Multi-source heterogeneous data fusion verification and analysis methods, equipment and storage media

By employing a multi-source heterogeneous data fusion and verification analysis method, the validity of monitoring data is verified using hidden Markov models and finite state machine models. By combining Markov blanket common feature learning and residual network for feature alignment and incremental learning, the problem of unifying multi-source pollution emission data to the same scale is solved, thereby improving the accuracy of emission estimation.

CN117370921BActive Publication Date: 2026-06-30CHINESE RES ACAD OF ENVIRONMENTAL SCI +2
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
CHINESE RES ACAD OF ENVIRONMENTAL SCI
Filing Date
2023-09-18
Publication Date
2026-06-30

AI Technical Summary

Technical Problem

Existing technologies cannot effectively unify and correlate mobile source pollution emission data obtained from multiple monitoring methods at the same scale, resulting in low emission estimation accuracy.

Method used

A multi-source heterogeneous data fusion verification and analysis method is adopted. The validity of the monitoring data is verified by using a hidden Markov model and a finite state machine model. Feature alignment is performed using a Markov blanket common feature learning model, and incremental learning is performed through a residual network to estimate the common features of multi-source pollution emission factors.

Benefits of technology

It enables effective correlation of emission data obtained from different monitoring technologies within the same feature space, thereby improving the accuracy of mobile source pollution emission estimation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117370921B_ABST
    Figure CN117370921B_ABST
Patent Text Reader

Abstract

This invention discloses a method, device, and storage medium for multi-source heterogeneous data fusion verification and analysis. The method includes verifying the validity of monitoring data collected by various monitoring methods; fusing vehicle operating condition information and pollution emission information collected by various detection devices; mapping the operating condition emission feature maps of different monitoring data to a high-dimensional space for feature alignment based on a Markov blanket common feature learning model; and incrementally learning the common features of multi-source pollution monitoring data to achieve consistent estimation of pollution emission factor levels from various monitoring data. This invention utilizes a Markov blanket for operating condition feature screening, extracting a subset of features related to pollution emissions, and constructing a common feature space for multi-source monitoring data using the feature extraction capabilities of residual networks. Finally, it uses a width-based learning structure for incremental learning of the common features of multi-source pollution monitoring data, effectively achieving deep feature alignment of multi-source heterogeneous data.
Need to check novelty before this filing date? Find Prior Art