Digital Twin Generation Using AI Synthetic Data for Sensor Anomalies
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Generating accurate digital resources for digital environments, such as metaverse environments, is challenging when hardware sensors malfunction, as it affects the overall system and prevents accurate capture of real-world resources.
Innovation Solution
A system using partial sensor data and artificial intelligence that includes a sensor data analyzer engine and an on-demand synthetic data generator to detect anomalies, generate synthetic data, and create a digital twin of a resource, ensuring accurate representation even with malfunctioning sensors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional sensor systems are used to capture real-world resources, then digital resources can be generated, but the system fails when sensors malfunction or provide incomplete data
Solution Approach 1:
An AI engine is introduced as an intermediary component between the sensor system and the digital resource generation process. The AI engine receives partial or anomalous sensor data, processes it through machine learning models, and generates complete digital representations. This intermediary layer allows the system to tolerate sensor failures while maintaining output quality, directly resolving the contradiction between reliability and information completeness.
Solution Approach 2:
The system changes the parameter of data completeness from a hard requirement to a probabilistic target. Instead of requiring 100% complete sensor data, the AI engine processes partial data and generates digital resources with confidence scores. This parameter change allows the system to operate reliably even when sensor data is incomplete, transforming the contradiction into a manageable trade-off.
2Quantity of substance
If multiple sensors are interconnected to gather system logs and telemetry data, then comprehensive data collection is achieved, but a single sensor malfunction impacts the overall system
Solution Approach 1:
The AI engine extracts and processes only the necessary data elements from the sensor network, rather than requiring all sensors to function simultaneously. When certain sensors fail, the AI engine extracts available data from functioning sensors and uses AI processing to compensate for missing elements, maintaining system stability while preserving data quantity.
Solution Approach 2:
The system implements beforehand cushioning by training AI models on diverse sensor data patterns including failure modes. The AI engine is pre-prepared to handle various sensor malfunction scenarios through prior training, allowing it to cushion the impact of sensor failures and maintain reliable operation despite individual component failures in the interconnected sensor network.
3Measurement precision
If manual intervention is used to handle sensor anomalies, then data accuracy can be maintained, but processing speed and efficiency decrease
Solution Approach 1:
The AI engine implements self-service by automatically detecting sensor anomalies, determining data quality, and generating digital resources without human intervention. The system autonomously processes partial or anomalous sensor data, maintaining measurement precision through AI-based validation while achieving high processing speeds through automated workflows, eliminating the trade-off between accuracy and productivity.
Solution Approach 2:
The patent replaces the mechanical system of manual data verification and processing with an AI-based intelligent system. The AI engine substitutes human operators in detecting anomalies and processing sensor data, maintaining or improving measurement precision through sophisticated algorithms while dramatically increasing productivity by eliminating manual intervention bottlenecks.
Data Source
AI summary
Systems, computer program products, and methods are described herein for generating a digital twin or a resource using partial sensor data and artificial intelligence. The present invention is configured to receive resource sensor data from a plurality of sensors, wherein the plurality of sensors is associated with a resource; apply a sensor data analyzer engine to the resource sensor data; determine, by the sensor data analyzer engine, whether at least one sensor anomaly of the resource sensor data is present; apply an on-demand synthetic data generator to the at least one sensor anomaly; generate synthetic sensor data associated with the resource, wherein the synthetic sensor data is based on a real-time pattern of the resource sensor data from the plurality of sensors; and generate, based on the resource sensor data from the plurality of sensors and the synthetic sensor data, a digital twin of the resource.