Sensor Interface Device Using Machine Learning for Characteristic Data Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sensor interface devices struggle to handle various types of sensors effectively, as they rely on hard-coded calculation methods for characteristic data, which may not be suitable for all connected sensors, and sending raw sensor data leads to communication band compression.
Innovation Solution
A sensor interface device that includes a data acquisition unit, storage unit, learning unit for machine learning, and communication unit to extract and send characteristic data suited to the connected sensors, using machine learning to construct a learning model and perform Fourier transformation on measurement data for efficient data representation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If hard-coded calculation methods are used for characteristic data, then processing is simple, but adaptability to various sensor types deteriorates
Solution Approach 1:
The sensor interface device automatically identifies the sensor type through self-diagnosis and autonomously selects the appropriate characteristic value calculation method without requiring manual configuration or hard-coding for each sensor type. The control unit performs self-service by adapting its processing based on the connected sensor's characteristics.
Solution Approach 2:
The calculation method is made dynamic rather than static. The control unit can change the characteristic value calculation method based on the identified sensor type, allowing the system to adapt its processing approach dynamically rather than being fixed to a single hard-coded method.
2Loss of information
If raw sensor data is transmitted, then data completeness is maintained, but communication band is compressed
Solution Approach 1:
Instead of transmitting all raw sensor data, the control unit extracts only the essential characteristic values that represent the sensor measurements. This extraction process removes unnecessary data while retaining the critical information needed for monitoring and analysis, thereby reducing communication bandwidth requirements.
Solution Approach 2:
The system transforms raw sensor data into different parameter representations (characteristic values) that convey the same essential information in a more compact form. By changing the data parameters from raw measurements to derived characteristic values, the system achieves efficient data compression while maintaining information quality.
3Loss of information
If characteristic data is extracted using machine learning, then data relevance is improved, but processing complexity increases
Solution Approach 1:
The control unit acts as an intermediary between the sensor and the higher-order device, performing machine learning-based characteristic value extraction. This intermediary function allows the system to maintain simple sensor and higher-order device components while achieving sophisticated data processing through the intermediate processing stage.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables the extraction and transmission of characteristic data tailored to specific sensors, reducing communication band usage and improving data abstraction, while preventing communication band compression and enhancing data relevance.
Implementation Method 1
a learning unit (130) for performing machine learning with a measurement data group stored by the storage unit (120) as an input, thereby extracting characteristic data, which is data representing a characteristic of the measurement data group
Implementation Method 2
a communication unit (140) for sending the characteristic data extracted by the learning unit (130) to a higher-order device (300)
Data Source
AI summary
In a case of sending characteristic data representing a characteristic of measurement data, rather than the measurement data itself from a sensor, characteristic data suited to the connected sensor is sent. A sensor interface device (100) which is connected in a communication path between a measurement means (200) and a higher-order device (300), includes: a data acquisition means (110) for acquiring measurement data, which is data based on a physical quantity measured by the measurement means (200); a storage means (1.20) for storing the measurement data thus acquired; a learning means (130) for performing machine learning with a measurement data group stored by the storage means (120) as an input, thereby performing extraction of characteristic data, which is data representing a characteristic of the measurement data group; and a communication means (140) for sending the characteristic data extracted by the learning means (130) to the higher-order device (300).