This readme file was generated on 2026-04-01 by Ya-Jing Wu

Title Dataset: Experiment Data for the Publication: Generalization of LSTM and CNN Autoencoders for Anomaly Detection Across Orthogonal and Longitudinal Turning
Dataset DOI: https://doi.org/10.15480/882.16923
Handle: https://hdl.handle.net/11420/62446
Publication DOI: pending

Author: Ya-Jing Wu, Ya-Jing.Wu@tuhh.de, ORCID: 0009-0001-5489-8667, affiliation: Hamburg University of Technology - Institute of Mathematics
Dataset creation date: 27.03.2026
Version: 1.0
License: Public Domain Mark 1.0 Universal
Date of data collection: 2024-02-04 to 2025-04-23 

Context:The data for development of the LSTM-based and  CNN-based autoencoders according to the time.
Information about the data:
- The folders of turning Process: OT, LT
OT: Orthogonal turning
LT: Longitudinal turning
--The subfolders of model development: Training, Validation, Test
--- Experiment Data: [CT0{AA}_{B}__vc[{v1}]_f[{v2}]_ap[{v3}].npz] or [WZ{AA}_B__vc[{v1}]_f[{v2}].npz]
AA: Cutting tool number
B = Cutting Edge Number, 
v1 = Value of the first cutting variable (vc, cutting velocity)
v2 = Value of the second cutting variable (f, feed rate)
v3 = Value of the third cutting variable (ap, Cutting depth)
Force sensor data of the experiments 
File format: .npz extension (a file format by Numpy that provides storage of array data using gzip compression)
Data structure: 
{force_axis}_{force_type}
>force_axis = Fx, Fy, Fz
>force_type = feeding, cutting, passive,


-- The subfolder of normalisation parameters: Normalisation
--- Nomalisation Data: "CT0{AA}_{B}__vc[{v1}]_f[{v2}]_ap[{v3}]_norm.npz" or "WZ{AA}_B__vc[{v1}]_f[{v2}.npz"
AA: Cutting tool number
B = Cutting Edge Number, 
v1 = Value of the first cutting variable (vc, cutting velocity)
v2 = Value of the second cutting variable (f, feed rate)
v3 = Value of the third cutting variable (ap, Cutting depth)
The normalisation parameters of force sensor data of the experiments 
Data structure: 
> mean: mean of three force channels (Fx, Fy, Fz)
> var: variance of three force channels (Fx, Fy, Fz)

-- The subfolder of the data for Cross-evaluation of anomaly detection models: Apply
--- Experiment Data: [CT0{AA}_{B}__vc[{v1}]_f[{v2}]_ap[{v3}].npz] or [WZ{AA}_B__vc[{v1}]_f[{v2}].npz]
AA: Cutting tool number
B = Cutting Edge Number, 
v1 = Value of the first cutting variable (vc, cutting velocity)
v2 = Value of the second cutting variable (f, feed rate)
v3 = Value of the third cutting variable (ap, Cutting depth)
Force sensor data of the experiments 
File format: .npz extension (a file format by Numpy that provides storage of array data using gzip compression)
Data structure: 
{force_axis}_{force_type}
> force_axis = Fx, Fy, Fz
> force_type = feeding, cutting, passive,