Unified Prediction Model for Cross-Device Anomaly Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems face challenges in performing device control and anomaly detection across different types of devices due to varying operational data characteristics, leading to prediction errors.
Innovation Solution
A system that corrects operational data to align with virtual operational data characteristics, using a prediction model trained with corrected data to perform device control and anomaly detection across multiple devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a prediction model is created for each device according to its characteristics, then prediction accuracy is improved, but device complexity and system complexity increase
Solution Approach 1:
Multiple device-specific prediction models are merged into a single unified prediction model. The system collects operational data from multiple devices with different characteristics, trains a single model on this diverse dataset, and uses it to predict outcomes for all devices. This eliminates the need to maintain separate models for each device while preserving prediction accuracy through comprehensive training data that captures various device characteristics.
Solution Approach 2:
A universal prediction model is developed that can handle multiple device types and characteristics simultaneously. The model is designed to be multi-functional, accepting operational data from various devices as input and providing accurate predictions across different device configurations. This universal approach reduces system complexity by replacing multiple specialized models with one adaptable model.
2Measurement precision
If operational data characteristics are corrected to match virtual operational data, then prediction accuracy across different devices is improved, but data processing complexity increases
Solution Approach 1:
The system transforms operational data parameters to match the characteristics of virtual operational data used during model training. This involves adjusting data distribution, scaling, or transformation of operational parameters so that real device data aligns with the training data characteristics. By changing parameters rather than creating separate models, the system maintains prediction accuracy while using a single unified model.
3Device complexity
If a single prediction model is used for multiple devices, then system complexity is reduced, but prediction accuracy deteriorates due to varying device characteristics
Solution Approach 1:
The system performs preliminary actions during the training phase by collecting and processing operational data from multiple devices with varying characteristics. This preparatory data collection and model training on diverse data ensures that the single prediction model is pre-adapted to handle different device characteristics, thereby maintaining prediction accuracy while using a unified model structure.
Data Source
AI summary
A system for performing device control or anomaly detection for a plurality of different devices using a prediction model trained with training data is provided. The system includes one or more processors; and memory storing a program that, when executed, causes the one or more processors to perform a process. The process includes: (a) correcting characteristics of operational data of the devices to approach characteristics of virtual operational data when creating the prediction model, and training the prediction model using the corrected operational data as training data; and (b) correcting characteristics of operational data of the devices in operation to approach the characteristics of the virtual operational data when operating the system, and inputting the corrected operational data to the prediction model.