Autonomously flying drones optimize asset management

Reliability and availability of plants and infrastructures are cornerstones of a company's profitability, and this without operating costs getting out of hand.

Innovative technologies such as autonomously flying drones and cognitive image processing enable new operational concepts. The benefits for companies: greater efficiency and process quality in inspection and maintenance.

Plant and maintenance management thus becomes a key discipline for avoiding cost-intensive downtime and increasing value creation.

With the importance of maintenance management, the strategic demands on activities that were previously primarily reactive are growing. Innovative concepts such as predictive maintenance - underpinned by techniques such as analytics, artificial intelligence/machine learning, and sensor technology - favor the design options for greater process efficiency and quality in asset management.

Enterprise Asset Management

A compelling example of a new value-creating service in asset management that operates in close coordination with SAP Enterprise Asset Management (EAM) is the integration of autonomously flying drones.

The flight systems, which are equipped with high-resolution cameras and sensor technology, are being used more and more in the area of surveillance and maintenance.

Due to their flexible flight characteristics, they are used specifically for control in areas that are difficult to access, large-scale facilities or for concealed infrastructures.

Compared to the previous "air support" by helicopter or airplane, unmanned aerial systems score with lower costs and more precise surveying.

The deployment of technicians on site is now only necessary in individual cases, as the inspection and fault detection is carried out virtually on the fly-by. The risks of accidents for the service personnel and also the general footprint of the mandatory inspections are reduced.

To fully unlock the optimization potential, it is essential to integrate lifecycle management of property, plant and equipment. Put more simply:

End-to-end process automation should be sought, spanning from drone operation to analysis of recorded data, which is in context with other asset information, to execution of services and control of technicians.

The core of the end-to-end process scenario is an integration and collaboration platform that organizes the exchange of messages and data between processes, services and people.

BTC implements such a next-level asset management environment that can be flexibly tailored to individual requirements. The basis for this is the SAP Cloud Platform and SAP Leonardo with the inclusion of SAP EAM services.

Drone Cloud Platforms

The description of an ideal process flow illustrates the interaction: At the start, the registration of the drone flight is initiated via an inspection application usually provided as a cloud service, which includes the official approval process.

Planning and management of flight routes can be done in-house or externally via services provided by highly specialized drone cloud platforms, as needed.

Geographical or spatial information systems help to define deployment plans and route guidance with just a few clicks. As part of standardized project control, partner companies automatically commission drones and pilots to carry out the work.

The orthophotography cameras, multispectral cameras and other sensor systems housed in the drones record current data as they fly overhead. Depending on the mission scenario, the images are evaluated.

In the case of periodically performed inspections, the analysis is typically done after the flight. In the case of an ad hoc inspection, which is carried out due to anomalies during operation, the data analysis can also be carried out directly, i.e. during the flight.

For the analysis process itself, cognitive evaluation methods of artificial intelligence come into play, with the help of which deviations from the usual normal state are detected from the state data.

Fiori Dashboard and SAP EAM

If the cognitive image analysis detects an anomaly, the service employee receives a corresponding notice in his Fiori/HTML5-based dashboard client or, within fully integrated scenarios, as an event in the SAP Asset Manager app on his iPad or iPhone.

In parallel, the fault is reported in the company's usual asset management environment, such as SAP EAM or S/4 Asset Management, and a job creation is automatically carried out. Subsequently, a mobile workforce management receives a push message with all necessary data including the route description from the geographical information system.

On site, augmented reality (AR) assistance systems guide the operator step-by-step through the work process via mobile display devices such as data glasses. Once the fault has been rectified, a message about the completion of the order is automatically generated in SAP EAM.

Two fundamental characteristics of future plant and maintenance management can be seen in the process flow outlined: Business processes are orchestrated across multiple application services, and data from different provenances is collected, combined and analyzed.

The offering provided by SAP promises, at least in perspective, a simple implementation of such next level asset management.

SCP and Cloud Foundry

The SAP cloud platform with the Cloud Foundry environment and the cloud platform integration as a software-technical basis make it comparatively easy to provide new services in addition to the existing asset management functions and to facilitate the exchange of data.
between different application functions.

Specific maintenance services such as drone management and cognitive image processing will be provided by BTC in the near future simply as tiles via the Fiori launchpad and can be combined variably with the existing application functions in the backend.

Asset Intelligence Network

Participation in the B2B business network SAP Asset Intelligence Network - which is a kind of Facebook for plant operators, manufacturers and service providers - on the one hand simplifies plant-related master data management; on the other hand, it promotes the exchange of utilization and error data as well as maintenance strategies.

This knowledge broadens the fundamental data base in asset management, for example when supplementary information about the usage profile, load and environmental conditions is incorporated in the event of an anomaly detection during the drone flight.

The information can be condensed into meaningful management indicators. The use of the SAP Geographical Enablement Framework for the spatial reference of operational data underscores the strategic note that asset management thus receives.

At a glance, the user learns about asset efficiency, failure rate or first-time-fix rate - i.e. problem solving at the first attempt. From the associated detailed information - also in comparison with other plants - conclusions can be drawn as to when, for example, the best time is for predictive maintenance and which steps need to be added to the asset management process in order to optimize plant efficiency and productivity.

One insight here may be to look more closely at the integration of drone management and cognitive image processing described above.

AI/deep learning technology detects anomalies

BTC's Unusual State Detection (USD) analysis environment works on the basis of deep learning methods. Specifically, it involves multilayer neural networks.

USD is characterized by its autonomous knowledge acquisition. To this end, it is fed historical data in a first step, from which AI experts at BTC derive a model that is trained to recognize implicit dependencies in the sensor data.

With the trained knowledge, the USD model is able to detect discrepancies as anomalies from normal behavior. This includes not only known but also unknown deviations.

Experts can retrieve the reported unknown pattern of a condition image and classify it with their expertise, thus broadening the knowledge base of the tool.

The tool becomes "smarter" in the assessment of deviations and error states over time, in the application scenario under consideration virtually with every flight.

In addition to the general image pattern matching function, the service is filled with "knowledge" on the part of BTC to automatically detect earthworks or construction machinery in the images, for example.