Battery Defect Detection Using Mask RCNN for Automated Shunting

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for detecting defects in single crystalline silicon solar cells, such as manual inspection and machine-assisted manual detection, are inefficient and limited in identifying complex defects, leading to high labor costs and reduced quality detection efficiency.

Innovation Solution

A battery detection method utilizing a deep neural network model based on the mask Region Convolutional Neural Network (RCNN) algorithm to identify defects in solar cells, which includes obtaining pictures from the production line, training the model with historical data, and automatically shunting defective batteries, allowing for both simple and complex defect identification without manual intervention.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If manual quality detection is used to observe single crystalline silicon solar cells, then labor cost is high and efficiency is poor, but the method can identify various defects with human judgment

Engineering Contradiction:
Improvedetection efficiencyVSAvoidmanual intervention level
Core Design Contradiction:
ProductivityVSExtent of automation

Solution Approach 1:

The patent replaces the manual mechanical inspection system with an automated machine vision system that captures images of solar cells and uses deep learning algorithms to automatically identify defects. The system substitutes human visual inspection with computational image analysis, achieving both high automation and maintained detection capability.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The defect detection model performs self-learning and self-improvement through continuous training with newly annotated defect data. The system automatically updates its detection capabilities without requiring manual reprogramming, enabling it to adapt to new defect types autonomously while maintaining high detection efficiency.

Inventive Principle:
Principle #25Self-service

2Adaptability or versatility

If machine-assisted manual quality detection with fixed defect definitions is used, then simple defects can be identified, but complex defects are difficult to identify and detection efficiency is reduced

Engineering Contradiction:
Improvedefect identification capabilityVSAvoiddetection efficiency
Core Design Contradiction:
Adaptability or versatilityVSProductivity

Solution Approach 1:

The patent implements a dynamic defect detection system where the defect definitions and detection parameters are not fixed but can be continuously updated. The system adapts its detection criteria based on new training data and evolving defect patterns, allowing it to handle both simple and complex defects efficiently without being constrained by predetermined rigid categories.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system changes its detection parameters and model configurations based on the specific defect types being analyzed. By adjusting model parameters, threshold values, and detection sensitivity based on training data characteristics, the system optimizes its performance for different defect complexities while maintaining overall detection efficiency.

Inventive Principle:
Principle #35Parameter changes

3Reliability

If a defect detection model is trained with historical training data, then the model can identify known defects, but recent defects cannot be identified without model updates

Engineering Contradiction:
Improvedefect identification accuracyVSAvoidmodel update mechanism
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent implements a feedback mechanism where detection results are continuously monitored and used to update the training data set. When new defect types or patterns are detected, the system automatically incorporates this information into the training data, retrains the model, and improves detection accuracy for future cases, creating a continuous improvement cycle.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The system performs preliminary data collection and annotation in advance to prepare updated training data before model retraining is needed. By proactively gathering and preparing new defect data samples and their annotations, the system ensures that model updates can be performed efficiently when required, maintaining reliability without excessive complexity.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3683572B1Battery detection method and device
Publication Date: 2022.06.01 BEIJING BAIDU NETCOM SCI & TECH CO LTD
  • EP3683572B1 patent drawingFigure 1~2
  • EP3683572B1 patent drawingFigure 3~5
  • EP3683572B1 patent drawingFigure 6

AI summary

The present disclosure provides a battery detection method and a battery detection device. The method includes: obtaining a picture of each battery on a battery production line, and obtaining a corresponding production node; inputting the picture into a preset defect detection model, and obtaining a detection result output by the defect detection model, and when the detection result denotes that there is the defect on the picture, sending a control instruction to a control device of the production node corresponding to the picture, to cause the control device to shunt the battery corresponding to the picture having the defect based on the control instruction. The detection result includes whether there is a defect, a defect type, and a defect position.