NeuroCPS4Maintenance – Neuromorphic anomaly detector in the edge for predictive maintenance
NeuroCPS4Maintenance is a project that aims to develop and demonstrate a neuromorphic edge anomaly detector that is robust against conceptual drift, alerts to faults early and provides fast, real-time response for predictive maintenance applications in high-demand industrial scenarios (industrial press). This anomaly detector will be based on deep learning algorithms (LSTM) and implemented on system-on-chips (SoC).
The innovation capacity of this neuromorphic anomaly detector prototype will favor the enablement of new technical solutions for predictive maintenance in high-demanding industrial environments, as it will open the possibility of future applications of the system to other types of machines or robotic units, especially in advanced manufacturing processes, giving rise to applications with high market potential.
This consortium can be a first seed within DIH4CPS for the development of solutions based on neuromorphic edge processing for predictive maintenance replicable across Europe, and a first step for new use cases in robotics applied to other non-industrial sectors:
- ITCL, technical leader of the consortium, will work on the design and development of deep learning-based time series analysis algorithms.
- DIBHU will coordinate the dissemination of the project, define the use cases and provide the application partners for experimentation.
- Intigia will be the technical partner in charge of implementing these algorithms in FPGAs and deploying them in the field.
The common goal of the project is to solve these difficulties by generating new solutions that can be implemented in a short period of time, through specific applications for each type of machine and maintenance problems. NeuroCPS4Maintenance aims to overcome the difficulties faced by SMEs in deploying predictive maintenance solutions due to the lack of data sets and cybersecurity issues.
The project will develop a neuromorphic anomaly detector. This detector will be deployed and evaluated in a relevant scenario.
Use of novel approaches to develop this neuromorphic anomaly detector.
The development and demonstration of the neuromorphic processor will make extensive use of CPES technologies.
The LSTM-drift algorithm and hardware accelerators will be developed to implement it in real time and deploy the prototype in an industrial press (relevant environment), where its components can be validated.
The innovation capacity of this neuromorphic anomaly detector prototype will favor the enabling of new technical solutions in predictive maintenance in highly demanding industrial environments.
- Develop a hardware prototype of a neuromorphic processor based on SoCs capable of detecting physical variables, interacting with it to avoid failures and displaying status information.
- To have an LSTM (deep learning algorithms) architecture for anomaly detection that adapts to the concept-Drift implemented in real time in the neuromorphic processor.
- To have a real-time estimation of machine status and alarm generation that can be accessed from the Internet and displayed on intuitive control panels.
- Perform a proof of concept applied to industrial stamp presses that operates in real time, detecting faults in advance and being able to issue alerts or stop the machine in case of detected faults.
- Evaluate the performance of the system in terms of accuracy, time to failure, robustness to machine changes over time and energy consumption.
Duration time: 2021 – 2022