Fulin Gao

Fulin Gao

East China University of Science and Technology
1 previous affiliation
3Machine learning papers tracked by Lacuna
Is this you?

Research Overview

Fulin Gao focuses on operational status monitoring, multi-sensor data fusion, and key performance indicator prediction for complex industrial systems. Early work developed ensemble feature selection methods to handle massive soft-sensing parameters in large-scale machinery. More recent efforts address decision fusion across multiple modules, emphasizing the management of evidence conflicts to improve equipment monitoring.

Research Trajectory

The available evidence indicates a focused research trajectory spanning 2021 to 2022, centered on the data-driven monitoring and performance prediction of complex industrial equipment. A primary challenge in modern large-scale machinery is the proliferation of sensor types and the resulting influx of massive soft-sensing parameters. While this data richness provides multiple perspectives on equipment health, it also introduces severe algorithmic challenges, including the curse of dimensionality, scale mismatches, and conflicting evidence.

In 2021, Gao led research addressing the feature selection bottleneck in key performance indicator (KPI) prediction. As first author of Improved Ensemble Feature Selection Based on DT for KPI Prediction, Gao developed an improved ensemble feature selection strategy based on decision trees (EFS-DT). This work aimed to streamline the regression of soft-sensing parameters essential for project scheduling and cost estimation. The proposed methodology utilized an ensemble of multi-criteria filtering results to broaden the feature selector's perspective without incurring prohibitive computational costs, supported by credibility and similarity analyses to refine the selected feature sets.

Concurrently, Gao's work expanded from specific KPI prediction to broader equipment health assessment. As first author of A status-relevant blocks fusion approach for operational status monitoring, Gao investigated methods for integrating diverse data streams to monitor operational status. This research established a foundation for handling heterogeneous industrial data by segmenting and fusing status-relevant information blocks.

By 2022, the focus shifted toward higher-level decision fusion to overcome the limitations of raw data integration. In Multi-Module Decision Fusion in Operational Status Monitoring, where Gao served as second author, the research addressed the difficulties of combining multi-module and multi-sensor data. Because direct data integration often suffers from scale mismatches and dimensionality issues, this work emphasized decision-level fusion. A core contribution of this research was the measurement and management of evidence conflicts—specifically addressing the irrationality of conflict factors in traditional evidence theory—to ensure that data from different sensors and modules could be reliably synthesized to reflect the overall operating status of the equipment.

Current Directions

Based on the most recent papers in the provided evidence, which date to 2022, Gao's research interests culminate in the area of multi-sensor decision fusion for industrial monitoring. Rather than attempting to integrate raw, high-dimensional sensor data directly, recent work investigates how to manage conflicting evidence at the decision level. This involves developing mathematical frameworks to reconcile contradictory signals from different equipment modules, ensuring robust operational status monitoring even when individual sensors degrade or disagree. The overarching theme is the extraction of reliable, actionable assessments of machine health from complex, heterogeneous, and potentially conflicting industrial data streams.

Profile updated

See something wrong? Report a mistake