Hierarchical neuromorphic sensor array with integrated learning for physicochemical property prediction

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing virtual sensing solutions face challenges in miniaturization, energy consumption, calibration, and scalability due to their sequential processing of data and decoupling of multi-sensor measurement, AI computation, and system memory, limiting adaptability to real-time operating conditions.

Innovation Solution

A modular, hierarchical artificial neural sensing system integrating physicochemical sensing networks with encoding-decoding networks and error feedback modules, fabricated in VLSI IC, operates in continuous time, and adapts to environmental changes through unsupervised learning, enabling miniaturization, low energy consumption, and scalable prediction of complex patterns.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If sequential processing is used in virtual sensing solutions, then data processing is simpler to implement, but adaptability to new operating conditions in real time deteriorates and productivity decreases

Engineering Contradiction:

Improveadaptability to new operating conditionsVSAvoiddata processing throughput

Core Design Contradiction:

Adaptability or versatilityVSProductivity

Solution Approach 1:

The system segments the sensing and processing functions into modular neural sensing units, each capable of independent operation. This segmentation enables parallel processing of multiple sensor data streams while maintaining adaptability through local learning capabilities in each unit, resolving the contradiction between simplified sequential processing and real-time adaptability requirements.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent implements dynamic processing through continuous-time operation of neural networks with adaptive learning rates. The system dynamically adjusts processing parameters based on incoming data characteristics, enabling real-time adaptation to new operating conditions while maintaining high throughput through efficient parallel computation architectures.

Inventive Principle:

Principle #15Dynamics

2Device complexity

If sequential processing with limited throughput is used, then system complexity is reduced, but miniaturisation and energy autonomy deteriorate

Engineering Contradiction:

Improvesystem architecture complexityVSAvoidsystem size for miniaturisation

Core Design Contradiction:

Device complexityVSVolume of moving object

Solution Approach 1:

The patent merges sensing elements, processing units, and memory into integrated neural sensing units. This consolidation reduces the overall system volume by eliminating separate components and interconnections, enabling miniaturisation while maintaining full processing capabilities through shared resources and distributed computation within each integrated unit.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The neural sensing units are designed with universal processing capabilities that can handle multiple sensor types and processing tasks. This multi-functionality reduces system complexity by using a standardized processing architecture across different sensing applications, while the modular design enables flexible configuration for specific miniaturised applications without requiring dedicated complex hardware for each function.

Inventive Principle:

Principle #6Universality (Multi-functionality)

3Use of energy by moving object

If more artificial neurons are used for sparse coding, then energy consumption is reduced and storage capacity expands, but device complexity increases

Engineering Contradiction:

Improveenergy consumptionVSAvoidneural network architecture complexity

Core Design Contradiction:

Use of energy by moving objectVSDevice complexity

Solution Approach 1:

The system implements local sparse coding where only specific subsets of neurons are activated for each processing task. This local quality approach reduces overall energy consumption by keeping most neurons in a low-power state while maintaining high computational capability when needed. The sparse activation pattern also reduces the effective complexity by limiting simultaneous interactions to only the necessary neural subsets.

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS12626090B2Hierarchical neuromorphic sensor array with integrated learning for physicochemical property prediction

Publication Date: 2026.05.12 UNIVERSITY OF ZURICH

AI summary

A modular artificial neural sensing system includes a hierarchical network of neural sensing units including a neuromimetic sensor array of artificial sensory synapses and sensory neurons for receiving physicochemical sensed signals and for outputting sensor output signals. An artificial neural network processor is adapted for processing the sensor output signals and includes processor neurons interconnected by processor synapses forming first connections and second connections. The processor outputs processor output signals. A first sensor interface feeds processed or unprocessed sensed signals into the processor. A second sensor interface receives output predicted signals from other neural sensing units and feeds processed or unprocessed output predicted signals into the processor. A signal decoder decodes the processor output signals and outputs decoder output signals. An error feedback module receives the decoder output signals and teaching signals for generating error signals depending on a difference between teaching signals and decoder output signals.