Drag to rotate · right-drag / Shift-drag to pan · scroll or pinch to zoom · select a 3D part or component below.
One representative NFT unit from the proposal's six-unit experimental design.
Proposal experiment: 6 independent NFT units (3 fixed-light controls, 3 biomass-adaptive), each with a separate reservoir and circulation loop. This viewer depicts one representative unit; current ch0 simulator data must not be interpreted as six independent physical units.
Sensor readings, source and connection status will appear here.
Virtual fresh mass (g) over simulated days; not an actual harvest result.
Separate native units: EC on the left (mS/cm), effective temperature on the right (°C). Not a DS18B20 water-temperature reading.
Green line: model PPFD. Dashed line: MQTT command setpoint, when available.
Delivered-packet latency samples (ms), not physical Edge/cloud measurements.
A concise audit for research presentations. The highlighted values below come from the selected policy's latest model step; charts use MQTT history. Symbols and constants follow the project code and configuration, not parameters calibrated on real plants.
μ = 0.22 × k × fT × fEC × fLBₜ₊Δ = min(160, Bₜ + Δ × μ × Bₜ × max(0, 1 − Bₜ/160))Δ = 0.25 day. k represents seeded plant-to-plant variation. Here, B is an internal simulator state, not measured harvest mass.
fT = exp(−((T − 22)/6)²)fEC = exp(−((EC − (ECopt + d))/0.6)²)ECopt = 1.2 + 0.7 × min(1, day/25)fL = (1 − exp(−PPFD/180))/(1 − exp(−300/180))d represents individual EC-demand variation; all coefficients remain model assumptions.
B̂ = max(0, B × (1 + 0.04 εcamera))Bs = 0.6 × Bs,previous + 0.4 × B̂LA = max(0, 18 × B^0.75 × (1 + 0.05 εLA))ε is seeded Gaussian noise. B̂ feeds the estimated-biomass chart; LA is virtual leaf area. The proposal's short-term growth prediction and RGB-D fusion model are not yet implemented.
Target(day) = 160/(1 + 79 × exp(−0.18 × day))Bs < 0.95×target → 340; Bs > 1.05×target → 260; otherwise → 300Static/Env use 300 µmol/m²/s; Bio evaluates roughly every 2 simulated days. Current PPFD is a model output/setpoint, not a physical PAR reading. The model photoperiod remains 16 hours/day.
ΔE = (PPFD/300) × 0.05 kW × 16 hours/day × Δ daysEtotal = ΣΔE; experimental efficiency = harvested fresh mass (g)/LED kWhEnergy here is a lamp-model result, not an electricity-meter reading. Experimental g/kWh is valid only when each unit's harvested mass and energy use are measured.
latency = max(1, nominal + N(0, 3 ms))delivery rate = delivered packets / attempted packets × 100%Nominal Edge: 28 ms (0.2% loss); cloud: 380 ms (4% loss). These are seeded model assumptions, not hardware benchmarks.
Proposal limitation: a valid experiment requires 6 independent NFT units, calibrated PAR, air temperature/humidity, nutrient-solution and energy measurements, biomass ground truth, and RGB-D model validation. The Bio policy currently also changes EC, so this comparison does not yet isolate the effect of LED control as the proposal intends.
| Register / Node | Physical Address | Sample Rate | Bus State |
|---|---|---|---|
| ESP32 Core 0 / 1 | 0x3FF44000 (RTC) | 240 MHz Dual | Firmware stub |
| DS18B20 ROM | 28-FF-64-1E-85-16-03-9B | 1.0 Hz (12-bit) | Spec only |
| RealSense ASIC | USB Bus 001 Dev 004 | 30 FPS RGB-D | Requires camera |
| Relay Optocouplers | ULN2803A Darlington | Event Dosed | Isolated 2.5 kV |