Manufacturing is complex and hard to control and existing technologies often fail at preempting production glitches. This is the motivation behind DataProphet’s OMNI product: an AI solution to reduce the cost of non-quality for heavy industries. OMNI enables effective orchestration of large numbers of control parameters across a production facility.
In manufacturing, each process step can have hundreds of controllable variables that affect the quality of the output of the current step, as well as the operation and effectiveness of all downstream process steps. With each additional variable, the number of possible configurations of process parameters increases and quickly diminishes the ability of even the most skilled experts to effectively manage and control the production process.
With traditional statistical process control software, control limits are programmed by human experts and often fail to capture the complex interactions between all process variables. As a result, even when parameters are within the defined control limits, there is still variability in plant performance and production quality that human experts cannot explain.
DataProphet’s solution learns from historical manufacturing data and exploits the collective knowledge and experience of all experts who have influenced the plant in the past. Moreover, OMNI preserves the experience of outgoing employees and institutionalises those years of operational experience for the benefit of the manufacturer’s current and future workforce. From historical observations, OMNI can isolate optimal plant configurations and calculate the parameter adjustments that are required to bring the production facility into a stable, optimised state. OMNI also continuously adapts to changing conditions and prescribes ongoing parameter adjustments, based on upstream and downstream variable changes, reducing variance in quality, as well as the cost of non-quality.
The large scale application of AI to industrial systems is extremely novel and required a number of innovative extensions to existing methods. Unsupervised AI systems have been used for decades to find patterns in complex datasets. In these machine learning models, the input data can be organised in a semi-interpretable format that retains crucial information about historical production trends.
By taking advantage of this capability of machine learning techniques, distinct operating paradigms can be discovered from the historical data of a given production facility ...More Here