Railroad Vehicle Abnormality Diagnostic Device Using Dynamic Sensor Data Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional abnormality diagnostic devices for railroad vehicles face challenges in accurately diagnosing issues due to dynamic and mutually dependent sensor data conditions such as time, location, weather, and passenger load, which complicates the diagnosis of components like brakes.
Innovation Solution
The diagnostic device learns data selection conditions based on sensor data from railroad vehicles, using a model that distinguishes between different operational conditions to selectively utilize relevant data for accurate abnormality diagnosis, incorporating a processing circuitry that generates a data selection condition to enhance diagnostic accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If arbitrary sensor data is applied to a diagnostic model without distinguishing operational conditions, then the diagnostic process is simple, but the diagnostic accuracy deteriorates
Solution Approach 1:
The patent segments sensor data based on operational conditions (time periods, traveling positions, weather conditions, passenger loads). By dividing the diagnostic process into condition-specific segments, the system selects only relevant sensor data for each diagnostic task, improving accuracy while maintaining operational simplicity through automated condition-based filtering.
Solution Approach 2:
The patent applies local quality by tailoring the diagnostic approach to specific operational conditions. Different sensor data selection criteria are applied locally according to the current operational context (e.g., different criteria for flat sections vs. inclined sections), ensuring that the diagnostic model receives high-quality, condition-appropriate data without requiring complex manual intervention.
2Quantity of substance
If all sensor data is utilized for diagnosis regardless of conditions, then data completeness is high, but diagnostic reliability deteriorates due to irrelevant data
Solution Approach 1:
The patent extracts only the relevant sensor data needed for specific diagnostic tasks by establishing data selection conditions based on operational context. For example, when diagnosing brake abnormalities, the system extracts sensor data only during periods when brakes are actually used, removing irrelevant data that would otherwise degrade diagnostic reliability while maintaining sufficient data quantity for accurate diagnosis.
Solution Approach 2:
The patent changes the parameters of data selection based on operational conditions. By dynamically adjusting which sensor data parameters are selected (time periods, positions, weather conditions, passenger loads), the system maintains high data completeness for relevant conditions while filtering out irrelevant data, thereby improving diagnostic reliability without sacrificing necessary data quantity.
3Adaptability or versatility
If sensor data from all operational conditions is used, then the diagnostic coverage is comprehensive, but the diagnostic precision deteriorates due to mixed characteristics
Solution Approach 1:
The patent implements dynamic data selection that adapts to current operational conditions. The system automatically adjusts which sensor data is used for diagnosis based on real-time conditions (time, position, weather, passenger load), maintaining comprehensive diagnostic coverage across different scenarios while ensuring high precision within each specific condition through context-appropriate data selection.
Data Source
AI summary
According to one embodiment, an abnormality diagnostic device includes processing circuitry. The processing circuitry learns, based on a model generated from sensor data of a diagnostic object in a railroad vehicle, a data selection condition for selecting the sensor data utilized to diagnose the diagnostic object. The processing circuitry diagnoses abnormality of the diagnostic object based on the sensor data satisfying the data selection condition and a diagnostic model representing a relation between the sensor data and the abnormality of the diagnostic object.