Sensor Placement Optimization in Fluid Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Optimizing the placement of sensors in fluid distribution networks to detect anomalies quickly and efficiently, as existing methods are either too complex or do not ensure optimal placement, leading to potential health and service disruptions due to undetected pollution or leaks.
Innovation Solution
A method using genetic algorithms to simulate scenarios, determine candidate sensor positions, and assign scores based on anomaly detection effectiveness, iteratively modifying these positions to achieve optimal placement with a limited number of sensors, considering network complexity and cost constraints.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the number of sensors is increased to improve anomaly detection coverage, then detection reliability is improved, but cost and device complexity increase
Solution Approach 1:
The patent changes the parameter of sensor quantity from unlimited to limited (e.g., exactly 3 sensors), transforming the optimization problem from finding the best configuration among many sensors to finding the optimal arrangement of a fixed small number of sensors. This parameter change reduces device complexity while maintaining detection reliability through strategic positioning.
Solution Approach 2:
The patent performs preliminary simulation of anomaly propagation scenarios before deploying sensors. By simulating various pollution events and their propagation paths in advance, the system identifies critical locations where sensors should be positioned to maximize detection coverage, thereby achieving reliable anomaly detection with minimal sensors.
2Measurement precision
If exhaustive testing of all sensor position combinations is performed to achieve optimal placement, then measurement precision is improved, but computational complexity becomes impractical
Solution Approach 1:
The patent changes the parameter of sensor quantity to a fixed small number (e.g., n=3), which transforms the combinatorial optimization problem from exponentially complex to computationally manageable. This parameter change enables exhaustive testing of all position combinations for the fixed number of sensors without requiring impractical computational resources.
Solution Approach 2:
Instead of performing exhaustive testing for all possible sensor quantities and positions, the patent applies partial action by limiting the analysis to a specific small number of sensors (n=3) and their optimal positions. This partial approach provides sufficiently precise optimization for practical deployment without the excessive computational burden of analyzing all possible configurations.
3Speed
If sensors are placed to maximize anomaly detection coverage, then detection speed is improved, but the number of sensors required increases
Solution Approach 1:
The patent changes the parameter of sensor quantity to a fixed minimal number (n=3), and simultaneously optimizes their positions to achieve maximal detection coverage. This parameter change enables the system to maintain fast anomaly detection speed while using the minimum possible number of sensors, resolving the contradiction between detection speed and sensor quantity.
Solution Approach 2:
The patent performs preliminary simulation of anomaly propagation to identify optimal sensor positions before deployment. By simulating various pollution scenarios and tracing their propagation paths in advance, the system determines the exact positions where 3 sensors should be placed to maximize detection speed across all possible anomaly locations, eliminating the need for additional sensors.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
The present invention aims to place sensors in a fluid distribution network by simulating network operating scenarios; by determining candidate sets of sensor positions; by determining the measurements of each sensor in each scenario, and the detection of related anomalies; by attributing a score to each candidate set of sensor positions, representing the capacity of the sensors placed in the positions of the set to perform a mission. Finally, the candidate sets are modified using algorithms referred to as genetic until a stoppage criteria is validated. The genetic algorithms may, for example, consist of crossing or mutating candidate sets.