More about the product
ADAPTIVE MACHINE LEARNING
One of the many steps in maximizing efficiency in utilities is data logging. Therefore, utilities often have a massive amount of data stored, and readily available. However, the utilities rarely have an effective way of utilizing all the data efficiently. When isolated and not properly analyzed, the data creates only little value.
At Sense Analytics, we collect data from different sources, both internally from the utility and externally. We then analyze it using adaptive machine learning. By applying advanced statistical methods, such as machine learning and Monte Carlo simulations, the models learn how to recognize underlying structures in the data.
Analyzing flow-data properly can, for example, be used to identify new leak occurrences on the pipes. The algorithms are trained on a large number of simulated data sets, and from that learn how to distinguish natural causes of higher consumption from actual leakages. The current approach to identifying abnormal consumption and leakages, is often to compare nighttime-consumption to static thresholds for what is considered normal levels of consumption. However, in complex systems, what is considered the 'normal' state is not static. Rather, it is dynamic and dependent on both internal and external factors. These simple solutions, therefore, often raise too many false positives. By using machine learning and more sophisticated analysis, we are able offer a solution that raises fewer false positives.
The results of our analysis are communicated for easy interpretation via our online platform. Further, the utility receives calls-to-action via SMS and email, when our analysis indicates that something is wrong with the network. The platform does not require any manual monitoring during the day.