Student Innovation-Innovative ideas that help manage and generate renewable /sustainable sources more efficiently

Description:	Student Innovation-Innovative ideas that help manage and generate renewable /sustainable sources more efficiently
Organization:	AICTE
Department:	AICTE, MIC-Student Innovation
Category:	Hardware
Theme:	Renewable / Sustainable Energytd>
Youtube Link:	Not provided
Dataset Link:	Not provided
Contact info:	Not provided

Smart India Hackathon (SIH) Solution Research Dossier

1. Executive Snapshot

Summary

The proposed solution, Aegis-Grid Hardware & Energy Management Unit, is an integrated, edge-AI-assisted hardware and software microgrid controller designed to maximize renewable energy generation and optimize power distribution [INFERENCE]. Addressing the critical problem of efficiency losses in localized renewable energy installations (e.g., solar PV and micro-wind turbines), Aegis-Grid combines a Dual-Axis Active Solar Tracker [RESEARCHED], an Adaptive Perturb & Observe Maximum Power Point Tracking (MPPT) Charger [RESEARCHED], an intelligent Battery Management System (BMS) [KNOWN], and automated Priority-Based Load Shedding [INFERENCE]. Operating on a dual-core ESP32 microcontroller with industrial Modbus RTU / RS-485 communications [RESEARCHED] and local SQLite / SPIFFS telemetry buffering [KNOWN], the system ensures zero data loss during grid/network disconnects. An lightweight Edge-AI model running locally provides short-term solar generation forecasting and real-time electrical fault/anomaly detection [INFERENCE], while a central web dashboard gives administrators microsecond-level monitoring and control [INFERENCE].

Metadata

Parameter	Details
PS Number	SIH26217 [PS]
Title	Student Innovation-Innovative ideas that help manage and generate renewable /sustainable sources more efficiently [PS]
Organization	AICTE [PS]
Department	AICTE, MIC-Student Innovation [PS]
Category	Hardware [PS]
Theme	Renewable / Sustainable Energy [PS]
Deadline	30 September 2026 [PS]
Ideas Counter	2/500 [PS]
Scraped Timestamp	2026-09-05 [PS]

Fundamental System Objective

To engineer an open-hardware, microcontroller-driven renewable energy management unit featuring dynamic MPPT, active dual-axis tracking, load prioritization, and edge-AI health diagnostics to optimize energy yield and microgrid stability in off-grid and semi-urban settings [INFERENCE].

2. Problem Deconstruction

The problem statement provided in the source JSON explicitly states: "Student Innovation-Innovative ideas that help manage and generate renewable /sustainable sources more efficiently" under the Hardware category [PS].

Atomic Requirements Breakdown

1. [REQ-GEN-01] Renewable Generation Efficiency Enhancement

   • Source Text: "generate renewable /sustainable sources more efficiently" [PS]
   • Interpretation: The physical generation mechanism must actively extract maximum possible energy from environmental inputs (e.g., solar radiation or wind movement) compared to static, unoptimized hardware setups [INFERENCE].
   • Implementation Implication: Build an active dual-axis solar tracking system powered by closed-loop LDR/sensor feedback and servo/stepper motors, integrated with a high-efficiency synchronous DC-DC buck-boost MPPT circuit [INFERENCE].
   • Confidence Tag: [PS] / [INFERENCE]

2. [REQ-MGT-01] Microgrid & Renewable Resource Management

   • Source Text: "help manage ... renewable /sustainable sources" [PS]
   • Interpretation: The solution must actively monitor, regulate, store, and dispatch energy generated by renewable sources, preventing battery degradation, overload, and unmanaged power outages [INFERENCE].
   • Implementation Implication: Implement a dynamic Battery Management System (BMS) with State-of-Charge (SoC) estimation, over-voltage/over-current/thermal cut-offs, and relay-based priority load shedding (separating critical vs. non-critical loads) [INFERENCE].
   • Confidence Tag: [PS] / [INFERENCE]

3. [REQ-HW-01] Hardware Prototype Realization

   • Source Text: "Category: Hardware" [PS]
   • Interpretation: The submission requires a physical, working hardware prototype, including sensor interfacing, power electronics, embedded firmware, and physical chassis design, rather than a purely software-based dashboard [INFERENCE].
   • Implementation Implication: Assemble a physical demonstration bench consisting of an ESP32 microcontroller, INA219/ACS712 current and voltage sensors, MOSFET switching matrices, solid-state relays, servo-driven panel mount, and battery pack [INFERENCE].
   • Confidence Tag: [PS] / [INFERENCE]

4. [REQ-SYS-01] Sustainability & Field Reliability

   • Source Text: "sustainable sources more efficiently" [PS]
   • Interpretation: The hardware unit must consume minimal auxiliary power (low self-consumption) and operate reliably under variable outdoor weather and low-network conditions [INFERENCE].
   • Implementation Implication: Firmware must use FreeRTOS low-power sleep states when generation is low, store telemetry locally during network outages, and sync to cloud/MQTT endpoints upon reconnection [INFERENCE].
   • Confidence Tag: [PS] / [INFERENCE]

3. Stakeholders & User Needs

Stakeholder	Goal	Current Pain	System Need
Off-Grid / Rural Facility Managers [INFERENCE]	Maintain uninterrupted power supply for critical equipment using local solar/battery microgrids [INFERENCE].	Unmanaged loads drain batteries prematurely; static solar panels lose 30-40% potential solar energy [RESEARCHED].	Automatic load shedding based on battery SoC, dual-axis panel alignment, and local status display [INFERENCE].
Institutional Microgrid Operators (Colleges/Hospitals) [INFERENCE]	Minimize grid electricity costs and monitor clean energy generation across multi-node installations [INFERENCE].	Lack of granular, real-time visibility into panel health, inverter status, and battery state-of-health (SoH) [KNOWN].	Industrial Modbus/RS-485 telemetry gateway connecting local node hardware to a central monitoring UI [RESEARCHED].
Field Maintenance Technicians [INFERENCE]	Rapidly diagnose electrical faults, sensor drift, and battery degradation before system failure [INFERENCE].	Time-consuming manual physical inspection of remote solar arrays and battery banks [KNOWN].	Edge-AI anomaly detection alerting on over-temperature, voltage sag, or panel shading/soiling [INFERENCE].
Disaster Response & Temporary Camps [INFERENCE]	Deploy rapidly deployable, self-sustaining renewable power hubs with zero cloud dependencies [INFERENCE].	Unreliable cloud connectivity renders smart controllers useless in remote field conditions [KNOWN].	Offline-first microgrid control logic executing locally on microcontroller hardware [INFERENCE].

4. Requirements Traceability

Proposed Feature / Module	Addressed Requirement	Technical Justification	Confidence Tag
Dual-Axis Light Tracking System	[REQ-GEN-01], [REQ-HW-01]	Rotates PV panel along azimuth and elevation axes using LDR differential logic, increasing daily irradiance capture by up to 35-40% [RESEARCHED].	[RESEARCHED]
Adaptive P&O MPPT Charge Controller	[REQ-GEN-01], [REQ-MGT-01]	Continuously adjusts DC-DC buck converter PWM duty cycle to lock onto the Maximum Power Point (MPP) of the PV array under changing temperature/irradiance [RESEARCHED].	[RESEARCHED]
Smart Priority Load Switcher (Relay Matrix)	[REQ-MGT-01]	Automatically disconnects non-essential loads (e.g., lighting/fans) when battery SoC drops below threshold (e.g., <30%), reserving power for critical loads [INFERENCE].	[INFERENCE]
Embedded BMS with Multi-Sensor Interfacing	[REQ-MGT-01], [REQ-SYS-01]	Monitors voltage, current (INA219), and cell temperature (DS18B20) in real time to prevent thermal runaway and deep discharge [RESEARCHED].	[RESEARCHED]
Modbus RTU / RS-485 Gateway + MQTT Engine	[REQ-MGT-01], [REQ-HW-01]	Provides noise-resistant industrial communication across long cable distances (up to 1 km) and bridges local hardware to MQTT endpoints [RESEARCHED].	[RESEARCHED]
Edge-AI Anomaly & Short-Term Yield Predictor	[REQ-GEN-01], [REQ-SYS-01]	Runs tinyML (TensorFlow Lite for Microcontrollers) on ESP32 to detect abnormal voltage-current curves and forecast next-hour yield [INFERENCE].	[INFERENCE]
Offline Telemetry Buffer (SPIFFS / Flash)	[REQ-SYS-01]	Caches up to 72 hours of sensor telemetry locally during Wi-Fi / GSM blackouts, preventing data loss [KNOWN].	[KNOWN]

5. Constraints & Environment Realities

1. Connectivity Realities: Rural microgrid sites in India frequently suffer from intermittent or non-existent cellular/Wi-Fi coverage [KNOWN]. The hardware must function completely autonomously without requiring active internet connections for primary control decisions [INFERENCE].
2. Thermal & Weather Exposure: Solar hardware deployed outdoors experiences ambient temperatures ranging from 0°C to 48°C and humidity up to 95% [KNOWN]. Hardware components must be housed in IP65-rated enclosures, and temperature compensation must be built into battery charging voltage thresholds [INFERENCE].
3. Power Self-Consumption: A microgrid controller that consumes excessive idle power defeats its own purpose [INFERENCE]. The overall hardware controller standby power must remain below 1.5 Watts [ASSUMPTION].
4. Dust & Soiling in Indian Conditions: Dust accumulation on solar panels in dusty regions reduces power yield by 15-25% [RESEARCHED]. The system should include dust detection heuristics based on expected vs. actual current output [INFERENCE].
5. Component Availability & Cost for Student Hackathons: Complex custom ASICs or high-end FPGA-based microcontrollers are infeasible for a student hackathon [INFERENCE]. Hardware components must rely on readily accessible off-the-shelf modules (ESP32, INA219, ACS712, SP3485 RS-485 transceiver, relay modules) [ASSUMPTION].

6. Research Questions

1. [RQ-01] What is the optimal balance between power consumed by tracking motors (servos/steppers) versus the additional solar energy harvested through dual-axis tracking? [INFERENCE]
2. [RQ-02] How can an ESP32 effectively calculate battery State-of-Charge (SoC) using open-circuit voltage and coulomb counting without expensive dedicated fuel gauge ICs? [RESEARCHED]
3. [RQ-03] What noise filter (e.g., Kalman Filter or Moving Average) is necessary to smooth LDR light intensity readings under fluctuating cloud cover? [RESEARCHED]
4. [RQ-04] What standard industrial protocol allows multi-node scalability across multiple solar panels and battery banks in an open-hardware ecosystem? [RESEARCHED]
5. [RQ-05] How can short-term solar irradiance yield be predicted on edge microcontrollers with <256KB RAM constraints? [INFERENCE]

7. Data Requirements & Sources

Data Type	Purpose	Ideal Real-World Source	Hackathon Substitute	Access Method	Availability	Privacy Sensitivity	Confidence
PV Voltage & Current	MPPT tracking, power calculation, health monitoring [INFERENCE].	High-precision Modbus power meter (e.g., Schneider IEM3000) [RESEARCHED].	INA219 I2C sensor / ACS712 current sensor + resistor divider [RESEARCHED].	I2C / ADC read on ESP32 [KNOWN].	High (Hardware generated) [KNOWN].	Non-sensitive [KNOWN].	[RESEARCHED]
Light Irradiance / Differential Light	Sun position tracking along azimuth/elevation [RESEARCHED].	Pyranometer sensor array [KNOWN].	4 x Light Dependent Resistors (LDRs) in cross configuration [RESEARCHED].	Analog pin read via ESP32 ADC [KNOWN].	High (Hardware generated) [KNOWN].	Non-sensitive [KNOWN].	[RESEARCHED]
Battery Voltage & Temperature	Battery SoC, cell protection, charge current adjustment [RESEARCHED].	Multi-cell Smart BMS with SMBus/CAN bus [RESEARCHED].	DS18B20 1-Wire temperature sensor + voltage divider [RESEARCHED].	GPIO 1-Wire / ADC [KNOWN].	High (Hardware generated) [KNOWN].	Non-sensitive [KNOWN].	[RESEARCHED]
Historical Solar Irradiance Data	Training Edge-AI forecasting model [INFERENCE].	NREL NSRDB (National Solar Radiation Database) API [RESEARCHED].	Kaggle Solar Power Generation Dataset (India Plants) [RESEARCHED].	REST API / CSV Download [KNOWN].	Open Public Data [KNOWN].	Non-sensitive [KNOWN].	[RESEARCHED]
Grid State / AC Sense	Grid availability detection for microgrid switching [INFERENCE].	Optocoupler AC voltage detector module [KNOWN].	Optocoupler module (PC817) or toggle switch simulator [INFERENCE].	GPIO interrupt pin [KNOWN].	High (Hardware generated) [KNOWN].	Non-sensitive [KNOWN].	[INFERENCE]

Categorized Data Summary

• A. Data Explicitly Mentioned: None explicitly enumerated in JSON (Student Innovation theme) [PS].
• B. Data Required by Inference: Panel V/I, Battery V/I/Temp, LDR Light Intensity, Relay Switching State [INFERENCE].
• C. Data That Must Be Collected: Real-time hardware telemetry during physical prototype testing [INFERENCE].
• D. Data That Can Be Synthetically Generated: Cloud cover simulation profiles, artificial load spikes, grid drop conditions [INFERENCE].
• E. Data That Still Needs Validation: Long-term degradation curves for specific battery chemistry (e.g., LiFePO4 vs. Lead-Acid) in high ambient Indian temperatures [NEEDS VALIDATION].

8. Existing Solutions & Gap Analysis

Existing Solution	Description	Strengths	Weaknesses	Relevance	Remaining Gap	Source / Confidence
Commercial Off-Grid MPPT Controllers (e.g., Victron SmartSolar)	Closed-source hardware controllers with Bluetooth/VE.Direct monitoring [RESEARCHED].	Highly reliable, high efficiency (>98%), robust thermal design [RESEARCHED].	High cost ($150-$500+), proprietary protocols, no integrated dual-axis tracking control [RESEARCHED].	High benchmark for MPPT efficiency [KNOWN].	Lack open hardware extensibility, integrated dual-axis tracking, and local programmable load shedding [INFERENCE].	Victron Energy Docs [RESEARCHED]
Industrial EMS Platforms (e.g., Schneider EcoStruxure Microgrid)	Enterprise SCADA and cloud-based energy optimization for large microgrids [RESEARCHED].	Comprehensive forecasting, grid-tied synchronization, multi-asset management [RESEARCHED].	Requires expensive PLC hardware, high subscription cost, cloud dependency [INFERENCE].	Enterprise reference architecture [KNOWN].	Oversized and prohibitively expensive for small-scale rural microgrids and localized installations [INFERENCE].	Schneider Electric [RESEARCHED]
Basic DIY Academic Solar Trackers	Basic Arduino projects moving a panel with LDRs and servo motors [RESEARCHED].	Simple design, low cost (<$20), easy to build [RESEARCHED].	No MPPT, no battery management, high standby power, no telemetry, unstable under cloudy sky jitter [INFERENCE].	Proof of concept for tracking mechanism [KNOWN].	Lack industrial power electronics, telemetry, load management, edge intelligence, and battery safety [INFERENCE].	IEEE Papers on Solar Trackers [RESEARCHED]

9. Candidate Solution Approaches

Approach 1: Basic Embedded Microcontroller with Static Logic (Threshold Rules)

• Mechanism: ESP32 running simple C++ code reading 4 LDRs for tracking, basic Perturb & Observe MPPT, and hardcoded battery voltage cut-offs [INFERENCE].
• Required Data: LDR voltage values, INA219 panel/battery voltage and current [INFERENCE].
• Architecture: Monolithic Arduino framework loop executing on ESP32 [INFERENCE].
• Complexity: Low [INFERENCE].
• Cost: Very Low (~₹2,500 / $30) [INFERENCE].
• Explainability: 100% rule-based deterministic logic [INFERENCE].
• Scalability: Poor; non-modular code and no standard industrial bus [INFERENCE].
• Hackathon Feasibility: Very High (3-4 days build) [INFERENCE].
• Production Feasibility: Low (Lacks fault tolerance, filtering, and industrial communications) [INFERENCE].
• Risks: LDR jitter under cloudy weather causes continuous servo hunt, rapidly draining battery power [RESEARCHED].

Approach 2: Cloud-Centric IoT Microgrid Controller with Cloud Machine Learning

• Mechanism: ESP32 acts solely as a telemetry dump node over Wi-Fi/GSM to AWS IoT Core / ThingSpeak. All processing, tracking offsets, load shedding decisions, and ML forecasting occur in the cloud [INFERENCE].
• Required Data: High-frequency cloud telemetry stream [INFERENCE].
• Architecture: Lightweight edge publisher + Cloud Serverless (AWS Lambda, DynamoDB, SageMaker) [INFERENCE].
• Complexity: High [INFERENCE].
• Cost: Medium (Ongoing cloud subscription & cellular data costs) [INFERENCE].
• Explainability: Medium (Cloud ML predictions dictate relay switching) [INFERENCE].
• Scalability: High cloud scale, but completely fragile at the edge [INFERENCE].
• Hackathon Feasibility: Medium [INFERENCE].
• Production Feasibility: Poor for off-grid rural sites due to network fragility [INFERENCE].
• Risks: Total microgrid loss of control during internet outages [INFERENCE].

Approach 3: Edge-AI Microgrid Management Unit with Hybrid MPPT, Dual-Axis Tracking, Modbus Telematics & Offline Buffering (Recommended)

• Mechanism: Dual-core ESP32 executing FreeRTOS [RESEARCHED]. Core 0 manages high-frequency power switching (MPPT PWM, BMS safety, LDR digital filtering) [INFERENCE]. Core 1 manages Modbus RTU / RS-485 telematics, local SQLite/SPIFFS data logging, and lightweight edge-AI inferencing (TensorFlow Lite for Microcontrollers) [RESEARCHED].
• Required Data: Real-time sensor readings + local historical buffer [INFERENCE].
• Architecture: Modular Edge Hardware + Industrial Bus + Web Local Dashboard + Cloud Sync Engine [INFERENCE].
• Complexity: Medium-High [INFERENCE].
• Cost: Low (~₹4,500 / $55) [INFERENCE].
• Explainability: High (Control loops are deterministic; AI is used for health prediction/yield forecast only) [INFERENCE].
• Scalability: Excellent via Modbus multi-drop architecture [RESEARCHED].
• Hackathon Feasibility: High (Achievable with standard modular prototype boards) [INFERENCE].
• Production Feasibility: High (Robust, offline-first, industrial protocol support) [INFERENCE].
• Risks: Complex firmware synchronization between FreeRTOS tasks [INFERENCE].

10. Recommended Approach & Why

Selected Solution

Approach 3: Edge-AI Microgrid Management Unit with Hybrid MPPT, Dual-Axis Tracking, Modbus Telematics & Offline Buffering [INFERENCE].

Justification & Rejection of Alternatives

• Why Approach 1 Was Rejected: Basic threshold logic suffers from severe sensor jitter under passing clouds, causing "servo hunting" which wastes harvested energy [RESEARCHED]. Furthermore, it lacks telemetry, industrial expandability, and battery health protections [INFERENCE].
• Why Approach 2 Was Rejected: Cloud-dependent microgrid control is fundamentally flawed for rural/off-grid deployment where internet access is erratic or non-existent [KNOWN]. A lost TCP connection must never result in an unmanaged battery overcharge or unexpected microgrid shutdown [INFERENCE].
• Why Approach 3 Wins:
  1. Autonomous Operation: Critical safety trips (over-voltage, short circuit, battery deep discharge) and MPPT control run locally on ESP32 Core 0 at millisecond rates [INFERENCE].
  2. Energy Harvest Maximization: Combining dual-axis active sun tracking (capturing ~35% more radiation) with dynamic MPPT yield optimization extracts peak power across all solar angles [RESEARCHED].
  3. Industrial Interoperability: Integrating Modbus RTU over RS-485 allows the prototype to interface directly with commercial energy meters and PLCs used in enterprise microgrids [RESEARCHED].
  4. TinyML Diagnostics: Edge-AI running locally alerts on panel degradation, soiling, and thermal anomalies without cloud processing latency [INFERENCE].

11. Technical Architecture

Layered Architecture Diagram

+-----------------------------------------------------------------------------------+
|                                 USER / OPERATOR LAYER                             |
|  +-----------------------------------+    +------------------------------------+  |
|  | Local OLED / TFT Display & Keypad |    | Web Dashboard (React/Streamlit)    |  |
|  +-----------------------------------+    +------------------------------------+  |
+------------------------------------------^----------------------------------------+
                                           | HTTP / WebSockets / Wi-Fi AP Mode
+------------------------------------------v----------------------------------------+
|                                  APPLICATION & GATEWAY LAYER                      |
|  +-----------------------------------------------------------------------------+  |
|  | ESP32 Core 1: Async Web Server, REST API, MQTT Publisher, Modbus Master      |  |
|  +-----------------------------------------------------------------------------+  |
|  | Local Storage: SPIFFS / Flash Memory Buffer (SQLite / Circular Ring Buffer) |  |
|  +-----------------------------------------------------------------------------+  |
|  | Edge-AI Engine: TensorFlow Lite for Microcontrollers (Fault / Yield ML Model) |  |
|  +-----------------------------------------------------------------------------+  |
+------------------------------------------^----------------------------------------+
                                           | FreeRTOS Inter-Task Queues
+------------------------------------------v----------------------------------------+
|                                HARDWARE & CONTROL LAYER                           |
|  +-----------------------------------------------------------------------------+  |
|  | ESP32 Core 0: Low-Level FreeRTOS Hardware Control Loop (100 Hz Task Engine)   |  |
|  |  - Kalman Filtered LDR Differential Tracker -> PWM Servo Control                |  |
|  |  - Adaptive Perturb & Observe MPPT -> Synchronous Buck PWM Driver             |  |
|  |  - BMS Safety Engine (SoC Calculation, Thermal Guard, Voltage Window Check)   |  |
|  |  - Priority Load Shedding Controller -> Multi-Channel MOSFET / Relay Module   |  |
|  +-----------------------------------------------------------------------------+  |
+------------------------------------------^----------------------------------------+
|                               PHYSICAL SENSOR & ACTUATOR LAYER                    |
|  +-------------------+  +------------------+  +-------------------+  +---------------+|
|  | PV Array & Wind   |  | INA219 / ACS712  |  | 4 x LDR Light     |  | Servo Motors  ||
|  | Generation Inputs |  | Voltage/Current  |  | Sensor Cross Array|  | (Azimuth/Tilt)||
|  +-------------------+  +------------------+  +-------------------+  +---------------+|
|  +-------------------+  +------------------+  +-------------------+  +---------------+|
|  | DS18B20 Temp      |  | Battery Pack     |  | SP3485 RS-485     |  | Relay Load    ||
|  | Sensors           |  | (LiFePO4 / Lead) |  | Modbus Transceiver|  | Switch Matrix ||
|  +-------------------+  +------------------+  +-------------------+  +---------------+|
+-----------------------------------------------------------------------------------+

Data-Flow Narrative

1. Sensors Acquisition: At 100 Hz frequency, ESP32 Core 0 reads 4 LDR analog voltages, INA219 I2C bus values (PV Voltage, PV Current, Battery Voltage, Battery Current), and DS18B20 temperature [INFERENCE].
2. Signal Filtering: LDR voltage values are passed through a software Kalman filter to eliminate high-frequency atmospheric noise and cloud flicker [RESEARCHED].
3. Active Tracking Execution: Core 0 calculates differential light error ($E_{azimuth} = LDR_{left} - LDR_{right}$, $E_{elevation} = LDR_{top} - LDR_{bottom}$). If $|E| > \text{deadband threshold}$, servo motors step panel orientation toward peak intensity [RESEARCHED].
4. MPPT Calculation: Core 0 calculates instantaneous PV power ($P = V \times I$). The Perturb & Observe algorithm adjusts DC-DC buck PWM duty cycle by $\Delta D$. If $\Delta P > 0$, direction of duty cycle adjustment is maintained; if $\Delta P < 0$, direction is reversed [RESEARCHED].
5. BMS & Load Shedding: Core 0 checks battery voltage against low-voltage cut-off (LVCO). If battery SoC drops below 30%, Relay Channel 2 (Non-Critical Load) trips immediately while Relay Channel 1 (Critical Load) remains online [INFERENCE].
6. Inter-Core Telemetry Sync: Processed metrics are written into a shared FreeRTOS queue and read by Core 1 [INFERENCE].
7. Storage & Edge-AI: Core 1 appends telemetry to local SPIFFS flash memory ring buffer [RESEARCHED]. Every 15 minutes, Core 1 executes the TFLite Micro model to generate next-hour yield forecast and check for V-I curve anomalies [INFERENCE].
8. Upstream Telemetry: If Wi-Fi/GSM connection is alive, Core 1 transmits buffered JSON telemetry payloads via MQTT to central broker; if disconnected, data remains safely buffered locally [RESEARCHED].

12. Technology Stack

Layer	Selected Technology	Why It Fits	Alternative Considered	Why Alternative Rejected
Microcontroller Hardware	ESP32-WROOM-32D (Dual Core 240MHz, Wi-Fi/BLE) [RESEARCHED]	Dual-core enables hardware control loop isolation from network/UI tasks; built-in Wi-Fi & low power [RESEARCHED].	Arduino Uno (ATmega328P)	8-bit, 16MHz single core lacks RAM/speed for networking, Modbus, and Edge-AI [KNOWN].
Real-Time Operating System	FreeRTOS (Native ESP-IDF / Arduino ESP32) [RESEARCHED]	Native preemptive multitasking ensures BMS protection tasks never stall due to network delay [INFERENCE].	Bare-metal loop()	Network calls (HTTP/MQTT) block timing-critical PWM and sensor polling [INFERENCE].
Local Storage / Buffer	ESP32 SPIFFS / LittleFS Flash File System [RESEARCHED]	Non-volatile flash storage on-chip (up to 1.5MB dedicated partition) [KNOWN].	External SD Card Module	External SD modules are prone to SPI bus disconnection in high-vibration field environments [INFERENCE].
Communication Protocol	Modbus RTU over RS-485 (SP3485 Transceiver) [RESEARCHED]	De facto industrial standard for energy meters, noise-immune differential signalling over 1km [RESEARCHED].	Pure I2C / SPI	I2C and SPI are limited to short PCB distances (<30cm) and succumb to industrial EMI noise [KNOWN].
Edge-AI Framework	TensorFlow Lite for Microcontrollers (TFLite Micro) [INFERENCE]	Optimized C++ runtime requiring <20KB memory footprint for quantized micro-models [INFERENCE].	Edge Impulse SDK	Edge Impulse cloud dependency for training; TFLite allows open-source offline C++ embedded compilation [INFERENCE].
Backend / Broker	Eclipse Mosquitto MQTT Broker + Node-RED [RESEARCHED]	Extremely lightweight, pub-sub architecture ideal for low-bandwidth IoT field links [KNOWN].	HTTP REST APIs	HTTP headers overhead wastes bandwidth; lacks real-time bi-directional telemetry streaming [KNOWN].
Dashboard Frontend	React.js / Streamlit Dashboard [KNOWN]	Provides responsive visualization of generation metrics, solar tracking angle, and manual overrides [INFERENCE].	ThingsBoard Cloud	Enterprise pricing; Streamlit / custom React allows 100% free hosting and offline local network operation [INFERENCE].

13. AI/ML Justification

Component 1: Edge-AI PV Health & Anomaly Detector

1. What exact problem does AI solve? Detects gradual degradation, soiling (dust cover), and partial shading faults by identifying deviations from theoretical $I-V$ characteristic power curves [INFERENCE].
2. What is the input? Real-time PV Voltage, PV Current, Surface Temperature, and Ambient LDR Light Intensity [INFERENCE].
3. What is the output? Health Anomaly Score ($0.0 - 1.0$) and status classification (NORMAL, DIRTY_PANEL, ELECTRICAL_FAULT) [INFERENCE].
4. Why can a simpler rule/system not solve it? Simple fixed thresholds fail because solar panel output varies non-linearly with temperature and solar irradiance; static rules produce excessive false alarms [RESEARCHED].
5. What data is required? Historical $V, I, T, \text{Irradiance}$ logs collected across clean and soiled panel conditions [INFERENCE].
6. How will it be evaluated? Classification accuracy, Precision, Recall, and confusion matrix on test dataset [INFERENCE].
7. What happens when it is wrong? Generates a minor warning flag on dashboard; does NOT interrupt main microgrid power flow [INFERENCE].
8. Does it require training? Yes, trained offline on host PC using Python (Scikit-Learn/TensorFlow) [INFERENCE].
9. Does it require fine-tuning? Optional post-deployment calibration [INFERENCE].
10. Does it require RAG? No [INFERENCE].
11. Does it require computer vision? No [INFERENCE].
12. Does it require speech processing? No [INFERENCE].
13. Does it require forecasting? No, real-time diagnostic classification [INFERENCE].
14. Is human oversight required? Yes, technician verifies panel condition prior to manual cleaning [INFERENCE].

Component 2: Short-Term Generation Yield Forecaster

1. What exact problem does AI solve? Forecasts expected solar energy generation for the next 1-3 hours to optimize load shedding schedules in advance [INFERENCE].
2. What is the input? Past 6 hours of rolling hourly average irradiance, temperature, and power output [INFERENCE].
3. What is the output? Predicted Power Output ($\text{Watts}$) for hour $t+1, t+2, t+3$ [INFERENCE].
4. Why can a simpler rule/system not solve it? Solar irradiance fluctuations due to moving cloud fronts exhibit complex time-series non-linear patterns that moving averages miss [RESEARCHED].
5. What data is required? Hourly weather and generation time-series datasets (e.g. Kaggle / NREL NSRDB) [RESEARCHED].
6. How will it be evaluated? Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) in Watts [INFERENCE].
7. What happens when it is wrong? Microgrid falls back to instant SoC battery threshold rules for load shedding [INFERENCE].
8. Does it require training? Yes, pre-trained Quantized LSTM / MLP neural network [INFERENCE].
9. Does it require fine-tuning? No [INFERENCE].
10. Does it require RAG? No [INFERENCE].
11. Does it require computer vision? No [INFERENCE].
12. Does it require speech processing? No [INFERENCE].
13. Does it require forecasting? Yes, time-series forecasting [INFERENCE].
14. Is human oversight required? No, automated advisory for load scheduler [INFERENCE].

Explicit Non-AI Exclusions

• Safety & Protection Circuitry: Battery over-voltage trip, short-circuit disconnect, over-temperature shutdown, and MPPT PWM adjustment MUST NOT use AI [INFERENCE]. They rely strictly on deterministic, hardcoded FreeRTOS C++ execution loops to guarantee microsecond response times and 100% deterministic safety [INFERENCE].

14. Data Architecture & Offline/Low-Network Design

Local Storage Schema (SPIFFS / Flash Ring Buffer)

-- Local Telemetry Table Structure (Stored in SPIFFS Flash Memory)
CREATE TABLE microgrid_telemetry (
    timestamp INTEGER PRIMARY KEY, -- UNIX Timestamp (seconds)
    pv_voltage REAL NOT NULL,      -- Solar Panel Voltage (Volts)
    pv_current REAL NOT NULL,      -- Solar Panel Current (Amperes)
    bat_voltage REAL NOT NULL,     -- Battery Bank Voltage (Volts)
    bat_current REAL NOT NULL,     -- Battery Charge/Discharge Current (A)
    bat_soc INTEGER NOT NULL,      -- Battery State of Charge (%)
    temp_c REAL NOT NULL,          -- Ambient / Cell Temperature (Celsius)
    azimuth_angle INTEGER NOT NULL,-- Servo Motor Azimuth Position (Degrees)
    tilt_angle INTEGER NOT NULL,   -- Servo Motor Tilt Position (Degrees)
    load1_state INTEGER NOT NULL,  -- Critical Load Relay (1=ON, 0=OFF)
    load2_state INTEGER NOT NULL,  -- Non-Critical Load Relay (1=ON, 0=OFF)
    synced_flag INTEGER DEFAULT 0  -- Cloud Sync Status (0=Pending, 1=Synced)
);

Caching, Synchronization & Retry Strategy

• Circular Ring Buffer Management: Flash memory writes are allocated to a 1MB SPIFFS ring buffer partition supporting up to 86,400 record entries (24 hours at 1 Hz, or 5 days at 5-minute sampling intervals) [INFERENCE]. Older synced records are automatically overwritten if flash capacity reaches 90% [INFERENCE].
• Network Availability Check: ESP32 Core 1 executes a background ping / MQTT heartbeat task every 30 seconds [INFERENCE].
• Batch Synchronization: Upon Wi-Fi connection establishment, Core 1 queries un-synced records (WHERE synced_flag = 0), bundles them into JSON arrays of 50 records per MQTT packet, and transmits to MQTT broker [INFERENCE].
• Exponential Backoff: If MQTT publishing fails or socket times out, retry delay doubles ($2s \rightarrow 4s \rightarrow 8s \rightarrow \dots \rightarrow \text{max } 300s$) to conserve system power and prevent network flooding [RESEARCHED].
• Bandwidth Optimization: Data payloads use short JSON key names ({"t":1725530000,"pv_v":18.4,"pv_c":2.1,"soc":85}) or binary CBOR encoding, reducing transmission payload size by up to 60% [RESEARCHED].

Offline UX Strategy

• Local Access via SoftAP Mode: When no external network exists, the ESP32 hosts its own local Wi-Fi Access Point (Aegis-Grid-AP) [RESEARCHED]. Technicians connect using a mobile browser to http://192.168.4.1 to view real-time gauges, download CSV logs directly from flash, and manually trigger panel positioning [RESEARCHED].

15. Security, Privacy & Governance

Personal Data & Location Privacy

• Zero PII Storage: System logs strictly collect electrical and environmental telemetry ($V, I, W, {^\circ}C, \text{Lux}$) [INFERENCE]. No personal user identifiers are collected or stored [INFERENCE].
• Coarse Location Configuration: Static GPS coordinates (if entered for astronomical sun calculations) are stored in encrypted NVS flash memory and never exposed on public MQTT topics [INFERENCE].

Authentication & Access Control

• Role-Based Web Access: Local web configuration interface uses HTTP Digest Authentication with CSRF token protection [RESEARCHED].
• MQTT TLS Security: Cloud telemetry uses MQTTS (Port 8883) with X.509 client certificates or username/strong token password authentication [KNOWN].

Electrical & Physical Safety Governance

• Hardware Interlocks: Dual independent hardware fuses (10A quick-blow) and optical isolation between microcontrollers and high-voltage AC/DC switching relays [INFERENCE].
• Prototype Disclaimer Label: As an SIH hackathon prototype, the hardware unit is explicitly labeled as a Concept Demonstrator. Full commercial grid synchronization requires formal compliance testing under IEC 61727 / IEEE 1547 standards before utility grid connection [RESEARCHED].

16. MVP Scope (Hackathon Build)

MUST BUILD (Core Hackathon Prototype)

• Physical dual-axis solar tracking mount driven by 2 x SG90/MG90S servos and 4 x LDR sensors [RESEARCHED].
• ESP32 microcontroller running dual-core FreeRTOS firmware [RESEARCHED].
• Working DC-DC MPPT charge control circuit using INA219 current/voltage sensors [RESEARCHED].
• 2-Channel Relay Module executing battery SoC-based priority load shedding (Critical vs Non-Critical Load) [INFERENCE].
• Local OLED/TFT display showing real-time power generation, tracking status, and battery percentage [INFERENCE].
• Embedded web server hosted on ESP32 in SoftAP mode for local control and log CSV download [RESEARCHED].
• Offline flash telemetry buffering demonstrating data recovery post-network drop [INFERENCE].

SHOULD BUILD (Targeted Enhancements if Time Permits)

• Modbus RTU / RS-485 interface connected to a secondary simulated energy node or meter [RESEARCHED].
• Pre-trained TFLite Micro anomaly detection model running on ESP32 to flag LDR sensor faults [INFERENCE].
• Central Streamlit / React live dashboard receiving live telemetry via MQTT broker [INFERENCE].

NICE TO HAVE (Post-Hackathon Roadmap)

• Automatic dust wiper motor integration for self-cleaning panels [INFERENCE].
• Anemometer wind-speed sensor integration to automatically stow panels flat during high-wind storms [RESEARCHED].
• Hardware PCB custom design (Gerber fabrication) replacing breadboard / jumper wire connections [INFERENCE].

DO NOT BUILD DURING HACKATHON

• High-voltage AC grid-tied inverter synchronization (high risk of electrical safety hazard) [INFERENCE].
• Custom silicon ASIC design [INFERENCE].
• Multi-year cloud billing infrastructure [INFERENCE].

17. Live Demo Simulation Plan

To deliver a flawless, high-impact demonstration during the SIH Grand Finale evaluation without relying on unpredictable ambient sunlight or live utility grid connections:

Hardware Environment Simulation Setup

1. Solar Irradiance Simulation: Use a controllable 100W Halogen Lamp or LED spotlight mounted on a moveable arm to act as the "Sun" [INFERENCE]. Moving the lamp across the LDR sensor array explicitly demonstrates real-time active dual-axis tracking response [INFERENCE].
2. Variable Solar Generation Simulation: Connect a variable DC bench power supply (0-20V, 0-3A) in series with a potentiometer to simulate passing cloud cover and varying PV irradiance curves [INFERENCE].
3. Load Shedding Demonstration: Connect two distinct DC loads (e.g., Load 1: 12V LED Strip as Critical Load; Load 2: 12V DC Fan as Non-Critical Load) [INFERENCE]. Adjust the simulated battery voltage potentiometer down to <11.5V (Low SoC) to visually demonstrate instant automatic disconnection of Load 2 while Load 1 stays lit [INFERENCE].
4. Network Drop & Sync Demonstration:
   • Disconnect the Wi-Fi router during active operation [INFERENCE].
   • Show on OLED display: Wi-Fi: Disconnected | Storage: Buffering to Flash [INFERENCE].
   • Reconnect Wi-Fi router [INFERENCE].
   • Show live dashboard updating instantly as stored SPIFFS logs sync without losing a single data point [INFERENCE].

Honesty Statement: The team will explicitly inform judges: "Due to indoor hackathon constraints, solar irradiance and battery discharge curves are physically driven by bench power supplies and halogen light sources reproducing real PV electrical profiles." [INFERENCE]

18. Implementation Roadmap

Target Deadline: 30 September 2026 [PS].

+-----------------------------------------------------------------------------------+
| PHASE 1: RESEARCH & CIRCUIT SCHEMATIC DESIGN (01 SEP 2026 - 05 SEP 2026)          |
|  - Finalize component selection & pinouts (ESP32, INA219, Servos, Relays, RS485)|
|  - Validate MPPT buck converter topology and sensor voltage divider math          |
+-----------------------------------------------------------------------------------+
                                         |
                                         v
+-----------------------------------------------------------------------------------+
| PHASE 2: HARDWARE ASSEMBLY & CORE FIRMWARE (06 SEP 2026 - 15 SEP 2026)            |
|  - Fabricate 2-axis acrylic mount and attach servos + 4-LDR sensor module         |
|  - Develop FreeRTOS Core 0 C++ tasks: Kalman filter LDR tracking & MPPT PWM loop  |
|  - Integrate INA219 current/voltage polling and OLED display UI                   |
+-----------------------------------------------------------------------------------+
                                         |
                                         v
+-----------------------------------------------------------------------------------+
| PHASE 3: TELEMETRY, OFFLINE BUFFER & WEB UI (16 SEP 2026 - 22 SEP 2026)           |
|  - Develop Core 1 tasks: SPIFFS ring-buffer logger & Modbus RTU / MQTT engine    |
|  - Build ESP32 embedded web server (SoftAP mode) & Streamlit cloud dashboard      |
|  - Deploy TFLite Micro pre-trained model for sensor fault detection               |
+-----------------------------------------------------------------------------------+
                                         |
                                         v
+-----------------------------------------------------------------------------------+
| PHASE 4: TESTING, SIMULATION DEMO & DOSSIER (23 SEP 2026 - 30 SEP 2026)          |
|  - Conduct artificial sun & cloudy sky hardware stress tests                      |
|  - Validate battery low-voltage load shedding and network recovery sync           |
|  - Finalize presentation slide deck, video demo, and repository documentation    |
+-----------------------------------------------------------------------------------+

19. Testing & Validation Plan

Test Category	Test Objective	Test Methodology / Tool	Acceptance Criteria
Functional Tracking Test	Verify dual-axis tracking aligns panel with moving light source [INFERENCE].	Sweep halogen spotlight across 180° azimuth and 90° tilt angles [INFERENCE].	Servo motors align panel normal to light vector within $\pm 3^\circ$ accuracy [RESEARCHED].
MPPT Efficiency Test	Verify MPPT extracts maximum power compared to direct panel connection [RESEARCHED].	Measure power output ($V \times I$) across varying load resistances [INFERENCE].	MPPT circuit yields $\ge 20%$ energy increase over direct panel connection [RESEARCHED].
Load Shedding Test	Verify priority load disconnect on low SoC [INFERENCE].	Lower battery sense voltage from 12.6V down to 11.2V using DC supply [INFERENCE].	Non-critical relay trips within $<500\text{ ms}$ of crossing threshold [INFERENCE].
Offline Recovery Test	Verify zero data loss during network blackout [INFERENCE].	Power off Wi-Fi router for 30 minutes during active logging, then reconnect [INFERENCE].	All un-synced logs in SPIFFS upload successfully; zero dropped timestamps [INFERENCE].
Industrial Noise Test	Verify RS-485 Modbus telemetry reliability near high-current relays [RESEARCHED].	Switch 10A inductive load repeatedly while polling Modbus registers [INFERENCE].	Modbus RTU CRC packet error rate remains $<0.1%$ [RESEARCHED].
False-Positive Anomaly Test	Ensure cloud shadows do not trigger false hardware fault alerts [INFERENCE].	Rapidly obscure LDR sensors for 5-second intervals [INFERENCE].	System classifies event as TEMPORARY_SHADING rather than SENSOR_FAILURE [INFERENCE].

20. Risk Register

Risk	Likelihood	Impact	Severity	Mitigation Strategy
Servo Motor Burnout / Over-Current	Medium	High	High	Implement software motion limits, deadband thresholds, and motor power cut-off when idle [INFERENCE].
Flash Memory Wear-Out	Medium	Medium	Medium	Use LittleFS wear-leveling algorithms and RAM write-buffering (flush to flash every 5 minutes) [RESEARCHED].
Sensor Calibration Drift	High	Medium	Medium	Include software baseline calibration routine accessible via local web UI button [INFERENCE].
ESP32 Core 0 Deadlock / Crash	Low	High	High	Enable hardware Watchdog Timer (WDT) with automatic 1-second system reset [RESEARCHED].
LDR Tracking Jitter Under Cloud Cover	High	Low	Medium	Apply software Kalman filtering and minimum light threshold interlocks before moving motors [RESEARCHED].

21. Cost & Scalability

Hackathon Prototype Budget (1 Unit)

• ESP32 Development Board: ₹450 ($5.50) [KNOWN]
• 2 x MG90S Metal Gear Servos & Acrylic Frame: ₹650 ($8.00) [KNOWN]
• 4 x LDR Sensors & Resistor Network: ₹50 ($0.60) [KNOWN]
• 2 x INA219 Current/Voltage Sensor Modules: ₹300 ($3.60) [KNOWN]
• 2-Channel 5V Optocoupled Relay Board: ₹150 ($1.80) [KNOWN]
• SP3485 RS-485 Modbus Transceiver Module: ₹120 ($1.50) [KNOWN]
• 0.96" I2C OLED Display: ₹220 ($2.70) [KNOWN]
• 12V Solar Panel (10W) + Small LiFePO4 / Lead Battery: ₹1,800 ($22.00) [KNOWN]
• Miscellaneous (Prototyping PCB, Wires, Enclosure): ₹400 ($5.00) [KNOWN]
• Total Prototype Cost: ~₹4,140 ($50.70) [INFERENCE]

Scaled Production Cost Model (1,000 Units)

• Integrated Custom PCB (ESP32 IC, MPPT MOSFETs, INA219 ICs, SP3485): ₹1,200 ($14.50) [ASSUMPTION].
• Industrial NEMA IP65 Weatherproof Enclosure & Connectors: ₹600 ($7.20) [ASSUMPTION].
• Heavy-Duty Worm-Gear Stepper Motors & Aluminium Brackets: ₹1,500 ($18.00) [ASSUMPTION].
• Total Unit Production Cost: ~₹3,300 ($40.00) per node [ASSUMPTION].

22. Differentiation

Dimension	Generic SIH Submissions	Proposed Aegis-Grid Solution	Requirement Direct Link
Hardware Control Architecture	Basic loop code on Arduino Uno with fixed light thresholds [INFERENCE].	Dual-Core FreeRTOS with Kalman-filtered LDR differential tracking and adaptive MPPT [RESEARCHED].	[REQ-GEN-01], [REQ-HW-01]
Battery & Load Management	No load management; battery runs until dead [INFERENCE].	Embedded BMS with dynamic State-of-Charge tracking and 2-tier priority load shedding [INFERENCE].	[REQ-MGT-01]
Communication Infrastructure	Basic Wi-Fi HTTP calls or local serial monitor output [INFERENCE].	Noise-immune RS-485 Modbus RTU industrial protocol + MQTTS telemetry bridge [RESEARCHED].	[REQ-MGT-01], [REQ-HW-01]
Network Failure Handling	System freezes or drops data when Wi-Fi drops [INFERENCE].	Offline-First design with SPIFFS ring-buffer caching and automatic batch sync [RESEARCHED].	[REQ-SYS-01]
Intelligence	Purely rule-based or generic cloud-based graphs [INFERENCE].	On-device TFLite Micro edge inference for sensor diagnostics and generation forecasting [INFERENCE].	[REQ-GEN-01], [REQ-SYS-01]

23. Future Scope

1. Anemometer & Storm Protection Integration: Adding wind speed sensing hardware to automatically tilt solar panels horizontally during high-wind gale events to prevent structural damage [RESEARCHED].
2. Micro-Wind Turbine Hybrid Interfacing: Incorporating a secondary AC-DC rectifier and MPPT input channel for small vertical-axis wind turbines (VAWT) to enable solar-wind hybrid charging during nighttime [RESEARCHED].
3. Peer-to-Peer (P2P) Microgrid Energy Trading: Expanding the Modbus/RS-485 network into an inter-node mesh network allowing neighboring microgrids to share excess battery energy via automated smart contracts [INFERENCE].

24. Assumptions & Open Items Log

Consolidated Assumptions Log

• [ASSUMPTION 01] Controller standby auxiliary power consumption remains below 1.5 Watts to avoid excessive self-discharge [ASSUMPTION].
• [ASSUMPTION 02] Prototype hardware component budget for student team is capped under ₹5,000 [ASSUMPTION].
• [ASSUMPTION 03] Industrial Modbus RTU baud rate of 9600 bps over RS-485 is sufficient for real-time telemetry polling [ASSUMPTION].

Consolidated Open Items & Validation Needed Log

• [NEEDS VALIDATION 01] Long-term battery degradation rates under high ambient Indian summer temperatures (>45°C) without active cooling [NEEDS VALIDATION].
• [NEEDS VALIDATION 02] Exact state regulatory compliance mandates (CEA / CERC) regarding small-scale off-grid solar controller reverse-current protection requirements [NEEDS VALIDATION].

Final Recommendation

The team should definitely proceed with building Aegis-Grid (Approach 3) [INFERENCE]. It directly fulfills the explicit SIH Hardware category mandate, delivers a tangibly demonstrable working physical prototype, and introduces clear engineering differentiators (FreeRTOS dual-core task isolation, active tracking, adaptive MPPT, priority load shedding, Modbus RTU, and offline telemetry buffering) that strongly appeal to hackathon judges [INFERENCE].

Recommended Stack

• Hardware Core: ESP32-WROOM-32D, INA219 Sensor, SP3485 RS-485 Transceiver, MG90S Servos, 2-Ch Relay Module [RESEARCHED].
• Firmware Runtime: C++ / ESP-IDF Framework on FreeRTOS [RESEARCHED].
• Embedded AI: TensorFlow Lite for Microcontrollers (TFLite Micro) [INFERENCE].
• Protocols: Modbus RTU (Serial RS-485), MQTT over TLS, HTTP REST (SoftAP) [RESEARCHED].
• Dashboard / Cloud: Streamlit / Node-RED + Mosquitto MQTT Broker [RESEARCHED].

Biggest Dependency

Availability and precision calibration of current/voltage sensors (INA219) and DC power switching components during prototype assembly [INFERENCE].

Biggest Risk

Servo motor mechanical binder/stalling or excessive current drain if tracking algorithms get caught in continuous oscillation during turbulent weather [RESEARCHED].

Biggest Technical Challenge

Implementing clean FreeRTOS inter-task queue synchronization between Core 0 high-frequency PWM switching logic and Core 1 network/flash-writing operations without triggering the hardware Watchdog Timer (WDT) [INFERENCE].

First 5 Actions

1. Action 1: Purchase standard hardware components (ESP32, INA219 sensor modules, MG90S metal-gear servos, 4 x LDRs, SP3485 module, 2-channel relay module) [INFERENCE].
2. Action 2: Wire and validate LDR differential readings on ESP32 breadboard and write a basic C++ Kalman filter smoothing routine [INFERENCE].
3. Action 3: Assemble 2-axis acrylic/3D-printed panel mount and attach servo motors; calibrate azimuth/elevation degree sweep limits [INFERENCE].
4. Action 4: Implement FreeRTOS dual-core structure: Core 0 running MPPT PWM and battery relay switching; Core 1 writing telemetry to SPIFFS flash [INFERENCE].
5. Action 5: Build local SoftAP web server interface (http://192.168.4.1) to verify direct CSV telemetry download and live browser control [INFERENCE].

Validation Before Coding

• Verify that the INA219 sensor I2C address does not conflict with the OLED display I2C address (0x40 vs 0x3C) [RESEARCHED].
• Measure idle current draw of servo motors to ensure logic circuit voltage regulators do not overheat [INFERENCE].

Buildability Score /100

92 / 100 [INFERENCE] (Extremely buildable by a 4-6 student engineering team within a hackathon timeframe using accessible off-the-shelf development modules and open-source embedded libraries).

Confidence

HIGH [INFERENCE] (Grounded in proven embedded engineering patterns, standard industrial Modbus protocols, established hardware solar tracking research, and exact problem requirements).

Research Sources

Title	URL	Source Type	Supported Claim	Reliability Assessment
Smart India Hackathon Official Portal	sih.gov.in	Government Portal	Problem statement metadata, deadline, AICTE MIC provider details [PS].	Official Government Source (100% Reliable) [1.1]
IoT-Enabled Dual-Axis Solar Tracking System Using ESP32	ResearchGate Publication	Academic Paper	Dual-axis tracking increases daily solar harvesting yield by up to 35-43% over static panels [RESEARCHED].	Peer-Reviewed Research (High Reliability) [2.1]
IoT and AI-Driven Approaches for Energy Optimization in Off-Grid Solar Systems	MDPI Journal Paper	Academic Journal	ESP32 and Raspberry Pi hybrid architectures for local microgrid monitoring and edge AI [RESEARCHED].	Peer-Reviewed Publication (High Reliability) [2.5]
Modbus RTU Communication with ESP32 and RS-485	NORVI Industrial IIoT Guide	Technical Documentation	Industrial RS-485 Modbus implementation patterns on ESP32 microcontrollers [RESEARCHED].	Official Hardware Vendor Docs (High Reliability) [3.2]
NLR PVWatts API & Solar Resource Models	NLR Developer Network	Official Lab API Docs	NREL/NLR solar energy estimation models and API endpoint structures [RESEARCHED].	Official National Lab Documentation (High Reliability) [4.1]
MNRE National Wind-Solar Hybrid Policy	MNRE Government Portal	Government Policy	Indian grid guidelines and technical conditions for hybrid renewable energy management [RESEARCHED].	Official Government Policy (100% Reliable) [4.1]