IEEE Computational Intelligence Magazine - February 2023 - 12

state and future asset usage. Furthermore,
the time-to-failure trajectories
in the training set are statistically right
censored, since maintenance interventions
interrupted the RUL trajectories.
To generate actionable predictions,
we need to reduce their variances.
Usually, this is achieved by deploying
an ensemble of diverse prediction
models.
Health Management - (4) Fault
Accommodation and (5)
Optimization
Health Assessment components could
be described as descriptive and predictive
analytics, as they provide assessments
and projections of the system's health
state. These assessments lead to prescriptive
analytics, as they determine the
onboard control actions and off-board
logistics, repair, and planning actions.
4) Fault accommodation refers to
onboard control actions, which are
usually implemented by intelligent
control systems. They are focused
on maintaining performance or
safety margins and are executed in
real time.
5) Off-board maintenance/repair actions
cover the offline decisions. They
require a decision support system
(DSS) performing multi-objective
optimizations, exploring Pareto
frontiers of corrective actions, and
combining them with preference
aggregations to generate the best decision
tradeoffs. Such decisions minimize
theimpactthatmaintenance and
repair events could cause on the profitable
operation ofthe assets.
PHM is one ofthe three topics covered
by the IEEE CAI 2023 vertical on
Industrial AI. The vertical will also
include Digital Twins to provide reliable,
actionable information by combining
data-driven and knowledgedriven
models (https://cai.ieee.org/
2023/digital-twins-dt-challenges), and
Cyber-Security to protect deployed
AI systems from being breached for
blackmailing, sabotage, and industrial
espionage (https://cai.ieee.org/2023/
cyber-security-challenges/). In the vertical,
we will also explore business models
to monetize AI-improved processes
or AI-enabled services or products;
KPIs for AI-enabled process improvement;
Productization of AI-enabled
services and products; AI Technology
development and deployment; Technology
sustainability, such as Model
maintenance, etc. We invite interested
readers to follow IEEE CAI 2023 developments
via https://cai.ieee.org/2023/
and attend the conference to benefit
directly from the interactions between
industrial thought leaders, practitioners,
and researchers.
References
[1] P. P. Bonissone and G. B. Fogel, " IEEE CAI
2023, an Industry-centered conference on AI, " IEEE
Comput. Intell. Mag., vol. 18, no. 4, pp. 6-7, Nov.
2022.
[2] P. P. Bonissone, " Machine learning applications, "
in Handbook ofComputational Intelligence, C. Alippi and
M. Polycarpou, Eds., New York, NY, USA: Springer,
2015, ch. 12, pp. 783-821.
[3] Zero Defect Manufacturing Platform. Accessed:
Oct. 18, 2022. [Online]. Available: www.zdmp.eu
[4] InterQ project. Accessed: Oct. 18, 2022. [Online].
Available: www.interq-project.eu
12 IEEE COMPUTATIONAL INTELLIGENCE MAGAZINE | FEBRUARY 2023

https://cai.ieee.org/2023/ https://cai.ieee.org/2023/digital-twins-dt-challenges https://cai.ieee.org/2023/digital-twins-dt-challenges http://www.zdmp.eu https://cai.ieee.org/2023/cyber-security-challenges/ http://www.interq-project.eu https://cai.ieee.org/2023/cyber-security-challenges/

IEEE Computational Intelligence Magazine - February 2023

Table of Contents for the Digital Edition of IEEE Computational Intelligence Magazine - February 2023