Pisa-based IProd, a software house that invented iProd MOP (Manufacturing Optimization Platform), the innovative online platform for the management of production and planning of company resources in a 4.0 perspective, offers the exclusive AiProd application. This proprietary and patented application is equipped with innovative Industry 4.0 application components that exploit the availability of new Artificial Intelligence and Machine Learning algorithms designed for manufacturing. AiProd, in fact, realizes in real time the combined integration of IoT data and audio files both acquired by the machines thanks to Alleantia's I4.0 "Plug & Play" technology with Leonardo's Audio Insights Artificial Intelligence system in cloud for training, refinement and recognition of the acoustic pattern. Specifically, AiProd, the first AI application of Acoustic Insights in manufacturing, interprets and harmonizes the IoT data generated by production assets during their operation, correlating them with the sounds emitted and recorded during processing and creates models with which to then compare. the data that will be collected during processing or testing of machinery.
Specifically, AiProd allows both the manufacturer and end user to have a predictive system to anticipate the real moment in which a fault occurs, thus planning maintenance according to the different strategies available and minimizing the impacts on production, preventing costly production waste or costly downtime for technical assistance.
The process parameters of the machine are recorded in time, while the operation of its main modules is monitored and compared continuously, keeping under control the trend over time of the main variables (tags) and their deviation from typical values.
Therefore, the data defined as "typical" that correspond to an effective and efficient use of the machine (which produces regularly and with optimal quality) are compared with the values assumed by the process variables of the machine when an anomaly occurs, associating them with acoustic phenomena generated during its operation. In this way, it becomes possible to build a predictive system that correlates the quality and efficiency aspects of the machine with those related to its maintenance.
If the tags are recognized, the Audio Insights technology analyzes the data with an already available neural network, comparing the data with the reference data and evaluating them through a score: high scores are synonymous with the functioning of the asset in compliance with the expected standard, while low scores are synonymous with error or malfunction of the machine. In this case, the App, depending on the score detected, may interrupt the operation of the machine or will promptly notify the staff in charge of production or maintenance or those involved in testing modules and assembled products of the anomaly. If the tags are new, however, the acquired data will be used for the automatic preparation and training of a new neural network.