For companies wanting to implement IoT in manufacturing, a perfect starting point is IoT-enabled condition monitoring as it drives value immediately and can be driven iteratively.

The key role of condition monitoring enabled by IoT is to provide a source of real-time data on the health of industrial machinery and their operating environment. IoT is expandable and has flexible evolution possibilities that make it ideal as a starting point for digital transformation.

Condition monitoring based on Internet of Things

IoT-driven condition monitoring leverages the data to monitor the current state of industrial machinery and identifies combinations of equipment’s ambient and operational parameters that could cause deterioration of product quality, or lead to potential equipment failures.
IoT-enabled condition monitoring’s capabilities are however much wider than that. With the correct approach to implementation, the solution gains advanced capabilities, which include deeper control and analytics, and can be used as a catalyst for other IoT initiatives. This will drive improvements at each stage of implementation.

IoT in manufacturing: Iteration drives value

When an IoT initiative is started with the basic architecture, it leads to a fast payoff and avoids excessive investment. Once it is established properly, it can be expanded further to enhance the functionality of the solution and create more use cases.

Prevent machine failure

Building the basics: Finding the pattern

As each asset type has its own unique normal behavioural pattern, the specifics of implementation will be different depending on the type of equipment used and the manufacturer. There are however four basic components used by IoT in manufacturing:

1. Gateways

Cloud and field gateways are used to connect industrial machinery to cloud software reliably to ensure that data can flow without interruption.

A cloud gateway enables secure data transmission between cloud data storage and field gateways.

A field gateways process sensor data before moving it to the cloud. This preprocessing in the field includes the filtering and aggregation of messages. Sensors on a welding machine may, for example, transmit vibration measurements that are relatively stable, but the measurement increases and decreases by 0.1% for every reading. Field gateways achieve more efficient data transmission by discarding the intermediate data points.

2. Sensors

IoT-driven condition monitoring uses ambient parameters (e.g. temperature and humidity) and equipment data (e.g., pressure, vibration and current). This data can be collected from sensors connected directly to the machine, a PLC, or a SCADA system.

3. Big data warehouse

Data required for insights about equipment conditions is extracted from a data lake, and then undergoes filtering, processing, organizing and modelling before it is loaded into a big data warehouse.

A big data warehouse not only stores the sensor data but also relevant contextual information, including operating parameters and equipment maintenance history collected from an ERP and other systems.

4. Data lake

All raw sensor data from numerous machines and their components are stored in a data lake without cleaning or processing it before the time.

Ramping up: Analytics and user applications

The IoT-driven condition monitoring solution can be taken to the next level with analytics and user applications once the basics are in place.

1. Analytics

To convert raw sensor data into equipment condition insights, data analytics applications are used by running sensor readings through analytics algorithms. The results can then be visualized and communicated to users via user apps.

Read also:  How to become a machinery trader with Exapro

A manufacturer may, for example, want to identify industrial batteries with substandard performance and identify batteries that will discharge soon. An analysis of the data from voltage and temperature sensors could identify batteries with a downward capacity trend that need to be replaced.

2. User apps

User apps facilitate bidirectional communication between users and the monitoring solution. User apps display data analytics’ insights via equipment health dashboards, diagrams, charts, etc.

User apps also allow additional data about a machine’s operating context to be entered into a big data warehouse. Maintenance technicians may for example log reasons for deviations detected in equipment parameters via a mobile app.

Evolving: Advanced architectural components

Once a condition monitoring solution has been established, it can be enhanced via advanced architectural components. The range of capabilities will be expanded to product quality control and predictive maintenance by integrating these components.

1. Product Quality Control

The advanced components can be used to enhance condition monitoring to product quality control.

Deep analytics

Deep analytics uses equipment condition data combined with yield quality data as well as the context and runs this dataset through a machine learning algorithm to detect specific machine condition patterns that will influence the products’ quality.

Control applications

When a machine learning algorithm detects a machine condition that may influence the products’ quality, users are informed and an output triggered. The machine may, for example, be switched off automatically to prevent equipment failure, or to prevent it from entering a state where it would produce faulty products.

2. Predictive Maintenance

Data collected with equipment condition monitoring is used to predict the conditions under which equipment is likely to fail and provide proactive recommendations to maintenance personnel.

Prevent machine failure

Enabled by expanding the condition monitoring solution with:

Deep analytics

Deep analytics is used to discover data patterns that indicate potential equipment deterioration. Machine learning algorithms are used on the sensor data combined with context and historical data to reflect the patterns in predictive models. A model is built, trained and tested until its accuracy is proven, after which it is used to forecast potential equipment failures. As more data becomes available from equipment, models are revised, updated and tested again to improve their accuracy.

Control applications

IoT in manufacturing can be used with control applications with a degree of automation. When a machine learning algorithms for example identifies that a specific combination of engine vibration and temperature could result in machine failure, the model can be used to trigger control applications that send commands to a machine and switches it to low-speed operation to prevent further wear. At the same time, an alert is triggered to inform users of a potential failure.

Business and process improvements with condition monitoring

When implementing IoT in manufacturing, IoT-driven condition monitoring can drive value in three ways:

  • Providing real-time data on the condition and health of industrial machines.
  • Making data-driven decisions for product quality improvement.
  • Paving the way for predictive maintenance.

Prevent machine failure

To implement these improvements with a reasonable investment and faster, it’s good to use iterative implementation, enabling the IoT-driven condition monitoring solution to bring both business and process improvements. This will also establish a solid technological foundation for future initiatives in digital manufacturing.

Leave a comment