Falkonry’s comprehensive pattern discovery system -- Finds patterns in operational data - unprocessed multivariate, time series data - providing detection and prediction capabilities that simple SPC or regression methods cannot match.
Explanation of findings provides the operations expert deeper understanding of the behavior causing each pattern. Unlike neural networks, Falkonry’s approach can determine which signals are most important to make the prediction, as well as how much each signal contributed to that result.
Streamlined UI and automation put operational experts in control -- Allows operational experts to rapidly scale use cases without the costs and delays of expanding data science teams.
Real-time visibility into factors which affect availability, performance and quality -- allows plant personnel to work in the “now” and address production inefficiencies before they impact business
Falkonry recognizes both frequency and spatial patterns present in time series. In other words, Falkonry takes into account how fast data varies and how much it varies over time.
Falkonry uses a multivariate, time series pattern based approach to discover, predict and explain behaviors of interest.
Unlike univariate, rule based approaches, patterns can see complex, subtle changes in behavior which span a number of subsystems in order to identify events of interest.
Unlike regression based approaches, patterns can identify behaviors which are sudden or that do not vary continuously over time.
Falkonry’s pattern discovery system automatically learns on a per equipment basis: Connect each one to Falkonry and let it run. This approach simplifies deployment. Whether one or one hundred assets, the process is the same.
The software is most commonly licensed according to the number of assets connected to the system. This license provides an unlimited number of dashboard users and use of an unlimited number of signals. Our software does not have the concept of a “model,” therefore it has no per-model charge. Rather there is one connection per asset which covers all the behavioral patterns of that asset.
Falkonry has shown the ability to find a wide range of behaviors, alert users when those behaviors occur and provide insight into what equipment behaviors are associated with them. Some examples include:
Detecting early part failure
Finding precursors which indicate need for early maintenance (e.g. cleaning)
In-line classification of product quality (binning)
Detection of specific process conditions which trigger operator actions
Identification of unknown changes in the environment indicating potential problem areas for investigation
Finding precursors which flag impending unsafe operating conditions or regulatory violations
Understanding equipment behavior characteristics quality differences for root cause investigation
No need for any custom APIs or custom software.
Falkonry has built-in connectors to pull data via OPC UA or MQTT protocols. Falkonry can also work with historians such as OSIsoft PI systems or file/log exports.
Identify at least one asset or process you want to monitor
Gain an understanding of the data sources for live connectivity
Falkonry has built-in connectors to pull data via OPC UA or MQTT protocols. Falkonry can also read near real time data from OSIsoft’s PI System or other historians.
If your systems do not include support for one of the above data transfer methods, contact us and an appropriate connector can be developed.
No historical data is required to get started.
There is no need to find, extract and interpret maintenance logs in order to find failure examples.
Similarly, historical sensor data is not required. Falkonry will build its understanding based only on new data sent to the system.
Compute and storage hardware is not required
Falkonry’s system is cloud hosted. For cases where an on-premise instance is required, we can deploy Falkonry on a specially designed edge server. In either case, there is no need for your IT department to provide hardware to get started.
Our experience has shown that looking at historical data in order to understand how a system like Falkonry will help your operational excellence plans leads to inactionable conclusions. Demos and very short term engagements lead to “POC
purgatory.” Instead, we have designed the Falkonry software to embed quickly and easily into your production operations. By using the software in this context, by using the insights it provides in your day-to-day operations, the real value of Falkonry’s capabilities can be demonstrated. The proof is concrete and actionable. It turns the question from: should we do this, into:
what do we lose when we take this out?