Machine Learning

1. Supervised learning
   
   1. Regression
      

      1. Gaussian process (GP) models

      2. Deep neural networks (DNN)

      3. Bayesian neural networks (BNN)

      4. Support vector machines

   2. Classification
      
      1. K-nearest neighbor algorithm
2. Unsupervised learning
   
   1. Dimension reduction
      
      1. Principal component analysis (PCA)
   2. Clustering
      
      1. K-mean
3. Reinforcement learning
4. Convolution neural networks (CNN)
5. Long short-term Memory (LSTM)
6. Convolutional LSTM
7. Graph neural network

1. Active/adaptive learning
2. Ensemble/hybrid learning
3. Transfer learning
4. Physics-informed machine learning (PIML)

We use machine learning for the following purposes: building surrogate models for expensive physics models,​ constructing data-driven machine learning models, learning multi-variate probability distributions, reducing the dimension of the problem, correcting the physics/data-driven model, and transferring information (model errors and discrepancies) from tested configurations to untested configurations.

• Developed machine learning (ML) surrogates for power grid decision-making (optimization) algorithms; used a novel hazard-aware loss function for model training
• Directly predicted quantities of interest (load shed, reserve capacity, etc.), which depend on the decisions (generator commitment and dispatch) made by solving multiple constrained optimization problems
• Developed a just-in-time learning methodology, where ML surrogates for each day are trained every day and continuously updated during the day-ahead market settlement and reliability assessment phase
• Used ML surrogates for real-time risk estimation given the latest (e.g., three hours-ahead) forecast of load and wind/solar generation, and improved grid operator’s situational awareness

• Gaussian process (GP) surrogate models for output quantities of interest (e.g., porosity, geometric accuracy, surface roughness, residual stress)
• GP surrogate models for model discrepancy, to correct the model prediction, trained using multi-physics simulations (fluid, thermal, mechanical).
• Computationally expensive high-dimensional multi-physics models are used for the simulation of laser powder bed fusion (LPBF) additive manufacturing.
• Computational efficiency in the construction of surrogate models is achieved by using dimension reduction techniques such as active subspace, principal component analysis (PCA), and singular value decomposition (SVD).
• Deep neural network (DNN) to model the bond formation process and mesostructured of printed fused filament fabrication (FFF) parts.
• Physics-informed machine learning (PIML) to improve the prediction accuracy and physics consistency of machine learning models.

Applying ML tools (e.g. artificial neural networks) to

• Create a hybrid physics-ML model, where ML is used to correct model form errors in the physics model
• Create surrogate ML counterparts of physics models of dynamic systems à use for non-intrusive evaluation of model errors
• Use ML models à transfer information (model errors and discrepancies) from tested configurations/locations/load cases to untested prediction configurations/locations/load cases.

• Aviation safety reporting system (ASRS) data
• Support vector machine (SVM) model to obtain risk category from event synopses
• Deep neural networks (DNN) to model the complex relations between event contextual features and event outcomes
• Fusion of two machine learning models to perform risk categorization

• FAA – system wide information management (SWIM) data for flight trajectories
• Deep neural network (DNN) for one-step-ahead flight trajectory prediction
• Deep Long Short-Term Memory (LSTM) neural network for longer-term flight trajectory prediction
• Multi-fidelity prediction by blending the two deep learning models

• DASHlink data on entire flight trajectory à extract landing information
• LSTM neural network for early warning of hard landing
• Bayesian neural network to capture epistemic uncertainty in model prediction

• Analyze the historical passenger airline accidents that happened from 1982 to 2006 as reported in the National Transportation Safety Board (NTSB) aviation database.
• Develop a Bayesian network representation of all the accidents by capturing the causal and dependent relationships among a wide variety of contributory factors and event sequences in terms of aircraft damage and personnel injury.
• Estimate the conditional probabilities in the Bayesian network by calibrating a monotonically increasing function, whose parameters are calibrated using the probability information on single events in the available data.
• Automate the generation of the Bayesian network in compliance with the XML format used in the commercial GeNIe modeler.

• SWIM Terminal Data Distribution System (STDDS) for surface movement event data
• Spark for constructing sequences of graphs to represent the trajectories of multiple objects on the airport ground
• Graph neural network for predicting the probability distribution of future ground movements
• Conflicting risk assessment using the model prediction

• Diagnosis, prognosis and decision-making for rotorcraft (inspection/maintenance/repair, mission planning, operational maneuvers).
• Training of machine learning models (Gaussian process (GP), DNN, LSTM, reinforcement learning) using rotorcraft dynamics and finite element-based stress analysis models.
• Sensor data used for damage diagnosis and model calibration.
• Mission planning: Optimize future mission profile for safe mission completion, based on diagnosis of current vehicle state and prognosis of future damage growth.
• Damage adaptive maneuver: Adjust flight maneuvers to minimize vibratory loads and redistribute component stress, in order to minimize damage growth.

• Supervised machine learning using training data constructed from 2-D FEA models for concrete structures with a single hidden crack.
• Two data-driven models constructed for damage diagnosis.
  • Prediction model that estimates the side-band sum (SBSum) at a sensor location, from vibro-acoustic modulation (VAM) testing. Bayesian damage diagnosis localizes the damage using observed VAM test data.
  • Classification model that uses VAM test parameters and the measured SBSum values at a particular sensor to classify the locations as damaged or undamaged.

• Damage detection with damage sensitive features identified by Singular Value Decomposition (SVD)
• Support Vector Machine classification model.
• Damage localization with crest factor and k-means clustering.

• Augment experimental data with 2D and 3D finite element models.
• Damage localization and quantification with transfer learning from sophisticated deep neural networks, such as VGG, ResNet, etc.
• Uncertainty quantification of damage diagnosis using Monte Carlo dropout.

• Data collection from cancer patients through
  • Passive surveillance with wearable sensors (activity, heart rate, and geolocation data)
  • Active surveillance with mobile app-based patient/caregiver reporting of non-routine events and patient-reported outcome measures
• Construction of predictive model of clinical deterioration in patient using machine learning methods such as random forest, artificial neural networks, etc. to predict the possible occurrence of unplanned treatment events
• Enabling timely response to clinical deterioration through early detection and prediction of clinical deterioration