# US 2026/0297223 A1 · Machine learning for property prediction and recipe recommendation of polyolefin-based polymer blends

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2026/0297223 A1
- Document ID: US20260297223A1
- Status: Published application
- Kind: A1
- Application number: 19/479,258
- Filed: 2024-04-26
- Published: 2026-10-01
- Earliest claimed: 2023-04-27
- Assignees: Borealis AG
- Applicants: Borealis GmbH
- Inventors: Juan Pablo Torres; Michael Jerabek; Daniela Mileva; Florian Kiehas; Martin Reiter
- Web page: https://patentgrep.com/patent/US20260297223A1

## CPC

- C08F 10/06: Propene
- C08L 23/10: Homopolymers or copolymers of propene
- C08L 2205/025: containing two or more polymers of the same hierarchy C08L, and differing only in parameters such as density, comonomer content, molecular weight, structure
- G06N 20/00: Machine learning
- C08F: MACROMOLECULAR COMPOUNDS OBTAINED BY REACTIONS ONLY INVOLVING CARBON-TO-CARBON UNSATURATED BONDS
- C08L: COMPOSITIONS OF MACROMOLECULAR COMPOUNDS (compositions based on polymerisable monomers C08F, C08G; artificial filaments or fibres D01F; textile treating compositions D06)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

A computer-implemented method for predicting attributes of a polypropylene composition. The method includes training a fingerprinting model with polypropylene training compositions by mapping contents of ingredients of the polypropylene training compositions to analytical features of the polypropylene training compositions, feeding the trained fingerprinting model, as an input, with contents of ingredients of a polypropylene test composition, and receiving, as an output, analytical features of the polypropylene test composition. The analytical features comprise at least one feature characteristic of the amorphous fraction or the crystalline fraction of the polypropylene composition. Also provided is a computer readable storage medium for storing computer program instructions defining the steps of the inventive method. Also encompassed is the use of analytical features comprising at least one feature characteristic of the amorphous fraction or the crystalline fraction of a polypropylene composition and contents of ingredients of polypropylene compositions for training a computer-implemented model.

## Claims

### Claim 1

A computer-implemented method for predicting attributes of a polypropylene composition, comprising:
Ia) training, via one or more processors, a fingerprinting model with polypropylene training compositions by mapping contents of ingredients of the polypropylene training compositions to analytical features of the polypropylene training compositions;
Ib) processing, via one or more processors, data corresponding to contents of ingredients of a polypropylene test composition using the trained fingerprinting model to generate output analytical features of the polypropylene test composition; and
Ic) receiving, via one or more processors, the output analytical features of the polypropylene test composition,
wherein the analytical features comprise at least one feature characteristic of an amorphous fraction or a crystalline fraction of the polypropylene composition.

### Claim 2 (depends on 1)

The computer-implemented method according to claim 1, wherein the polypropylene training compositions and the polypropylene test composition comprise at least one polypropylene, wherein the at least one polypropylene is preferably selected from the group consisting of propylene homopolymer, random propylene copolymer, heterophasic propylene copolymer and combinations or blends thereof.

### Claim 3 (depends on 1)

The computer-implemented method according to claim 1, wherein the polypropylene training compositions and/or the polypropylene test composition comprise(s) at least one further ingredient selected from the group consisting of:
impact modifier,
filler, and
polyethylene.

### Claim 4 (depends on 1)

The computer-implemented method according to claim 1, wherein the at least one feature characteristic of the amorphous fraction or the crystalline fraction is selected from the group consisting of:
an amorphous fraction percentage,
a crystalline fraction percentage,
a ratio of amorphous fraction percentage:crystalline fraction percentage,
an ethylene content in the amorphous fraction,
an ethylene content in the crystalline fraction,
a ratio of ethylene content in the amorphous fraction:ethylene content in the crystalline fraction,
an intrinsic viscosity in the amorphous fraction,
an intrinsic viscosity in the crystalline fraction, and
a ratio of intrinsic viscosity in the amorphous fraction:intrinsic viscosity in the crystalline fraction.

### Claim 5 (depends on 1)

The computer-implemented method according to claim 1, wherein the analytical features further comprise at least one of a total ethylene weight fraction and an intrinsic viscosity of the polypropylene composition.

### Claim 6 (depends on 1)

The computer-implemented method according to claim 1, wherein the fingerprinting model is a regression-based algorithm,
wherein the regression-based algorithm preferably comprises a linear regression and, optionally, a quantile regression.

### Claim 7 (depends on 1)

The computer-implemented method according to claim 1, further comprising:
IIa) training, via one or more processors, a property predicting model with polypropylene training compositions by mapping analytical features and at least contents of non-polymeric ingredients of polypropylene training compositions to at least one property of the polypropylene training compositions;
IIb) processing, via one or more processors, analytical features of the polypropylene test composition as received in step Ic) and at least contents of non-polymeric ingredients of the polypropylene test composition using the trained property predicting model to generate at least one output property of the polypropylene test composition,
IIc) receiving, via one or more processors, the at least one output property of the polypropylene test composition,
wherein the at least one output property is selected from the group consisting of chemical and physical properties.

### Claim 8 (depends on 7)

The computer-implemented method according to claim 7, wherein the chemical and physical properties are selected from the group consisting of mechanical properties, rheological properties, processing properties and surface properties,
wherein the chemical and physical properties preferably comprise a plurality of properties including:
Young's modulus, preferably at room temperature or at −30° C.;
Yield stress, preferably at room temperature or at −30° C.;
Notched Charpy impact strength, preferably at room temperature or at −20° C.;
Impact puncture energy, preferably at room temperature or at −30° C.;
Coefficient of linear thermal expansion, preferably in flow direction or in transverse direction;
Melt flow rate, preferably at 2.16 kg and 230° C. or at 21.6 kg and 230° C.;
Shrinkage, preferably after injection in the flow direction and/or perpendicular to the flow direction;
a value indicative of tiger stripes, preferably after production of an injection mould with a filling time of 3.0 s; and
scaled versions of the plurality of properties, more preferably scaled versions obtained after a hyperbolic-tangent transformation or a logarithmic transformation.

### Claim 9 (depends on 7)

The computer-implemented method according to claim 7, wherein the property predicting model is a Gaussian Process Regression (GPR).

### Claim 10 (depends on 1)

The computer-implemented method according to claim 1, further comprising an optimization algorithm,
preferably an optimization algorithm for predicting at least one recipe of a polypropylene composition which comes closest to at least one predetermined target property,
more preferably a genetic algorithm for predicting at least one recipe of a polypropylene composition which comes closest to at least one predetermined target property.

### Claim 11 (depends on 10)

The computer-implemented method according to claim 10, wherein the optimization algorithm comprises:
IIIa) receiving at least one target property as an input,
IIIb) comparing the at least one property of the polypropylene test composition as received in step IIc) with the at least one target property, and
IIIc) receiving, as an output, the contents of ingredients of the polypropylene test composition of as a recipe.

### Claim 12 (depends on 10)

The computer-implemented method according to claim 10, wherein the optimization algorithm is a non-dominated sorting based multi-objective evolutionary algorithm (NSGAII) receiving more than one target property as an input.

### Claim 13 (depends on 1)

The computer-implemented method of claim 1, wherein the output is used for controlling a process in a development of application-tailored polypropylene composition.

### Claim 14

A non-transitory computer readable storage medium having stored thereon computer program instructions that when executed by a processor, cause a computer to:
Ia) train, via one or more processors, a fingerprinting model with polypropylene training compositions by mapping contents of ingredients of the polypropylene training compositions to analytical features of the polypropylene training compositions;
Ib) process, via one or more processors, data corresponding to contents of ingredients of a polypropylene test composition using the trained fingerprinting model to generate output analytical features of the polypropylene test composition; and
Ic) receive, via one or more processors, the output analytical features of the polypropylene test composition,
wherein the analytical features comprise at least one feature characteristic of an amorphous fraction or a crystalline fraction of the polypropylene composition.

### Claim 15 (depends on 1)

The computer-implemented method of claim 1,
wherein training, via one or more processors, the fingerprinting model with polypropylene training compositions by mapping the contents of the ingredients of the polypropylene training compositions to the analytical features of the polypropylene training compositions includes mapping analytical features comprising at least one feature characteristic of the amorphous fraction or the crystalline fraction of a polypropylene composition and contents of ingredients of polypropylene compositions.

### Claim 16

A computing system, comprising:
one or more processors; and
one or more memories having stored thereon instructions that, when executed by the one or more processors, cause the computing system to:
Ia) train, via one or more processors, a fingerprinting model with polypropylene training compositions by mapping contents of ingredients of the polypropylene training compositions to analytical features of the polypropylene training compositions;
Ib) process, via one or more processors, data corresponding to contents of ingredients of a polypropylene test composition using the trained fingerprinting model to generate output analytical features of the polypropylene test composition; and
Ic) receive, via one or more processors, the output analytical features of the polypropylene test composition,
wherein the analytical features comprise at least one feature characteristic of an amorphous fraction or a crystalline fraction of the polypropylene composition.

### Claim 17 (depends on 16)

The computing system of claim 16, the memories having stored thereon instructions that, when executed by the one or more processors, cause the computing system to:
IIa) train, via one or more processors, a property predicting model with polypropylene training compositions by mapping analytical features and at least contents of non-polymeric ingredients of polypropylene training compositions to at least one property of the polypropylene training compositions;
IIb) process, via one or more processors, analytical features of the polypropylene test composition as received in step Ic) and at least contents of non-polymeric ingredients of the polypropylene test composition using the trained property predicting model to generate at least one output property of the polypropylene test composition,
IIc) receive, via one or more processors, the at least one output property of the polypropylene test composition,
wherein the at least one output property is selected from the group consisting of chemical and physical properties.

### Claim 18 (depends on 16)

The computing system of claim 16, the memories having stored thereon instructions that, when executed by the one or more processors, cause the computing system to:
predict at least one recipe of a polypropylene composition using one or both of (i) an optimization algorithm, and (ii) a genetic algorithm.

### Claim 19 (depends on 16)

The computing system of claim 16, the memories having stored thereon instructions that, when executed by the one or more processors, cause the computing system to:
IIIa) receive at least one target property as an input,
IIIb) compare the at least one property of the polypropylene test composition as received in step IIc) with the at least one target property, and
IIIc) receive, as an output, the contents of ingredients of the polypropylene test composition as a recipe.

### Claim 20 (depends on 16)

The computing system of claim 16, the memories having stored thereon instructions that, when executed by the one or more processors, cause the computing system to:
control, via one or more processors, a process in a development of application-tailored polypropylene composition.
