Power Load Prediction Using 3D Time-Scale Operation Modes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current power load prediction methods based on growth rate are limited by fixed user tags, which fail to reflect the latest conditions, resulting in inaccurate predictions.
Innovation Solution
A power load prediction method that converts historical one-dimensional time series data into a three-dimensional matrix, allowing for the division into operation modes based on time scales, and uses these to derive value bands for predicting future power load data, incorporating external data relevance for improved accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If power load prediction is based on growth rate calculated from fixed user tags, then the prediction method is simple to implement, but the prediction accuracy is limited because the tags cannot reflect the latest conditions
Solution Approach 1:
The patent segments the historical power load data into different time scales (e.g., yearly, monthly, weekly) and further divides each time scale into different operation modes based on characteristics such as weekdays, weekends, and holidays. This segmentation allows the system to capture nuanced patterns in power consumption that fixed user tags cannot reflect, thereby improving prediction accuracy while maintaining computational feasibility through structured data organization.
Solution Approach 2:
The patent transforms the traditional one-dimensional time series data into a multi-dimensional data structure by introducing time scale and operation mode dimensions. This dimensional expansion enables the system to analyze power load patterns across multiple granularities and contexts simultaneously, capturing complex temporal patterns that single-dimensional approaches miss, thus resolving the accuracy-implementation complexity trade-off.
2Measurement precision
If historical power load data is analyzed in detail with multiple time scales and operation modes, then prediction accuracy is improved, but the data processing complexity increases
Solution Approach 1:
The patent performs preliminary actions by pre-processing historical power load data into a structured multi-dimensional format with defined time scales and operation modes before actual prediction. This pre-organization of data including the identification and categorization of different operation modes (weekdays, weekends, holidays) reduces the computational burden during the prediction phase, as the complex pattern recognition has already been partially accomplished during data preparation.
Solution Approach 2:
The patent changes the parameters of data representation by transforming raw time series data into a multi-dimensional structure with specific parameters including time scale, operation mode, and aggregated load values. This parameter transformation simplifies subsequent analysis by converting complex temporal patterns into structured data that can be more efficiently processed and compared across different time scales and operational contexts.
Data Source
AI summary
Disclosed are a power grid user classification method and device, and a computer-readable storage medium. The method includes: determining user power consumption data of each time period within a time interval, the user power consumption data of each time period including user power consumption data of each time granularity within the time period; generating a power consumption pattern image of a user within the time interval based upon the user power consumption data of each time period within the time interval; and classifying the user based upon an image recognition result of the power consumption pattern image. A user is classified by performing image recognition on a power consumption pattern image, and file information does not need to be manually inputted, thereby reducing a manual workload. In addition, a completion defect caused by incomplete file information is avoided, thereby improving the accuracy of classification.