# US 2005/0038538 A1 · Computer-aided modeling and manufacture of products

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

- Number: US 2005/0038538 A1
- Document ID: US20050038538A1
- Status: Published application
- Kind: A1
- Application number: 10/884,622
- Filed: 2004-07-02
- Published: 2005-02-17
- Earliest claimed: 2003-08-13
- Assignees: Cargill, Incorporated
- Inventors: John McDonald; Sean Smith; Norman Weeks; Kevin Anderson
- Web page: https://patentgrep.com/patent/US20050038538A1

## CPC

- G05B 13/0255: {the criterion being a time-optimal performance criterion}
- G05B 17/02: electric
- G05B: CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS (systems for controlling or regulating non-electric variables G05D; systems for regulating electric or magnetic variables G05F; control devices or systems insofar as characterised by mechanical features only G05G)

## Abstract

Disclosed are modeling and process control techniques for manufacturing products. More specifically, computer-aided modeling techniques are described that allow the manufacturer to predict a profile for a multivariate output that is necessary to achieve a target performance property for a manufactured product.

## Claims

### Claim 1

A method comprising:
developing a first regression model to calibrate a multivariate output with a preselected property of a product;
developing a second regression model to calibrate the multivariate output with at least one operating parameter that affects the value of the preselected property;
developing a third regression model to predict a version of the multivariate output from the preselected property; and
predicting the operating parameter from the multivariate output to yield the preselected property.

### Claim 2 (depends on 1)

The method of claim 1, wherein developing a third regression model comprises developing a third regression model to predict a compressed, statistically constrained version of the multivariate output from the preselected property.

### Claim 3 (depends on 1)

The method of claim 1, wherein predicting the operating parameter comprises:
convolving the predicted multivariate output with a predetermined template of the multivariate output to create a de-compressed vector; and
using the de-compressed vector of the predicted multivariate output to predict the at least one operating parameter to yield the preselected property.

### Claim 4 (depends on 3)

The method of claim 3, further comprising:
selecting a desired value of the preselected property and calculating the predicted multivariate output using the third regression model;
applying the second regression model to the predicted multivariate output to calculate the operating parameter; and
using the operating parameter to produce the desired value of the preselected property.

### Claim 5 (depends on 1)

The method of claim 1, wherein the predicted multivariate output is selected from the group consisting of particle size distributions, rheology profiles, texture profiles, gas chromatography outputs, liquid chromatography outputs, thermal analysis outputs, infra red spectroscopy outputs, Raman spectroscopy outputs, and optical absorption spectroscopy outputs.

### Claim 6 (depends on 1)

The method of claim 1, wherein the at least one operating parameter is selected from the group consisting of a reaction time, a reaction temperature, an addition rate for the reactants, a concentration of the reactants, and mixing intensity of the reaction.

### Claim 7 (depends on 1)

The method of claim 1, wherein the product is selected from a group consisting of monomers, ionomers, oligomers, polymers, inorganic particles, and network polymers including polysaccharides.

### Claim 8 (depends on 1)

The method of claim 1, wherein the product comprises a paper product, and the preselected property of the paper product is selected from the group consisting of surface strength, internal bond, burst strength, tensile strength, tear strength, porosity, short span compression (SCT), dynamic contact angle, and starch retention.

### Claim 9 (depends on 4)

The method of claim 4, wherein selecting the desired value of the preselected property comprises selecting the desired value based on data selected from a group consisting of pricing for the product, sales volumes of the product, a geographic location of a facility for manufacturing the product, customer preference information for the product, and a current level of inventory of the product.

### Claim 10 (depends on 4)

The method of claim 4, wherein calculating the predicted multivariate output comprises developing a reverse model in which multivariate output data serves as a predicted variable, and the preselected property for the product serves as a prediction variable.

### Claim 11 (depends on 4)

The method of claim 4, wherein calculating the predicted multivariate output comprises developing a computation model from which the multivariate output can be predicted based on the performance properties of the manufactured product.

### Claim 12 (depends on 4)

The method of claim 4, wherein calculating the predicted multivariate output comprises:
producing the product;
analyzing the product to generate multivariate output data for the product;
conducting performance tests on the product to generate performance property data for the preselected property; and
applying chemometric techniques to develop a correlation between the multivariate output and the performance property data.

### Claim 13 (depends on 12)

The method of claim 12, wherein applying chemometric techniques comprises:
representing the multivariate output data as a polynomial; and
regressing the multivariate output polynomial against the performance property data.

### Claim 14 (depends on 13)

The method of claim 13, wherein regressing the coefficients of the polynomial comprises using a Partial Least Squares regression algorithm.

### Claim 15 (depends on 12)

The method of claim 12, wherein applying chemometric techniques comprises:
estimating a range for an order for the multivariate output polynomial;
estimating a range for an order for a regression polynomial;
iterating over the estimated ranges to generate combinations of multivariate output polynomials and regression polynomials;
calculating residuals for the combinations; and
selecting one of the combinations based on the residuals.

### Claim 16

A method comprising:
receiving a communication from a business unit via a computer network, wherein the communication specifies a value for a performance property of a product;
invoking a reverse chemometric model to calculate a multivariate output based on the specified value and a operating parameter to achieve the performance property; and
communicating the operating parameter to a remote facility for manufacturing the product according to the operating parameter.

### Claim 17 (depends on 16)

The method of claim 16, wherein the reverse chemometric model calculates a multivariate output by calculating a particle size distribution (PSD) for the product.

### Claim 18

A method comprising:
determining a set of price levels for a product, wherein the price levels correspond to values of a performance property of the product;
calculating multivariate output profiles based on the values of the performance property; and
manufacturing the product that achieves one of the calculated multivariate output profiles.

### Claim 19 (depends on 18)

The method of claim 18, further including selling the product according to the respective pricing levels.

### Claim 20

A system comprising:
a server storing a chemometric model that correlates a performance property of a product with a multivariate output; and
a software module executing on the server to present an interface to receive a selection of the performance property.

### Claim 21 (depends on 20)

The system of claim 20, wherein the software module comprises a web server and the interface comprises a user interface.

### Claim 22 (depends on 20)

The system of claim 20, wherein the software module comprises an application programming interface (API) to programmatically receive the selection.

### Claim 23 (depends on 20)

The system of claim 20, further comprising a client computer located within a manufacturing facility and coupled to the server via a network, wherein the server communicates a operating parameter to the client computer based on the selection.

### Claim 24 (depends on 20)

The system of claim 20, wherein the operating parameter is selected from the group consisting of a reaction time, a reaction temperature, an addition rate for the reactants, a concentration of the reactants, and mixing intensity for the reaction.

### Claim 25 (depends on 20)

The system of claim 20, wherein the product comprises a paper product, and the preselected property of the paper product is selected from the group consisting of surface strength, internal bond, burst strength, tensile strength, tear strength, porosity, short span compression (SCT), dynamic contact angle, and starch retention.

### Claim 26 (depends on 20)

The system of claim 20, wherein the interface provides input areas for a user to provide one selected from a group consisting of pricing for the product, sales volumes of the product, a geographic location of a facility for manufacturing the product, customer preference information for the product, and a current level of inventory of the product.

### Claim 27

A computer-readable medium comprising instructions therein to cause a computing device to:
select a value for a performance property of a product;
calculate a multivariate output based on the selected value;
select an operating parameter based on the calculated multivariate output; and
communicate the operating parameter to a facility for manufacturing the product according to the selected operating parameter.

### Claim 28 (depends on 27)

The computer-readable medium of claim 27, further comprising instruction causing the computing device to develop a reverse model in which multivariate output data serves as a predicted variable, and the performance property for the product servers as a prediction variable.

### Claim 29 (depends on 27)

The computer-readable medium of claim 27, further comprising instruction causing the computing device to develop a computation model from which the multivariate output can be predicted based on the performance properties of the product.

### Claim 30 (depends on 27)

The computer-readable medium of claim 27, further comprising instruction causing the computing device to apply chemometric techniques to develop a correlation between the multivariate output data and the performance property data.

### Claim 31 (depends on 27)

The computer-readable medium of claim 27, further comprising instruction causing the computing device to:
receive input defining a range for an order for the multivariate output polynomial;
receive input defining a range for an order for a regression polynomial;
iterate over the ranges to generate combinations of multivariate output polynomials and regression polynomials;
calculate residuals for the combinations; and
select one of the combinations based on the residuals.

### Claim 32

A method comprising:
a. developing a first regression model to calibrate a multivariate output with a preselected property of a product;
b. developing a second regression model to calibrate the multivariate output with at least one operating parameter that affects the value of the preselected property;
c. developing a third regression model to predict a compressed, statistically constrained version of the multivariate output from the preselected property and convolving the predicted multivariate output with a predetermined template of the multivariate output to create a de-compressed vector of the predicted multivariate output that is used to predict the at least one operating parameter to yield the preselected property;
d. selecting a desired value of the preselected property and calculating the predicted multivariate output using the third regression model and the template;
e. applying the second regression model of step (b) to the predicted multivariate output of step (d), to calculate the at least one operating parameter; and
f. using the at least one operating parameter from step (e) to produce the desired value of the preselected property of step (d).
