GridDefense – AI Anomaly Detection

CNN-LSTM Model Analysis & Threat Detection

The Model Output

Prediction Summary

Status: Attack Detected
Type: Replay Attack
Attack Probability
89%
Normal Probability
11%

Anomaly Detection

⚠ Suspicious ● Anomalies

Key Contributing Factors

Sequence Deviation High Impact
Signal Drift Moderate Impact
Temporal Pattern Moderate Impact
Amplitude Spike Low Impact

Convolutional Neural Network detected spatial anomalies in voltage and frequency patterns. Primary features: sudden spikes, pattern deviations.

Long Short-Term Memory identified temporal inconsistencies. The model flagged unusual sequence patterns over the past 30 samples.

Detected Replay Attack based on temporal sequence and signal drift anomalies.