ParticleTensorAI

ParticleTensorAI® (PTA) is an AI-based optical measurement platform for advanced characterization of droplets, particles, sprays, and material-related properties. It combines time-resolved light-scattering measurements with high-speed data acquisition and artificial intelligence.

Instead of reducing every detected particle or droplet immediately to only a few numerical values, ParticleTensorAI® preserves larger sections of the original optical signals. These signal sequences are transformed into multidimensional data structures — particle tensors — that can be analyzed by AI models.

This approach enables ParticleTensorAI® to investigate complex relationships within sprays and particle flows, including signal shape, intensity, temporal structure, particle-event density, detector-to-detector relationships, and material-related optical properties.

Conventional optical particle measurement methods normally identify individual particle events and calculate predefined quantities such as particle size or velocity from each event.

ParticleTensorAI® follows a different strategy. Complete time-resolved signal sequences can be retained and analyzed together. This means that information does not necessarily have to be reduced to individual particle events before the AI evaluation begins.

The method is therefore particularly interesting for complex sprays, dense particle sequences, overlapping signals, material characterization, and applications in which changes in the complete signal structure contain useful process information.

Learn more about the ParticleTensorAI® measurement principle

Droplets or particles pass through a defined optical measurement region and generate time-dependent light-scattering signals. Fast optical detectors record these signals with high temporal resolution.

ParticleTensorAI® buffers the detector signals over a defined measurement interval. The recorded data are then arranged into a multidimensional representation that preserves information about signal amplitudes, signal shapes, temporal relationships, and the distribution of many particle events within the measurement window.

An AI model processes this representation and determines the output required for the measurement task.

How ParticleTensorAI® creates and analyzes tensors

Classical particle-counting systems normally require a trigger condition to identify individual particle events. This can become difficult when many droplets are present simultaneously, signals overlap, or signal amplitudes vary strongly.

ParticleTensorAI® can analyze complete buffered signal sequences without first detecting and separating every individual event. This makes it possible to include complex and overlapping signal structures in the AI evaluation.

Trigger-free analysis is therefore one of the fundamental differences between ParticleTensorAI® and conventional counting-based particle measurement methods.

ParticleTensorAI® is available in three standard configurations. The systems use the same fundamental evaluation concept but differ in the number of optical detector channels.

System	Optical Probe	Control Unit	Detector Channels	Evaluation
PTA1	LSS1	ZEON	1	ParticleTensorAI®
PTA2	LSS2	IMEA	2	ParticleTensorAI®
PTA4	LSS4	CLEON	4	ParticleTensorAI®

ParticleTensorAI® is part of the modular ai-quanton measurement platform. PTA1 and PTA2 use the same basic optical and electronic hardware platforms as SprayQuantAI® SQA1 and SQA2. PTA4 uses the LSS4 and CLEON hardware platform also used for TSTOF measurements.

The difference is primarily in the acquisition and evaluation concept. In ParticleTensorAI® operation, time-resolved detector signals are retained over defined measurement intervals and processed as multidimensional data structures by AI models.

This modular architecture allows the optical measurement hardware and the data-analysis concept to be developed independently for different measurement tasks.

The available output depends on the optical configuration, measurement data, reference information, and the AI model developed for the application.

ParticleTensorAI® research and applications can include:

• classification of sprays, droplets, or particles
• detection of changes in spray conditions
• comparison of different operating conditions
• analysis of material-related optical properties
• droplet composition analysis
• refractive-index-related characterization
• analysis of high-concentration sprays
• monitoring of stable and unstable process conditions

Explore ParticleTensorAI® applications

Light-scattering signals are influenced not only by particle size and velocity. Optical and material properties of droplets and particles can also change the recorded signal structure.

ParticleTensorAI® investigates how these complex differences can be identified directly from time-resolved optical measurements. This creates new possibilities for detecting changes in material composition while droplets or particles are moving through a spray process.

ParticleTensorAI® uses AI models developed for defined measurement tasks. Training and validation therefore require suitable measurement and reference data.

The development process can include measurements under controlled reference conditions, creation of ParticleTensorAI® datasets, model training, independent validation, and implementation of the resulting AI model in the measurement software.

For customer-specific applications, the optical configuration and AI model can therefore be developed together as one measurement solution.

Learn more about AI model development

ParticleTensorAI® can be adapted to different laboratory, research, and industrial measurement tasks. Depending on the application, parameters such as optical geometry, working distance, detector configuration, acquisition interval, tensor structure, AI model, output parameters, and process interfaces can be adapted.

The development of a measurement solution therefore starts with the physical measurement task: the spray or particle system, material, process conditions, required measurement information, available reference data, and integration environment.

SprayQuantAI® and ParticleTensorAI® use related optical measurement technologies but follow different evaluation concepts.

	SprayQuantAI®	ParticleTensorAI®
Primary approach	Individual-event measurement	Buffered signal-sequence analysis
Particle detection	Individual events	Individual separation not always required
Data representation	Particle parameters / signal features	Multidimensional tensor
Evaluation	Classical or AI-based	AI-based
Typical focus	Size, velocity, number and spray monitoring	Complex signal patterns, material and process characterization

ParticleTensorAI® builds on research in time-resolved light-scattering measurement, TSTOF, machine-learning-based particle characterization, material characterization, and the analysis of complex sprays.

Scientific publications, conference contributions, and patent activities document the development of the underlying optical and AI-based measurement concepts.

ParticleTensorAI® research and publications

• Measurement Principle
• Tensor & AI Analysis
• PTA1
• PTA2
• PTA4
• Applications
• AI Model Development
• Research & Publications

For application-specific measurements or industrial integration, contact ai-quanton with information about your spray, particles, material, process conditions, and required measurement results.