This project implements a Functional Multilayer Perceptron (FMLP) using Functional Data Analysis (FDA) in a centralized data scenario and a Federated Learning scenario with horizontal data partitioning setting. Experiments conducted in the novel maintenance strategy called Predictive Maintenance (PM) can be replicated by executing the files written in Live Code File Format Matlab R2020+.
It is assumed the installation of the Parallel Computing Toolbox to execute the present code.
Predictive Maintenance (PM) is a novel maintenance strategy that uses Condition Monitoring (CM) to predict the Remaining Useful Life (RUL) of industrial machinery. Using RUL estimations, maintenance interventions are performed when needed. Particularly, the code available is related to the RUL prediction of turbofan engines using NASA C-MAPSS datasets (https://www.nasa.gov/intelligent-systems-division).
Conditional monitoring data C-MAPSS, located in the data folder, is composed of four FD00x datasets. Each dataset was created after monitoring multiple turbofan engines through 21 sensors and three operating settings. Particularly, the regression problem implemented here is related to predicting the target value (RUL) using just the 21 sensors. More details of how to replicate those experiments in a centralized scenarion can be found in:
- Centralized data scenario: CentralizedData_FMLP.mlx
NOTE: Data of this repository was pre-processed using the criteria described in [1]. This reference also mathematically explains the implemented Functional Multilayer Perceptron.
Decentralized data scenarios are common in industrial applications. Companies present difficulties gathering many failure instances in isolated systems (e.g., air fleets measuring aircraft engines) to develop accurate PM applications separately. Therefore, it is assumed that different parties J (e.g. multiple aircraft) partially monitored the turbofan engines described in a particular FD00x dataset using the same variables. Federated Learning experiments can be replicated by setting the number of parties J and selecting the FD00x dataset. It is configured in the following Live Code File:
- Federated Learning: CentralizedData_FMLP.mlx
NOTE: More details about data partitioning and the Federated Averaging (FedAvg) algorithm can be found in Pending Reference
If you use this content in a scientific publication, we would appreciate using the following citation:
@InProceedings{LLasagRosero2023,
author={Llasag Rosero, Ra{\'u}l
and Silva, Catarina
and Ribeiro, Bernardete},
editor={Iliadis, Lazaros
and Maglogiannis, Ilias
and Alonso, Serafin
and Jayne, Chrisina
and Pimenidis, Elias},
title={Forecasting Functional Time Series Using Federated Learning},
booktitle={Engineering Applications of Neural Networks},
year={2023},
publisher={Springer Nature Switzerland},
address={Cham},
pages={491--504},
doi={10.1007/978-3-031-34204-2_40},
isbn={978-3-031-34204-2}
}
[1] Llasag Rosero, R., Silva, C., Ribeiro, B. (2023). Forecasting Functional Time Series Using Federated Learning. In: Iliadis, L., Maglogiannis, I., Alonso, S., Jayne, C., Pimenidis, E. (eds) Engineering Applications of Neural Networks. EANN 2023. Communications in Computer and Information Science, vol 1826. Springer, Cham. https://doi.org/10.1007/978-3-031-34204-2_40
McMahan, H.B., Moore, E., Ramage, D., Arcas, B.A.: Federated learning of deep networks using model averaging (2016) https://arxiv.org/abs/1602.05629
Liu, W., Chen, L., Chen, Y., Zhang, W: Accelerating federated learning via momentum gradient descent (2019) https://arxiv.org/abs/1910.03197
Li, Q., Wen, Z., Wu, Z., Hu, S., Wang, N., Liu, X., He, B.: A survey on federated learning systems: Vision, hype and reality for data privacy and protection (2019). https://arxiv.org/abs/1907.09693
Yang, L., Yan, K., Xinwei, Z., Liping, L., Yong, C., Tianjian, C., Mingyi, H., Qiang, Y.: A communication efficient vertical federated learning framework (2019) https://arxiv.org/abs/1912.11187
Rosero, R.L., Silva, C., Ribeiro, B.: Remaining useful life estimation in aircraft components with federated learning. International Journal of Prognostics and Health Management 2020(5) (2020). https://doi.org/10.](https://doi.org/10.36001/phme.2020.v5i1.1228
M., P., Priya, R.M.S., Pham, Q., Dev, K., Maddikunta, P.K.R., Gadekallu, T.R., Huynh-The, T.: Fusion of federated learning and industrial internet of things: A survey (2021). https://arxiv.org/abs/2101.00798
Ning, G., Guanghao, L., Li, Z., Yi, L.: Failure prediction in production line based on federated learning: an empirical study. Journal of Intelligent Manufacturing 33(8), 2277–2294 (2022). https://doi.org/10.1007/s10845-021-01775-2