Table 11 Performance metrics of proposed JO-LSTM controller for PWM and HBO-LSTM for BMS.
Parameter | statistical test | Test statistic (value) | p-value | Interpretation |
|---|---|---|---|---|
Energy generation (PV & Wind) | T-test/ANOVA | F = 5.62 | 0.014 | Significant variation in energy output under different conditions (p < 0.05) |
Environmental data impact | Pearson Correlation | r = 0.85 | 0.001 | Strong correlation between solar irradiance and energy output (p < 0.05) |
Grid stability (voltage/frequency variations) | F-test (Levene’s Test) | F = 2.15 | 0.087 | No significant variance in grid stability (p > 0.05) |
Battery performance (SOC, DOD, efficiency) | Paired T-test | t = 3.98 | 0.008 | Significant improvement in battery performance after optimization (p < 0.05) |
IoT-based monitoring accuracy | Wilcoxon Signed-Rank Test | W = 175 | 0.022 | IoT-based monitoring accuracy improved significantly (p < 0.05) |
Harmonic distortions in grid | Kruskal–Wallis Test | H = 10.21 | 0.018 | Significant reduction in harmonic distortions post-filtering (p < 0.05) |
Deep learning algorithm performance (JO-LSTM, HBO-LSTM, SCO-LSTM) | ANOVA (Post-hoc Tukey Test) | F = 7.89 | 0.004 | Significant difference in algorithm accuracy, favoring SCO-LSTM (p < 0.05) |
Crowd-sensing data reliability | Cohen’s Kappa/Fleiss’ Kappa | κ = 0.78 | – | High agreement among crowd-sensing sources (κ > 0.75 is considered strong) |
Overall system efficiency | Regression Analysis | R2 = 0.92 | < 0.001 | Strong predictive power of key parameters affecting system optimization |
