Adopting a Deep Learning Split-protocol Based

Predictive Maintenance Management System for

Industrial Manufacturing Operations

Authors:
Biswan Senapati, Bharat S Rawal
Contents & Agenda

• Introduction to Digital Manufacturing & Predictive Maintenance

• AI, Split protocol based Predictive Maintenance in Manufacturing Facilities
• Implementation strategy adoption of PdM
• Computational Model(AI,ML, SAP EAM- for the Predictive maintaince and asset management)
• Reliable Centered Maintenance -RCM in Industrial manufacturing
• Q&A
Digital Manufacturing and Predictive maintaince:
Digital Manufacturing and Predictive maintaince:
In the digital manufacturing sites, advancement of production sites are shopfloor activities are controlled by
equipment's, IIoT devices, sensors and large machines. All these smart devices continue to play a key role,
and their breakdowns or downtime can halt production, may hamper the demand and supply issue from
customers(MTO/MTS) and finally incur heavy costs with Low RoI. As per the recent survey, by ISA-
international society of automation, the downtimes and predictive maintaince cost is $ 647 billion globally
every year and as per Gartner estimates a loss of $540,000 every hour of downtime while dealing with
Industrial manufacturing sites.  To avoid these maintenance costs, adopting quantum computing's, AI, ML,
and high-end technologies could save money and save cost to the operational activities and increase the
efficiency of the manufacturing production sites.

According to a recent survey, 10% to 25% of downtime and production losses occur frequently due to
predictive maintenance issues (PdM) within heavy industrial manufacturing sites, resulting in large losses
and costly deals for the business. This would be important when considering a Make to Order (MTO) and
Make to Stock (MTS) scenario which is very tedious and critical to manage through a standard operation
activities. Certainly, a computational methods, AI,ML and spilt technology could potentially help to manage
the PdM in production and manufacturing sites.
Artificial Intelligence Conducts Predictive Maintenance in Manufacturing
In the manufacturing sites, smart equipment's and machines have a setup routines, and historical data
which needs regular checkup and predictive maintaince as per the day-to-day activities. Generally, it
needs the inspection, repairs and maintaince tasks which could be a part of operational tasks in
production units. Maintaining a manual task or maintaince planning could be as difficult for a plant
operator or a site supervisors as it could be well managed by computational tools, which may be effective
and accurate in the manufacturing sites.

AI, and ML can be a cost-effective approach for iterations required to identify data that can be used for
the detection of failure patterns. Data necessary for predictive maintenance collected using sensors
include temperature, humidity, pressure and vibration, which then need to be measured and analyzed.
Even when SCADA (Supervision and Data Acquisition) systems are used to collect data, visualize, and
trigger alerts, it requires a lot of human intervention for operation. Manually interpreting this data in real-
time to detect anomalies is exceedingly difficult. Both AI, ML can analyze the data in real-time and detect
failure patterns, preventing downtime and optimizing production.
AI-enabled Split-Migration Architecture In Industrial Manufacturing:

 Sudden Failure State (000): A sudden (but noticeable) breakdown

occurs while the production line is running well. This includes things
like broken tools, snapped bands, melted wire, etc.

 Intermittent Failure State (010); The "complete" machine failure is

typically on the path to intermittent failures, which come and go. By
their very nature, these intermittent or random failures might be
challenging to pinpoint. Regular maintenance can frequently stop
intermittent breakdowns.

 Gradual Failure State (100); These are the malfunctions that become
apparent over time as a machine’s usefulness gradually deteriorates.
Regular maintenance can frequently stop Gradual breakdowns. For this
work, we assume an intermittent failure state and a gradual failure state
as a degrading state (100).

Component failure state diagram

 Data Types in Predictive Maintenance
Here are the data type used in the Predictive Maintaince(PdM) in industrial manufacturing sites, production units.
Predictive maintenance of pieces of equipment would require experimentation to determine which combination data can
be helpful to predict machine failure.

 Data from the Core ERP-SAP S/4 HANA: Most of the core manufacturing, supply chain, device and IIoT sensors data are driving
from SAP system and from the SAP IAM core systems within the manufacturing sites .

 PLC(Programable logic Data): All these data are captured from the human and equipment's and machine interaction, and which can be
used to analyze human inputs and the machine’s output.

 IIoTs/Sensor Data on Equipment: Data are coming from the Industrial IoT devices, data from the devices( data on heat, humidity,
vibrations, etc., ), these data have a contributory role in machinery breakdown and maintaince KPIs. AI, ML can be used to analyze this data,
along with other data types to detect failure patterns and prevent machines from breaking down in any industrial sites.

 Historical Data:  Data on previous malfunctions and breakdowns can be used along with sensors to accurately assess the current state of the
machine and determine when it is likely to develop similar issues.

 External Data and Computer Vision Data: External data are from the equipment and general environment, such as weather data,
can be used in combination with other data types to measure impact on equipment’s functioning at the manufacturing sites, and the information
coming from the camera as it watches equipment function and does a qualitative assessment of products on the assembly line.
Implementation Strategy to adopt in Predictive maintenance

A few of the key critical steps which need to be followed to obtain the PdM are listed below:
 Benefits of Predictive Maintaince(PdM)

There are multi dimensions benefits while applying the best machine learning and AI prediction model to
understand the real time application in Industrial operations. Below are a few benefits:

 Reduce downtime of the machine/equipment’s/devices/IIoT

 Increase productivity/operational effectiveness of machines

 Reduce the TCO and Increase the ROI for the enterprise operations.

 Improve safety and identify the root causes of failures

 Derive the Intelligent assets management by adopting the QML, QAI, ML, and Artificial Intelligence

 Reduced unplanned downtime, and Increase production uptime in industrial manufacturing operation

 Enhance the resource productivity (including human resources and machines) and Improve product quality
AI, ML model in Predictive Maintaince (PdM):
Application of Deep Leaning/Machine Learning Model in PdM

 Enhancing operating equipment’s

reliability and reducing cost expenses
 Improving operating equipment
efficiency in manufacturing sites
 Reducing environmental impact
 Smart baselining and prediction-based
anomaly detection in manufacturing
sites
 Anomaly detection for monitoring and
control

 Continuous support and early alerts

What is new in Fourth level of Predictive maintaince(PdM 4.0)
• Maximize utilization of Industrial
resources and minimize the cost of
maintaince and downtime
• Replace the close to failure machine and
components
• Enable Just in time, as per MTO/MTS
• Discover the patterns for the problem
• KPI and ROI of asset conditions and
PAM
• RISK reductions
The fourth level is called Predictive Maintenance 4.0 is known as
PdM 4.0:
Level 1: Visual inspections: periodic physical inspections.
During this inspection, conclusions are solely based on the
expertise of the inspector.
Level 2: Instrumental inspections: periodic inspections. During
these inspections, conclusions are based on the expertise of the
inspector, combined with the measured value of our
instruments.
Level 3: Real-time condition monitoring: continuous real-time
condition monitoring of assets, during which alerts are being
send out if a predetermined critical level is reached.
Level 4. PdM 4.0: continuous real-time monitoring of assets.
Critical levels are based on predictive techniques, such as
regression analysis, and are most based on a combination or
correlation of different measurement values.
Best practice and Industrial catalyst to adopt in Predictive maintenance(PdM)

 Identify and prioritize goals.

 Create, track, and monitor the KPIs.

 Update the stakeholder buy-in

 Adoption of the right technology or software and Set up PdM mechanism update

 Training on the best implementation methods for preventive maintenance plans in Industrial sites.

 Develop the preventive maintenance checklist.

 Fine-tune plan based on results in Production sites.

 Smart Manufacturing Sites and Automation

 IIoT, sensors, and smart factories.

Reliability-Centered Maintenance (RCM)

In the industrial manufacturing space, reliability-centered maintenance (RCM) offers a wide range of service-based
maintenance and ensures that all systems can run at optimal efficiency and accuracy as per the business needs. It
offers best practices and adaptive models to save money, improve safety, and enhance machine performance and
reliability in industrial manufacturing sites.
Benefits of the RCM approach:
 Cost efficiency
 Improved performance:
 Enhanced safety
 Equipment longevity
 Production predictability.

The RCM-based predictive maintenance could be assessed and achieved with the following steps, and it should be
applied to the Industrial applications and production units.
SAP Digital Predictive maintaince Service:
Benefits of SAP Digital Predictive maintaince Service

 Reduce downtime of the machine/equipment’s/devices/IIoT

 Increase productivity/operational effectiveness of machines

 Reduce the TCO and Increase the ROI for the enterprise operations.

 Improve safety and identify the root causes of failures

 Derive the Intelligent assets management by adopting the QML, QAI, ML, and Artificial Intelligence

 Reduced unplanned downtime, and Increase production uptime in industrial manufacturing operation

 Enhance the resource productivity (including human resources and machines) and Improve product quality
Intelligent Asset Management (IAM and RCM-controlled predictive
maintenance:
Benefits of intelligence asset management:

 Organizations can achieve operational

excellence by using cloud-based solutions
that fore- see possible operational issues
before they arise.
 Cost management effectively within the
production sites.
 Management of Assets and Enterprise asset
management
 Interconnective, transparency and easy way
to achieve the predictive maintaince in
industrial manufacturing sites
Future Scope and Q &A