Historically, data analytics focused on volume and variety, but “Big Data” analytics now considers extremely large diverse data sets delivered at high velocities (e.g., terabytes in seconds). Instead of hourly whole house load consumption data, smart metering now floods the data stream with appliance-specific data collected per second.
Leveraging prior experience in image processing (unstructured point clouds), econometrics (time series analysis), and similar projects, MHI has developed innovative solutions in pattern recognition, signal decomposition, and data visualization such as:
- 3D point cloud image data processing for rapid urban structure modeling
- Wavelet transforms and clustering to assess battery performance and degradation
- Monte Carlo simulations and A/B testing to generate dynamic train and test series for fault management
- Machine learning to classify signal interactions of large data series