In today’s factories, 14 million hours of unplanned downtime happen due to system failures in the US alone, resulting in billions of dollars lost to the industry.

In order to prevent such events, manufacturing operations typically employ an expensive route-based approach where an expert gathers data to assess the health of the equipment or utilize a range of suboptimal sensing solutions that do not reliably detect all potential failures that could occur in these systems.

Why Predictive Maintenance?

Now that Industry 4.0 (also known as the Industrial Internet of Things, IIoT) wave is well under way, industrial customers are more focused on deploying solutions that increase equipment uptime, reduce operational cost, extend equipment lifetime, and improve worker productivity.

Predictive maintenance solutions combine sensing technologies to gather equipment data and employ advanced analytics and algorithms to draw actionable insights into the health of the equipment. As a result, this approach is expected to increase customers’ overall productivity by more than 30%.

Customers are asking for a full turnkey wireless solution that combines hardware and software and is easy to install and use. They need a solution that does not require experts to manually gather data and/or install and maintain dedicated networks.

The ADI Solution

ADI has created a flexible, out-of-the-box end-to-end secure, wireless predictive maintenance solution that combines ADI software, hardware, and domain expertise in electric motors to create a secure scalable offering for predictive maintenance of electric motors.

The Otosense SMS solution consists of an edge device called Smart Motor Sensor (SMS), Android and iOS mobile applications to provision the device, web dashboards to visualize the data, and AI based analytics and REST API services hosted in the cloud.

Battery operated, the SMS device combines ADI’s MEMS sensors, precision converters and signal chains. Firmware embedded in the SMS device captures various parameters of the motor (vibration, temperature, speed, magnetic flux) and sends this data securely over a Wi-Fi connection to the backend cloud for processing. An Artificial Intelligence (AI) engine which runs on the cloud analyzes the data and monitors the health of the motors.

The system can predict nine different electrical and mechanical failures that commonly occur in motors and upon detecting one, sends push notifications and/or emails and informs the customer on the appropriate action to be taken. The Smart Motor Sensor (SMS) product suite is available direct to customers as an end-to-end solution or via a RESTful API.

The combination of ADI’s advanced technologies like MEMS sensors, precision converters, signal chains along with analytics, algorithms and domain expertise is a key differentiator that makes ADI the ideal partner for our customers.

Kevin Carlin, Vice President, CbM-OtoSense AI business unit says “The OtoSense SMS predictive maintenance solution has been deployed across multiple industries and geographies and has proven to be the most capable platform available in its field. The individual framework components that the team has developed coupled with ADI’s best in class sensors, measurement techniques and domain knowledge are enabling us to solve our customers’ most challenging problems.”

The sensor to cloud secure software framework helped to meet customers desire for complete solutions, without the frustration or the complexity of putting together a full system to capture actionable insight from real-world data.

Greater reusability doesn’t just drive faster time to market, but also security and quality goals. Core turnkey, production grade components included in the framework enabled faster time to market, so more of the development schedule was focused on delivering the domain specific solution for any given product. The integration of best-in-class billing and software license infrastructure allows for the realization, and on-going renewal, of software as a service (SaaS) based revenue.

Providing Software as a Service (SaaS) solutions means exposure to the full onslaught of hackers and digital thieves. Security as part of the framework is not only critical for our customers, but for our brand; any early security breach would greatly diminish future success.

Along with functionality and scalability, prescriptive best practices and processes included as part of the framework ensure the most robust and secure solutions are delivered. The Software Framework is an embodiment of recommended best practices and proven processes; from approaching each new customer engagement from a user-centric perspective to automation of DevOps principles to achieve efficiency and compliance.

SMS cloud and web

The OtoSense SMS cloud backend is built on and extends the secure software framework. The application is multi-tenanted to support onboarding multiple customers and ensuring data segregation.

We also offer a unique custom label partner version which allows customers such as motor repair workshops, motor distributors and others to easily add their logo and branding and sell SMS directly to their customers. Partners can seamlessly create new customer tenants with just a few clicks and all the necessary architecture is automatically generated and onboarding email invitations sent to the customer.

The OtoSense SMS mobile applications have been architected with code re-use as a priority. The core application consists of several modules, including an SMS Communication Module, API Communication Module and a Core Application Module (views and navigation).

These modules represent everything that can be re-used across all custom label partner applications. For each new custom label application, a new configuration set is created which contains the partners branded UI components and API configuration. This software architecture enables the mobile development team to create new partner applications quickly and easily, as well as release new features and updates to all partner applications efficiently.

SMS AI engine

The ADI-OtoSense AI platform can address any Sensing Interpretation challenge. This platform has been leveraged to implement the machine learning components of the Smart Motor Sensor solution.

ADI’s deep domain expertise in electric motors has been used to create a digital blueprint of an electric motor’s optimal behavior, across all ranges of loads, rotation speeds and temperatures. However, each motor once sold deviates from this initial model. Its environment, usage and maintenance operations make it unique.

After installation on a motor, OtoSense SMS initially acquires data to learn how this motor differs from the blueprint, creating a unique digital twin for the motor the SMS edge device is attached to. From there, OtoSense SMS compares the data coming from this motor to its digital twin running in the exact same operating conditions.

If there is a substantial difference in behavior between the physical motor and its digital twin, OtoSense SMS analyzes this difference to determine the most probable root cause of this discrepancy, informing the user about the existence of a fault, its origin, and the action to perform to resolve it.

SMS firmware

SMS firmware embedded in the SMS captures various parameters of the motor (vibration, temperature, speed, magnetic flux) that it is attached to and sends this data securely over a Wi-Fi connection to the backend cloud for processing. SMS firmware connects with iOS or Android smartphone via the Wi-Fi interface to enable device installation and provisioning. The firmware provides diagnostics information for troubleshooting the device during and after deployment and can be updated securely Over the Air.

SMS firmware was built leveraging a portable Service Oriented Architecture based framework. This Secure Device Framework (SDF) consists of modular, reusable functional blocks wrapped in application agnostic service API’s to enable reuse for other IoT applications with a short Time to Market cycle. It also enables application developers to easily integrate application specific signal processing functions and algorithms along with service APIs into the product.

Sensor to cloud software testing

A comprehensive test plan was set in place to ensure we deliver the highest quality solution to our customers.

In addition to unit tests and integration tests performed by developers as part of the development cycle, a dedicated, independent test team was tasked with performing end-to-end solution testing. The majority of the tests were automated to ensure quick turnaround testing and to perform regression testing as the software evolved over time.

A test set up employs a variety of popular open-source test frameworks integrated into a single cohesive test automation framework. This enabled us to perform all end-to-end functional tests for all the critical functions of that require interaction between the device, mobile application and web application.

Several hours of “golden” data was recorded to test the Analytics engine to ensure robustness in performance with minimal false positives and negatives, and appropriate notifications are sent to users when failure events are detected. Test methodology included both positive and negative cases as well as testing under different network conditions to ensure reliability, robustness, and graceful recovery under severe conditions.

Shankar Malladi: Director, Embedded Software & Product Quality; Jason Griffin: Director, Technology Solutions; and Sebastien Christian: Product Line Director, CbM-Otosense