AI Telemetry Channel Identification for Mixed Construction Fleets
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing fleet management systems struggle to identify and manage telemetry data from construction machines of uncertain origin, particularly when machines are from different manufacturers, due to varying telemetry channel identifications and proprietary data practices.
Innovation Solution
An apparatus and process utilizing artificial neural networks (ANN) to identify telemetry data by configuring weights based on decision criteria and deep knowledge of construction machinery, assigning candidate channel IDs, and computing confidence measures to ensure accurate identification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional fleet management systems are used to manage telemetry data from diverse construction machines, then systems can manage machines from single manufacturers with known telemetry formats, but systems cannot accurately identify telemetry data from machines of uncertain origin or different manufacturers
Solution Approach 1:
The patent introduces an artificial neural network as an intermediary component between the diverse telemetry data sources and the fleet management system. The ANN learns to map telemetry data from various manufacturers to standardized channel IDs, acting as a mediator that translates different proprietary formats into a unified structure without requiring direct integration with each manufacturer's specific protocol.
Solution Approach 2:
The system transforms the identification approach by changing from deterministic rule-based matching to probabilistic pattern recognition. The neural network learns parameter relationships between different telemetry formats and standardizes them through weighted transformations, allowing accurate identification despite format variations across manufacturers.
2Productivity
If manual identification methods are used for telemetry data, then systems can maintain simple processing logic, but systems require substantial human effort and time for data identification
Solution Approach 1:
The neural network enables the system to automatically identify and classify telemetry data without human intervention. The ANN self-adjusts its weights through training and then autonomously performs channel ID assignment, eliminating the need for manual data identification while maintaining high accuracy across diverse machine fleets.
Solution Approach 2:
The system performs preliminary training of the neural network using labeled telemetry data before deployment. This preliminary action allows the ANN to learn identification patterns in advance, so that during actual operation, the system can rapidly identify new telemetry data without requiring real-time human analysis or manual configuration.
3Reliability
If proprietary telemetry formats from different manufacturers are maintained separately, then systems can preserve manufacturer-specific data integrity, but systems cannot provide unified fleet management interface
Solution Approach 1:
The patent segments the telemetry identification process into distinct functional layers: the neural network processing layer that handles manufacturer-specific format variations, and the unified fleet management interface layer that presents standardized data. This segmentation allows each layer to optimize for its specific function while maintaining overall system coherence and data integrity.
Solution Approach 2:
The neural network serves as a universal translator that can handle multiple manufacturer formats through a single unified interface. The ANN's weighted connection structure allows it to adapt to different manufacturers' telemetry protocols while outputting standardized channel IDs, providing multi-functionality without requiring separate processing paths for each manufacturer.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A computer-implemented fleet management system for managing a diverse fleet of construction machines (20) can comprise: an AI processor (350) to: configure an artificial neural network (ANN) (360) with processor-adaptable weights (358) on edges (364) thereof that are applied to decision results thereof; train the ANN (360) to assign values to the weights (358) based on known telemetry data (248) as identified by respective known telemetry channel IDs in a native ontology; execute the trained ANN (360) on a set of telemetry data series (248) assembled from the construction machine telemetry data received from a member construction machine of the fleet of construction machines (20) to generate thereby one or more candidate telemetry channel IDs (332) for the respective telemetry data series; and compute a confidence measure indicative of a likelihood that the candidate telemetry channel IDs (332) correctly identify the telemetry data series. The system can also comprise a threshold comparator (380) to compare the computed confidence measure with a confidence threshold (382) and, if the computed confidence measure meets the confidence threshold (382), the candidate telemetry channel IDs (332) are provided as the telemetry channel IDs of the telemetry data series arriving from the member construction machine; and a telematics processor (60) to perform fleet management operations for the fleet of construction machines (20) based thereon the telemetry channel IDs.