Skip to main content

Stream Processing Blocks

Stream processing blocks are used when values should be analyzed continuously instead of only being mapped or renamed.

They are especially useful for analytics use cases, including:

  • statistical calculations
  • condition monitoring
  • anomaly detection
  • AI-assisted scoring or detection

How stream processing works​

A stream processing block receives one or more input values from connectors, data mappers, or other flow blocks.

The block then:

  1. buffers the incoming values
  2. applies the selected stream processing function
  3. calculates the result continuously as new values arrive
  4. exposes the calculated result on the block outputs

This makes stream processing blocks well suited for calculations that depend on time history, moving windows, or continuously updated context.

Inputs, buffering, and outputs​

  • Inputs are connected to the block from the left side of the flow.
  • Incoming values are buffered by the stream processor based on the selected function and its settings.
  • The function calculates one or more output values from that buffered stream.
  • Those outputs can then be connected to enterprise connectors, data mappers, dashboards, or OPC UA outputs.

In practice, this means the block behaves like an online analytics component inside the flow.

Settings​

Each stream processing block is based on a selected function.

That function defines:

  • the required input parameters
  • the available output parameters
  • the configurable settings

Depending on the selected function, settings can include things like:

  • window size
  • thresholds
  • model parameters
  • update intervals

Typical examples​

Averaging​

Use a stream processing block to calculate a moving average of a noisy sensor value such as vibration, temperature, or power consumption.

This is useful when you want to smooth the raw signal before sending it to a dashboard or alarm workflow.

Z-score​

Use a Z-score based block to compare the current value with its recent distribution.

This helps identify when a value deviates strongly from its recent normal behavior, even when the absolute threshold is not fixed.

Anomaly detectors​

Anomaly detection blocks use buffered history and a detection method to determine whether the current behavior looks abnormal.

Typical uses include:

  • unexpected vibration patterns
  • abnormal temperature rise
  • cycle time deviations
  • unusual power consumption

These blocks are a good fit for predictive maintenance and machine condition monitoring.

When to use stream processing instead of a data mapper​

Use a data mapper when:

  • you want to rename, combine, or transform values directly
  • the logic is mostly deterministic and immediate
  • no historical buffer is required

Use a stream processing block when:

  • the result depends on buffered history
  • the logic is analytical or statistical
  • the calculation should update continuously over time
  • you want to support AI or anomaly detection use cases

Flow usage​

You can add a stream processing block from the flow menu.

After adding the block:

  1. select the stream processing function
  2. configure the function settings
  3. connect source values to the block inputs
  4. connect the block outputs to the next destination in the flow

This allows analytics logic to run directly inside the Reniway flow architecture.