Detectiv.ai®
Industrial decision support system
Ready-to-production artificial intelligence solution for predictive maintenance, in-line quality control and automatic detection of process drifts.
Detectiv.ai® easily identifies anomalous behaviors in real-time industrial data collected from sensors, PLCs and production processes, to anticipate failures and downtime.
Contact us for more information
Benefits
Increase profitability of the monitored systems:
driving and implementing corrective actions as well as providing warning in case of process drifts.
Improving product and services:
identification of unwanted behaviors and their causes that can be avoided with product reviews or allow for an upgrade of one’s service.
Protect and value the experience:
assimilate the ability of experts to interpret data and allows you to use it in an unlimited way.
3 reasons to choose us
Vibration analysis expertise:
hardware and software for in-depth analysis and understanding of mechanical phenomena and drifts.
Scaled proposal according to the specific integration needs:
Detectiv.ai® can be integrated as a server or internally manage the logic with which to access the data.
Safe investment:
standardization of the runtime and workflow enable fast feasibility analysis, short time-to-market and measurable ROI.
Modules
ANOMALY DETECTION
Used to identify conditions that deviate from normality/optimality, working under the paradigm “train on normal data only” (when there are no known examples or very few examples of anomalies, or the types of anomalies and drifts might be very different one from each other).
EVENT CLASSIFICATION
Used when there is the opportunity to assign a label to a specific event (continuous or discrete), for example if we need to provide information regarding a fingerprint cycle of the machine, or to assign a severity label of an anomalous event.
KEY INDICATOR PREDICTION
Used to correlate the measurements within industrial processes with an effect on the final product, in order to timely predict its value and preventively implement corrective actions.
Key features
CREATE AI MODELS
A single instance of Detectiv.ai® can handle multiple models, pre-processing and post-processing functions.
Each model might be dedicated to a specific subsystem of the analysed machine.
LISTEN TO YOUR MACHINES!
Detectiv.ai® can be used in both batch (fingerprint cycle) and real-time scenarios, analysing streams of data incoming from data sources directly. It is possible to process multiple signals together which may come from accelerometers, PLCs, servodrives and so on.
RUN AND MANAGE AI MODELS
Detectiv.ai® can be integrated and used directly by client through its library. However, we are able to provide integration activities with the external system, so that the client can receive a fully functioning solution embedded in its systems.
Detectiv.ai® provides synthetic and intuitive dashboard representing how much an observation is considered anomalous (anomaly detection), or to which class the analysed time series belongs (event classification), or the key indicators predicted.
How it works
01. Collect data
from the machine.
03. Train the model
(anomaly detection, event classification and other upcoming problem types).
START
FINISH
02. Choose training data
that represent normal working conditions of the machine (anomaly detetction models) or the events that are under analysis (event classification models).
04. Call inferences
on the trained model with new data coming from the machine.
01. Collect data
from the machine.
02. Choose training data
that represent normal working conditions of the machine (anomaly detetction models) or the events that are under analysis (event classification models).
03. Train the model
(anomaly detection, event classification and other upcoming problem types).
04. Call inferences
on the trained model with new data coming from the machine.
PREDICTIVE MAINTENANCE AND QUALITY, FOOD & BEVERAGE
Anomaly detection on data from electric motors, for a major OEM, a new machine generation equipped with sensors and AI to anticipate potential malfunctions and downtime.
PREDICTIVE QUALITY AND MAINTENANCE, DIE-CASTING
Real time machine monitoring to improve the operation of the aluminium die-casting machines, to allow the prevention of failures and unexpected machines downtimes and the planning of maintenance operations.
KEY INDICATOR PREDICTION, FOOD & BEVERAGE
Predict the product final quality with only the machine’s input parameters available by identifying relevant variables and using them within an artificial intelligence model.