Method for monitoring and/or controlling phase separation in chemical processes and samples
The present invention relates to determining phase information of a fluid sample. At least one image of the fluid sample including the phase information of the fluid sample is provided. Additionally, a data driven model which comprises at least one output channel for the phase information is provided. The at least one output channel includes at least one output channel for classifying a boundary between two phases of the fluid sample, such that the phase information includes information about a property of the boundary between the two phases, such as a height, a volume, a type, or a strength of the boundary. The phase information of the fluid sample is derived based on the data driven model and the at least one image of the fluid sample including the phase information of the fluid sample.
1 . A phase analysis system for determining phase information of a fluid sample, wherein the phase analysis system comprises:
a data driven model comprising at least one output channel for the phase information, wherein the at least one output channel includes at least one output channel for classifying a boundary between two phases of the fluid sample, wherein the phase information includes information about a property of the boundary between the two phases,
wherein the data driven model is used for deriving the phase information of the fluid sample based on the data driven model and on at least one image of the fluid sample that comprises the phase information of the fluid sample,
wherein the data driven model further comprises a single model including output channels for phases and boundaries of the phases, and
wherein an output phase analysis results in output phase monitoring or control.
2 . The phase analysis system according to claim 1 , wherein the phase analysis system further is configured for:
generating the at least one image of the fluid sample including the phase information of the fluid sample by superimposing at least two images of the fluid sample obtained at different imaging conditions.
3 . The phase analysis system according to claim 2 , wherein the different imaging conditions include different exposure times and wherein the at least two images are obtained for different exposure times and otherwise identical imaging conditions.
4 . The phase analysis system according to claim 2 , wherein the different imaging conditions include different angles of incidence for light impinging on the fluid sample, different lighting conditions, different exposure times, or any combination thereof.
5 . The phase analysis system according to claim 1 , wherein the information about the property of the boundary between the two phases includes one or more of:
a height of the boundary between the two phases of the fluid sample,
a volume of the boundary between the two phases of the fluid sample,
a strength of the boundary between the two phases of the fluid sample, and
a type of the boundary between the two phases of the fluid sample.
6 . The phase analysis system according to at claim 1 , wherein the phase information additionally includes one or more of:
a number of different phases of the fluid sample,
a type of one or more of the phases of the fluid sample,
one or more gradients in one or more of the phases of the fluid sample,
a turbidity of the fluid sample,
a turbidity at a specific position of the fluid sample,
a position of one or more interfaces between different phases of the fluid sample,
a height of one or more of the phases of the fluid sample,
a volume of one or more of the phases of the fluid sample, and
a bubble size distribution in case that one of the phases of the fluid sample is a foam.
7 . The phase analysis system according to claim 1 , wherein the data driven model comprises a single model including output channels for phases and boundaries of the phases.
8 . The phase analysis system according to claim 1 , wherein the data driven model is a neural network trained based on labeled training data of fluid samples, wherein the labeled training data include the phase information of the fluid samples.
9 . The phase analysis system according to claim 1 , wherein the phase analysis system comprises a light source, an image sensor, and a processor, wherein the light source is configured for providing incident light to the fluid sample, wherein the image sensor is configured for obtaining one or more images of the fluid sample, wherein the phase analysis system is configured for generating the at least one image of the fluid sample including the phase information of the fluid sample based on the one or more images of the fluid sample obtained by the image sensor, and wherein the processor is configured for deriving the phase information based on the data driven model and the at least one image of the fluid sample including the phase information of the fluid sample.
10 . A method for determining phase information of a fluid sample, comprising:
deriving the phase information of the fluid sample based on a data driven model and on at least one image of the fluid sample including the phase information of the fluid sample,
wherein the data driven model comprises at least one output channel for the phase information, wherein the at least one output channel includes at least one output channel for classifying a boundary between two phases of the fluid sample, and wherein the phase information includes information about a property of the boundary between the two phases,
wherein the data driven model further comprises a single model including output channels for phases and boundaries of the phases, and
wherein an output phase analysis results in output phase monitoring or control.
11 . The method according to claim 10 , further comprising one or more of:
generating at least two images of the fluid sample obtained at different imaging conditions,
generating the at least one image of the fluid sample including the phase information of the fluid sample by superimposing the at least two images of the fluid sample,
obtaining the at least two images of the fluid sample at different exposure times for otherwise identical imaging conditions,
training the data driven model based on labeled training data of fluid samples, wherein the labeled training data include the phase information of the fluid samples,
causing a light source to provide incident light to the fluid sample,
causing an image sensor to provide one or more images of the fluid sample,
generating the at least one image of the fluid sample including the phase information of the fluid sample based on the one or more images of the fluid sample provided by the image sensor, and
generating the at least one image of the fluid sample including the phase information of the fluid sample as image with a highest contrast compared to other images of the fluid sample.
12 . The method according to claim 10 , comprising optimizing a property of the fluid sample including one or more of:
optimizing a foam volume of the fluid sample,
optimizing a foam height of the fluid sample,
optimizing a phase stability of the fluid sample, and
optimizing shelf life of the fluid sample.
13 . The method according to claim 10 , comprising monitoring a fluid sample including one or more of:
monitoring emulsion-polymerisation during a reaction within the fluid sample, and
monitoring evolution of the fluid sample over time.
14 . A computer program product for determining phase information of a fluid sample, wherein the computer program product comprises program code means for causing a processor to carry out the method according to claim 11 , when the computer program product is run on the processor.
15 . A computer-readable medium having stored the computer program product of claim 14 .