracy can be improved, but a large amount of record data is required which may not be available due to limited probe cars
Solution Approach 1:
The patent segments the prediction approach by creating individual models for each probe car rather than aggregating data into group behavior models. This allows each model to be trained on limited individual data while collectively providing comprehensive coverage. The segmentation principle resolves the contradiction by shifting from requiring large aggregated datasets to utilizing multiple small individual datasets.
Solution Approach 2:
The patent introduces a new dimension by modeling at the individual vehicle level rather than the group level. This dimensional shift from aggregate statistics to individual instances allows the system to achieve prediction accuracy without requiring large volumes of aggregated record data, as each individual model contributes uniquely to the overall prediction capability.
2Measurement precision
If individual models are generated for each vehicle to reflect user decision-making features, then prediction accuracy is improved, but the complexity of the prediction system increases
Solution Approach 1:
The patent merges multiple individual models into a unified prediction framework. While individual models are created for each probe car, they are combined and integrated to produce collective predictions. This merging approach maintains high prediction accuracy by preserving individual characteristics while reducing overall system complexity through unified data structures and processing pipelines.
Solution Approach 2:
The patent creates individual models with universal applicability - each model uses the same structural framework and can process similar types of input data. This universality allows the system to handle multiple vehicles with consistent processing logic, reducing complexity compared to having entirely separate specialized systems for each vehicle.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
This electricity-demand prediction device (100) is provided with a data reception unit (101), an individual-model generation unit (112), and a demand-prediction computation unit (113). Vehicle probe data in which specific driving states are recorded is inputted to the data reception unit (101), and on the basis of said vehicle probe data, the individual-model generation unit (112) generates an individual model for each specific vehicle or for each user of a specific vehicle. Said individual models represent correlations between historical values associated with factor information regarding factors in decisions to charge specific vehicles at a specific charging facility and historical values representing electricity demand for said specific vehicles at said specific charging facility. The demand-prediction computation unit (113) computes a predicted electricity demand for the specific vehicles at the specific charging facility on the basis of the generated individual models.