Package neqsim.process.safety.risk.ml

package neqsim.process.safety.risk.ml

Machine Learning Integration Package for Risk Assessment.

This package provides a standardized interface for integrating external machine learning models with the NeqSim risk framework. It enables data-driven risk assessment and prediction.

Key Classes

• RiskMLInterface - Main interface for ML model management and prediction

ML Use Cases

• Failure Prediction: Predict equipment failures before they occur
• Anomaly Detection: Identify unusual patterns indicating potential problems
• RUL Prediction: Estimate remaining useful life of equipment
• Risk Scoring: ML-based risk scoring for complex scenarios
• Optimization: Optimize operations under risk constraints

Integration Example

// Create ML interface
RiskMLInterface mlInterface = new RiskMLInterface("Platform Risk ML");

// Register a failure prediction model
RiskMLInterface.MLModel model =
    mlInterface.createFailurePredictionModel("pump-failure-v1", "Pump Failure Predictor");
model.setVersion("1.0.0");
model.setAccuracy(0.92);

// Set up predictor (e.g., calling external Python/TensorFlow service)
model.setPredictor(new MLPredictor() {
  public MLPrediction predict(Map features) {
    // Call ML service
    RiskMLInterface.MLPrediction pred = new RiskMLInterface.MLPrediction(model.getModelId());
    pred.setPrediction(callMLService(features)); // Your ML service
    pred.setConfidence(0.85);
    return pred;
  }
});

// Register feature extractor
mlInterface.registerFeatureExtractor("process", new FeatureExtractor() {
  public Map extractFeatures(Map processData) {
    Map features = new HashMap();
    features.put("pressure", processData.get("PT-001"));
    features.put("temperature", processData.get("TT-001"));
    features.put("vibration", processData.get("VT-001"));
    return features;
  }
});

// Make prediction
Map processData = getLatestProcessData();
RiskMLInterface.MLPrediction prediction =
    mlInterface.predictWithExtraction("pump-failure-v1", "process", processData);

if (prediction.getPrediction() > 0.7) {
  // High failure probability - trigger alert
  generateMaintenanceWorkOrder();
}

Python Integration

The ML interface is designed for easy integration with Python-based ML models via REST API or direct JPype/Py4J bridging.

Version:

1.0

Author:

NeqSim Development Team

See Also:

• ConditionBasedReliability

Related Packages

Package	Description
neqsim.process.safety.risk	Probabilistic risk analysis framework for process safety.

Class	Description
MLIntegrationExamples	Examples and templates for integrating external ML frameworks with the risk system.
MLIntegrationExamples.BaseMLAdapter	Base adapter class with common functionality.
MLIntegrationExamples.MLModelAdapter	Interface for ML model adapters.
MLIntegrationExamples.OnnxAdapter	Adapter for ONNX Runtime models.
MLIntegrationExamples.RestApiAdapter	Adapter for REST API-based model serving.
MLIntegrationExamples.TensorFlowAdapter	Adapter for TensorFlow SavedModel format.
MLIntegrationExamples.ThresholdModel	Simple threshold-based model for testing and fallback.
RiskMLInterface	Machine Learning Integration Interface for Risk Assessment.
RiskMLInterface.FeatureExtractor	Feature extractor for process data.
RiskMLInterface.MLModel	Machine learning model wrapper.
RiskMLInterface.MLModel.ModelType
RiskMLInterface.MLPrediction	ML prediction result.
RiskMLInterface.MLPredictor	Functional interface for ML model prediction.
RiskMLInterface.ModelPerformanceMetrics	Model performance metrics.
RiskMLInterface.PredictionRecord	Prediction record for history tracking.