AI Dash - Wildfire mitigation system

Explore

Link to page

Context -

ElectriCo incurs approximately $50 million annually in wildfire-related damages, with outages impacting 50,000 users each year.

With worsening climate conditions and the expansion of urban-wildland interfaces, company leadership has recognised wildfire risk as a critical challenge. They are now prioritising proactive measures to enhance public safety, safeguard infrastructure, and ensure reliable service.

Objective -

AI Dash aims to equip ElectriCo’s leadership with an AI-driven wildfire mitigation system to predict, prevent, and respond to wildfire threats, enhancing resilience, safety, and operational efficiency while achieving the following goals -

⁠

⁠

Understanding Wildfire -

Wildfires can be triggered by a combination of heat, oxygen, and fuel, with various natural and external influences contributing to the risk. Understanding these factors would help in predicting and mitigating wildfire threats effectively -

Factors influencing wildfire

Probable reason

Indicator

Description

HEAT 🔥

High temperature (Natural)

Temperature trends

Persistent heatwaves increase fire susceptibility

Dry climatic condition (Natural)

Drought index (Palmer Drought severity index)

Relative Humidity

Red flag warning days

Prolonged droughts lower moisture levels, making vegetation highly flammable

Lightning activity (Natural)

Lightning detection networks

A major natural ignition source, especially in storm-prone regions

Human activity (External)

Human activity index

Unattended campfires, discarded cigarettes, and industrial sparks

Power line faults (External)

SCADA + Arc fault sensors

Electrical arcing from high-tension wires can ignite dry vegetation

OXYGEN 💨

Wind speed & direction

Remote automated weather station forecast

Strong winds can rapidly spread flames, making suppression efforts difficult

FUEL 🍂

Vegetation type

Satellite data and NDVI (Normalised Difference Vegetation Index)

Invasive weed species growth

% Dead fuel

Certain plant species, like invasive weeds, burn more easily Dead vegetation, fallen branches, and accumulated organic debris provide fuel for wildfires

Fuel moisture level

Live fuel moisture level

Dead fuel moisture level

The drier the vegetation, the higher the ignition risk

⁠

Understanding target user’s personas

ElectriCo’s wildfire mitigation efforts focus on addressing the needs of key stakeholders to ensure operational efficiency, regulatory compliance, safety, and resilience.

Utility Operations Team

🎯 Need : Real-time wildfire monitoring with automation 📌 Why? : To minimise operational risks, enhance system reliability, and reduce downtime

Field Engineers & Maintenance Crews

🎯 Need : Predictive maintenance and risk alerts 📌 Why? : To anticipate potential threats, plan proactive responses, and reduce emergency repairs

Local Communities & Emergency Responders

🎯 Need : Early wildfire detection and power reliability 📌 Why? : To protect public infrastructure, enhance safety, and ensure efficient coordination during wildfire incidents

Regulatory Agencies & Government

🎯 Need : Compliance with fire safety regulations 📌 Why? : To proactively mitigate risks, align with industry standards, and ensure public safety

Solution Design & Core Components -

Incidence response - Emergency integration

AI Dash will enhance ElectriCo’s emergency response by ensuring rapid wildfire detection and stakeholder coordination, minimising damage and response time.

Core components of the build

AI-Powered Fire Detection → Satellites & IoT sensors enable real-time heat/smoke detection for early wildfire identification

Automated Communication Platform → Instant wildfire alerts via mobile app, SMS, and calls to notify ElectricCo and relevant stakeholders

Emergency Services Collaboration → Seamless data sharing with fire departments & local authorities for coordinated wildfire containment

Risk mitigation - Proactively preventing wildfire

While climatic / atmospheric conditions can’t be prevented, AI Dash can still help ElectricCo prevent other natural fuel driven / external heat creation incidences which could potentially lead to a wildfire.

Core components of the build

AI-Powered Vegetation Monitoring → Satellite imagery & drones detect encroachment, enabling dynamic trimming schedules to manage risk hotspots

AI-Driven Human Activity Detection → Monitors human presence in high-risk zones using satellite imagery & aerial surveillance to reduce ignition risks

Grid Hardening Strategies → Advising ElectriCo on underground wiring, covered conductors, and SCADA / GIS integration for automated line de-energisation in wildfire-prone areas

Risk assessment - Accurately predicting wildfire

AI Dash will develop a Wildfire Risk & Burn Probability Index based on -

Historical wildfire trends

Natural factors driving wildfire risk

External factors contributing to combustibility / fire spread

Core Components of the Build

Precision Instrumentation → Deploying sensors / drones / satellites and data sources to measure wildfire risk factors across different geographies

AI-Driven Risk Modelling → Training a data science model to analyse causality and correlations between wildfire incidents and contributing factors

Simplified Burn Probability Score → Developing a universal, easy-to-interpret risk index that can be integrated into a digital product for external stakeholders

Prioritisation of solves -

While sub KRs are prioritised basis RICE, guiding principle focuses on first responding to ongoing wildfire risks, then move towards proactive mitigation and finally risk prediction.

Sub KR

Final priority

Reach

Impact

Confidence

Effort

Incidence response

AI powered fire detection

Automated communication platform

Emergency services collaboration

Prevention

AI powered vegetation monitoring

Ai driven human activity detection

Grid hardening strategies

Risk prediction

Precision instrumentation

AI driven risk model

Burn probability UX

⁠

Roadmap phasing -

⁠

⁠

All risk prediction and mitigation updates will be communicated to ElectricCo via automated communications platform. UX revamp for integration is not included in the current roadmap phase.

Success criteria

North Star Objective : Measure overall impact on wildfire risk reduction and operational efficiency

💰 Annual Wildfire-Related Costs → Reduction in $ spent on wildfire damages

⚡ User Impact → Reduction in # of customers affected by wildfire-related outages

🚒 Response Efficiency → % decrease in response time and improvement in response effectiveness

🌿 Wildfires Prevented → Predictive count of wildfires avoided through dynamic vegetation management and other intervention

Incident Response Metrics Objective : Assess adoption, accuracy, and effectiveness of real-time wildfire detection and response

Communication Platform Engagement →

📲 Alert Click-Through Rate (CTR) → % of users interacting with alerts

✅ Alert Acknowledgment Rate → % of alerts acknowledged by recipients

Detection Accuracy →

🔥 Fire Detection Precision & Recall → Accuracy of AI models in identifying wildfires

🌿 Vegetation Encroachment Detection → Precision & recall of AI models for vegetation risk assessment

Risk Prediction Metrics Objective : Evaluate the effectiveness of AI-driven risk modelling and its adoption

Wildfire Risk Prediction Accuracy

🔥 Precision & Recall of Risk Models → Effectiveness of AI in predicting wildfire risk

Burn Probability Score Adoption

📊 Usage & Engagement → Adoption rate of burn probability scores by ElectricCo and stakeholders

Potential challenges -

Initiative

Challenge

Mitigation Strategy

AI-Powered Fire Detection

Sensor accuracy & false positives in heat/smoke detection

Enhance AI model training with real-world data Integrate multi-sensor validation

Automated Communication Platform

Low adoption & engagement of alerts by stakeholders

Optimise alert UX, use multi-channel notifications (SMS, app, calls)

Emergency Services Collaboration

Coordination delays with fire departments & local authorities

Establish direct API integrations & real-time alerting dashboards

AI-Powered Vegetation Monitoring

High cost & complexity of continuous satellite/drones monitoring

Use AI to prioritise high-risk zones Deploy cost-effective periodic monitoring

Burn Probability Score

Low adoption by external stakeholders due to complexity

Ensure a simplified risk index with clear, actionable insights

Precision Instrumentation

Scalability issues with deploying sensors across vast geographies

Use a hybrid model combining ground sensors, drones, and satellite data

AI-Driven Human Activity Detection

Privacy concerns in tracking human movement in risk zones

Implement anonymised data processing & ensure regulatory compliance

There are no rows in this table

⁠

Want to print your doc?
This is not the way.

Try clicking the ⋯ next to your doc name or using a keyboard shortcut (

CtrlP

) instead.