Air particulate classification

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing air sensors face challenges in identifying complex combinations of air properties, including the presence of specific aerosols and filter quality, which hinders effective air quality monitoring and management.

Innovation Solution

A system utilizing machine learning models trained with air sensor data to classify particulate matter and filter properties, enabling the identification of aerosols and filter exhaustion levels through machine learning models integrated with air sensor systems and control systems.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional air sensor methods are used to measure air properties, then basic measurements can be obtained, but the ability to identify complex combinations of properties including specific aerosols and filter quality is insufficient

Engineering Contradiction:

Improveidentification accuracy of complex air propertiesVSAvoidsystem complexity for complex property identification

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

A machine learning model is introduced as an intermediary between the air sensor data and the identification of complex air properties. The model processes raw sensor measurements and outputs classified information about aerosol types, filter quality, and complex property combinations, enabling accurate identification without directly increasing sensor complexity

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The system transforms raw sensor measurements into classified categorical outputs through the machine learning model. By changing the parameter representation from continuous sensor values to discrete classified categories (e.g., aerosol types, filter quality levels), the system achieves better identification of complex properties while managing system complexity

Inventive Principle:

Principle #35Parameter changes

2Measurement precision

If machine learning models are integrated to classify air particulate data, then accurate identification of aerosols and filter quality is achieved, but computational processing requirements increase

Engineering Contradiction:

Improveclassification accuracy of particulate matterVSAvoidcomputational energy consumption

Core Design Contradiction:

Measurement precisionVSUse of energy by stationary object

Solution Approach 1:

The machine learning model is trained in advance with comprehensive air sensor data to learn patterns of aerosol types and filter quality. This preliminary training action enables the model to make accurate classifications during operation without requiring complex real-time computations, reducing energy consumption during actual air quality monitoring

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system uses a trained machine learning model that has copied and stored knowledge about air particulate patterns during the training phase. During operation, the model applies this copied knowledge to classify new sensor data efficiently, avoiding the need for energy-intensive real-time analysis of all possible particulate combinations

Inventive Principle:

Principle #26Copying

Data Source

PatentUS12067052B2Air particulate classification

Publication Date: 2024.08.20 ALGOLOOK INC

AI summary

A method classifies air particulate data. Air sensor data is received from an air sensor. The air sensor data comprises a particulate vector. A machine learning model generates output from the air sensor data using a machine learning model. A notification that uses the machine learning model output is transmitted by a control system.