Predictive analysis and Forecasting
With Predictto we want to help you make the best decisions for your maintenance management system. To do this, we use advanced analytics to create predictive models and give forecasts about the future status of your assets. Our algorithms use Machine Learning together with statistical models to give you a complete forecast, mathematically informed but easy to understand, so that you can make informed decisions about your asset maintenance plans.
First, it’s important to clarify what reliability is: Also called survival probability, reliability estimates the probability that an asset will work without failures, from the beginning of its operation to a determined moment in time. Reliability decreases over time, until the asset reaches the end of its life.
Reliability models describe the behavior of systems, and are built using mathematical tools and statistics that show, explain and predict the capability of the systems to operate without failures over time.
Predictto recommends using reliability models for assets that are more susceptible to failures and that are not continuously monitored. This is because reliability models use failure records (obtained from the work order history of the assets) as training data.
In simple terms, deterioration models allow you to estimate how the status of an asset evolves over time, through relevant physical and/or chemical variable analysis measured over time. Using Machine Learning and advanced analytics, we can predict the values of these variables in the future and offer a forecast for the state of the equipment.
With Predictto, we use an algorithm that allows us to establish a probabilistic model of the behavior of one or more variables over time, using measurements of these variables as training data. With this algorithm, it’s possible to estimate the behavior of the variables in the future and thereby create forecasts for the state of the asset over time. These forecasts are reliable because they are based on mathematically informed analysis. This way, we don’t offer blind predictions. Rather, we give reliable forecasts.
It's important to mention that if you want to perform this type of analysis, you need to have monitoring data, whether through meters connected to Fracttal or other external meters integrated into your Predictto account.