Method, device, equipment and medium for judging consistency of MEMS tunable fabry-perot cavity chip

Abstract

The application discloses a kind of judging MEMS adjustable Fabry-Perot cavity chip consistency method, device, equipment and medium, method includes respectively collecting standard chip and sample spectrum data of chip to be measured;Standard chip and sample spectrum data collected by chip to be measured are divided into modeling set and judging set;Spectrum modeling is carried out to standard chip and chip to be measured modeling set spectrum data using the same modeling mode, obtain standard chip spectrum model and chip to be measured spectrum model;Standard chip spectrum model and chip to be measured spectrum model are used respectively to predict judging set sample, and obtain standard predicted value and chip to be measured predicted value respectively;The overall offset of standard predicted value and chip to be measured predicted value is calculated, and the consistency of standard chip and chip to be measured is judged in combination with overall offset.The application can greatly improve the pass rate of spectral chip under the premise of ensuring the consistency of spectral chip prediction ability, and can effectively solve the problem of low pass rate and chip waste in traditional chip consistency judgment method.

Description

Technical Field

[0001] This invention relates to the field of MEMS Fabry-Perot cavity chip technology, and in particular to a method, apparatus, device and medium for determining the consistency of MEMS tunable Fabry-Perot cavity chips. Background Technology

[0002] With the development of microelectromechanical systems (MEMS) technology, miniaturized near-infrared spectrometers based on MEMS Fabry-Perot cavity interferometric tunable filter chips have been widely used in various fields such as food, beverage, medicine, and coal in recent years. The optical principle of the MEMS tunable Fabry-Perot cavity chip is based on the Fabry-Perot interferometry principle. The spectrometer, fabricated using semiconductor integrated circuit technology, is driven by different voltages, allowing for real-time changes in the cavity length to obtain different narrow-wavelength spectra. Compared to traditional large-scale Fourier transform spectral analysis systems, miniaturized near-infrared spectrometers offer advantages such as simple structure, low cost, portability, rapid detection, and real-time results, and have become a popular product in the field of spectroscopy.

[0003] With the development of miniaturized near-infrared spectroscopy technology, there are high requirements for the consistency of miniaturized near-infrared spectroscopy equipment in practical detection applications. To ensure the consistency of miniaturized near-infrared spectroscopy equipment, the most important thing is to ensure the consistency of the tunable Fabry-Perot cavity chip of the spectral MEMS. The current traditional method for judging the consistency of tunable Fabry-Perot cavity chips of spectral MEMS is as follows: Apply the same driving voltage to both the standard tunable Fabry-Perot cavity chip of the spectral MEMS (standard chip) and the tunable Fabry-Perot cavity chip of the spectral MEMS under test (chip under test). Determine whether the wavelength value of the characteristic wavelength point obtained by the chip under test through narrowband filtering is the same as that of the standard chip under the same voltage, and whether the light intensity value collected at the characteristic wavelength point is the same as that of the standard chip under the same voltage. If the characteristic wavelength point and the light intensity value are the same or have small differences, it indicates that the chip under test has good consistency with the standard chip and can be used in practical rapid spectral detection applications. If the difference is large, it indicates that the consistency difference of the chip under test is large and cannot be used in practical rapid spectral detection applications. Traditional chip consistency assessment methods place extremely high demands on chip manufacturing processes and precision, easily leading to a significant decrease in the yield rate of tunable FAP-Cavity chips in spectral MEMS, resulting in chip waste. In practical rapid spectral detection applications, even if the standard chip and the chip under test (DUT) have significant differences in light intensity values ​​at characteristic wavelength points or the same characteristic wavelength point, if the DUT and standard chips have consistent predictive capabilities, they can still be judged to have good consistency and be used in practical rapid spectral detection applications. Therefore, it is urgent to overcome the problems of traditional chip consistency assessment methods, improve the yield rate of tunable FAP-Cavity chips in spectral MEMS, and reduce chip waste. Summary of the Invention

[0004] This invention provides a method, apparatus, device, and medium for determining the consistency of MEMS adjustable Faber cavity chips, in order to solve the problems of low pass rate and chip waste in traditional chip consistency determination methods.

[0005] The technical solution adopted in this invention is: to provide a method for determining the consistency of a MEMS tunable Fabry-Perot cavity chip, comprising:

[0006] Spectral data of the standard chip and the chip under test were collected separately.

[0007] The sample spectral data collected from the standard chip and the chip under test are divided into a modeling set and a judgment set;

[0008] The same modeling method was used to perform spectral modeling on the spectral data of the standard chip and the chip under test to obtain the spectral model of the standard chip and the spectral model of the chip under test.

[0009] The standard chip spectral model and the test chip spectral model were used to predict the judgment set samples, and the standard predicted value and the test predicted value were obtained respectively.

[0010] Calculate the overall offset between the standard predicted value and the predicted value to be tested, and use the overall offset to determine the consistency between the standard chip and the chip to be tested.

[0011] Furthermore, the method for separately acquiring the spectral data of the standard chip and the chip under test includes:

[0012] By applying different voltages to the chip and changing the chip's cavity length, different narrowband spectra can be obtained. When the chip's narrowband filter reaches a characteristic wavelength point, the light intensity value at that characteristic wavelength point is collected.

[0013] Furthermore, the modeling set is a collection of sample spectral data with known calibration values, used to establish spectral models for the standard chip and the chip under test, wherein the samples included in the modeling sets of the standard chip and the chip under test are completely identical; the judgment set is a collection of sample spectral data with unknown calibration values, used to verify whether the predictive capabilities of the standard chip and the chip under test are similar, wherein the samples included in the judgment sets of the standard chip and the chip under test are completely identical.

[0014] Furthermore, when performing spectral modeling on the spectral data of the standard chip and the chip under test modeling set, a quantitative model based on the sample component content and a qualitative model based on the sample category are established respectively; that is, the standard chip spectral model actually includes the standard chip quantitative model and the standard chip qualitative model, and the chip under test model actually includes the chip under test quantitative model and the chip under test qualitative model.

[0015] Spectral modeling was performed on the standard chip and the chip under test respectively to obtain the quantitative model and qualitative model of the standard chip, as well as the quantitative model and qualitative model of the chip under test.

[0016] Furthermore, the method for standard chip spectral modeling includes: matching the spectral data of the modeling set samples collected by the standard chip with the sample component content calibration values ​​one by one; performing specific preprocessing on the spectral data of the modeling set samples; and using partial least squares method to linearly fit the spectral data of the modeling set and its corresponding component content calibration values ​​to establish a linear fitting equation between the light intensity value at the characteristic wavelength point of the standard chip and the sample component content calibration value, thereby obtaining a quantitative model of the standard chip.

[0017] The spectral data of the model set samples collected by the standard chip are matched one by one with the sample category calibration values. After preprocessing the spectral data of the model set samples, a classification model between the light intensity value at the characteristic wavelength point of the standard chip and the sample category calibration value is established by using the linear discriminant analysis method. This is the qualitative model of the standard chip. The classification model is used to predict the unknown category samples in the discrimination set and obtain their category prediction values.

[0018] The method for spectral modeling of the chip under test includes: mapping the spectral data of the modeling set samples collected from the chip under test to the calibration values ​​of the sample component content one by one; performing specific preprocessing on the spectral data of the modeling set samples; and using partial least squares method to linearly fit the spectral data of the modeling set and its corresponding component content calibration values ​​to establish a linear fitting equation between the light intensity value at the characteristic wavelength point of the chip under test and the sample component content calibration value, thereby obtaining a quantitative model of the chip under test.

[0019] The spectral data of the model set samples collected by the chip under test are matched one by one with the sample category calibration values. After preprocessing the spectral data of the model set samples, a classification model between the light intensity value at the characteristic wavelength point of the chip under test and the sample category calibration value is established by using the linear discriminant analysis method. This is the qualitative model of the chip under test. The classification model is used to predict the unknown category samples in the discrimination set and obtain their category prediction values.

[0020] Furthermore, the standard chip quantitative model is used to predict the spectral data of each sample in the judgment set acquired by the standard chip, and the predicted values ​​of the component content of each sample in the judgment set are obtained. Then, the standard chip qualitative model is used to predict the spectral data of each sample in the judgment set acquired by the standard chip, and the predicted values ​​of the categories of each sample in the judgment set are obtained. Similarly, the quantitative and qualitative models of the chip to be tested are used to perform the same predictive analysis to obtain the predicted values ​​of the chip to be tested.

[0021] Furthermore, the method for calculating the overall offset between the standard predicted value and the predicted value to be tested, and for judging the consistency between the standard chip and the chip to be tested based on the overall offset, includes:

[0022] For quantitative analysis of the judgment set samples: calculate the quantitative offset between the predicted component content of the standard chip and the predicted component content of the chip to be tested; determine the consistency between the chip to be tested and the standard chip by comparing the set quantitative offset threshold with the quantitative offset.

[0023] Qualitative analysis of the judgment set samples: Calculate the qualitative offset between the predicted class value of the standard chip and the predicted class value of the chip under test; determine the consistency between the chip under test and the standard chip based on the comparison between the set qualitative offset threshold and the qualitative offset.

[0024] This invention also discloses a device for determining the consistency of a MEMS tunable Fabry-Perot cavity chip, comprising:

[0025] The acquisition module is used to acquire spectral data of the standard chip and the chip under test, respectively.

[0026] The partitioning module is used to divide the sample spectral data collected from the standard chip and the chip under test into a modeling set and a judgment set;

[0027] The modeling module is used to perform spectral modeling on the spectral data of the standard chip and the chip under test modeling set using the same modeling method, and to obtain the spectral model of the standard chip and the spectral model of the chip under test.

[0028] The prediction module is used to predict the judgment set samples using the standard chip spectral model and the test chip spectral model respectively, and obtain the standard prediction value and the test prediction value respectively.

[0029] The adjustment module is used to calculate the overall offset between the standard predicted value and the predicted value to be tested, and to judge the consistency between the standard chip and the chip to be tested based on the overall offset.

[0030] The present invention also discloses an electronic device, including a processor coupled to a memory; the processor is configured to execute a computer program stored in the memory, such that the electronic device performs the method described above for determining the consistency of a MEMS tunable Fapper cavity chip.

[0031] The present invention also discloses a computer-readable storage medium including a computer program or instructions that, when the computer program or instructions are run on a computer, cause the computer to perform the method described above for determining the consistency of a MEMS tunable Fapper cavity chip.

[0032] The beneficial effects of this invention are as follows: This invention combines the consistency judgment standard of spectral chips in actual spectral rapid detection applications. By comparing the qualitative and quantitative prediction capabilities of two types of spectral MEMS tunable Fabry-Perot cavity chips, the consistency judgment of the chip under test and the standard chip is completed. Under the premise of ensuring the consistency of the spectral chip prediction capabilities, the pass rate of spectral chips is greatly improved. This can effectively solve the problems of low pass rate and chip waste in traditional chip consistency judgment methods. Attached Figure Description

[0033] Figure 1 This is a flowchart illustrating the method for determining the consistency of a MEMS tunable Faper cavity chip disclosed in this invention. Detailed Implementation

[0034] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be described in further detail below with reference to the accompanying drawings, but the embodiments of the present invention are not limited thereto.

[0035] Example 1:

[0036] See Figure 1 This embodiment discloses a method for determining the consistency of a MEMS tunable Fabry-Perot cavity chip, including the following steps:

[0037] 101: Collect spectral data of the standard chip and the chip under test respectively.

[0038] The standard chip refers to a standard spectral MEMS tunable Fabry-Perot cavity chip, and the chip under test refers to the spectral MEMS tunable Fabry-Perot cavity chip to be tested. The spectral MEMS tunable Fabry-Perot cavity chip is driven by different voltages, changing the cavity length in real time to obtain different narrow-band spectra. Since theoretically the driving voltage can be infinitely divided, the corresponding cavity length of the MEMS Fabry-Perot cavity chip can also be infinitely divided. Therefore, the spectral data acquired from the sample can contain light intensity values ​​at an infinite number of wavelengths. However, too many wavelengths will severely affect the acquisition and analysis efficiency, while too few wavelengths will affect the analysis accuracy. Therefore, the sample spectral data acquired after narrowband filtering by the spectral MEMS tunable Fabry-Perot cavity chip is used to select characteristic wavelengths according to actual needs. That is, when the narrowband filtering of the spectral MEMS tunable Fabry-Perot cavity chip is a characteristic wavelength, the light intensity value at that characteristic wavelength is acquired, effectively improving the efficiency of sample spectral data acquisition and analysis.

[0039] In this embodiment, the spectral MEMS tunable Fabry-Perot cavity chip is limited by existing chip technology, and its wavelength range is limited. The commonly used spectral MEMS tunable Fabry-Perot cavity chip covers a wavelength range of 400nm. In order to balance the efficiency and accuracy of spectral analysis, this embodiment selects 41 equidistant spectral characteristic wavelength points as the characteristic wavelength points of the sample spectral data. That is, the sample spectral data collected after narrowband filtering by the spectral MEMS tunable Fabry-Perot cavity chip contains 41 characteristic wavelength points, and the wavelength difference between any two adjacent characteristic wavelength points is 400 / (41-1) = 10nm. The spectral data of the sample is actually represented by the light intensity values ​​at these 41 characteristic wavelength points.

[0040] 102: Divide the sample spectral data collected from the standard chip and the chip under test into a modeling set and a judgment set.

[0041] The sample spectral data are categorized into a modeling set and a judgment set according to a certain proportion. The modeling set consists of sample spectral data with known calibration values, used to establish spectral models for the standard chip and the chip under test. The samples in the modeling set of the standard chip and the chip under test are completely identical. The judgment set consists of sample spectral data with unknown calibration values, used to verify whether the predictive capabilities of the standard chip and the chip under test are similar. The samples in the judgment set of the standard chip and the chip under test are completely identical.

[0042] In this embodiment, to ensure the spectral model established from the sample spectral data of the modeling set has good predictive ability and stability, the number of samples in the modeling set needs to be no less than 50. Simultaneously, to reduce prediction error, the number of samples in the judgment set needs to be no less than 20. Therefore, in this embodiment, spectral data from 70 samples are collected for both the standard chip and the chip under test, and the sample spectral data are divided into a modeling set and a judgment set at a 5:2 ratio. That is, the modeling sets of both the standard chip and the chip under test each contain 50 samples, all of which have known calibration values ​​and are identical, and are used to establish the spectral models of the standard chip and the chip under test. Similarly, the judgment sets of both the standard chip and the chip under test each contain 20 samples, all of which have unknown calibration values ​​and are identical, and are used to verify whether the predictive abilities of the standard chip and the chip under test are similar.

[0043] 103: Using the same modeling method, spectral modeling was performed on the spectral data of the standard chip and the chip under test (DUT) modeling set to obtain the standard chip spectral model and the DUT spectral model. Miniaturized near-infrared spectroscopy rapid detection technology is mainly used for quantitative and qualitative analysis in practical applications. To ensure the versatility of the tunable Fapper cavity chip in spectral MEMS, when performing spectral modeling on the spectral data of the standard chip and the DUT modeling set, a quantitative model based on sample component content and a qualitative model based on sample category were established respectively. That is, the standard chip spectral model actually includes both the standard chip quantitative model and the standard chip qualitative model, and the DUT model actually includes both the DUT quantitative model and the DUT qualitative model.

[0044] In this embodiment, for standard chip spectral modeling, the spectral data of the modeling set samples collected by the standard chip are mapped one-to-one with the sample component content calibration values. After performing a specific preprocessing method on the spectral data of the modeling set samples, partial least squares (PLS) is used to linearly fit the spectral data of the modeling set and its corresponding component content calibration values, establishing the light intensity values ​​(P1, P2, ... P1) at 41 characteristic wavelength points of the standard chip. 41 The linear fitting equation between the sample component content calibration value M and the sample component content calibration value is as follows:

[0045] M = t1*P1 + t2*P2 + t3*P3 + ... + t 41 *P 41 +b

[0046] Among them, t i , (i = 1, 2, 3, ..., 41) are the partial least squares linear fitting parameters, and b is a constant term. Partial least squares is a regression modeling method for multiple dependent variables on multiple independent variables, and it is the most commonly used modeling method in quantitative spectral analysis. Obtaining a standard chip quantitative model through spectral modeling using partial least squares yields reliable and comprehensive analytical conclusions.

[0047] Furthermore, the spectral data of the modeling set samples collected by the standard chip are matched one by one with the sample category calibration values. After performing specific preprocessing on the spectral data of the modeling set samples, the linear discriminant analysis (LDA) method is used to establish a classification model between the light intensity values ​​at 41 characteristic wavelength points of the standard chip and the sample category calibration values. This is the qualitative model of the standard chip. The classification model is used to predict the unknown category samples in the discrimination set and obtain their category prediction values.

[0048] Similarly, for the spectral modeling of the chip under test, the same spectral preprocessing and modeling methods as for the standard chip are used to establish both a quantitative model and a qualitative model of the chip under test. Specifically, the method for spectral modeling of the chip under test includes: mapping the spectral data of the modeling set samples collected from the chip under test one-to-one with the sample component content calibration values; performing specific preprocessing on the spectral data of the modeling set samples; and then using partial least squares method to linearly fit the spectral data of the modeling set and its corresponding component content calibration values ​​to establish a linear fitting equation between the light intensity value at the characteristic wavelength point of the chip under test and the sample component content calibration values, thereby obtaining the quantitative model of the chip under test.

[0049] The spectral data of the model set samples collected by the chip under test are matched one by one with the sample category calibration values. After preprocessing the spectral data of the model set samples, a classification model between the light intensity value at the characteristic wavelength point of the chip under test and the sample category calibration value is established by using the linear discriminant analysis method. This is the qualitative model of the chip under test. The classification model is used to predict the unknown category samples in the discrimination set and obtain their category prediction values.

[0050] 104: The standard chip spectral model and the test chip spectral model were used to predict the judgment set samples, and the standard prediction value and the test prediction value were obtained respectively.

[0051] The standard chip quantitative model is used to predict the component content of each sample in the judgment set acquired by the standard chip. Then, the standard chip qualitative model is used to predict the category of each sample in the judgment set. Similarly, the quantitative and qualitative models of the target chip are used to perform the same predictive analysis to obtain the predicted value. The predicted values ​​in this step all consist of two parts: predicted component content and predicted sample type.

[0052] In this embodiment, the 20 samples in the label judgment set are numbered (A1, A2, ... A...). 20 Using a standard chip-based quantitative model, predictions were made for 20 samples, resulting in predicted component content values ​​for the 20 samples in the judgment set (B1, B2, ..., B...). 20 ), where sample Aj (i = 1, 2, 3, ..., 20) and component content prediction value B j (i = 1, 2, 3, ..., 20) have a one-to-one correspondence. Similarly, the standard chip qualitative model is used to predict the categories of the 20 samples, and the predicted categories of the 20 samples in the judgment set are (C1, C2, ..., C...). 20 ), of which sample A j (i = 1, 2, 3, ..., 20) and the category prediction value Cj (i = 1, 2, 3, ..., 20) are in a one-to-one correspondence.

[0053] Furthermore, the prediction method for the chip under test is consistent with that for the standard chip, and the component contents of the 20 samples in the judgment set are obtained as (D1, D2, ... D...). 20 The predicted values ​​for each category are (E1, E2, ..., E...). 20 ), of which sample A j (i = 1, 2, 3, ..., 20) and the predicted component content D j (i = 1, 2, 3, ..., 20) and category prediction value E j (i = 1, 2, 3, ..., 20) all have a one-to-one correspondence.

[0054] 105: Calculate the overall offset between the standard predicted value and the predicted value to be tested, and judge the consistency between the standard chip and the chip to be tested based on the overall offset.

[0055] For the predicted values ​​of component content, the quantitative offset between the standard predicted value and the predicted value to be tested is calculated. The smaller the offset, the better the consistency between the standard chip and the chip to be tested. For the predicted values ​​of sample type, the standard predicted value and the predicted value to be tested are compared to see if they are consistent, and the qualitative offset is calculated. The smaller the qualitative offset, the better the consistency between the standard chip and the chip to be tested.

[0056] In this embodiment, for the judgment set samples (A1, A2, ... A... 20 Quantitative analysis of the standard chip showed that the predicted component content values ​​were (B1, B2, ... B). 20 The predicted component content of the chip under test is (D1, D2, ... D). 20 The quantitative offset X1 between the two is calculated as follows:

[0057]

[0058] In quantitative analysis, the smaller the quantitative offset, the better the consistency between the standard chip and the chip under test. Based on the actual needs of spectral rapid detection applications, the quantitative offset threshold is set to H. When the quantitative offset X≤H, the chip under test and the standard chip have good consistency and can be applied to spectral quantitative rapid detection. When the quantitative offset X>H, the chip under test and the standard chip have poor consistency and cannot be applied to spectral quantitative rapid detection.

[0059] For the judgment set sample (A1, A2, ... A 20 Qualitative analysis of the standard chip class predicts the values ​​as (C1, C2, ..., C...). 20The predicted category value of the chip under test is (E1, E2, ..., E...). 20 The qualitative offset X2 between the two is calculated as follows:

[0060]

[0061] Where equal(C k E k ) is a qualitative judgment function, when C k =E k equal(C k E k The output is 0 when C k ≠E k At that time, e q ual(C k E k The output is 1.

[0062] In qualitative analysis, the smaller the qualitative offset, the better the consistency between the standard chip and the chip under test. Based on the actual needs of spectral rapid detection applications, the qualitative offset threshold is set to Y. When the qualitative offset X≤Y, the chip under test and the standard chip have good consistency and can be applied to spectral qualitative rapid detection; when the qualitative offset X>Y, the chip under test and the standard chip have poor consistency and cannot be applied to spectral qualitative rapid detection.

[0063] Furthermore, while ensuring the predictive capability of the tunable Fabry-Perot cavity chip in spectral MEMS, in order to maximize the yield and utilization rate of the tunable Fabry-Perot cavity chip in spectral MEMS, when the actual spectral rapid detection requirement is quantitative analysis, it is only necessary to ensure that the quantitative prediction capability of the standard chip and the chip under test is consistent. This indicates good consistency between the chip under test and the standard chip, and it can be applied to quantitative rapid spectral detection. When the actual spectral rapid detection requirement is qualitative analysis, it is only necessary to ensure that the qualitative prediction capability of the standard chip and the chip under test is consistent. This indicates good consistency between the chip under test and the standard chip, and it can be applied to qualitative rapid spectral detection. When the actual spectral rapid detection requirement is both qualitative and quantitative analysis, it is necessary to ensure that the quantitative and qualitative prediction capabilities of the standard chip and the chip under test are consistent simultaneously. This indicates good consistency between the chip under test and the standard chip, and it can be applied to rapid spectral detection.

[0064] Example 2:

[0065] This embodiment discloses a device for determining the consistency of a MEMS tunable Fapper cavity chip, comprising:

[0066] The acquisition module is used to acquire spectral data of the standard chip and the chip under test, respectively.

[0067] The partitioning module is used to divide the sample spectral data collected from the standard chip and the chip under test into a modeling set and a judgment set;

[0068] The modeling module is used to perform spectral modeling on the spectral data of the standard chip and the chip under test modeling set using the same modeling method, and to obtain the spectral model of the standard chip and the spectral model of the chip under test.

[0069] The prediction module is used to predict the judgment set samples using the standard chip spectral model and the test chip spectral model respectively, and obtain the standard prediction value and the test prediction value respectively.

[0070] The adjustment module is used to calculate the overall offset between the standard predicted value and the predicted value to be tested, and to judge the consistency between the standard chip and the chip to be tested based on the overall offset.

[0071] Example 3:

[0072] This embodiment discloses an electronic device, including a processor coupled to a memory; the processor is used to execute a computer program stored in the memory, so that the electronic device performs the method for determining the consistency of a MEMS tunable Fapper cavity chip as described in Embodiment 1.

[0073] Example 4:

[0074] A computer-readable storage medium includes a computer program or instructions that, when executed on a computer, cause the computer to perform a method for determining the consistency of a MEMS tunable Fapper cavity chip as described in Embodiment 1.

[0075] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for determining the consistency of a MEMS tunable Fabry-Perot cavity chip, characterized in that, include: Spectral data of the standard chip and the chip under test were collected separately. The sample spectral data collected from the standard chip and the chip under test are divided into a modeling set and a judgment set; The same modeling method was used to perform spectral modeling on the spectral data of the standard chip and the chip under test to obtain the spectral model of the standard chip and the spectral model of the chip under test. The standard chip spectral model and the test chip spectral model were used to predict the judgment set samples, and the standard predicted value and the test predicted value were obtained respectively. Calculate the overall offset between the standard predicted value and the predicted value to be tested, and use the overall offset to determine the consistency between the standard chip and the chip to be tested.

2. The method for determining the consistency of a MEMS tunable Fabry-Perot cavity chip according to claim 1, characterized in that, The method for separately acquiring spectral data of the standard chip and the chip under test includes: By applying different voltages to the chip and changing the chip's cavity length, different narrowband spectra can be obtained. When the chip's narrowband filter reaches a characteristic wavelength point, the light intensity value at that characteristic wavelength point is collected.

3. The method for determining the consistency of a MEMS tunable Fabry-Perot cavity chip according to claim 1, characterized in that, The modeling set is a collection of sample spectral data with known calibration values, used to establish spectral models for the standard chip and the chip under test, wherein the samples included in the modeling sets of the standard chip and the chip under test are completely identical; the judgment set is a collection of sample spectral data with unknown calibration values, used to verify whether the predictive capabilities of the standard chip and the chip under test are similar, wherein the samples included in the judgment sets of the standard chip and the chip under test are completely identical.

4. The method for determining the consistency of a MEMS tunable Fabry-Perot cavity chip according to claim 1, characterized in that, When performing spectral modeling on the spectral data of the standard chip and the chip under test modeling set, a quantitative model based on the sample component content and a qualitative model based on the sample category are established respectively; that is, the standard chip spectral model actually includes the standard chip quantitative model and the standard chip qualitative model, and the chip under test model actually includes the chip under test quantitative model and the chip under test qualitative model. Spectral modeling was performed on the standard chip and the chip under test respectively to obtain the quantitative model and qualitative model of the standard chip, as well as the quantitative model and qualitative model of the chip under test.

5. The method for determining the consistency of a MEMS tunable Fabry-Perot cavity chip according to claim 4, characterized in that, The method for standard chip spectral modeling includes: mapping the spectral data of the modeling set samples collected by the standard chip to the sample component content calibration values ​​one by one; performing specific preprocessing on the spectral data of the modeling set samples; using partial least squares method to linearly fit the spectral data of the modeling set and its corresponding component content calibration values; establishing a linear fitting equation between the light intensity value at the characteristic wavelength point of the standard chip and the sample component content calibration values; and thus obtaining a quantitative model of the standard chip. The spectral data of the model set samples collected by the standard chip are matched one by one with the sample category calibration values. After preprocessing the spectral data of the model set samples, a classification model between the light intensity value at the characteristic wavelength point of the standard chip and the sample category calibration value is established by using the linear discriminant analysis method. This is the qualitative model of the standard chip. The classification model is used to predict the unknown category samples in the discrimination set and obtain their category prediction values. The method for spectral modeling of the chip under test includes: mapping the spectral data of the modeling set samples collected from the chip under test to the calibration values ​​of the sample component content one by one; performing specific preprocessing on the spectral data of the modeling set samples; and using partial least squares method to linearly fit the spectral data of the modeling set and its corresponding component content calibration values ​​to establish a linear fitting equation between the light intensity value at the characteristic wavelength point of the chip under test and the sample component content calibration value, thereby obtaining a quantitative model of the chip under test. The spectral data of the model set samples collected by the chip under test are matched one by one with the sample category calibration values. After preprocessing the spectral data of the model set samples, a classification model between the light intensity value at the characteristic wavelength point of the chip under test and the sample category calibration value is established by using the linear discriminant analysis method. This is the qualitative model of the chip under test. The classification model is used to predict the unknown category samples in the discrimination set and obtain their category prediction values.

6. The method for determining the consistency of a MEMS tunable Fabry-Perot cavity chip according to claim 4 or 5, characterized in that, The standard chip quantitative model is used to predict the spectral data of each sample in the judgment set acquired by the standard chip, and the predicted values ​​of the component content of each sample in the judgment set are obtained. Then, the standard chip qualitative model is used to predict the spectral data of each sample in the judgment set acquired by the standard chip, and the predicted values ​​of the categories of each sample in the judgment set are obtained. Similarly, the quantitative model and qualitative model of the chip to be tested are used to perform the same predictive analysis to obtain the predicted values ​​of the chip to be tested.

7. The method for determining the consistency of a MEMS tunable Fabry-Perot cavity chip according to claim 6, characterized in that, The method for calculating the overall offset between the standard predicted value and the predicted value to be tested, and for determining the consistency between the standard chip and the chip to be tested based on the overall offset, includes: For quantitative analysis of the judgment set samples: calculate the quantitative offset between the predicted component content of the standard chip and the predicted component content of the chip to be tested; determine the consistency between the chip to be tested and the standard chip by comparing the set quantitative offset threshold with the quantitative offset. Qualitative analysis of the judgment set samples: Calculate the qualitative offset between the predicted class value of the standard chip and the predicted class value of the chip under test; determine the consistency between the chip under test and the standard chip based on the comparison between the set qualitative offset threshold and the qualitative offset.

8. A device for determining the consistency of a MEMS tunable Fabry-Perot cavity chip, characterized in that, include: The acquisition module is used to acquire spectral data of the standard chip and the chip under test, respectively. The partitioning module is used to divide the sample spectral data collected from the standard chip and the chip under test into a modeling set and a judgment set; The modeling module is used to perform spectral modeling on the spectral data of the standard chip and the chip under test modeling set using the same modeling method, and to obtain the spectral model of the standard chip and the spectral model of the chip under test. The prediction module is used to predict the judgment set samples using the standard chip spectral model and the test chip spectral model respectively, and obtain the standard prediction value and the test prediction value respectively. The adjustment module is used to calculate the overall offset between the standard predicted value and the predicted value to be tested, and to judge the consistency between the standard chip and the chip to be tested based on the overall offset.

9. An electronic device, characterized in that, The device includes a processor coupled to a memory; the processor is configured to execute a computer program stored in the memory to cause the electronic device to perform the method for determining the consistency of a MEMS tunable Fabry-Perot cavity chip as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, Includes a computer program or instructions that, when run on a computer, cause the computer to perform the method for determining the consistency of a MEMS tunable Fapper cavity chip as described in any one of claims 1-7.

Images