EV Charging Pile Real-Time Monitoring System

01 · Project Information

• Project Duration: February 2023 - October 2023

• Role: Team Member

• Institution: Southwest Jiaotong University

• Project Type: Provincial-Level Innovation Project

02 · Project Overview

This project focused on developing an AI-driven, cloud-based monitoring system for electric vehicle charging infrastructure.

The system integrated artificial intelligence, the Internet of Things, edge computing, and cloud technologies to support real-time monitoring, anomaly detection, predictive maintenance, and dynamic load balancing across EV charging piles.

The project aimed to reduce charging-equipment failures, improve energy-distribution efficiency, and support sustainable energy management within 5G smart-grid environments.

03 · Project Background

As the number of electric vehicles and charging facilities increases, charging networks face growing challenges in equipment management, fluctuating energy demand, and operational complexity.

Traditional maintenance approaches frequently rely on scheduled inspections or repairs after a failure occurs. This can create several limitations:

• Early anomalies may not be detected

• Fault diagnosis can depend heavily on manual experience

• Large-scale equipment inspection is expensive

• Fluctuating charging demand may increase grid pressure

• Centralized data processing can introduce latency

• Power may not be allocated efficiently across charging piles

This project applied AI to charging-equipment data and used IoT and edge computing to support a transition from reactive maintenance to predictive maintenance.

04 · Project Objectives

The primary objectives were to:

1. Develop an AI-powered EV charging monitoring system with real-time analytics

2. Implement anomaly detection and predictive maintenance

3. Collect charging-equipment data through IoT sensors

4. Reduce real-time processing latency through edge computing

5. Optimize load balancing in a 5G smart-grid environment

6. Improve energy efficiency and charging-network stability

7. Validate the commercial feasibility of AI-enabled energy management

8. Develop a scalable architecture for larger charging networks

05 · System Architecture

The system consisted of the following modules:

• Charging-pile data acquisition

• IoT communication

• Edge computing nodes

• Cloud data-management platform

• AI anomaly detection

• Predictive maintenance

• Energy-demand forecasting

• Dynamic load balancing

• Real-time monitoring and alerts

IoT sensors collected charging-equipment data, while edge nodes performed part of the preprocessing and real-time evaluation. The processed data was then sent to the cloud platform for storage, analysis, and model prediction.

06 · System Workflow

The system followed the process below:

1. Collect charging-pile operating data through IoT sensors

2. Receive and preprocess the data at edge nodes

3. Check data integrity and communication status

4. Store and analyze the data in the cloud

5. Identify abnormal operating patterns using AI models

6. Predict potential equipment failures and maintenance requirements

7. Analyze real-time power demand across charging piles

8. Dynamically allocate energy according to grid conditions

9. Notify operators of anomalies and maintenance requirements

10. Continuously update equipment status and allocation strategies

07 · Technical Implementation

1. AI-Driven Cloud Monitoring

A cloud-based monitoring architecture was designed to integrate real-time analytics with AI-driven predictive maintenance.

The platform was responsible for:

• Receiving real-time charging-pile data

• Storing equipment status and historical records

• Analyzing operating trends

• Identifying abnormal conditions

• Predicting potential equipment failures

• Generating maintenance recommendations

• Displaying charging-pile status

• Sending fault notifications to operators

Anomaly-detection algorithms identified unusual patterns before they developed into serious equipment failures.

The predictive maintenance mechanism contributed to a 25% reduction in system failure rates.

2. IoT Data Acquisition

IoT sensors were integrated into the charging piles to continuously collect equipment and energy-consumption data.

The collected data represented:

• Charging-pile operating status

• Power output

• Equipment load

• Energy demand

• Equipment anomalies

• Communication status

• Historical operating trends

IoT connectivity linked distributed charging devices to a unified monitoring platform, providing the data foundation for real-time analytics and centralized management.

3. Edge Computing

Transmitting all raw data to the cloud can be affected by bandwidth limitations and network latency.

Edge computing nodes were therefore deployed closer to the charging piles to perform local data processing and rapid condition evaluation.

The edge nodes were used to:

• Clean and organize real-time sensor data

• Filter duplicate or invalid information

• Identify anomalies requiring immediate action

• Reduce the amount of data transmitted to the cloud

• Shorten response time

• Maintain selected local functions during unstable network conditions

Processing data closer to its source enabled the system to respond more quickly to demand fluctuations and equipment anomalies.

4. AI Anomaly Detection and Predictive Maintenance

Machine learning models analyzed real-time and historical charging-pile data to identify patterns associated with potential failures.

The system evaluated:

• Whether operating parameters deviated from normal ranges

• Whether an abnormal condition persisted

• Whether multiple indicators changed simultaneously

• Whether current data resembled previous fault patterns

• Whether equipment performance was declining

• Whether preventive maintenance should be scheduled

Predictive maintenance allowed operators to inspect or repair equipment before a failure occurred, reducing downtime and the impact of reactive maintenance.

5. Real-Time Load Balancing

Charging demand varies according to vehicle volume, charging duration, and individual charging-pile utilization.

An adaptive energy-allocation model was developed to distribute available power across multiple charging piles.

The system considered:

• Current power demand

• Vehicle charging status

• Real-time charging-pile load

• Available grid capacity

• Historical energy consumption

• Forecast short-term demand

The dynamic load-balancing algorithm adjusted power allocation as demand changed, reducing local grid strain while maintaining charging services.

6. Energy-Efficiency Optimization

Dynamic allocation was used to improve energy efficiency across the charging network.

The optimization focused on:

• Reducing unnecessary power allocation

• Avoiding simultaneous overload across multiple charging piles

• Adjusting supply strategies according to demand forecasts

• Redistributing power as demand changed

• Improving overall charging-network efficiency

• Maintaining smart-grid stability

This design supported more flexible and sustainable energy management.

7. Support for 5G Smart Grids

The project explored real-time interaction between charging infrastructure and smart grids using the low-latency and high-connectivity capabilities of 5G networks.

Potential functions included:

• Large-scale charging-device connectivity

• Real-time status transmission

• Rapid communication of load changes

• Remote fault diagnosis

• Dynamic energy dispatch

• Coordination between distributed edge nodes

The integration of 5G, IoT, and edge computing improved the grid's ability to respond to changes in EV charging demand.

08 · Challenges and Solutions

HK$1.6 million

09 · My Responsibilities

As a team member, I contributed to:

• Analyzing monitoring requirements for EV charging infrastructure

• Participating in the overall functional design

• Supporting the real-time charging-data workflow

• Participating in the IoT data-acquisition design

• Supporting AI anomaly-detection and predictive-maintenance research

• Supporting coordination between edge and cloud components

• Analyzing energy demand and load-balancing logic

• Participating in system testing and performance evaluation

• Preparing technical documentation and milestone results

• Supporting project demonstrations and business-feasibility analysis

10 · Key Contributions and Outcomes

• Developed an AI-powered real-time EV charging monitoring system

• Integrated IoT sensors for continuous equipment monitoring

• Applied edge computing to reduce processing latency

• Developed anomaly-detection and predictive-maintenance capabilities

• Reduced system failure rates by 25%

• Designed a dynamic load-balancing and energy-allocation model

• Improved responsiveness to changing energy demand

• Supported connectivity within a 5G smart-grid environment

• Designed a scalable system architecture

• Secured a contract valued at HK$1.6 million

• Established partnerships with energy companies for smart-grid deployment

11 · Business and Market Impact

The project evaluated both the technical and commercial feasibility of its real-time monitoring and energy-optimization solution.

A contract valued at HK$1.6 million demonstrated market demand for AI-enabled energy management and predictive maintenance.

The solution has potential B2B applications for:

• EV charging operators

• Energy-management companies

• Commercial parking facilities

• Urban charging networks

• Industrial parks

• Fleet operators

• Smart-grid service providers

Its modular architecture allows the system to be adapted according to network scale, energy demand, and deployment environment.

12 · Project Impact

The project integrated AI, IoT, edge computing, and smart-grid technologies into EV charging-infrastructure management.

Its primary value included:

• Supporting the transition from reactive to predictive maintenance

• Improving charging-pile reliability

• Reducing large-scale maintenance pressure

• Optimizing real-time power allocation

• Reducing excessive local energy loads

• Improving smart-grid responsiveness

• Supporting sustainable EV infrastructure

13 · Conclusion

This project developed an AI-powered real-time monitoring and energy-management system for EV charging piles.

IoT sensors collected equipment data, edge computing reduced processing latency, and cloud-based AI models supported anomaly detection, failure prediction, and dynamic load balancing.

Through the project, I developed a stronger understanding of EV charging infrastructure, smart grids, IoT data acquisition, edge computing, predictive maintenance, and energy allocation. I also gained experience in cross-functional collaboration, system testing, and commercial-feasibility analysis.

14 · Future Work

Future development could focus on:

• Applying reinforcement learning to adaptive energy allocation

• Expanding the system to larger charging networks

• Integrating renewable energy sources

• Improving failure-prediction accuracy

• Increasing the explainability of anomaly-detection models

• Developing more detailed equipment-health evaluation

• Strengthening smart-grid data-transmission security

• Improving the independent operation of edge nodes during network outages