Industrialize Machine Learning For Operations
TrendMiner MLHub accelerates the collaboration between local operational experts and central analytics groups.
Existing ML models can be easily deployed with this separate module. Facilitate your machine learning projects, with all the capabilities of the TrendMiner NextGen Production Client, like lightning-fast searches, on-click importing of data frames, and immediate visualization of notebook outputs in DashHub.
MLHub – Key Capabilities
CLEAN DATA COLLECTION
Make the most of TrendMiner views. Use saved views that contain interesting fragments of time-series and contextual data and open them as data frames in MLHub, where they can be further pre-processed and analyzed.
MLHub support multiple ways of working. From building and training models, in notebooks to the deployment of models from 3rd parties (like AWS Sagemaker or Microsoft Azure) or from your own data science stack
Our Jupyter Notebook environment comes pre-loaded with the most common toolkits for data processing in addition to importing your own custom tools or consulting the extensive libraries of the world’s largest open-source communities.
By converting them to open standards (PMML or ONNX), the most diverse machine learning models in MLHub are made available for efficient scoring by TrendMiner’s internal ML engine.
TrendMiner ML Model Tags offer a no-code interface to leverage the machine learning results for any operations user, like any other tag (sensor reading), for greater insights into process behavior.
Augment operational analytics by using machine learning model outputs throughout TrendMiner. Leverage Machine Learning model tags, enrich process events, and see advanced visualization tiles.
MLHub – Use Case Areas
TrendMiner is versatile software capable of handling a wide variety of use cases to analyze, monitor, and predict process and asset performance within their operational context. With the addition of MLHub, machine learning capabilities become available throughout the solution. These include clustering, classification, regression, and dimension reduction. MLHub provides the resources to solve new end-to-end use cases in the following areas:
Soft sensors can reduce costs, create redundancy for crucial assets, and save time and manpower when resources are short. Use cases include modus operandi classification, clustering of seasonal behavior, and estimated steam consumption.
TrendMiner offers enhanced anomaly detection with fingerprinting and thresholding. MLHub extends the capabilities with new principles such as isolation forest and K-nearest neighbor. The Self-Organizing-Maps algorithm can handle fluctuating durations in processes.
Extend TrendMiner’s predictive maintenance capabilities. Use cases include survival analysis for events that haven’t happened yet, simulation for most durable runtime process parameters, classification of failure modes, and predicting batch and quality.
Process Data Classification
MLHub enables users to identify various set points by analyzing the quality features of raw materials. This helps to cluster similar processing conditions and predict yield, aiding in process optimization.
LEVELED PLAYING FIELD
More than ever before, data science is becoming a team sport. MLHub levels the playing field and bridges the gap between data science groups and operations.
EASY TO GET STARTED
A model is only good if it is used. MLHub makes it easy and fast to create, train, and deploy machine learning models into use, which helps improves the adoption rate among users.
MLHub contributes to sustainability, plant safety, and overall profitability goals by helping to solve more use cases and increase an organization’s data maturity level.