Predictive sensor system, method, and computer program product
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing time series forecasting devices struggle to accurately predict changes in data over long periods due to limitations in kernel function selection, which restricts their ability to analyze data beyond a certain time segment.
Innovation Solution
A predictive sensor system that employs two trained models, one suitable for near-future forecasting and another for distant future forecasting, which are combined using a hyperbolic function to generate predictions for intermediate time segments, allowing for accurate forecasting over extended periods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single kernel function is selected for a single time segment, then forecasting accuracy is improved for that specific segment, but the ability to forecast accurately over long periods deteriorates
Solution Approach 1:
The patent divides the forecasting task into multiple time segments (first forecasting segment and second forecasting segment), each handled by a specialized trained model. The first trained model focuses on near-future segments while the second trained model handles distant future segments, allowing each model to optimize for its specific time range rather than attempting to cover all periods with a single model
Solution Approach 2:
The forecasting device is designed to perform multiple forecasting functions by integrating two different trained models. The system can forecast both near-future and distant future time segments using a unified device architecture that selects and applies the appropriate model based on the target forecasting period
2Measurement precision
If multiple trained models are used for different time segments, then forecasting accuracy over long periods is improved, but device complexity increases
Solution Approach 1:
The system dynamically selects which trained model to apply based on the characteristics of the forecasting segment. The device can adaptively choose between the first trained model for near-future segments and the second trained model for distant future segments, making the system flexible and intelligent rather than statically complex
Solution Approach 2:
The patent introduces a selection mechanism that acts as an intermediary between the multiple trained models and the forecasting task. This mediator determines which model is most appropriate for the given forecasting segment, simplifying the overall system architecture by providing a clear decision-making framework rather than requiring complex interactions between models
Data Source
AI summary
A predictive sensor system has a memory that collects time series data from a sensor that detects a parameter of an atmosphere in an inhabitable space. Circuitry implements a trained predictive sensor model and generates predicted sensor measurement data for a future segment of time. The trained predictive sensor model includes a first trained model and a second trained model. The first trained model is more suitable for forecasting the predicted data in a first forecasting segment than the second trained model. The second trained model is more suitable for forecasting the predicted data in a second forecasting segment than the first trained model wherein the second forecasting segment comes later than the first forecasting segment. The circuitry also forecasts the predicted data in another time segment between the first forecasting segment and the second forecasting segment based on the time series data and the trained predictive sensor model.