Open Access
Review
Table 2
Summary of the related works based on the prediction model for daylight artificial light integration.
| Literature | Data | Window blind control | Algorithm details | Model evaluation | Performance |
|---|---|---|---|---|---|
| Xie and Sawyer (2021) | Predictors: Irradiance, sky ratio, glare; Response: Window Blind Position; Data Source: Experimental real-time field data | Type: Venetian Blind; Position: East, West, North, and South; Control: Up–down, intermediate position with static slat angle of 45° | Algorithms: KNN, SVM, and RF; Parameter Tuning: Hyperparameter Optimization | Statistical: NA; Real-time: window venetian blind Slat angle control 0°, 15°, 30°, and 45° for glare prediction below 0°, 15°, and 30° | Could prevent 86.5%–96.9% of the glare and potentially reduce lighting energy use by 80.8% |
| Sanjeev Kumar et al. (2020ª) | Predictors: Indoor and outdoor window illuminance, work plane/ceiling illuminance, vertical illuminance; Response: Window Blind Position DGI, DGP, and GS; Field data collected from test room | Type: Venetian Blind; Position: East; Control: Up–down and intermediate position with static slat angle of 45°. | Algorithms: ET, ANN, GPR, and SVM; Parameter Tuning: Bayesian Hyperparameter Optimization; Feature Selection: DT; Type: Regression for DGP/DGI and classification for GS | Regression: MSE, RMSE, MAE, MAPE, and R2; Classification: Accuracy, PE, AUC-RoC, Precision, and Recall; Hypothesis: Friedman’s ranking test for model selection | ET model performs better than other models. The accuracy of DGP, DGI and GS are 99.84%, 99.39% and 94.4%; Total energy Consumes 17% less power than the uncontrolled system and 15% less power than the baseline system. |
| Chiesa et al. (2020) | Predictors: Illuminance (indoor/outdoor) and temperature Response: Blind Control, and LED; Data Source: Experimental (50 × 50 × 50 cm) Set up | Type: Venetian Blind; Position: East, and West; Control: Slat Angle Control (No up-down control) | Algorithms: Fuzzy Logic; Parameter Tuning: NA; Feature Selection: NA; Type: NA | Real-time: Indoor illuminance (300 lux), outdoor illuminance (1000 lux), and Lighting Energy. | When illumination is controlled by zones, Zones 1 and 2 consume power of 0.7 W and 0.69 W respectively compared to 4 W and 1 W without control |
| Yeon et al. (2019) | Predictors: Outdoor temperature, Relative Humidity, Solar Altitude, Solar Radiation, zone people occupant count; Response: Blind Slat angle (0 to 180°); | Type: Venetian Blind; Position: South; Control: Slat angle control (no up and down control) | Algorithms: ANN; Parameter Tuning: N/S; Feature Selection: N/S; Type: Regression | Statistical: RMSE; | The overall energy consumption was 9.1% lower than the baseline scenario of blind angle set at 50°. |
| Sadeghi et al. (2017) | Predictors: environmental variables, human attributes; Response: Blind up and down position; Data Source: Field study conducted in private offices | Type: motorized roller shades; Position: South; Control: Raising and lowering (25%, 50%, 75%, and 100%) | Algorithms: Bayesian modeling and BR models; Parameter Tuning: N/S; Feature Selection: N/S; Type: Regression | Statistical: RMSE | BMBCL models predicted shade raising/lowering and electric light dimming actions. BR models predicted the intermediate operating states of the systems. |
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.