Frequent power outages are expensive not only for the power utility companies and their customers but also for the whole economy. For e.g., outages cost the US economy $150 billion annually (US Department of Energy, 2018). Despite utility investments in grid modernization and reliability improvement, power outages are increasing in both number and cost. Factors causing outages could be either internal (aging assets, poor network design, monitoring practices etc.) or external (extreme weather events, storms, and wildfires). Grid reliability and resiliency are critical grid characteristics, based on which several reliability indices are defined and closely monitored by the regulator. Besides affecting the bottom line, outages also result in reduced customer satisfaction levels.

Utilities are in constant search for technological innovations designed to increase grid uptime and reduce outages. One such innovation discussed here is the adoption of IoT-enabled autonomous drones and AI/ML to achieve optimization in outage management process and improve grid reliability.

Drone technologies combined with IoT and AI/ML have a great potential in the power utility industry. Distribution of linear assets across a wide geographic area and the associated difficulty and safety concerns make it a good case for adoption of IoT drones and automated processing technologies. In fact, this technology is already being pursued by utilities for business processes like overhead asset inspection, vegetation encroachment detection and for carrying out damage assessment in case of wide-spread damage caused by natural calamities.

Outage management is another potential area where this technology could be adopted to reduce outage restoration time, minimize human dependency, and improve grid reliability.

Current challenges with outage management

Outage management process includes steps such as outage identification, device fault prediction, outage restoration, and call closure. Outage restoration is a major process consuming a significant amount of time in the overall outage lifecycle management. Some of the challenges inherent in this process include:

• Inconsistency in device fault prediction – Prediction accuracy depends on several factors such as percentage of outage reported by customers and AMI system, up-to-date network model and device status, efficiency of prediction algorithm, and system configuration.
• Multiple field trips required by crew to fix fault and restore outage – Fault identification, inspection, and repair may require multiple field trips involving different type of crews.
• People dependency in executing the process - Efficiency of field operations depend on crew expertise, availability, distance, and accessibility of faulty device.
• Health & safety concerns - Crew must traverse through multiple locations to investigate and pinpoint the location of fault. This could be remote and difficult to reach locations such as river crossings, middle of forest, hilly regions etc. They may have to climb poles and towers as part of investigation which is a big safety concern.

Autonomous drones and image analytics to address these challenges

Utilities have started to use drones to capture images of distribution grid assets for periodic inspection. In most cases, these are offered as ‘data-as-a-service’ or ‘insights-as-a service’ by drone operators. However, using drones for outage management has different needs. The image capture requirement is ad-hoc and on real-time basis. Regular drones need to be carried by field crew to the location of interest and require assembling and navigation using the Ground Control System (GCS) at the site. Even though it reduces manual effort in image capture and address some of the safety concerns, it still requires field crew visit to remote locations and involves manual efforts in flying the drone.

In contrast autonomous industrial drones, also known as “drone-in-a-box” systems can be operated remotely from a central command center. Here the flight plan is created and delivered to respective Drone Docking Station (DDS) as per need and drone flight is triggered almost instantaneously. Images captured by drone can be processed in real-time using advanced analytics and AI/ML algorithms both at the edge and the central platform to automatically detect the faulty assets. Feeding these insights to the work management system helps in quicker resolution of outages in the power distribution network.

Autonomous process addresses the health and safety concerns for the crew. It reduces the field visits as the insights provided help the crew to prepare well with necessary tools and items prior to field visit. AI/ML-based fault detection is consistent, and accuracy improves overtime.

Autonomous Drones and Drone Command Center (DCC) setup

Figure 1: Autonomous drones and Drone Command Center (DCC) setup

Power utilities can strategically position drone docking stations across the operational area to ensure coverage to critical sections of the power network. Drones must be pre-fitted with necessary sensors for image capture and communicate to the GCS. Flight plan can be created centrally in the DCC and pushed to the edge device residing at the docking stations. The edge device can also enable local processing such as health monitoring of devices, pre-processing of captured images, and trigger notification in case of any threshold breach.

Enhanced outage management process

Figure 2: Outage management process

The above schematic depicts the enhanced outage management process using autonomous drones combined with AI/ML processing.

For the customer outages detected by smart meters, OMS system predicts the device outage and forwards relevant data to DCC. Here, the central team will identify the respective docking stations to be invoked and creates the drone flight plan based on the area to be surveyed by the drone. The flight plan is then pushed to the edge device situated at the respective DDS. The GCS system at the edge will fly the drone and controls the flight path, which can be viewed from the DCC. Image data captured by the drone is pushed to the edge device which performs analytics to validate the quality of the image. If the image is not of the right quality, it notifies the DCC wherein the same is investigated and necessary corrections can be made to the flight plan. Once the new plan is pushed to the edge device, drone is flown again to re-capture the images. Once the image quality validation succeeds, edge device transmits the data to image analytics application. Here the faulty device is identified using AI/ML and the anomaly data is forwarded to field mobility or Work Management System (WMS) for dispatch of crew to the field for fault resolution. The insights on the fault helps the crew to get better understanding of the issue and carry necessary tools and equipment to the field to enable timely resolution. Once the fault is fixed and power restored, respective outage is cleared in OMS system. This way, DCC helps in avoiding multiple field trips for the crew by autonomously and accurately identifying faulty device and enables quicker outage restoration.

Solution architecture

Depicted below is the proposed solution architecture covering end-to-end IoT solution.

IoT enabled autonomous drones and AI/ML can help achieve optimization in outage management process and improve grid reliability.

Figure 3: Solution architecture

The solution components are spread across edge and central layers.

Central layer

Solution leverages HCLTech’s “Dynamic Ecosystem of Connected Devices (DECoDe)” which is a platform-agnostic device management solution to manage, monitor, configure, and troubleshoot IoT devices remotely from a consolidated dashboard. The features offered at this layer include:

• Flight planning – Centralized planning of drone flight path and push the flight plan to GCS via the edge gateway.
• Real-time drone monitoring – Dashboard to monitor the real-time streaming video from GCS. Other parameters such as weather, wind-speed, temperature etc., are also transmitted in real-time to closely monitor the real-time conditions.
• Image analytics & AI/ML processing – Image processing using AI/ML algorithms to identify fault in power distribution network.
• Enterprise integration – API-based integration with outage management and field mobility systems to receive drone flight request and forward fault data.
• Edge device management – Devices at the edge include drone, edge gateway, GCS, and docking station. Solution provides full lifecycle management comprising of services such as remote monitoring, configuration, control, remote firmware management, and updates.
• Alerts & notifications – Indicates device anomalies such as drone battery warning, firmware upgrade failure, accidental damage etc. Also notifies on data-related exceptions such as image not of right quality, corrupt image file, transfer failure etc.
• Security -- Device and gateway access based on roles, including security – secure connection and on-boarding of IoT devices with authentication and authorization services. Also, leverage enhance security services provided by cloud platforms.

Edge layer

• Drone flight is initiated and controlled by the GCS system at the edge. It receives flight plan from central layer via the gateway device and monitors real-time parameters while the drone is in flight. This data is transmitted in real-time to central layer via the gateway for real-time drone monitoring.
• High-definition image / video data captured by the drone is pushed to edge device once the flight is complete and drone returns to the docking center.
• Edge device holds the DECoDe agent which communicates with DECoDe solution on central layer. Analytical models deployed at the edge do pre-processing of images captured to validate the image quality and flag notification. All operational and non-operational data flows via the gateway to the central layer.

The solution can be hosted either on-premise or on any cloud platform.

Solution benefits

Automation introduced by the solution improves overall efficiency of power outage management process and boosts field crew productivity. Following are the list of benefits that can be leveraged from this solution.

Figure 4: Solution benefits

In addition to achieving process efficiency, the images captured would also help in auditing at a later point in time and can be used during operator training programs.

Conclusion

Autonomous drones and image processing technologies holds a promising future for utilities in effectively managing power outages and improving network reliability.

Nowadays, several vendors have started to provide autonomous drone solutions for defense and industrial needs. Utilities can set up drone docking stations across the operational area and own drones of specific configurations addressing their image capture needs. The command center can be part of the control room operations used for real-time monitoring of assets using SCADA systems. Automated processing of images using advances analytics and AI/ML is another important component of the solution. A bespoke analytics platform or a third-party solution can fit in here, which can integrate with drone solution and enterprise systems for real-time data processing and leveraging insights to derive value in the outage management process.

Utility customers can adopt a staggered approach by implementing solution to cover critical areas of the network and then slowly scale it to cover rest of their service area.

References

1. https://www.energy.gov/ne/articles/department-energy-report-explores-us-advanced-small-modular-reactors-boost-grid
2. https://www.geospatialworld.net
3. https://www.faa.gov/uas

HCLTech IoT WoRKS™ is a dedicated Internet of Things (IoT) business unit of HCLTech. Our award winning, best-in-class, customer and industry specific, deployment ready solutions co-created with customers, enable them to maximize effectiveness and returns on their asset investments. HCLTech has rich experience in image analytics space and dedicated Data Analytics practice who develop solutions in image processing using AI/ML and Deep learning technologies. You can visit HCLTech website to know more about our solutions and offerings around IoT and Image Analytics.

HCLTech offers a cloud-based image analytics solution called as “Geospatial Linear Asset Management”, which provides a platform-based approach for managing image data at scale and provides deep learning algorithms for processing images to identify different types of anomalies. It also provides standard based APIs for integration with enterprise systems.

In addition, HCLTech also offers a centralized device lifecycle management solution called as DECODE (“Dynamic Ecosystem of Connected Devices”). This solution can be leveraged to manage devices such as Drones, Edge devices, Image capture sensors and GCS systems from a unified platform.