A dataset from a hydraulically actuated forestry crane for data-driven stress estimation (PATU655Stress)

Area of Science:

• Mechanical Engineering
• Robotics
• Data Science

Background:

• Hydraulic forestry cranes face challenges in performance and safety due to complex interactions and limited state estimation.
• Accurate prediction of structural stresses is crucial for fatigue life assessment but hindered by nonlinearities and scarce data.
• Publicly available experimental datasets for real-world hydraulic cranes are notably limited.

Purpose of the Study:

• To present PatuCrane655, a novel experimental dataset from a real PATU 655 forestry crane.
• To facilitate data-driven modeling, state estimation, and structural stress prediction for hydraulic cranes.
• To address the scarcity of real-world operational data for hydraulic machinery.

Main Methods:

• Collected time-synchronized sensor data, including joystick inputs, hydraulic pressures, actuator displacements, boom angles, and strain gauge signals.
• Recorded data under diverse operating conditions: multiple payloads, hydraulic settings, and crane configurations.
• Utilized a sequence-based LSTM model to demonstrate the dataset's utility for predicting structural strains.

Main Results:

• The PatuCrane655 dataset comprises comprehensive, synchronized measurements from a functional forestry crane.
• A baseline LSTM model successfully predicted structural strains, validating the dataset's suitability for data-driven stress estimation.
• The dataset supports the development of advanced models for internal state and stress prediction.

Conclusions:

• The PatuCrane655 dataset is a significant contribution to research on hydraulic forestry cranes.
• It enables advancements in data-driven modeling for improved performance, safety, and fatigue life prediction.
• The dataset will foster further research in state estimation and structural health monitoring for heavy machinery.