poly-nomial calibration model maintained focused linear model correction and linear model correction

Abstract

This application relates to focused linear model correction and linear model correction of multivariate calibration model maintenance. A device can obtain master beta coefficients of a master calibration model associated with a master instrument. The master beta coefficients can be at a grid of a target instrument. The device can perform a constrained optimization of an objective function according to a set of constraints in order to determine a pair of transfer beta coefficients. The constrained optimization can be performed based on a pair of initial transfer beta coefficients, the master beta coefficients, and spectra associated with a scout set. The device can determine a transfer beta coefficient based on the pair of transfer beta coefficients. The device can determine a final transfer beta coefficient based on a set of transfer beta coefficients including the transfer beta coefficient. The final transfer beta coefficient can be associated with generating a transfer calibration model corresponding to the master calibration model for use by the target instrument.

Description

[0001] Divisional Application Instructions

[0002] This application is a divisional application of application No. 202210541483.0, filed on July 10, 2019, entitled "Focusing Linear Model Correction and Linear Model Correction for Multivariate Calibration Model Maintenance", wherein application No. 202210541483.0 is a divisional application of application No. 201910621447.3, filed on July 10, 2019, entitled "Focusing Linear Model Correction and Linear Model Correction for Multivariate Calibration Model Maintenance".

[0003] background

[0004] Spectroscopic instruments can be configured with calibration models for calibrating spectral measurements performed by the instrument. Calibration models are typically generated based on reference values ​​corresponding to known samples and the spectra measured by the instrument corresponding to those known samples.

[0005] Overview

[0006] According to some possible implementations, a method may include: obtaining, by a device, principal beta coefficients of a principal calibration model associated with a master instrument, wherein the principal beta coefficients are located at the grid of the target instrument; performing constraint optimization of an objective function by the device according to a set of constraints to determine a pair of transfer beta coefficients, wherein the constraint optimization is performed based on a pair of initial transfer beta coefficients, principal beta coefficients, and a spectrum associated with a scouting set; determining transfer beta coefficients by the device and based on the pair of transfer beta coefficients; and determining final transfer beta coefficients by the device based on a set of transfer beta coefficients including the transfer beta coefficients, wherein the final transfer beta coefficients are associated with generating a transfer calibration model corresponding to the principal calibration model for use by the target instrument.

[0007] According to some possible implementations, a method may include: determining, by a device, that the mesh of a master instrument associated with a master calibration model does not match the mesh of a target instrument, for which a transfer calibration model corresponding to the master calibration model is generated; interpolating the β coefficients of the master calibration model to the mesh of the target instrument by the device and based on the determination that the mesh of the master instrument does not match the mesh of the target instrument; and determining, by the device, the master β coefficients associated with the generation of the transfer calibration model based on the result of interpolating the β coefficients of the master calibration model to the mesh of the target instrument.

[0008] According to some possible implementations, a method may include: obtaining a reconnaissance set associated with an updated calibration model by a device, wherein the reconnaissance set includes spectra associated with a set of samples, on which the calibration model will be updated; determining β coefficients associated with the calibration model by the device; determining updated β coefficients associated with the updated calibration model by the device and based on the β coefficients using a linear model correction (LMC) technique; and updating the calibration model by the device based on the updated β coefficients.

[0009] The aspects of this disclosure may be implemented in one or more of the following embodiments:

[0010] 1) A method comprising:

[0011] The device obtains the principal β coefficients of the principal calibration model associated with the master instrument, wherein the principal β coefficients are located at the grid of the target instrument;

[0012] The device performs constraint optimization of the objective function according to a set of constraints in order to determine a pair of transfer β coefficients, wherein the constraint optimization is performed based on a pair of initial transfer β coefficients, the principal β coefficients, and the spectrum associated with the reconnaissance set;

[0013] The transmission β coefficients are determined using the device and based on the pair of transmission β coefficients; and

[0014] The device determines the final transfer β coefficient based on a set of transfer β coefficients including the transfer β coefficient, wherein the final transfer β coefficient is associated with generating a transfer calibration model corresponding to the master calibration model for use by the target instrument.

[0015] 2) According to the method described in 1), wherein the transfer calibration model is generated based on the final transfer β coefficient.

[0016] 3) According to the method of 1), wherein the set of transfer β coefficients includes at least one other transfer β coefficient, each other transfer β coefficient being determined based on a corresponding execution of constraint optimization of the objective function according to the initial transfer β coefficients of the corresponding pair.

[0017] 4) According to the method in 1), obtaining the principal β coefficients includes:

[0018] Determine that the mesh of the master instrument matches the mesh of the target instrument; and

[0019] The β coefficients of the master calibration model are identified as the master β coefficients.

[0020] 5) According to the method in 1), obtaining the principal β coefficients includes:

[0021] It is determined that the mesh of the master instrument does not match the mesh of the target instrument;

[0022] Based on the determination that the grid of the master instrument does not match the grid of the target instrument, the master calibration set is interpolated to the grid of the target instrument in order to create interpolated calibration data;

[0023] A regression model is generated based on the interpolated calibration data; and

[0024] The principal beta coefficient is determined as the beta coefficient of the regression model.

[0025] 6) According to the method in 1), wherein obtaining the principal β coefficients includes:

[0026] It is determined that the mesh of the master instrument does not match the mesh of the target instrument;

[0027] Based on the determination that the mesh of the master instrument does not match the mesh of the target instrument, the β coefficients of the master calibration model are interpolated to the mesh of the target instrument; and

[0028] The master β coefficients are determined based on the result of interpolating the β coefficients of the master calibration model to the grid of the target instrument.

[0029] 7) According to the method in 6), wherein the β coefficients of the master calibration model are interpolated to the grid of the target instrument based on the determination that the master calibration set associated with the master calibration model is unavailable.

[0030] 8) The method according to 1), wherein the set of constraints includes relevant constraints associated with each of the principal β coefficient and the pair of transitive β coefficients.

[0031] 9) The method according to 1), wherein the set of constraints includes a slope constraint associated with each of the principal β coefficient and the pair of transitive β coefficients.

[0032] 10) According to the method of 1), wherein the set of constraints includes calibration range constraints for the predicted values ​​associated with the reconnaissance set.

[0033] 11) The method according to 1) further includes generating the pair of initial transfer β coefficients based on at least one of the following:

[0034] The random generation of the pair of initial transfer β coefficients,

[0035] Apply the linear function associated with the random value to the principal beta coefficient, or

[0036] Add random values ​​to the principal β coefficients.

[0037] 12) A method comprising:

[0038] The device determines that the mesh of the master instrument associated with the master calibration model does not match the mesh of the target instrument, and a transfer calibration model corresponding to the master calibration model is generated for the target instrument.

[0039] Using the device and based on the determination that the grid of the master instrument does not match the grid of the target instrument, the β coefficients of the master calibration model are interpolated to the grid of the target instrument; and

[0040] The device determines the master β coefficients associated with generating the transfer calibration model based on the result of interpolating the β coefficients of the master calibration model to the grid of the target instrument.

[0041] 13) According to the method in 12), wherein the β coefficients of the master calibration model are interpolated to the grid of the target instrument based on the determination that the master calibration set associated with the master calibration model is unavailable.

[0042] 14) The method according to 12) further includes:

[0043] Constraint optimization of the objective function is performed based on a set of constraints in order to determine a pair of transitive β coefficients, wherein the constraint optimization is performed based on a pair of initial transitive β coefficients, the principal β coefficients, and the spectrum associated with the reconnaissance set;

[0044] The transmission β coefficients are determined based on the pair of transmission β coefficients;

[0045] The final transfer β coefficient is determined based on a set of transfer β coefficients including the aforementioned transfer β coefficient; and

[0046] The transfer calibration model is generated based on the final transfer β coefficient.

[0047] 15) The method according to 14), wherein the set of constraints includes:

[0048] The associated constraints are related to the principal β coefficient and each of the pair of transitive β coefficients.

[0049] The slope constraint associated with the principal β coefficient and each of the pair of transferred β coefficients, and

[0050] Calibration range constraints for the predicted values ​​associated with the reconnaissance set.

[0051] 16) The method according to 14) further includes generating the pair of initial transfer β coefficients based on at least one of the following:

[0052] The random generation of the pair of initial transfer β coefficients,

[0053] Apply the linear function associated with the random value to the principal beta coefficient, or

[0054] Add random values ​​to the principal β coefficients.

[0055] 17) The method according to 12) further includes:

[0056] Based on the principal β coefficients and using linear model correction (LMC) techniques, determine the transfer β coefficients associated with generating the transfer calibration model; and

[0057] The transfer calibration model is generated based on the transfer β coefficient.

[0058] 18) The method according to 17), wherein a reference value for the reconnaissance set associated with the use of the LMC technique is predicted based on the master calibration model and the master transfer set.

[0059] 19) A method comprising:

[0060] The device obtains a reconnaissance set associated with updating the calibration model, wherein the reconnaissance set includes spectra associated with a set of samples, and the calibration model will be updated based on the set of samples;

[0061] The device determines the β coefficient associated with the calibration model;

[0062] The updated β coefficients associated with updating the calibration model are determined using the device and based on the β coefficients using linear model correction (LMC) techniques; and

[0063] The device updates the calibration model based on the updated β coefficient.

[0064] 20) The method according to 19), wherein the update of the calibration model is performed during operation of the device. Attached Figure Description

[0066] Figures 1A-1C This is an overview illustration of the example implementation described herein.

[0067] Figure 2 This is a diagram of an example environment in which the systems and / or methods described herein can be implemented.

[0068] Figure 3 yes Figure 2 An illustration of example components of one or more devices.

[0069] Figure 4This is a flowchart of an example process for a focused linear model calibration technique as described in this article, which is associated with determining the transfer β coefficients used to generate a transfer calibration model.

[0070] Figures 5A-5C Is with Figure 4 An example illustration associated with the focused linear model correction technique.

[0071] Figure 6 This is a flowchart of an example process for interpolating the β coefficients of the master calibration model to the grid of the target instrument, as described herein, in order to determine the master β coefficients for use in a focusing linear model correction technique or a linear model correction technique.

[0072] Figures 7A-7C as well as Figure 8A and Figure 8B This is a diagram relating the interpolation of the β coefficients of the master calibration model to the grid of the target instrument, and using the LMC and fLMC techniques associated with the execution of the calibration model transfer.

[0073] Figures 9A-9D , Figure 10A , Figure 10B , Figure 11A and Figure 11B This is a diagram illustrating example results associated with standardizing calibration models across multiple instruments.

[0074] Figure 12 This is a flowchart of an example process for updating a model using linear model correction techniques as described in this article.

[0075] Figure 13A and Figure 13B This is a diagram illustrating example results of performing a calibration model update using the linear model correction technique.

[0076] Figure 14A and Figure 14B This is a diagram illustrating example results associated with predicting reference values ​​using the master calibration model and master transfer set.

[0077] Detailed description

[0078] The following detailed description of exemplary embodiments is with reference to the accompanying drawings. The same reference numerals in different figures may identify the same or similar elements.

[0079] Calibration model transfer and calibration model update are two important areas concerning the maintenance of multivariate calibration models for spectral applications, such as those in the near-infrared (NIR) region.

[0080] In some cases, when using a multivariate calibration model developed on a first spectrometer (or under one environmental condition) to predict calibration properties for samples measured on a second spectrometer (or by the first spectrometer under different environmental conditions), the results are unacceptable. Furthermore, even for the same spectrometer, the signal may drift over time, meaning that the existing calibration model will need to be updated. To avoid the cumbersome and costly task of recollecting data and recalibrating the existing calibration model when updating the calibration model, calibration transfer techniques can be implemented to transfer the calibration model from one condition to another, regardless of the source of the drift.

[0081] A common requirement of calibration model transfer techniques is obtaining a transfer dataset, which includes spectra from the same set of samples collected by a first instrument (e.g., the master instrument from which the calibration model will be transferred, or a given instrument under original conditions) and a second instrument (e.g., the target instrument to which the calibration model will be transferred, or a given instrument under target conditions). In some cases, obtaining a transfer dataset is difficult or impossible. For example, obtaining a transfer dataset may be impossible when a calibration model for perishable materials needs to be transferred from a master instrument located in one country to a target instrument located in another country.

[0082] Linear model correction (LMC) techniques address this problem by requiring only a small set of spectra collected solely by the target instrument. The set of spectra used by LMC techniques is called the reconnaissance set. However, for LMC to function, reference values ​​for the reconnaissance set are needed (e.g., actual values ​​measured in a chemistry lab). In some cases, obtaining these reference values ​​can be very time-consuming and / or expensive.

[0083] Some implementations described herein provide a focused LMC (fLMC) technique that can be used in conjunction with calibration model transfer. Similar to LMC, fLMC requires only a reconnaissance set collected by the target instrument. However, unlike LMC, fLMC does not require reference values ​​for the reconnaissance set. Therefore, using fLMC in conjunction with calibration model transfer reduces the cost, difficulty, and / or complexity of calibration model transfer (e.g., compared to LMC and the general calibration model transfer techniques described above).

[0084] Furthermore, calibration model transfer is frequently encountered from a master instrument with relatively high spectral resolution and / or a relatively wide wavelength range to a target instrument with relatively low spectral resolution and / or a relatively narrow wavelength range (e.g., when transferring a calibration model from a benchtop instrument to a portable instrument). For calibration model transfer in this situation, typical calibration transfer techniques require a complete master calibration set (e.g., a set of spectra associated with a set of samples measured by the master instrument) to initiate the calibration model transfer process. Here, the spectra of the master calibration set are interpolated into the grid of the target instrument, and then an intermediate model is developed for transfer to the target instrument.

[0085] However, accessing the master calibration set is not always possible. Even when the master calibration set is accessible, the master database may be large and / or have a long maintenance history in some cases. Therefore, obtaining a clean master calibration set from the database can be difficult and / or time-consuming.

[0086] Some implementations described herein provide a process in which the fLMC technique or LMC technique is used in conjunction with performing calibration model transfer, employing the beta coefficients of the master calibration model without requiring a master calibration set. The use of beta coefficients (instead of a master calibration set) reduces the cost, difficulty, and / or complexity of calibration model transfer.

[0087] Furthermore, when a calibration model developed on a master instrument is to be deployed on multiple other instruments (e.g., multiple different target instruments) that may have inter-instrumental variations, performing calibration model transfer using conventional calibration model transfer techniques can be difficult (e.g., when the target instruments are located far from the master instrument). In some implementations, to address this issue, LMC or fLMC techniques can be configured on multiple target instruments. When the master calibration model is transferred to the target instruments, the user only needs to collect a reconnaissance set (e.g., spectra from several samples associated with a given application). The calibration model can (e.g., when reference values ​​are available) combine LMC techniques or combine fLMC techniques (e.g., regardless of whether reference values ​​are available) to automatically correct using these spectra.

[0088] Furthermore, as mentioned above, after a calibration model is deployed on a given instrument, it may need to be updated (e.g., due to changes in samples, measurement environment, etc.). A common technique for performing calibration model updates is to add new samples to the existing calibration set and then rebuild the calibration model. However, this technique may require a large number of samples to adapt the calibration model to the new samples or conditions. Moreover, this technique requires all calibration data to be available. Furthermore, when the calibration database is large, especially when the spectral range is wide and the spectral resolution is high, rebuilding the calibration model can consume significant time and / or resources (e.g., processor resources, battery power, etc.). Therefore, updating the calibration model during online operation of the instrument may be impossible.

[0089] Some implementations described herein provide techniques for using LMC (Local Motion Control) to update calibration models. Essentially, LMC requires a relatively small number of samples to perform a calibration model update. In some implementations, the update set (i.e., the reconnaissance set associated with performing the calibration model update) may include samples representing different conditions for future samples to make future predictions more accurate. Furthermore, calibration model updates using LMC are relatively faster than the general update techniques described above. For example, a calibration model update using an LMC procedure can be performed within seconds, enabling calibration model updates during online operations.

[0090] Furthermore, in some cases, transfer sets from both the master and target instruments may be available, but the reference value for the transfer set may not be available. In such cases, as described herein, the transfer set from the target instrument can be used as the reconnaissance set, and the reference value predicted by the master calibration model from the transfer set from the master instrument can be used as the reference value to perform the LMC technique. Therefore, the LMC technique can be performed when spectral data is available; otherwise, the spectral data can be used to perform other conventional calibration transfer techniques.

[0091] Figures 1A-1C This is an illustration of an example implementation described herein. Figure 1A and Figure 1B This is an illustration of an example implementation 100 associated with a focus linear model correction (fLMC) correction technique, which is associated with generating a transfer calibration model corresponding to a master calibration model associated with a master instrument for configuration on a target instrument.

[0092] against Figure 1A and Figure 1B The objective of Example Implementation 100 is to transfer the master calibration model configured on the master instrument to the target instrument. In other words, a transfer calibration model corresponding to the master calibration model configured on the master instrument is generated for use by the target instrument. Example Implementation 100 describes the use of fLMC technology in association with the generation of the transfer calibration model.

[0093] As in Figure 1A As shown in reference numeral 105, the modeling device (e.g., the device associated with generating the transfer calibration model) can obtain the principal β coefficients of the principal calibration model at the grid of the target instrument.

[0094] The principal beta coefficients may include a set of coefficients associated with the master calibration model. For example, the principal beta coefficients may include a vector of regression coefficients associated with a partial least squares (PLS) regression calibration model configured on the master instrument.

[0095] As described above, the principal β coefficient resides on the grid of the target instrument. The grid of the target instrument is a parameter of the target instrument defined by its spectral resolution and wavelength range. Similarly, the grid of the master instrument is a parameter of the master instrument defined by its spectral resolution and wavelength range. In some implementations, the grid of the master instrument may differ from that of the target instrument (e.g., when the master instrument has a relatively higher spectral resolution and / or a wider wavelength range compared to the target instrument). Alternatively, the grid of the master instrument may match that of the target instrument (e.g., when the spectral resolution and wavelength range of the master instrument match those of the target instrument within a threshold amount).

[0096] In some implementations, the modeling device obtains the master β coefficients based on whether the mesh of the master instrument matches the mesh of the target instrument.

[0097] For example, the modeling device can determine whether the mesh of the master instrument matches the mesh of the target instrument (e.g., based on information provided by the master instrument and / or the target instrument, or based on information stored or accessible by the modeling device). In some implementations, if the modeling device determines that the mesh of the master instrument matches the mesh of the target instrument, the modeling device can identify the β coefficients of the master calibration model as the master β coefficients. In other words, when the mesh of the master instrument matches the mesh of the target instrument, the modeling device can directly use the β coefficients of the master instrument as the master β coefficients (e.g., because the β coefficients of the master calibration model are already on the mesh of the target instrument). In this case, the β coefficients of the master calibration model can be used as the master β coefficients, regardless of whether the master calibration set associated with the master calibration model is available.

[0098] In some implementations, if the modeling apparatus determines that the mesh of the master instrument does not match the mesh of the target instrument, the master instrument can obtain principal beta coefficients based on the principal calibration set associated with the master calibration model. For example, if the mesh of the master instrument does not match the mesh of the target instrument, the modeling apparatus can interpolate the principal calibration set to the mesh of the target instrument to create interpolated calibration data (i.e., the spectrum of the principal calibration set interpolated to the mesh of the target instrument). Here, the modeling apparatus can generate a regression model (e.g., a PLS model, a principal component regression (PCR) model, etc.) based on the interpolated calibration data, and the principal beta coefficients can be determined as the beta coefficients of the regression model. In some implementations, the modeling apparatus can obtain the principal beta coefficients in this manner when the principal calibration set is available. For example, the modeling apparatus can determine that the principal calibration set is available (e.g., accessible, not exceeding a threshold size or complexity level), and can proceed as described above.

[0099] In some implementations, if the modeling equipment determines that the mesh of the master instrument does not match the mesh of the target instrument, the master instrument can obtain the master β coefficients based on the β coefficients of the master calibration model, an example of which is... Figure 1CAs shown in the image.

[0100] Figure 1C This is an illustration of an example implementation 150 associated with interpolating the β coefficients of the master calibration model to the grid of the target instrument to obtain the master β coefficients. As shown by reference numeral 155, the modeling device can determine that the grid of the master instrument does not match the grid of the target instrument. As shown by reference numeral 160, based on this determination, the modeling device can interpolate the β coefficients of the master calibration model to the grid of the target instrument. As shown by reference numeral 165, the result of interpolating the β coefficients of the master calibration model to the grid of the target instrument can be used as the master β coefficients. In some implementations, the modeling device can obtain the master β coefficients in this manner when the master calibration set is unavailable. For example, the modeling device can determine that the master calibration set is unavailable (e.g., inaccessible, exceeds a threshold size, or complexity level) and can continue as described above.

[0101] In some implementations, the modeling apparatus may determine the primary β coefficients by using an fLMC technique (as described in conjunction with Example Implementation 100) for calibration model transfer, in association with interpolating the β coefficients of the primary calibration model to the grid of the target instrument. Alternatively, the modeling apparatus may determine the primary β coefficients by using an LMC technique in association with interpolating the β coefficients of the primary calibration model to the grid of the target instrument. In other words, the interpolation of the β coefficients of the primary calibration model to the grid of the target instrument may be used in association with performing either an fLMC technique or an LMC technique for calibration model transfer.

[0102] Returning to the fLMC technique associated with example implementation 100, in some implementations, the modeling device may determine the final transfer β coefficients based on a set of transfer β coefficients. The final transfer β coefficients are the β coefficients used to generate the transfer calibration model. In some implementations, as described below, the modeling device may determine each of the set of transfer β coefficients based on the corresponding iteration of the constraint optimization of the objective function.

[0103] In some implementations, for each iteration, the modeling device can perform constrained optimization of the following objective function:

[0104]

[0105] It has the following constraints:

[0106] corr(b transA ,b 主 )≥r (1)

[0107] corr(b transB ,b 主 )≥r (2)

[0108] slope(b transA,b 主 )≥r (3)

[0109] slope(b transB ,b 主 )≥r (4)

[0110] minY cal <<X 侦察 b transA <<maxY cal (5)

[0111] minY cal <<X 侦察 b transB <<maxY cal (6)

[0112] Where X 侦察 It is a reconnaissance set (e.g., the spectrum of a reconnaissance set measured by the target instrument), b transA and b transB These are a pair of transitive beta coefficients, b, associated with a given iteration. 主 Here, is the principal beta coefficient, r is the constraint threshold, and minY cal and maxY cal Define the calibration range associated with the target instrument.

[0113] In some implementations, the constraint threshold r (e.g., the relevant constraints and / or slope constraints as described in the equations above) can be optimized using a validation set. In this case, a set of constraint thresholds r can be used iteratively, and the optimal r (e.g., determined based on the root mean square error of prediction (RMSEP) of the validation set) can be used in association with the determined transfer β coefficient. In some implementations, this constraint threshold optimization can be used in association with fLMC or LMC techniques.

[0114] To establish the objective function, the concept of reproducibility is introduced. Assume a pair of transitive β coefficients b. transA and b transB Each of them can be suitable for the reconnaissance set, using b transA and b transB The differences in the predicted values ​​of the reconnaissance set should be small. Therefore, the objective function is to minimize the difference using b. transA and b transB The squared difference of the predicted values ​​of the reconnaissance set. By using this reproducibility concept, the need for a reference value for the reconnaissance set is eliminated. In other words, due to this reproducibility concept, the fLMC technique does not require a reference value for the reconnaissance set (unlike the LMC technique).

[0115] To obtain meaningful results, the minimization of the objective function needs to be performed under a set of constraints. For example, this set of constraints could include the principal beta coefficient (b... 主 ) and each of the pair of transmission β coefficients (b transA and b transB The associated constraints are related to the given iteration of constrained optimization of the transfer β coefficients and the objective function. According to these constraints, b transA and b 主 The correlation between them and b transB and b 主 The correlation between them should satisfy a threshold (e.g., indicated by equations (1) and (2) respectively, where, for example, the value r can be greater than or equal to 0.95).

[0116] As another example, this set of constraints could include a slope constraint associated with each of the principal beta coefficient and a pair of transitive beta coefficients, which are associated with a given iteration of constrained optimization of the objective function. According to this slope constraint, b transA and b 主 The slope between and b transB and b 主 The slope between them should satisfy a threshold (e.g., indicated by equations (3) and (4) respectively, where, for example, the value r can be greater than or equal to 0.95).

[0117] As another example, this set of constraints could include a calibration range constraint on the predicted values ​​associated with the reconnaissance set. Based on this calibration constraint, using b... transA The predicted value of the reconnaissance set (i.e., X) 侦察 b transA ) and by b transB The predicted value of the reconnaissance set (i.e., X) 侦察 b transB The values ​​should be within the calibration range (e.g., indicated by equations (5) and (6) respectively), or within the range close to the reference values ​​of the reconnaissance set.

[0118] To begin a given iteration of the constrained optimization process described above, b is required. transA and b transB The initial values ​​(i.e., b) are respectively transA0 and b transB0 In some implementations, the modeling device may generate a pair of initial transferred β coefficients based on the random generation of a pair of initial transferred β coefficients. Alternatively, the modeling device may generate the principal β coefficients based on applying a linear function associated with random values ​​to the principal β coefficients (e.g., b...). transA0 ,b transB0 =m×b 主+n, where m and n are random numbers) to generate a pair of initial transfer β coefficients. Alternatively, the modeling device can generate a pair of initial transfer β coefficients based on adding random values ​​to the master β coefficients (e.g., b +n, where m and n are random numbers). transA0 ,b transB0 =b 主 +n, where n is a random number, to generate a pair of initial transfer β coefficients.

[0119] For a given iteration of constrained optimization, the modeling apparatus can generate a pair of initial transfer β coefficients (e.g., b for iteration i). transAi0 and b transBi0 And b for iteration k transAk0 and b transBk0 Furthermore, it can perform constrained optimization of the objective function to determine a pair of transitive β coefficients (e.g., b for iteration i). transAi and b transBi And b for iteration k transAk and b transBk Then, the modeling device can determine the transfer β coefficients (e.g., b for iteration i) based on this pair of transfer β coefficients. transi and b for iteration k transk For example, as shown in reference numeral 110 regarding iteration i, the modeling device can generate b. transAi0 and b transBi0 Perform constrained optimization of the objective function to determine b. transAi and b transBi And based on this pair of propagation β coefficients (e.g., based on the average b... transAi and b transBi ) to determine the transfer β coefficient (b) associated with iteration i. transi As another example, as shown in reference number 115 regarding iteration k, the modeling device can generate b. transAk0 and b transBk0 Perform constrained optimization of the objective function to determine b. transAk and b transBk And based on this pair of propagation β coefficients (e.g., based on the average b... transAk and b transBk ) to determine the transfer β coefficient (b) associated with iteration k. transk Here, b transi and b transk Included in a set of transfer β coefficients, the modeling device can determine the final transfer β coefficients (b) based on this set of transfer β coefficients. trans ).

[0120] In some implementations, the modeling device can be configured to perform multiple (e.g., 5, 20, 100, etc.) iterations of constrained optimization of the objective function (e.g., to avoid biased results based on the randomized nature of a pair of initial transfer β coefficients).

[0121] like Figure 1B As shown by reference numeral 120, the modeling device can determine the final transfer β coefficient (b) based on this set of transfer β coefficients. trans For example, the modeling device can determine the final transfer β coefficient as equal to that set of transfer β coefficients (e.g., b). transi to b transk The average, median, and mode of a number.

[0122] As shown in reference numeral 125, the modeling device can generate a transfer calibration model based on the final transfer β coefficient. For example, the modeling device can generate a regression model (e.g., a PLS model, a PCR model, etc.) based on the final transfer β coefficient. As shown in reference numeral 130, the modeling device can provide a transfer calibration model to the target instrument (e.g., enabling the target instrument to use the transfer calibration model). In this way, the modeling device can be configured to use the fLMC technique, which allows the modeling device to generate a transfer calibration model using the spectra associated with the reconnaissance set without requiring reference values ​​for the reconnaissance set.

[0123] As mentioned above, Figures 1A-1C This is provided as an example only. Other examples are possible and may differ from those provided. Figures 1A-1C Example of the description.

[0124] Figure 2 This is a diagram of an example environment 200 in which the systems and / or methods described herein may be implemented. Figure 2 As shown, environment 200 may include a main instrument 205, a target instrument 210, a modeling device 215, and a network 220. The devices in environment 200 may be interconnected via wired connections, wireless connections, or a combination of wired and wireless connections.

[0125] The master instrument 205 includes a device configured with a master calibration model capable of performing spectral measurements on a sample. For example, the master instrument 205 may include a benchtop (i.e., non-handheld) spectrometer capable of performing spectral methods (e.g., vibrational spectroscopy, such as near-infrared (NIR) spectroscopy, mid-infrared spectroscopy, Raman spectroscopy, etc.). In some embodiments, the master instrument 205 may be able to obtain spectral measurements at a higher resolution than those obtained by the target instrument 210 (i.e., the master instrument 205 may be a high-resolution device, while the target instrument 210 may be a low-resolution device). For example, the master instrument 205 may be able to obtain spectral measurements on 400 channels, while the target instrument 210 may be able to obtain spectral measurements on 125 channels. In some embodiments, the master instrument 205 may be configured with a master calibration model for calibrating the spectral measurements obtained by the master instrument 205. In some embodiments, the master instrument 205 may receive and / or transmit information to another device in the environment 200 (such as modeling device 215).

[0126] Target instrument 210 includes a device capable of performing spectral measurements on a sample based on a target calibration model, wherein the target calibration model can be generated based on information associated with a master calibration model as described herein, which is associated with master instrument 205. For example, target instrument 210 may include a mobile spectrometer device or a handheld spectrometer device performing spectroscopy. In some embodiments, target instrument 210 may be able to obtain spectral measurements at a lower resolution than those obtained by master instrument 205. In some embodiments, target instrument 210 may receive and / or transmit information to another device in environment 200, such as modeling device 215.

[0127] Modeling device 215 includes a device capable of performing operations associated with: transferring a master calibration model from master instrument 205 to target instrument 210 (i.e., generating a transfer calibration model corresponding to the master calibration model) and / or updating a calibration model configured on a given instrument (e.g., master instrument 205 or target instrument 210) as described herein. For example, modeling device 215 may include a server, a set of servers, a computer, a cloud computing device, etc. In some embodiments, modeling device 215 may receive and / or transmit information to another device in environment 200 (e.g., master instrument 205 and / or target instrument 210). In some embodiments, modeling device 215 and master instrument 205 may be implemented within a single device. Alternatively, in some embodiments, modeling device 215 and target instrument 210 may be implemented within a single device.

[0128] Network 220 includes one or more wired and / or wireless networks. For example, network 220 may include cellular networks (e.g., New Radio / 5G networks, Long Term Evolution (LTE) networks, 3G networks, Code Division Multiple Access (CDMA) networks, etc.), Public Land Mobile Networks (PLMN), Local Area Networks (LAN), Wide Area Networks (WAN), Metropolitan Area Networks (MAN), Telephone Networks (e.g., Public Switched Telephone Network (PSTN)), Private Networks, Ad Hoc Networks, Intranets, the Internet, Fiber-based Networks, Cloud Computing Networks, etc., and / or combinations of these or other types of networks.

[0129] Figure 2 The number and layout of the devices and networks shown are provided as examples. In reality, with... Figure 2 Compared to the devices and networks shown, there may be additional devices and / or networks, fewer devices and / or networks, different devices and / or networks, or devices and / or networks with different arrangements.

[0130] also, Figure 2 The two or more devices shown can be implemented within a single device, or Figure 2 The single device shown can be implemented as multiple distributed devices. For example, although the master instrument 205 and the modeling device 215 are described as two independent devices, the master instrument 205 and the modeling device 215 can be implemented within a single device. As another example, the target instrument 210 and the modeling device 215 can be implemented within a single device. Alternatively or additionally, a set of devices in environment 200 (e.g., one or more devices) can perform one or more functions described as being performed by another set of devices in environment 200.

[0131] Figure 3 This is an illustration of example components of device 300. Device 300 may correspond to main instrument 205, target instrument 210, and / or modeling device 215. In some embodiments, main instrument 205, target instrument 210, and / or modeling device 215 may include one or more devices 300 and / or one or more components of device 300. Figure 3 As shown, device 300 may include bus 310, processor 320, memory 330, storage unit 340, input unit 350, output unit 360 and communication interface 370.

[0132] Bus 310 includes components that allow communication among the parts of device 300. Processor 320 is implemented in hardware, firmware, or a combination of hardware and software. Processor 320 may take the form of a central processing unit (CPU), graphics processing unit (GPU), accelerated processing unit (APU), microprocessor, microcontroller, field-programmable gate array (FPGA), application-specific integrated circuit (ASIC), or another type of processing unit. In some embodiments, processor 320 includes one or more processors that can be programmed to perform functions. Memory 330 includes random access memory (RAM), read-only memory (ROM), and / or another type of dynamic or static storage device (e.g., flash memory, magnetic storage, and / or optical storage) for use by processor 320.

[0133] Storage component 340 stores information and / or software related to the operation and use of device 300. For example, storage component 340 may include hard disk (e.g., magnetic disk, optical disk, magneto-optical disk, and / or solid-state disk), compressed optical disk (CD), digital universal disk (DVD), floppy disk, cartridge, magnetic tape, and / or another type of non-transitory computer-readable medium along with a corresponding drive.

[0134] Input component 350 includes components that allow device 300 to receive information, for example, via user input (e.g., a touchscreen display, keyboard, keypad, mouse, button, switch, and / or microphone). Alternatively, input component 350 may include sensors for sensing information (e.g., a Global Positioning System (GPS) component, accelerometer, gyroscope, and / or actuator). Output component 360 includes components that provide output information from device 300 (e.g., a display, speaker, and / or one or more light-emitting diodes (LEDs)).

[0135] Communication interface 370 includes transceiver-like components (e.g., a transceiver and / or separate receiver and transmitter) that enable device 300 to communicate with other devices, for example, via a wired connection, a wireless connection, or a combination of wired and wireless connections. Communication interface 370 may allow device 300 to receive information from another device and / or provide information to another device. For example, communication interface 370 may include an Ethernet interface, an optical interface, a coaxial interface, an infrared interface, a radio frequency (RF) interface, a universal serial bus (USB) interface, a Wi-Fi interface, a cellular network interface, etc.

[0136] Device 300 can perform one or more of the processes described herein. Device 300 can perform these processes based on software instructions stored in non-transitory computer-readable media (e.g., memory 330 and / or storage unit 340) executed by processor 320. Computer-readable media are defined herein as non-transitory memory devices. Memory devices include storage space within a single physical storage device or storage space distributed across multiple physical storage devices.

[0137] Software instructions may be read from another computer-readable medium or from another device into memory 330 and / or storage unit 340 via communication interface 370. When executed, the software instructions stored in memory 330 and / or storage unit 340 may cause processor 320 to perform one or more processes described herein. Alternatively or additionally, hard-wired circuitry may be used in place of or in combination with software instructions to perform one or more processes described herein. Therefore, the embodiments described herein are not limited to any particular combination of hardware circuitry and software.

[0138] Figure 3 The number and arrangement of the components shown are provided as an example. In fact, with... Figure 3 Compared to the components shown, device 300 may include additional components, fewer components, different components, or components arranged differently. Alternatively, a set of components of device 300 (e.g., one or more components) may perform one or more functions described as being performed by another set of components of device 300.

[0139] Figure 4 This is a flowchart of an example process 400 for focusing linear model correction (fLMC) techniques associated with determining the transfer β coefficients used to generate the transfer calibration model, as described herein. In some implementations, Figure 4 One or more process blocks may be executed by the modeling device 215. In some implementations, Figure 4 One or more process blocks may be executed by another device or a group of devices (such as master instrument 205 and / or target instrument 210) that is separate from or includes modeling device 215.

[0140] like Figure 4 As shown, process 400 may include obtaining the principal beta coefficients of the principal calibration model associated with the master instrument, wherein the principal beta coefficients are located at the grid of the target instrument (block 410). For example, modeling device 215 may obtain the principal beta coefficients of the principal calibration model associated with master instrument 205, wherein, as described above, the principal beta coefficients are located at the grid of target instrument 210.

[0141] like Figure 4As further illustrated, process 400 may include performing constraint optimization of the objective function based on a set of constraints to determine a pair of transfer β coefficients, wherein the constraint optimization is performed based on a pair of initial transfer β coefficients, principal β coefficients, and the spectrum associated with the reconnaissance set (block 420). For example, modeling device 215 may perform constraint optimization of the objective function based on a set of constraints to determine a pair of transfer β coefficients, wherein, as described above, the constraint optimization is performed based on a pair of initial transfer β coefficients, principal β coefficients, and the spectrum associated with the reconnaissance set.

[0142] like Figure 4 As further illustrated, process 400 may include determining the transfer β coefficients based on the pair of transfer β coefficients (block 430). For example, as described above, modeling device 215 may determine the transfer β coefficients based on the pair of transfer β coefficients.

[0143] like Figure 4 As further illustrated, process 400 may include determining a final transfer β coefficient based on a set of transfer β coefficients, including transfer β coefficients, wherein the final transfer β coefficient is associated with generating a transfer calibration model corresponding to the master calibration model for use by the target instrument (block 440). For example, modeling device 215 may determine a final transfer β coefficient based on a set of transfer β coefficients, including transfer β coefficients, wherein the final transfer β coefficient is associated with generating a transfer calibration model corresponding to the master calibration model for use by the target instrument 210.

[0144] Process 400 may include additional implementations, such as any single implementation or any combination of implementations, such as those described below and / or any other processes described elsewhere herein.

[0145] In some implementations, the modeling device 215 and / or the target instrument 210 may generate a transfer calibration model based on the final transfer β coefficient.

[0146] In some implementations, a set of transitive β coefficients includes at least one other transitive β coefficient, each of which is determined based on a corresponding execution of constraint optimization of the objective function according to the initial transitive β coefficients of the corresponding pair.

[0147] In some implementations, when obtaining the master β coefficients, the modeling device 215 can determine that the mesh of the master instrument 205 matches the mesh of the target instrument 210 and identify the β coefficients of the master calibration model as master β coefficients.

[0148] In some implementations, when obtaining the principal beta coefficients, the modeling device 215 may determine that the mesh of the master instrument 205 does not match the mesh of the target instrument 210; based on the determination that the mesh of the master instrument 205 does not match the mesh of the target instrument 210, the master calibration set is interpolated to the mesh of the target instrument 210 to create interpolated calibration data; a regression model is generated based on the interpolated calibration data; and the principal beta coefficients are determined as the beta coefficients of the regression model.

[0149] In some implementations, when obtaining the principal β coefficients, modeling device 215 may determine that the mesh of master instrument 205 does not match the mesh of target instrument 210; based on the determination that the mesh of master instrument 205 does not match the mesh of target instrument 210, the β coefficients of the master calibration model are interpolated to the mesh of target instrument 210; and the principal β coefficients are determined based on the result of interpolating the β coefficients of the master calibration model to the mesh of target instrument 210. In some implementations, the β coefficients of the master calibration model are interpolated to the mesh of target instrument 210 based on the determination that the master calibration set associated with the master calibration model is unavailable.

[0150] In some implementations, in addition to the calibration range constraints of the predicted values ​​associated with the reconnaissance set, a set of constraints includes correlation constraints associated with each of the principal β coefficient and a pair of transitive β coefficients, and / or slope constraints associated with each of the principal β coefficient and a pair of transitive β coefficients.

[0151] In some implementations, the modeling device 215 may generate a pair of initial transfer β coefficients based on the random generation of a pair of initial transfer β coefficients, applying a linear function associated with the random values ​​to the principal β coefficients, and / or adding random values ​​to the principal β coefficients.

[0152] Although Figure 4 An example block of process 400 is shown, but in some implementations, it is different from... Figure 4 Compared to the blocks depicted, process 400 may include additional blocks, fewer blocks, different blocks, or blocks with different arrangements. Alternatively, two or more blocks of process 400 may be executed in parallel.

[0153] To demonstrate the effectiveness of fLMC technology, the PLS regression model of Brix of sugarcane was transferred from the benchtop FOSS NIR master instrument to the portable MicroNIR target instrument. Figures 5A-5C This is a diagram relating to the results passed by this example calibration model using the fLMC technique.

[0154] A total of 1712 FOSS spectra were used to construct the master calibration model. These spectra were first interpolated to a MicroNIR grid. Interpolated calibration data were used to construct an intermediate master calibration model, and the resulting β coefficients were used as b...主 The MicroNIR instrument collected 126 spectra, of which 15 were randomly selected as the reconnaissance set for performing fLMC. The remaining 111 spectra were used as an external validation set to validate the transfer calibration model. The predictive performance of the transfer calibration model was compared with that of the master calibration model using the FOSS validation set from the same 111 samples.

[0155] like Figure 5A As shown, without performing calibration model transfer, the root mean square error (RMSEP) of the prediction is high when using an intermediate master calibration model to predict the MicroNIR validation set. However, as Figure 5B As shown, the RMSEP is significantly reduced when using the fLMC technique for calibration model transfer. The RMSEP for the FOSS validation set using the original FOSS model was also calculated and used as a benchmark for evaluating the performance of the transferred calibration model. Figure 5C As can be seen, the residuals between the predicted and laboratory Brix values ​​for the validation set obtained by the transferred calibration model remain within approximately ±2 RMSEP of the original FOSS master calibration model, indicating that the transferred MicroNIR calibration is within the original limits of the FOSS calibration with approximately 95% confidence. These results demonstrate that the performance of the transferred calibration model generated using fLMC technology is close to that of the original FOSS master calibration model (e.g., with a relatively wide wavelength range and relatively high spectral resolution).

[0156] Furthermore, for comparison, the same FOSS master calibration model was transferred to MicroNIR using Mean Difference Correction (MDC) and Piecewise Direct Normalization (PDS) techniques (two typical techniques for calibration model transfer). To apply these two techniques, a transfer set consisting of 15 spectra from both the master and target instruments was used. These spectra were from the same samples used in the reconnaissance set when using fLMC. Using the transferred calibration models with MDC and PDS, the RMSEP on the same validation set was 1.80 and 0.72, respectively. Therefore, in this case, fLMC performed better than MDC but worse than PDS. However, unlike PDS, fLMC does not require a transfer set on the master instrument, making it relatively less costly and / or complex while achieving similar performance.

[0157] As mentioned above, Figures 5A-5C This is provided as an example only. Other examples are possible and may differ from those provided. Figures 5A-5C Example of the description.

[0158] Figure 6This is a flowchart of an example process 600 for interpolating the β coefficients of the master calibration model to the grid of the target instrument to determine the master β coefficients used for fLMC or LMC techniques. In some implementations, Figure 6 One or more process blocks may be executed by the modeling device 215. In some implementations, Figure 6 One or more process blocks may be executed by another device or a group of devices (such as master instrument 205 and / or target instrument 210) that is separate from or includes modeling device 215.

[0159] like Figure 6 As shown, process 600 may include determining that the mesh of the master instrument associated with the master calibration model does not match the mesh of the target instrument, and generating a transfer calibration model corresponding to the master calibration model for the target instrument (block 610). For example, as described above, modeling device 215 may determine that the mesh of the master instrument 205 associated with the master calibration model does not match the mesh of the target instrument 210, and generating a transfer calibration model corresponding to the master calibration model for the target instrument 210.

[0160] like Figure 6 As further illustrated, process 600 may include interpolating the β coefficients of the master calibration model to the grid of the target instrument based on the determination that the grid of the master instrument does not match the grid of the target instrument (block 620). For example, as described above, modeling device 215 may interpolate the β coefficients of the master calibration model to the grid of the target instrument 210 based on the determination that the grid of the master instrument 205 does not match the grid of the target instrument 210.

[0161] like Figure 6 As further illustrated, process 600 may include determining the primary β coefficients (block 630) associated with the generation of the transfer calibration model based on the result of interpolating the β coefficients of the primary calibration model to the grid of the target instrument. For example, as described above, modeling device 215 may determine the primary β coefficients associated with the generation of the transfer calibration model based on the result of interpolating the β coefficients of the primary calibration model to the grid of the target instrument 210.

[0162] Process 600 may include additional implementations, such as any single implementation or any combination of implementations, such as those described below and / or any other processes described elsewhere herein.

[0163] In some implementations, based on the determination that the master calibration set associated with the master calibration model is unavailable, the β coefficients of the master calibration model are interpolated to the grid of the target instrument 210.

[0164] In some implementations, the modeling device 215 can perform constrained optimization of the objective function based on a set of constraints to determine a pair of transfer β coefficients, wherein the constrained optimization is performed based on a pair of initial transfer β coefficients, principal β coefficients, and a spectrum associated with the reconnaissance set. Here, the modeling device can determine the transfer β coefficients based on this pair of transfer β coefficients; it can determine the final transfer β coefficients based on a set of transfer β coefficients including these transfer β coefficients. In other words, in some implementations, the modeling device 215 can use the fLMC technique to determine the final transfer β coefficients. In some implementations, the set of constraints includes correlation constraints associated with each of the principal β coefficients and the pair of transfer β coefficients, slope constraints associated with each of the principal β coefficients and the pair of transfer β coefficients, and calibration range constraints for the predicted values ​​associated with the reconnaissance set. In some implementations, the modeling device 215 can generate a pair of initial transfer β coefficients based on the random generation of the pair of initial transfer β coefficients, by applying a linear function associated with random values ​​to the principal β coefficients, or by adding random values ​​to the principal β coefficients.

[0165] In some implementations, modeling device 215 may determine the transfer beta coefficients associated with the generated transfer calibration model based on the master beta coefficients and using linear model correction (LMC) techniques. In other words, in some implementations, modeling device 215 may use LMC techniques to determine the final transfer beta coefficients. In some implementations, reference values ​​for the reconnaissance set associated with the LMC technique are predicted based on the master calibration model and the master transfer set.

[0166] Although Figure 6 An example block of process 600 is shown, but in some implementations, it is different from... Figure 6 Compared to the blocks depicted, process 600 may include additional blocks, fewer blocks, different blocks, or blocks with different arrangements. Alternatively, two or more blocks of process 600 may be executed in parallel.

[0167] In some implementations, as described above, the β coefficients of the master calibration model can be interpolated into the grid of the target instrument 210 and used as master β coefficients. For example, in some implementations, this technique can be used in conjunction with LMC or fLMC techniques. Figures 7A-7C as well as Figure 8A and Figure 8B This is a diagram relating the interpolation of the β coefficients of the master calibration model to the grid of the target instrument, and using the LMC and fLMC techniques associated with the execution of the calibration model transfer.

[0168] Use the above about Figures 5A-5C Using the same dataset described, the LMC technique was performed with the results of interpolating the β coefficients of the master calibration model to the grid of the target instrument as the master β coefficients, and the results are shown in... Figures 7A-7C In this case, since there is no available master calibration set, it is impossible to construct an intermediate master calibration model. For example... Figure 7A As shown, when the interpolated β coefficients are used directly to predict the validation set on the target instrument, the resulting RMSEP is high. Figure 7B As shown, when the interpolated β coefficients are used as the principal β coefficients and the LMC technique is performed, the RMSEP is significantly reduced. Furthermore, as... Figure 7C As shown, the residuals between the predicted and laboratory Brix values ​​of the calibration model on the validation set are kept within ±2 RMSEP of the original FOSS master calibration model, with limited exceptions.

[0169] In addition, the above-mentioned Figures 5A-5C Using the same dataset described, the fLMC technique was performed using the results of interpolating the β coefficients of the master calibration model to the grid of the target instrument as the master β coefficients. The results are shown in... Figure 8A and Figure 8B In the middle. Although the performance is slightly lower compared to using LMC technology, it is still comparable to (e.g. Figure 5A Compared to the case where calibration model transfer is not performed (as shown), RMSEP is significantly reduced. Figure 8A As shown, the RMSEP is quite low, with a normalized RMSEP of 7.7% (normalized to the mean Brix value of the validation set). Furthermore, as... Figure 8B As shown, most of the residuals between the predicted and laboratory Brix values ​​for the validation set obtained by transferring the calibration model remain within ±2 RMSEP of the original FOSS master calibration model. It is noteworthy that the fLMC technique is the only technique that can be used when the master calibration set is unavailable, the grids of the master and target instruments are different, only the reconnaissance set collected by the target instrument is used for transfer, and there is no reference value for the reconnaissance set. In some implementations, the performance of the transferred calibration model can be further improved as the calibration model is updated.

[0170] As mentioned above, Figures 7A-7C as well as Figure 8A and Figure 8B Provided as an example only. Other examples are possible and may differ from those regarding 7A- Figure 7C as well as Figure 8A and Figure 8B Example of the description.

[0171] In some implementations, the techniques described herein can be used to standardize calibration models across multiple instruments. As mentioned above, instruments or devices of the same type often exhibit inter-instrumental differences. Therefore, when a calibration model is developed on one instrument but needs to be deployed on multiple (e.g., hundreds, millions, etc.) instruments, these inter-instrumental differences can lead to inconsistent performance. For this problem, performing calibration model transfer using general methods may be impractical, especially when instruments are located in different locations. To address this issue, LMC and fLMC techniques can be configured on the instrument. Here, when the master calibration model is transferred to the target instrument, only spectra from a small sample size need to be collected. The calibration model can be automatically corrected using LMC (e.g., when reference values ​​for the reconnaissance set are available) or fLMC (e.g., regardless of whether reference values ​​for the reconnaissance set are available).

[0172] Figures 9A-9D , Figure 10A , Figure 10B , Figure 11A and Figure 11B This is a diagram illustrating example results associated with the standardization of calibration models across multiple instruments. Figures 9A-9D , Figure 10A , Figure 10B , Figure 11A and Figure 11B In a related example, raw data from the MicroNIR device was calibrated in two different ways (Data A and Data B) to simulate inter-instrumental variations. 759 spectra from 38 mixture samples were used to build a calibration model to predict caffeine content. 200 spectra from another 10 mixture samples were used as a validation set. Figure 9A and Figure 9B As shown, the performance is similar when using calibration model A to predict validation data A, or when using calibration model B to predict validation data B. However, as... Figure 9C As shown, performance degrades when calibration model A is used to predict validation data B. Figure 9D As shown, many residuals between the predicted values ​​and laboratory values ​​of validation set B exceed the ±2RMSEP baseline used to predict validation A using model A.

[0173] Ten samples with three replicate spectra were randomly selected from calibration set B as a reconnaissance set to perform LMC and fLMC techniques. Figure 10A As shown, using LMC technology, RMSEP is significantly reduced. Figure 10B As shown, all predicted residuals are within the ±2RMSEP baseline used to predict and validate model A. Figure 11A As shown, RMSEP is similarly reduced using the fLMC technique. For example... Figure 11BAs shown, most of the predicted residuals are within ±2 RMSEP of the validation model A when using model A. Therefore, it is effective to use LMC or fLMC techniques to correct for inter-instrumental differences in model performance.

[0174] In fact, LMC and fLMC techniques are effective when using as few as eight samples as the reconnaissance set. It is worth noting that... Figure 10A , Figure 10B , Figure 11A and Figure 11B The results shown are examples of moderate performance. The reconnaissance samples are randomly selected to simulate real-world test scenarios on the user side. The final performance results are affected by which samples are used as the reconnaissance set. Similarly, fLMC performs worse than LMC. However, when no reference values ​​are available for the reconnaissance set, fLMC is the only technique that can be used to calibrate model propagation.

[0175] As mentioned above, Figures 9A-9D , Figure 10A , Figure 10B , Figure 11A and Figure 11B This is provided as an example only. Other examples are possible and may differ from those provided. Figures 9A-9D , Figure 10A , Figure 10B , Figure 11A and Figure 11B The example described.

[0176] As described above, in some cases, LMC techniques can be applied to calibration model updates by using updated samples as a reconnaissance set. It is noteworthy that this does not require all calibration data (e.g., as is required by general model update techniques that add updated samples to the calibration set and recalibrate the model) and takes relatively little time, allowing calibration model updates to be performed during online operation of the instruments (e.g., master instrument 205, target instrument 210).

[0177] Figure 12 This is a flowchart of an example process 1200 for performing calibration model updates using LMC technology. In some implementations, Figure 12 One or more process blocks may be executed by the modeling device 215. In some implementations, Figure 12 One or more process blocks may be executed by another device or a group of devices (such as master instrument 205 and / or target instrument 210) that is separate from or includes modeling device 215.

[0178] like Figure 12As shown, process 1200 may include obtaining a reconnaissance set associated with updating the calibration model, wherein the reconnaissance set includes spectra associated with a set of samples, on which the calibration model will be updated (block 1210). For example, modeling device 215 may obtain a reconnaissance set associated with updating the calibration model, wherein the reconnaissance set includes spectra associated with a set of samples, on which the calibration model will be updated.

[0179] like Figure 12 As further illustrated, process 1200 can determine the β coefficients associated with the calibration model (block 1220). For example, modeling device 215 can determine the β coefficients associated with the calibration model.

[0180] like Figure 12 As further illustrated, process 1200 may include determining updated β coefficients associated with the updated calibration model based on the β coefficients and using LMC techniques (block 1230). For example, modeling device 215 may determine updated β coefficients associated with the updated calibration model based on the β coefficients and using LMC techniques.

[0181] like Figure 12 As further illustrated, process 1200 may include updating the calibration model based on the updated β coefficients (block 1240). For example, modeling device 215 may update the calibration model based on the updated β coefficients (e.g., such that the updated calibration model uses the updated β coefficients associated with performing the calibration).

[0182] Process 1200 may include additional implementations, such as any single implementation or any combination of implementations, such as those described below and / or any other process described elsewhere herein.

[0183] In some implementations, the calibration model is updated during the operation of the instruments (e.g., master instrument 205, target instrument 210) without taking the equipment offline.

[0184] Although Figure 12 An example block of process 1200 is shown, but in some implementations, it is different from... Figure 12 Compared to the blocks depicted, process 1200 may include additional blocks, fewer blocks, different blocks, or blocks with different arrangements. Alternatively, two or more blocks of process 1200 may be executed in parallel.

[0185] Figure 13A and Figure 13B This is a diagram illustrating example results of performing a calibration model update using the linear model correction technique.

[0186] In order to update (combined with the above) Figure 7BThe described sugarcane Brix model uses an additional 30 MicroNIR spectra as an update set. Here, LMC technology is applied to update the calibration model. Figure 13A It is aimed at and Figure 7B The same validation set used is illustrated in the diagram of the predictions associated with this update. As shown, model performance is improved, where... Figure 13B As shown, RMSEP decreases and the prediction residual decreases.

[0187] As indicated above, Figure 13A and Figure 13B This is provided as an example only. Other examples are possible and may differ from those provided. Figure 13A and Figure 13B Example of the description.

[0188] As mentioned above, LMC technology requires reference values ​​from the reconnaissance set. When both the transfer sets from the master instrument 205 and the target instrument 210 are available, but the reference values ​​for these samples are unavailable, the master calibration model and the master transfer set can be used to predict the reference values ​​so that LMC technology can be made available.

[0189] Figure 14A and Figure 14B This is a diagram illustrating example results associated with predicting reference values ​​using the master calibration model and master transfer set. (Example...) Figure 14A As shown, using and Figures 5A-5C For related datasets that are identical, the RMSEP is 0.44 when using the true reference value. For example... Figure 14B As shown, the RMSEP is 0.80 when using reference values ​​predicted using the master calibration model and master transfer set. Although the performance of LMC techniques using predicted reference values ​​is not as good as that of LMC techniques using true reference values, as mentioned above, the performance is improved compared to using MDC or fLMC techniques and is similar to that using PDS techniques.

[0190] As indicated above, Figure 14A and Figure 14B This is provided as an example only. Other examples are possible and may differ from those provided. Figure 14A and Figure 14B Example of the description.

[0191] Some implementations described herein provide a focused LMC (fLMC) technique that can be used in conjunction with calibration model transfer. Similar to LMC, fLMC requires only a reconnaissance set collected by the target instrument. However, unlike LMC, fLMC does not require reference values ​​for the reconnaissance set. Therefore, using fLMC in conjunction with calibration model transfer reduces the cost, difficulty, and / or complexity of calibration model transfer (e.g., compared to LMC and the general calibration model transfer techniques described above).

[0192] Some implementations described herein provide a process in which the fLMC technique or LMC technique uses the β coefficients of the master calibration model associated with the execution calibration model transfer, without requiring a master calibration set.

[0193] Some implementation methods described in this article provide a process for updating models using LMC technology.

[0194] The foregoing disclosure provides illustrations and descriptions, but is not intended to be exhaustive or to limit the embodiments to the precise forms disclosed. In view of the above disclosure, modifications and variations are possible or can be obtained from practice of the embodiments.

[0195] As used herein, the term component is defined to be broadly interpreted as hardware, firmware, and / or a combination of hardware and software.

[0196] This article describes some implementation methods in conjunction with thresholds. As used herein, a threshold can refer to a value that is greater than, more than, higher than, greater than or equal to, less than, less than, lower than, less than or equal to, or equal to the threshold.

[0197] It will be apparent that the systems and / or methods described herein can be implemented in various forms, including hardware, firmware, or a combination of firmware and software. The actual dedicated control hardware or software code used to implement these systems and / or methods is not a limitation on the implementation. Therefore, while the operation and behavior of the systems and / or methods are described herein without reference to specific software code, it should be understood that software and hardware can be designed to implement the systems and / or methods based on the description herein.

[0198] Although specific combinations of features are stated in the claims and / or disclosed in the specification, these combinations are not intended to limit the disclosure of possible embodiments. In fact, many of these features can be combined in ways not specifically stated in the claims and / or not disclosed in the specification. Although each dependent claim listed below may be directly subordinated to only one claim, the disclosure of possible embodiments includes each dependent claim in combination with each other claim in the claim set.

[0199] No element, action, or instruction used herein should be construed as essential or necessary unless explicitly stated otherwise. Furthermore, the articles “a” and “an” as used herein are intended to include one or more items and are interchangeable with “one or more.” Additionally, the term “set” as used herein is intended to include one or more items (e.g., related items, unrelated items, a combination of related and unrelated items, etc.) and is interchangeable with “one or more.” The term “one” or similar language is used where only one item is intended to be described. Furthermore, the terms “has,” “have,” and “having,” as used herein, are intended to be open-ended terms. Additionally, unless explicitly stated otherwise, the phrase “based on” is intended to mean “at least partially based on.”

Claims

1. A method for updating a calibration model, comprising: After the calibration model is deployed to the spectrometer, the device receives an update set, which includes new samples representing different conditions; as well as After the calibration model is deployed to the spectrometer and during the online operation of the spectrometer, the device updates the calibration model using linear model correction (LMC) technology and the update set. The calibration model is updated based on the β coefficient associated with the calibration model.

2. The method of claim 1, wherein the calibration model is updated without using all calibration data of the existing calibration set.

3. The method of claim 1, wherein the calibration model is updated without adding the new sample to the existing calibration set.

4. The method of claim 1, wherein the update set is a reconnaissance set associated with performing a calibration model update.

5. The method according to claim 1, further comprising: Receive the master calibration model from the master instrument; Generate a transfer calibration model corresponding to the master calibration model; as well as The transfer calibration model is transmitted to the target instrument. The spectroscopic instrument mentioned therein is either the main instrument or the target instrument, and The calibration model mentioned therein is either the master calibration model or the transfer calibration model.

6. The method of claim 1, wherein the spectrometer is implemented within the device.

7. The method according to claim 1, further comprising: The device determines the β coefficient associated with the calibration model.

8. The method according to claim 1, further comprising: The β coefficient is determined using the LMC technique and the update set, based on another β coefficient associated with the calibration model.

9. An apparatus for updating a calibration model, comprising: One or more memory units; as well as One or more processors, coupled to the one or more memories, are configured to enable the device to: After the calibration model is deployed to the spectroscopic instrument, an update set is received, which includes new samples representing different conditions. as well as After the calibration model is deployed to the spectrometer and during the online operation of the spectrometer, the calibration model is updated using the linear model correction (LMC) technique and the update set, based on the β coefficients associated with the calibration model.

10. The device of claim 9, wherein the one or more processors are configured to cause the device to update the calibration model as follows: The calibration model is updated without using all calibration data from the existing calibration set.

11. The device of claim 9, wherein the one or more processors are configured to cause the device to update the calibration model as follows: The calibration model is updated without adding the new sample to the existing calibration set.

12. The device of claim 9, wherein the update set is a reconnaissance set associated with performing a calibration model update.

13. The apparatus of claim 9, wherein the spectrometer is a target instrument, and The calibration model mentioned above is a transfer calibration model.

14. The apparatus of claim 9, wherein the spectroscopic instrument is implemented within the apparatus.

15. The device of claim 9, wherein the one or more processors are further configured to cause the device to: Receive the master calibration model from the master instrument; Generate a transfer calibration model corresponding to the master calibration model; and The transfer calibration model is transmitted to the target instrument. The spectroscopic instrument mentioned therein is either the main instrument or the target instrument, and The calibration model mentioned therein is either the master calibration model or the transfer calibration model.

16. The device of claim 9, wherein the one or more processors are further configured to cause the device to: Determine the β coefficient associated with the calibration model.

17. The device of claim 9, wherein the one or more processors are further configured to cause the device to: The β coefficient is determined using the LMC technique and the update set, based on another β coefficient associated with the calibration model.

18. A non-transitory computer-readable medium storing an instruction set, the instruction set comprising: One or more instructions, which, when executed by one or more processors of the device, cause the device to: After the calibration model is deployed to the spectroscopic instrument, an update set is received, which includes new samples representing different conditions. as well as After the calibration model is deployed to the spectrometer and during the online operation of the spectrometer, the calibration model is updated using the linear model correction (LMC) technique and the update set, based on the β coefficients associated with the calibration model.

19. The non-transitory computer-readable medium of claim 18, wherein the one or more processors are configured to cause the device to update the calibration model as follows: The calibration model is updated without using all calibration data from the existing calibration set, and without adding the new sample to the existing calibration set.

20. The non-transitory computer-readable medium of claim 18, wherein the spectroscopic instrument is a master instrument, and The calibration model mentioned therein is the master calibration model.

Images