Abnormality Score Normalization for Heterogeneous Manufacturing Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In the manufacturing industry, detecting product defects or failures in manufacturing devices at an early stage is challenging due to the complexity and variety of data types involved, such as continuous, discrete, and text data, which requires manual analysis and is difficult to process efficiently, especially with the increasing volume of manufacturing IoT data.
Innovation Solution
An abnormality score calculation apparatus that acquires manufacturing data, calculates abnormality scores for various modes, and converts their scale to make them equal in occurrence degree, using feature quantities and statistical modeling to normalize and compare scores across different types and sources, facilitating intuitive analysis and decision-making.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If statistical processing is performed on various types of manufacturing data (continuous, discrete, text data with different decimal point accuracies), then the ability to detect abnormalities and estimate causes is improved, but the complexity of data processing and analysis burden increases significantly
Solution Approach 1:
The patent transforms heterogeneous manufacturing data (continuous, discrete, text) with different properties and decimal point accuracies into a unified parameter format. By converting all data types to a common numerical representation with standardized precision, the system enables consistent statistical processing across diverse data sources while reducing processing complexity.
Solution Approach 2:
The patent segments the complex data processing task into distinct stages: data acquisition with type identification, decimal point accuracy standardization, abnormality score calculation for each data type, and aggregate score computation. This segmentation allows each stage to handle specific data characteristics independently, reducing overall processing complexity.
2Measurement precision
If manual analysis is performed on huge amounts of manufacturing data, then detailed inspection is possible, but the analysis burden on operators becomes significant and efficiency is extremely difficult to maintain
Solution Approach 1:
The patent replaces manual mechanical analysis with an automated computational system that calculates abnormality scores using statistical processing. The system automatically acquires manufacturing data, standardizes formats, computes scores for each data type, and generates aggregate results, eliminating the need for operators to manually analyze huge amounts of data while maintaining thoroughness.
Solution Approach 2:
The patent introduces abnormality scores as an intermediary metric between raw manufacturing data and operator decision-making. Instead of requiring operators to directly interpret complex multi-type data, the system converts all data into standardized abnormality scores that indicate deviation from normal conditions, simplifying the analysis interface while preserving detailed information.
3Adaptability or versatility
If data with different properties (continuous value data, discrete value data, text data) are processed together, then comprehensive abnormality detection is achieved, but the difficulty of processing and standardizing data increases
Solution Approach 1:
The patent creates a universal data processing framework that handles multiple data types (continuous, discrete, text) through a single standardized abnormality score calculation mechanism. Each data type is processed according to its specific characteristics but ultimately converted to the same numerical score format, enabling comprehensive detection while simplifying standardization.
Solution Approach 2:
The patent applies parameter transformation by converting diverse data types into a unified abnormality score parameter. Continuous data, discrete data, and text data each undergo type-appropriate transformations but all result in standardized numerical scores with consistent decimal point accuracy, reducing the difficulty of processing heterogeneous data together.
Data Source
AI summary
An abnormality score calculation apparatus according to an embodiment includes a processing circuit configured to: acquire first data concerning a status of a product or a manufacturing process; calculate based on the first data an abnormality score for a respective one of a plurality of abnormality modes or for a respective one of a plurality of pieces of the first data of various types; and convert a scale of a respective one of a plurality of abnormality scores including the abnormality score in such a manner that the abnormality scores become substantially equal in occurrence degree.