Non-Volatile Memory Fault Detection Across Multiple Control Lines
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing fault detection methods for non-volatile memory are inaccurate and limited, primarily focusing on Word Lines, failing to provide comprehensive fault information and unable to identify specific bad lines, leading to inefficient and resource-intensive manual threshold setting.
Innovation Solution
A method involving data dimensionality reduction and machine learning models to analyze threshold voltage distribution data across multiple control lines, predicting failure probabilities and identifying bad lines through fusion models and abnormality scoring, enhancing detection accuracy and automation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If only Word Line control lines are detected in traditional fault detection schemes, then the detection process is simple, but the fault information obtained is limited and detection accuracy is poor
Solution Approach 1:
The patent segments the control lines into different types (Word Lines, Bit Lines, Select Lines, etc.) and performs separate analysis on each type. This segmentation allows comprehensive fault detection across all control lines while managing complexity through systematic categorization and independent processing of each control line type.
2Measurement precision
If comprehensive fault analysis of all control lines is performed, then detection accuracy improves, but the detection process becomes complex and time-consuming
Solution Approach 1:
The patent performs preliminary actions by pre-processing control line data, identifying control line types, and preparing fault analysis frameworks before actual fault detection. This preliminary preparation enables faster real-time fault detection by having analysis templates and processing pipelines ready in advance.
Solution Approach 2:
The patent implements dynamic fault detection by adaptively selecting analysis methods based on control line types and detected anomaly patterns. The system dynamically adjusts its analysis depth and focus areas, performing detailed analysis only where faults are suspected, thereby reducing overall detection time while maintaining accuracy.
3Extent of automation
If manual threshold setting is used for fault detection, then the system is easier to implement, but it requires significant manual intervention and resources
Solution Approach 1:
The patent implements self-service through automated threshold determination algorithms that learn optimal detection thresholds from historical fault data and control line characteristics. The system automatically adjusts thresholds based on observed patterns without requiring manual calibration, reducing manual intervention while maintaining detection effectiveness.
Solution Approach 2:
The patent incorporates feedback mechanisms where detection results are continuously fed back to refine threshold settings and improve detection algorithms. This closed-loop approach allows the system to automatically learn from past detections and adjustments, reducing manual tuning requirements over time while improving accuracy.
4Measurement precision
If detailed analysis of each control line is performed to identify specific bad lines, then fault localization precision improves, but the computational resources required increase
Solution Approach 1:
The patent applies partial action by performing detailed analysis only on control lines that exhibit suspicious characteristics or fall into high-risk categories. Rather than uniformly analyzing all control lines at maximum detail, the system focuses computational resources on promising candidates for faults, achieving good localization precision with reduced overall computational burden.
Data Source
AI summary
A fault detection method for a non-volatile memory, an apparatus, an electronic device and a storage medium are provided. The fault detection method for a non-volatile memory includes obtaining threshold voltage distribution data for a non-volatile memory to be detected, obtaining, from the threshold voltage distribution data, a data feature of each of a plurality of control line types, predicting a possibility of failure of each control line type based on the data feature of each control line type, to obtain a type prediction result, and performing a fault detection operation in the non-volatile memory based on the type prediction result of each control line type.