Probabilistic Data Analysis for Reliable Abnormal State Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data analysis systems for industrial machinery face challenges in accurately identifying and estimating abnormal phenomena without relying on experienced technicians, as the reliability of analysis results is unclear due to varying conditions and uncertainties in control data.
Innovation Solution
A data analysis system that repeatedly inputs control data into a data analysis model with a parameter including a random variable, calculating estimates based on the distribution of output values to enhance accuracy by reflecting uncertainty and identifying unit phenomena to isolate specific abnormal states.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If data analysis is performed using a deterministic model, then the analysis process is simple and fast, but the reliability of analysis results is unclear due to varying conditions and uncertainties in control data
Solution Approach 1:
The patent transforms the deterministic model parameters into probabilistic parameters with random variables. The weight parameters w are changed from fixed values to random variables following probability distributions, allowing the model to capture uncertainties in the control data and provide more reliable analysis results under varying conditions.
Solution Approach 2:
The patent introduces dynamic sampling of weight parameters from probability distributions during the analysis process. Instead of using fixed deterministic weights, the model dynamically samples different weight configurations to reflect the varying conditions and uncertainties, making the analysis more adaptable to changing environments.
2Measurement precision
If a data analysis model with random variables is used, then the accuracy and reliability of analysis improve, but the computational complexity and processing time increase due to repeated inputs
Solution Approach 1:
The patent performs repeated model inputs (excessive action) to capture the distribution of output values, but only uses essential statistics (mean and variance) from these repetitions rather than processing every individual output. This partial processing approach maintains accuracy while reducing the computational burden of full repeated analysis.
Solution Approach 2:
The patent extracts key statistical features (mean and variance) from the distribution of repeated output values. By taking out only the essential characteristics needed for accurate abnormal phenomenon identification rather than analyzing the complete distribution, the system achieves high accuracy with reduced processing requirements.
3Reliability
If repeated model inputs are performed to capture output distribution, then the estimation accuracy improves by reflecting uncertainty, but the computational load and processing time increase
Solution Approach 1:
The patent extracts only the mean and variance from the distribution of repeated output values, rather than processing the complete distribution or every individual output. This extraction of essential statistical features maintains the reliability benefits of uncertainty incorporation while significantly improving processing efficiency.
Solution Approach 2:
The patent performs repeated model inputs to capture uncertainty (excessive action) but processes only a partial representation (mean and variance) of the output distribution. This partial processing approach maintains estimation reliability while reducing the computational load compared to analyzing the full distribution.
Data Source
AI summary
A data analysis system comprises a memory configured to store a data analysis model trained in advance by using training data configured to output an output value indicating whether a target to be analyzed is in a specific state in response to the data analysis model receives input data on the target to be analyzed, the data analysis model including a parameter that includes a random variable; and circuitry configured to input the input data to the data analysis model repeatedly a plurality of times and calculate an estimate indicating a degree of being in the specific state based on a distribution of a plurality of the output values output from the data analysis model for the plurality of times.