AI Arc Energy Detection for Series DC Solar Circuits
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In solar photovoltaic systems, arcs in series DC circuits are difficult to detect due to their characteristics similar to normal current patterns, leading to challenges in assessing arc energy and managing associated risks, which can result in system damage and safety hazards.
Innovation Solution
An arc risk management method using a combination of pre-processing current measurement values and an artificial intelligence network comprising a dilated convolutional neural network and a recurrent neural network, including a transformer layer and long short-term memory (LSTM), to estimate arc energy and indicate arc risks quantitatively, enabling real-time monitoring and pre-arc detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If physical protective relay systems (fuse, breaker) are used to detect arcs, then arc detection is relatively easy in parallel circuits, but these systems are ineffective for series DC circuits where arcs have no zero crossing point and resemble normal current patterns
Solution Approach 1:
The patent transforms the detection parameter from simple current magnitude (which fails for series DC arcs) to arc energy calculated through integration of voltage and current over time. This parameter transformation enables differentiation between normal current and arc current in series DC circuits where traditional zero-crossing detection is ineffective.
Solution Approach 2:
The patent replaces physical protective relay systems with an artificial intelligence-based detection system that uses neural networks to analyze current patterns. This substitution enables the system to detect arcs in series DC circuits by learning complex patterns that resemble normal current, achieving both high accuracy and versatility across different circuit types.
2Reliability
If traditional arc detection methods are used, then arcs can be detected in some cases, but it is very difficult to detect arcs in series DC circuits with inverters due to noise and current ripple
Solution Approach 1:
The patent introduces an intermediary processing layer that calculates arc energy by integrating voltage and current over time, and applies noise filtering techniques. This intermediary processing separates the arc detection signal from the noise and ripple generated by inverters, enabling reliable detection without requiring overly complex hardware modifications.
Solution Approach 2:
The patent performs preliminary signal processing including noise filtering and arc energy calculation before feeding data to the detection algorithm. By preprocessing the signals to enhance arc-related features and suppress noise, the system achieves reliable detection while keeping the overall device complexity manageable.
3Productivity
If arcs are not detected rapidly and accurately, then system operation continues, but this leads to continuous damage, shortened product life, and potential industrial accidents or loss of human life
Solution Approach 1:
The patent implements a feedback mechanism where the AI detection system continuously monitors current patterns, identifies arcs in real-time, and triggers immediate protective actions. This closed-loop feedback enables rapid detection and response, preventing continuous damage while allowing normal system operation to continue when no arcs are present.
Solution Approach 2:
The patent applies preliminary anti-action by implementing predictive analytics that can detect pre-arc conditions before actual arcs occur. The system takes preventive measures by identifying early signs of arc formation and triggering protective actions in advance, thereby preventing the harmful effects of arcs before they cause damage or safety hazards.
Data Source
AI summary
An embodiment of the present disclosure provides an arc risk management method comprising: pre-processing measurement values of currents flowing into an electric apparatus; estimating a level of arc energy in the electric apparatus by inputting the measurement values into one artificial intelligence network comprising a first layer including a dilated convolutional neural network and a second layer including a recurrent neural network; and indicating an arc risk to the electric apparatus in a quantitative way according to the level of arc energy.