Method of controlling a production process using prediction model output directly linking interacted probe radiation information with morphological process step control parameters
A method of controlling a production process including a process step for the morphological modification of a bio-material matrix comprises obtaining digital input data acquired during each of a plurality of production runs of the process, which input data includes information from radiation within a portion of the electromagnetic or acoustic spectrum having interacted with the matrix at one or more locations within the process together with a process control parameter and production event data for the associated production run; generating in a computer a prediction model from a multivariate analysis of the digital input data, which model links the information directly with one or more of process control parameters, production run events and process control settings; and applying in the computer the prediction model to interacted information obtained from a new production run to generate as an output one or more of a process control parameter a process control event and a predicted production run event for the new production run for use in controlling the production process.
1. A method of controlling a production process including a morphological modification process step in which is employed a morphological modification device adapted to perform a morphological modification of a bio-material matrix based feedstock, the method comprising:
obtaining digital input data acquired during each of a plurality of production runs of the production process, which input data includes, as interacted probe radiation information, information from probe radiation having interacted with the feedstock at one or more locations within the production process together with one or more of a process control parameter, a process control setting and production event data for an associated production run;
generating one or more prediction models from a multivariate analysis of the digital input data, where a particular model links the interacted probe radiation information directly with a particular process control parameter, process control setting or production run event;
making the one or more prediction models available to a data processor;
obtaining new interacted probe radiation information for the feedstock during a new production run of the production process, the new interacted probe radiation information being information from probe radiation having interacted with the feedstock at one or more locations within the new production run of the production process;
applying, in the data processor, one or more of the prediction models to the new interacted probe radiation information to generate, as an output, one or more of a process control parameter, a process control setting and a predicted production run event for the new production run predicted in order to achieve a target performance of the production process; and
controlling the morphological modification process step using the output,
wherein the interacted probe radiation information is obtained from radiation within a microwave to X-ray portion of an electromagnetic spectrum.
2. A method as claimed in claim 1 wherein controlling the morphological modification process step includes controlling one or both of a rate of supply or a condition of feedstock employed in the morphological modification process step.
3. A method as claimed in claim 1 wherein controlling the morphological modification process step includes controlling an operation of the morphological modification device.
4. A method as claimed in claim 1 wherein the output includes a predicted production run event and in that method further comprises generating a dependent sensory signal.
5. A method as claimed claim 1 wherein the morphological modification device includes one or more of a pellet pressing unit, an extruder unit, an expander unit, a cutter unit, a sieving unit, and a milling unit.
6. A method as claimed in claim 1 wherein the interacted probe radiation information is obtained from a location in a part of the production process preceding the morphological modification device.
7. A method as claimed in claim 6 wherein the interacted probe radiation information is used in a feed-forward control of the morphological modification process step.
8. A method as claimed in claim 1 wherein the interacted probe radiation information is obtained from radiation within a near infra-red portion of the electromagnetic spectrum.
9. A method as claimed in claim 1 wherein the interacted probe radiation information is representative of intensity variations of the probe radiation indexed against an expression of a wavelength of the probe radiation.
10. A method as claimed in claim 1 wherein,
the multivariate analysis includes applying algorithms implementing methodologies including at least one of
derivatives,
standard normal variate,
multiplicative signal correction to remove effects from the interacted probe radiation information which are unrelated to the process control parameter, and
process control setting or production run event.
11. A method as claimed in claim 10 wherein,
the multivariate analysis includes applying algorithms implementing methodologies including
linear modelling algorithms including partial least squares,
principal components regression,
multiple linear regression, or ridge regression, or
non-linear modelling algorithms including artificial neural networks or support vector machines to generate a predictive model.
12. A method as claimed in claim 1 wherein the interacted probe radiation information is obtained from radiation within ultrasound portion of an acoustic spectrum.
13. A method as claimed in claim 1 wherein the method further comprises:
storing the digital input data in an updatable electronic database,
wherein the generating the one or more prediction models includes performing a multivariate analysis of the stored digital input data.
14. A method as claimed in claim 1 wherein the bio-material matrix is a substantially biological material matrix including a foodstuff matrix, an animal-feed matrix, a petfood matrix, an aqua-feed matrix, a biomass matrix, and/or an intermediate of, or a component for, any preceding matrix.
15. A method of controlling a production process including a morphological modification process step in which is employed a morphological modification device adapted to perform a morphological modification of a bio-material matrix based feedstock, the method comprising:
obtaining digital input data acquired during each of a plurality of production runs of the production process, which input data includes, as interacted probe radiation information, information from probe radiation having interacted with the feedstock at one or more locations within the production process together with one or more of a process control parameter, a process control setting and production event data for an associated production run;
generating one or more prediction models from a multivariate analysis of the digital input data, where a particular model links the interacted probe radiation information directly with a particular process control parameter, process control setting or production run event;
obtaining new interacted probe radiation information for the feedstock during a new production run of the production process, the new interacted probe radiation information being information from probe radiation having interacted with the feedstock at one or more locations within the new production run of the production process;
applying, in a data processor, one or more of the prediction models to the new interacted probe radiation information to generate, as an output, one or more of a process control parameter, a process control setting and a predicted production run event for the new production run, wherein the output includes the predicted production run event; and
generating a dependent sensory signal based on the output.
16. A method of controlling a production process including a morphological modification process step in which is employed a morphological modification device adapted to perform a morphological modification of a bio-material matrix based feedstock, the method comprising:
obtaining digital input data acquired during each of a plurality of production runs of the production process, which input data includes, as interacted probe radiation information, information from probe radiation having interacted with the feedstock at one or more locations within the production process together with one or more of a process control parameter, a process control setting and production event data for an associated production run, wherein the interacted probe radiation information is obtained from radiation within a microwave to X-ray portion of an electromagnetic spectrum;
generating one or more prediction models from a multivariate analysis of the digital input data, where a particular model links the interacted probe radiation information directly with a particular process control parameter, process control setting or production run event;
obtaining new interacted probe radiation information for the feedstock during a new production run of the production process, the new interacted probe radiation information being information from probe radiation having interacted with the feedstock at one or more locations within the new production run of the production process; and
applying, in a data processor, one or more of the prediction models to the new interacted probe radiation information to generate, as an output, one or more of a process control parameter, a process control setting and a predicted production run event for the new production run.
17. A method of controlling a production process including a morphological modification process step in which is employed a morphological modification device adapted to perform a morphological modification of a bio-material matrix based feedstock, the method comprising:
obtaining digital input data acquired during each of a plurality of production runs of the production process, which input data includes, as interacted probe radiation information, information from probe radiation having interacted with the feedstock at one or more locations within the production process together with one or more of a process control parameter, a process control setting and production event data for an associated production run, wherein the interacted probe radiation information is obtained from radiation within ultrasound portion of an acoustic spectrum;
generating one or more prediction models from a multivariate analysis of the digital input data, where a particular model links the interacted probe radiation information directly with a particular process control parameter, process control setting or production run event;
obtaining new interacted probe radiation information for the feedstock during a new production run of the production process, the new interacted probe radiation information being information from probe radiation having interacted with the feedstock at one or more locations within the new production run of the production process; and
applying, in a data processor, one or more of the prediction models to the new interacted probe radiation information to generate, as an output, one or more of a process control parameter, a process control setting and a predicted production run event for the new production run.