Analyze your process data

Why is this happening? How is this related? Search for hidden information and create actionable insight reports providing well explainable and interpretable results, that trigger “aha” moments and enable immediate efficiency improvements.

Let aivis search your historical process data (time-series or tabular) for hidden patterns and relationships to find root causes, predict quantities, or find influencing factors.

Just define your goal like “predict this quantity” or “find root causes for this type of event”. aivis then fully automated analyses your entire raw, uncleaned, unfiltered, and unsynchronized tabular or time-series data.

The result is an easy-to-understand report summarizing aivis’ findings in an explainable and actionable manner.

Use Cases

Analyzing process data with aivis can be done towards various use cases, each of which comes with its own type of report.

Find signal dependencies

Datatype: Time-series

Can this signal be predicted from the other signals? If yes, which of the other signals correlates most? Find out how signals are connected to each other and how well a signal can be predicted from the other signals. Check links between any available time-series signal like temperature, torque, speed, pressure, etc. and find out if they can be derived from others or show any unexpected correlations.

aivis presents the answer as a report, which lists all contributing signals, sorted by their correlation. This can also be used to check if any signal qualifies for a soft sensor.

Find influencing factors

Datatype: Tabular

Can this variable be well estimated based on other available variables solely? Could there be a model for this highly nonlinear relationship? Find out how variables are connected to each other and how well you can determine a critical variable from other variables within your tabular data. Check links between variables and find out if they show any unexpected correlations.

aivis presents the answer as a report, which lists all contributing variables, sorted by their correlation. This can also be used to check if any variable qualifies for an estimator.

Find root causes for disruptions

Datatype: Time-series & tabular

How can I prevent this regularly occurring process disruption? What are the reasons that lead to this disruption? Is there only one reason, or several? Let aivis autonomously search unfiltered and uncleaned historical raw data of possibly thousands of sensors to find all independent critical dynamics (segments) that lead to the same disruption. Only a simple mathematical expression is needed tell aivis when in the past the disruption was present and when it was not.

aivis presents the answer as a report, which lists all found dynamics (segments), directly related to the disruption including contributing signals and how they differ from their usual behavior.

Application examples

quality Assurance – error detection – root cause indicators – WASTE reduction

Check out this selection of application examples, where aivis has already successfully been applied in the past. Since aivis is industry-agnostic, it can be applied to countless other scenarios as well.

Pulp & Paper

Goal: Understanding paper quality

During the continuous process of papermaking, paper quality is a critical process parameter. Unfortunately, this parameter can not be measured directly because it has to be determined in the lab. Since it takes about one hour from sampling to the result, the operator is flying blind, so to speak, until the next lab value is available. 

Can the paper quality also be determined indirectly based on other, easier approachable signals?

aivis validated that the paper quality can be derived very well (r 2 = 97,7%) from other signals and revealed all contributing signals. Since those other signals are all measurable live and inline, the paper quality could be determined live as well using a soft sensor.

Oil & Gas

Goal: Understanding oil viscosity

The dead oil viscosity is a critical process parameter that heavily depends on the type of oil. Even slight differences in composition can dramatically impact the viscosity, making it almost impossible to address this issue with classical black oil correlations.

Can the oil viscosity reliably predicted based on other, more accessible quantities?

aivis validated that the oil viscosity can be derived very well (Mean absolute error 0.42) from other quantities and revealed all contributing quantities. This means that a model can be created, which, just like a reference book, can be used to determine the viscosity at any time when presented with those contributing quantities.

Pulp & Paper

Goal: Understanding paper breaks

Paper breaks during the continuous process of industrial paper manufacturing are hard to control. These breaks can occur at any time without any warning. The sheet or paper web breaks inside the machine. Subsequently, there are various cleaning and retreading steps required before production can be continued. Sheet breaks have a significant negative impact on a paper machine’s productivity.

What are critical dynamics that lead to the paper break? Which signals are relevant and how are they behaving different?

After about one hour of computing time, aivis had completed an event analysis report. With this analysis report, the paper experts were able to identify not only dynamics already known as a possible cause, but also a variety of conditions and their changes that had never been considered as a cause of paper breakage. These new findings led to immediate improvements that reduced paper breaks by 87%.

Gear Boxes

Goal: Understanding slippage times

Gearboxes are produced in a discrete manufacturing process. After each gear box is completed, its slippage time is measured to determine if the box is within the required tolerance.

Can the slipping time be reconstructed from the data? How do the torque curves that are within tolerance differ from those that are outside?

aivis was able to reconstruct the slippage time from other available data and extracted properties that could be used to identify and sort out bad gear boxes earlier in the process.

Get started and analyze your process data!