Distribution Network Planning Evaluation Using Live Maps and U-Net
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional distribution network planning evaluation methods face inefficiencies due to manual intervention, inconsistent standards, and lack of integration with machine learning, leading to suboptimal power distribution system performance.
Innovation Solution
A live digital map driven distribution network planning scheme using a U-Net convolutional neural network trained with an awakening-sleep mechanism, combined with data-driven information entropy weight calculation, to evaluate planning schemes objectively and efficiently.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional distribution network planning evaluation methods are used, then manual intervention is required which ensures some level of control, but the evaluation efficiency is restricted and too much manual intervention is required
Solution Approach 1:
The system enables self-service evaluation by automatically processing planning schemes through the neural network model. The model independently performs feature extraction, evaluation metric calculation, and scheme ranking without requiring manual intervention at each step, thus dramatically improving evaluation efficiency while maintaining consistency
Solution Approach 2:
The patent replaces the mechanical manual evaluation process with an intelligent system based on neural networks and automated algorithms. The mechanical system of human reviewers manually assessing each scheme is substituted with an automated computational system that processes schemes consistently and efficiently
2Reliability
If traditional evaluation methods are used, then planning standards can be maintained, but inconsistent planning standards and subjective influence occur
Solution Approach 1:
The patent transforms the evaluation process from subjective qualitative assessment to objective quantitative analysis by changing the parameters used for evaluation. The system uses standardized numerical metrics and weighted scoring based on multiple evaluation dimensions, eliminating subjective influence and ensuring consistent planning standards across all evaluations
Solution Approach 2:
The neural network model serves multiple functions: it extracts features from diverse input data, evaluates different planning schemes against unified standards, and provides consistent rankings. This universal evaluation framework ensures that the same objective criteria are applied to all schemes regardless of complexity or scale
3Measurement precision
If machine learning methods are not fully utilized, then simpler evaluation processes are maintained, but the intelligent level and accuracy of distribution network evaluation need to be further improved
Solution Approach 1:
The patent introduces an intermediary neural network model that bridges the gap between raw planning data and evaluation results. This intermediary system processes complex relationships in the data that cannot be directly analyzed, extracting meaningful patterns and features that improve evaluation accuracy while managing model complexity through structured architecture
4Productivity
If automated evaluation is implemented, then evaluation efficiency is improved, but the system requires sophisticated machine learning models
Solution Approach 1:
The patent segments the evaluation system into distinct functional modules: data preprocessing, feature extraction, metric calculation, and result generation. This segmentation allows the sophisticated neural network to be broken down into manageable components, improving evaluation speed through parallel processing while making the overall system complexity more controllable and understandable
Data Source
AI summary
The present invention discloses a live digital map driven distribution network planning scheme intelligent evaluation method and system, relating to the field of intelligent distribution network planning and evaluation. The method comprises: acquiring a live digital map image, acquiring plot information by using U-net convolutional neural network plot identification strategy based on a stacked auto-encoder, and selecting alternative planning schemes according to index hierarchy structures and data driving evaluation. By adopting the present invention, the precision and efficiency of planning scheme evaluation is improved.