A method for calibrating optical power of a DDMI optical module based on secondary temperature compensation

Abstract

The application relates to the technical field of optical communication, in particular to a light module DDMI light power calibration method based on quadratic temperature compensation, which comprises the following steps: initializing and configuring parameters of a light module and test equipment, collecting standard light power true values and module original DDMI values under different conditions, recording mean values and variances of calibration light power points to remove abnormal data points, and assigning weights to the remaining data points; a quadratic polynomial of the light power true values, the original DDMI values and temperature is established and fitted, and an initial solution of a coefficient vector is calculated; residual errors and standard deviations are calculated and abnormal points are removed; fitting and abnormal point removal are repeatedly performed to obtain an optimal solution of the coefficient vector; and the optimal solution is written into a nonvolatile memory of the light module and used for real-time light power calibration calculation; the application combines a quadratic temperature compensation model, improves the reliability and precision of calibration, completes abnormal processing and data cleaning without manual intervention, and can better adapt to production conditions.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of optical communication technology, in particular to a light module DDMI light power calibration method based on quadratic temperature compensation. BACKGROUND

[0002] Now, the optical modules provided by the suppliers on the market basically can support the digital diagnostic monitoring (DDM) function, can monitor the optical module parameters (such as working temperature, transmitting light power and receiving light power, etc.) in real time, and help the network management unit to predict and locate faults. The light power is an important monitoring parameter for measuring whether the current optical module works normally, and its accuracy and response speed will directly reflect the working condition of the device in the current optical module.

[0003] At present, the digital diagnostic monitoring interface (DDMI) light power calibration for optical modules on the market mainly includes the following three schemes, scheme one is to directly read an actual light power at the current wavelength through the external light power device, and the value is directly written into the Flash storage. Scheme two is to perform interpolation calculation on the monitoring ADC value and the read value of the external light power device based on scheme one, and the monitored ADC value can linearly reflect the change of the DDMI light power of the current optical module. The coefficients obtained by fitting calculation are written into the Flash storage. Scheme three is to improve the interpolation formula based on scheme two, and add quadratic interpolation, so that the light power calibration accuracy is further improved.

[0004] However, the above schemes have their own shortcomings: scheme one is time-saving, simple and fast, and only one set of data is needed for calibration, but the error control is poor, and it is only used in low-precision occasions; scheme two is improved, and only one set of data is needed for calibration, but the light power curve of the optical device cannot be well adapted, that is, there is a large error at the non-calibration point under linear calibration; scheme three uses quadratic terms for calibration, which improves the calibration accuracy to a certain extent, but there is no obvious improvement for the temperature drift under different temperatures and the error of the sampling point itself. SUMMARY

[0005] The purpose of the present application is to solve the problem of decline of fitting accuracy caused by non-uniform distribution of measurement error and abnormal points in the traditional calibration method in the DDMI calibration of the optical module, and the temperature drift characteristics of the light power under different temperatures can also be better captured, thereby improving the calibration accuracy of the entire module under normal working conditions, and providing a light module DDMI light power calibration method based on quadratic temperature compensation.

[0006] In order to achieve the above-mentioned purpose of the application, the embodiments of the present application provide the following technical solutions:

[0007] A light module DDMI light power calibration method based on quadratic temperature compensation, comprising the following steps:

[0008] S1: Initialize the test equipment and configure the parameters. Under multiple different calibration temperature points and different calibration optical power points, continuously measure multiple times, synchronously collect the true optical power value and the original DDMI value of the module, and record the mean and variance of each collected true optical power value in real time.

[0009] S2: Based on the mean and variance obtained in S1, outlier data points are removed, and weights are assigned to the remaining data points. The weights of the data points are negatively correlated with their variances.

[0010] S3: Integrate the data points obtained in S2 into a dataset, establish a quadratic polynomial model of the true value of optical power, the original DDMI value of the module, and the temperature, and construct the feature matrix, observation vector, and weight matrix;

[0011] S4: Use weighted least squares to fit the quadratic polynomial model obtained in S3 with the weights allocated in S2, and calculate the initial solution of the coefficient vector in the quadratic polynomial model; based on the initial solution, traverse the feature matrix, observation vector, and weight matrix to calculate the residuals, and identify and remove outlier data points again according to the standard deviation of the residuals; repeat the steps of fitting and removing outlier data points until the standard deviation of the residuals converges or no new outlier data points are identified, thereby obtaining the optimal solution of the coefficient vector;

[0012] S5: Write the optimal solution of the obtained coefficient vector into the non-volatile memory of the optical module for real-time optical power calibration calculation when the optical module is working.

[0013] In step S1, the testing equipment includes a testing host, an optical module under test, an optical power meter, a temperature control chamber, an optical switch, an adjustable optical attenuator, and a light source. Specifically, S1 includes:

[0014] S11: Initialize the test equipment, where:

[0015] ,

[0016] ,

[0017] Define a list of m calibration temperature points (in °C). express The j-th calibration temperature point in the data, j=1,2,...,m;

[0018] Define a list of n calibration optical power points. express The i-th calibration optical power point, i=1,2,...,n;

[0019] M defines the number of continuous measurements under the condition of each (i, j), M≥3;

[0020] S12: Set multiple calibration temperature points and multiple calibration optical power points, the setting process includes outer loop and inner loop:

[0021] Outer loop: traverse the calibration temperature points , j = 1, 2,..., m; control the temperature control box to set the environment temperature to , and wait for the environment temperature to be stable;

[0022] Inner loop: at the current environment temperature , traverse the calibration optical power points , i = 1, 2,..., n; then control the adjustable optical attenuator or light source to adjust the input optical power to the target value , continuously measure M times, and synchronously record the optical power true value P m (i, j, k) and the module original DDMI value R raw (i, j, k) measured M times, where k = 1, 2,..., M;

[0023] Real-time calculation of the mean and variance of each optical power true value P m (i, j, k) collected:

[0024] ;

[0025] ;

[0026] ;

[0027] ;

[0028] wherein, represents the mean of the i-th calibration optical power true value measured M times at the temperature ; represents the mean of the i-th module original DDMI value measured M times at the temperature ;

[0029] represents the variance of the i-th optical power true value P m (i, j, k) measured M times at the temperature ; represents the variance of the i-th module original DDMI value R raw (i, j, k) measured M times at the temperature .

[0030] In the above scheme, by constructing an automated test platform, intensive data collection is carried out under the conditions of multiple calibration temperature points and calibration optical power points, and the control logic of nested loops is adopted to realize controllable sampling in the temperature-optical power two-dimensional space, which adapts to the conditions for establishing a high-precision calibration model; at the same time, by continuously measuring multiple times and calculating the mean and variance, the stability of each data point is quantitatively evaluated in the data collection stage.

[0031] In S2, for each calibration optical power point , if the variance of M measurements is greater than a preset threshold, it is judged that the data point is unstable, and re-measurement or rejection of abnormal data points is performed; wherein the list of calibration optical power points after re-measurement or rejection of abnormal data points is .

[0032] In the above scheme, a preliminary abnormal data point judgment criterion based on a preset variance threshold is provided, which can reject abnormal data points in the preliminary processing process, thereby reducing the influence of the error of the sampling point itself on the final calibration coefficient.

[0033] In S3, the list and the list are traversed, and the data points obtained in S2 are integrated into a data set, a list of m calibration temperature points is defined, denotes the jth calibration temperature point in , j = 1, 2,..., m; a list of n1 calibration optical power points is defined, denotes the i th calibration optical power point in , , n1 ≤ n; n1 is the number of calibration optical power points in the list after rejection of abnormal data points; the number of data points in the data set is L1 = n1*m, and the quadratic polynomial model constructed is:

[0034] ;

[0035] wherein t is the temperature, k0 is the base slope coefficient of R raw , k1 is the first-order coefficient of R raw coupled with temperature, k2 is the second-order coefficient of R raw coupled with temperature, k3 is the first-order offset coefficient of temperature, k4 is the second-order offset coefficient of temperature, and b is the overall constant offset,

[0036] the coefficient vector β can be obtained:

[0037] ;

[0038] ​Wherein, β represents a coefficient vector; T represents a matrix transpose.

[0039] In the above scheme, the constructed quadratic polynomial model can accurately describe the non-linear behavior of the temperature drift of the to-be-tested optical module under different temperature conditions, and high-precision compensation of the temperature drift phenomenon within a certain temperature range is realized.

[0040] In S3, the feature matrix, the observation vector and the weight matrix are specifically:

[0041] The observation vector Y is constructed, and the dimension is L1*1:

[0042] ;

[0043] Wherein, represents the average value of the jth module original DDMI value measured M times at the calibration temperature point , j = 1, 2,..., m; ;

[0044] The feature matrix X is constructed, and the dimension is L1*6: for each data measured at the calibration temperature point and the calibration optical power point , a set of 6-dimensional feature values can be obtained:

[0045] ;

[0046] The L1 sets of feature values obtained above are integrated to obtain the feature matrix X:

[0047] ;

[0048] Wherein, L1 = n1*m, represents the jth calibration temperature point in , j = 1, 2,..., m; represents the average value of the jth module original DDMI value measured M times at the calibration temperature point ;

[0049] The observation vector Y and the feature matrix X are constructed into a matrix equation:

[0050] ;

[0051] The weight of the data point is defined, and the variance of the data point is inversely proportional to the weight, that is, the greater the variance, the lower the reliability, and the smaller the corresponding weight:

[0052] ;

[0053] wherein, denotes the calibration temperature point and the calibration optical power point , and denotes the variance;

[0054] Based on the respective weights of the L1 data points, an L1xL1 diagonal weight matrix W is constructed, wherein the main diagonal elements of the matrix W are the weight values of the data points:

[0055] ;

[0056] wherein, W is an L1*L1 diagonal matrix, w1=w(1,1), w2=w(2,1),..., w L1 =w(n1,m).

[0057] In the above scheme, the calibration problem is converted into a matrix equation form: , which provides a premise for applying the weighted least squares method, and by constructing a diagonal weight matrix W with the reciprocal of the variance as the element, a large weight is given to the data points with low variance and high quality, so that the final calibration coefficient can suppress the influence of the low-quality data points, thereby improving the robustness and accuracy of the model under the inevitable measurement noise.

[0058] In the S4, the solution formula of the weighted least squares method is:

[0059] ;

[0060] The initial solution β of the coefficient vector can be obtained: first :

[0061] ;

[0062] The calculation method of the residual error ε is:

[0063] ;

[0064] The standard deviation calculation formula of the residual error is:

[0065] ;

[0066] wherein, denotes the standard deviation of the residual error, ε ι denotes the residual error of the i-th data point; L1 is the total number of effective data points in the current iteration, i=1,2,...,L1; p is the number of parameters in the model.

[0067] In the above scheme, the specific solution formula of the weighted least squares method and the residual calculation method are provided, so that the core step of "weighted fitting" can be accurately implemented. By calculating the residual, a quantitative tool is provided for further identifying and removing hidden abnormal data points that pass the primary variance screening but have systematic deviations from the overall model.

[0068] The condition for judging a data point as an abnormal data point by the residual is:

[0069] ;

[0070] If an abnormal point is detected, the elements corresponding to the abnormal point in the feature matrix, observation vector and weight matrix are removed, and new feature matrix, observation vector and weight matrix are formed, and data quantity check is performed, if the number of remaining data points L1 is insufficient, the process is terminated.

[0071] In the above scheme, potential abnormal points are dynamically and multiple times identified and removed during the fitting process, forming a dynamic and self-optimized closed-loop calibration process, which enhances the tolerance and exclusion ability of the algorithm for hidden abnormal points. The data quantity check mechanism ensures the reliability of the iteration process and avoids model failure due to insufficient data.

[0072] After the loop is terminated, the optimal solution of the coefficient vector is obtained:

[0073] ;

[0074] Where, β final represents the optimal solution of the coefficient vector; T represents the matrix transpose.

[0075] In the above scheme, the end point and final output of the iterative optimization process, i.e. the optimal solution of the coefficient vector, are clearly defined. After writing the optimal solution of the coefficient vector into the optical module, it can ensure that the original DDMI value and the true value of the optical power of the optical module remain high precision in actual work.

[0076] Compared with the prior art, the beneficial effects of the present application are:

[0077] The present application combines a quadratic temperature compensation model, without the need for separate hardware costs, to accurately compensate for the nonlinear drift characteristics of the optical module internal temperature variation, while the dynamic weighting strategy can automatically reduce the weight of unstable measurement points, and the iterative abnormal data point removal function can effectively filter out bad data points caused by accidental factors, improving the reliability and accuracy of calibration. At the same time, the self-diagnosis and optimization function in the present scheme can predict data quality, without the need for manual intervention to complete abnormal processing and data cleaning, and is suitable for complex and variable production conditions in mass production environment. BRIEF DESCRIPTION OF DRAWINGS

[0078] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some of the embodiments of the present application, and therefore should not be regarded as a limitation on the scope, and for those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0079] Figure 1 Flow chart for the method of embodiment 1 of the present application;

[0080] Figure 2 Connection diagram of the test equipment of the present application.

[0081] Main component symbol explanation: temperature control box 1; optical module to be tested 2; bus 3; test host 4; optical switch 5; adjustable optical attenuator 6; optical power meter 7; light source 8. DETAILED DESCRIPTION

[0082] The technical solutions of the embodiments of the present application will be described clearly and completely in the following with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all. The components of the embodiments of the present application described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of the present application.

[0083] The technical solution is to collect the true value of optical power and the original DDMI value of the module at different calibration temperature points and different calibration optical power points, to eliminate abnormal data points in the collection process through mean and variance calculation, to assign different weights to data points with different variances by using a weight distribution strategy, to obtain a globally optimal calibration coefficient by solving a weighted least squares method, to perform abnormal diagnosis again through residual error estimation and residual standard deviation estimation of the fitted data points, until a set rule is met, to obtain the final converged calibration coefficient and write it into the EEPROM of the optical module.

[0084] Embodiment 1:

[0085] As shown in Figure 1 , 2 , a light module DDMI optical power calibration method based on quadratic temperature compensation includes the following steps:

[0086] S1: initialize and configure parameters of the test equipment, continuously measure multiple times under the condition of multiple different calibration temperature points and different calibration optical power points, synchronously collect the true value of optical power and the original DDMI value of the module, and record the mean and variance of each collected optical power true value in real time.

[0087] The test equipment comprises a temperature control box 1, a light module to be tested 2, an optical switch 5, an adjustable optical attenuator 6, an optical power meter 7, a light source 8 and a test host 4; the light module to be tested 2 is placed in the temperature control box 1, and the light module to be tested 2, the optical switch 5, the adjustable optical attenuator 6 and the optical power meter 7 are connected in sequence; the light source 8 is communicated with the optical switch 5, the adjustable optical attenuator 6 and the optical power meter 7; and the test host 4 sends control instructions to the temperature control box 1, the optical switch 5, the adjustable optical attenuator 6, the optical power meter 7 and the light source 8 through a bus 3.

[0088] Since the light source 8 is communicated with the optical switch 5, the adjustable optical attenuator 6 and the optical power meter 7, the connection mode therebetween is mostly flange plus fiber head, and there is often insertion loss at the interface, resulting in a certain error delta P between the actual reading value of the optical power meter 7 and the true test value, so the light source 8 emits standard optical power P1, and the actual measurement value P2 of the optical power meter 7 at this time is recorded, and then delta P = P1-P2 is the insertion loss value of the entire link. When the power of the light module to be tested is tested subsequently, only the insertion loss delta P needs to be added to the existing reading value to obtain the actual measurement value.

[0089] Initialize the test equipment, wherein:

[0090] ;

[0091] ;

[0092] Define a list of m calibration temperature points (unit: ℃), such as [-10, 0, 25, 45, 70], denotes the jth calibration temperature point in , j = 1, 2,..., m;

[0093] Define a list of n calibration optical power points (unit: dBm), such as [2, 0, -3, -6, -9, -12, -15], denotes the ith calibration optical power point in , i = 1, 2,..., n;

[0094] M defines the number of continuous measurements under each (i, j) condition, and M ≥ 3.

[0095] Set multiple calibration temperature points and multiple calibration optical power points, and the setting process comprises an outer loop and an inner loop:

[0096] Outer loop: traverse the calibration temperature points , j = 1, 2, …, m; control the temperature control box 1 to set the ambient temperature to , and wait for the ambient temperature to stabilize;

[0097] Inner loop: at the current ambient temperature , traverse the calibration optical power points , i = 1, 2, …, n; then control the adjustable optical attenuator 6 or the light source 8 to adjust the input optical power to the target value , continuously measure M times, and synchronously record the true optical power P m (i, j, k) and the module raw DDMI value R raw (i, j, k), where k = 1, 2, …, M.

[0098] Real-time calculation of the mean and variance of each true optical power P m (i, j, k) collected:

[0099] ;

[0100] ;

[0101] ;

[0102] ;

[0103] wherein represents the mean of the i-th calibration optical power true value measured M times at the temperature ; represents the mean of the i-th module raw DDMI value measured M times at the temperature ;

[0104] represents the variance of P m (i, j, k) after the i-th optical power true value is measured M times at the temperature ; represents the variance of R raw (i, j, k) after the i-th module raw DDMI value is measured M times at the temperature .

[0105] S2: based on the mean and variance obtained in S1, remove abnormal data points, and assign weights to the remaining data points, wherein the weight of a data point is negatively correlated with its variance.

[0106] Stability judgment: for each calibration optical power point If either the true variance of the optical power measured in M ​​measurements or the variance of the original DDMI value of the module is greater than twice the corresponding expected accuracy, then the measurement of that data point is considered unstable. The corresponding true value of optical power P m (i,j,k) Measurement is unstable;

[0107] Action performed: The testing equipment automatically alarms and prompts the operator to check the calibration optical power point. The measurement involves the optical path connection, light source stability, or the status of the optical module under test. Data points that are unstable are marked as abnormal and need to be remeasured or directly discarded, and will not be included in the subsequent fitting process.

[0108] After remeasurement or removal of outlier data points, weights are assigned to the remaining data points, with the weights of the data points being negatively correlated with their variances.

[0109] The list of calibrated optical power points (in dBm) after remeasurement or removal of outlier data points is as follows: , express The Middle One calibration optical power point, .

[0110] S3: Integrate the data points obtained in S2 into a dataset and establish the true value of the optical power P. m Original DDMI value R of the module raw A quadratic polynomial model of temperature t is constructed; feature matrix X, observation vector Y, and weight matrix W are constructed.

[0111] Traversing the list and list Then, all data points are integrated into a dataset. Define a list of m calibration temperature points (in °C). express The j-th calibration temperature point in the data, j=1,2,...,m; Define a list of n1 calibration optical power points (in dBm). express The Middle One calibration optical power point, n1≤n, where n1 is the list after removing outlier data points. The number of calibration optical power points in the dataset; the number of data points in this dataset is L1=n1*m.

[0112] A temperature compensation model is established. During the temperature calibration process, at any calibration temperature point and any calibration optical power point, the true value P of the optical power of the optical module under test 2 is obtained. m Compared with the original DDMI value R of the module rawThere is a nonlinear relationship, but the slope K and the intercept B of the relationship will drift with the change of temperature t, This drift relationship is not a simple linear relationship, but presents a certain nonlinearity, so the present application adopts a quadratic polynomial to more accurately describe the relationship between the slope K(t) and the intercept B(t). Based on this idea, the following function relationship is established:

[0113] The slope K(t) is a function of temperature t:

[0114] (1);

[0115] The intercept B(t) is a function of temperature t:

[0116] (2);

[0117] Wherein, k0 represents the basic slope coefficient of R raw , which determines the main linear influence of R raw on P m ; k1 represents the first-order coefficient of R raw coupled with temperature t, which describes the degree of linear change of the slope K(t) with temperature; k2 represents the second-order coefficient of R raw coupled with temperature t, which describes the degree of nonlinear (curved) change of the slope K(t) with temperature; k3 represents the first-order offset coefficient of temperature t, which describes the degree of linear change of the intercept B(t) with temperature, such as the drift with direct current bias; k4 represents the second-order offset coefficient of temperature t, which describes the degree of nonlinear change of the intercept B(t) with temperature; b represents the overall constant offset, which together with other terms constitutes B(t).

[0118] Based on the above function relationship, the relationship between P m and R raw at any temperature t can be derived as:

[0119] (3);

[0120] Substituting equation (1) and equation (2) into equation (3), we get:

[0121] (4);

[0122] Expand the parentheses:

[0123] ;

[0124] The coefficient vector β is:

[0125] ;

[0126] Wherein, β represents the coefficient vector; T represents the matrix transpose.

[0127] The observation vector Y (dimension L1*1) is constructed:

[0128] ;

[0129] wherein, represents the jth calibration temperature point in the calibration temperature set The mean value of the jth optical power true value measurement after M times of measurement at the calibration temperature point , j = 1, 2, …, m; ;

[0130] The feature matrix X (dimension L1*6) is constructed: for each data measured at the calibration temperature point and the calibration optical power point , a set of 6-dimensional feature values can be obtained:

[0131] ;

[0132] The feature matrix X of the feature values of L1 data points can be integrated:

[0133] ;

[0134] wherein, represents the jth calibration temperature point in the calibration temperature set The mean value of the jth module original DDMI value measurement after M times of measurement at the calibration temperature point , j = 1, 2, …, m; ; The observation vector Y and the feature matrix X are constructed into a matrix equation:

[0135] ;

[0136] ;

[0137] The objective is to solve the coefficient vector β so that the predicted value of the model is closest to the true observation vector Y.

[0138] In order to better quantify the reliability of the data points, for the data points that pass the stability check, the data point weight is defined, which is inversely proportional to the variance of the data point , the greater the variance, the lower the reliability, and the smaller the corresponding weight:

[0139] ;

[0140] wherein, represents the calibration temperature point and the calibration optical power point​ the weight under the following formula, denotes the variance;

[0141] Based on the respective weights of the L1 data points, an L1xL1 diagonal weight matrix W is constructed, wherein the main diagonal elements of the matrix W are the weight values of the data points:

[0142] ,

[0143] wherein W is an L1*L1 diagonal matrix, w1=w(1,1), w2=w(2,1),..., w L1 =w(n1,m).

[0144] S4: using the weighted least squares method to fit the quadratic polynomial model obtained in S3 with the weights assigned in S2, calculate the initial solution of the coefficient vector of the quadratic polynomial model; based on the initial solution, traverse the feature matrix, observation vector, and weight matrix to calculate the residual, and according to the standard deviation of the residual, identify and eliminate abnormal data points again; repeat the steps of fitting and eliminating abnormal data points until the standard deviation of the residual converges or no new abnormal data points are identified, thereby obtaining the optimal solution of the coefficient vector.

[0145] wherein the solution formula of the weighted least squares method is:

[0146] ;

[0147] The initial solution of the coefficient vector β first :

[0148] ;

[0149] The calculation method of the residual ε is:

[0150] ;

[0151] The standard deviation of the residual is calculated according to the following formula:

[0152] ;

[0153] wherein denotes the standard deviation of the residual; ε ι denotes the residual of the ith data point; L1 is the total number of effective data points in the current iteration, i=1,2,...,L1; p is the number of parameters in the model, in the present embodiment, p=6, including (k0, k1, k2, k3, k4, b) 6 parameters.

[0154] The condition for judging that a data point is abnormal through the residual is:

[0155] ;

[0156] If an abnormal data point is detected, the element corresponding to the abnormal data point in the feature matrix X, the observation vector Y and the weight matrix W is removed, and a new observation vector Y', a weight matrix W' and a feature matrix X' are formed. At the same time, a data amount check is performed. If the number of remaining data points is insufficient, an error prompt 'calibration failure: insufficient valid data' is given, and the process is terminated. If the number of remaining data points is sufficient, the next round of iterative fitting is performed. In this embodiment, if the number of remaining data points L1<20, an error prompt 'calibration failure: insufficient valid data' is given, and the process is terminated. If the number of remaining data points L1≥20, the next round of iterative fitting is performed.

[0157] When no new abnormal data point is found after this iteration, the loop is terminated, and the optimal solution β of the coefficient vector is obtained. final

[0158]

[0159] S5: The optimal solution of the coefficient vector obtained is written into the non-volatile memory of the optical module, used for real-time optical power calibration calculation of the optical module during operation, and a calibration report is generated, recording key information such as the production code serial number generated in advance by the production line production system, the calibration time, the calibration temperature point, and the final coefficient vector.

[0160] In actual operation of the optical module, the accuracy of the optical module during DDMI optical power calibration is not high, and the calibration accuracy will deviate at different temperatures. In this embodiment, under the condition of multiple different calibration temperature points , different calibration optical power points , the optical power true value P m and the original DDMI value R raw of the module are measured continuously M times (M=5), and the mean and variance of each calibration optical power point collected are recorded in real time. The present application combines a quadratic temperature compensation model, which can accurately compensate for the nonlinear drift characteristics of the optical module internal with temperature changes without separately increasing hardware costs. At the same time, the dynamic weighting strategy can automatically reduce the weight of unstable data points, and the iterative abnormal data point removal function can effectively filter out bad data points caused by accidental factors such as fiber end face pollution and human interference, greatly improving the reliability and accuracy of calibration. At the same time, the automatic diagnosis and optimization function in the present application can predict data quality, and can complete abnormal processing and data cleaning without manual intervention, adapting to complex and variable production conditions in mass production environment.

[0161] ​​The above merely illustrates the specific embodiments of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of the changes or replacements within the technical range disclosed by the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for calibrating optical power of a DDMI optical module based on secondary temperature compensation, characterized in that, Comprise the following steps: S1: initialize and configure parameters of the test equipment, continuously measure multiple times under the condition of multiple different calibration temperature points, different calibration optical power points, synchronously collect the true value of optical power and the original DDMI value of the module, and record the mean and variance of each collected true value of optical power in real time; S2: based on the mean and variance obtained in S1, remove abnormal data points, and assign weights to the remaining data points, wherein the weight of a data point is negatively correlated with its variance; S3: integrate the data points obtained in S2 into a data set, establish a quadratic polynomial model of the true value of optical power, the original DDMI value of the module and temperature, construct a feature matrix, an observation vector and a weight matrix; The S3, traversing the list and list After that, the data points obtained in S2 are integrated into a data set, define a list of m calibration temperature points, represent the jth calibration temperature point in the list, j = 1, 2, …, m; define a list of n1 calibration optical power points, represent the nth calibration optical power point in the list, , n1≤n; n1 is the number of calibration optical power points in the list after removing abnormal data points; the number of data points in the data set is L1 = n1*m, and the quadratic polynomial model constructed is:​ ； Where t is temperature, and k0 is R raw The basic slope coefficient, k1 is R raw The first-order coefficients coupled with temperature, k2 is R raw The second-order coefficients coupled with temperature, k3 is the first-order temperature offset coefficient, k4 is the second-order temperature offset coefficient, and b is the global constant offset. The coefficient vector β can be obtained: ； Wherein, β represents the coefficient vector; T represents matrix transposition; In S3, the construction of the feature matrix, the observation vector and the weight matrix is as follows: An observation vector Y with a dimension of L1*1 is constructed: ； wherein represents the mean value of the jth light power true value measurement after m times, j = 1, 2,..., m at a temperature of ;​ The characteristic matrix X is constructed, dimension L1*6: for each of the data measured at the calibration temperature points and the calibration optical power points a set of 6-dimensional characteristic values is obtained: ； The L1 groups of feature values obtained above are integrated to obtain a feature matrix X: ； wherein L1 = n1*m, represents the jth calibration temperature point in represents the jth calibration temperature point in the average of the original DDMI values of the jth module measured m times at the temperature of ;​​ The observation vector Y and the feature matrix X are constructed into a matrix equation: ； The weight of a data point is inversely proportional to the variance of the data point is defined as the variance of the data point The greater the variance, the lower the reliability and the smaller the corresponding weight. ； wherein, represents the weight at the calibration temperature point and the calibration optical power point represents the variance;​ Based on the weight of each of the L1 data points, a L1*L1 diagonal weight matrix W is constructed, wherein the main diagonal elements of the matrix W are the weight values of the data points: ； where W is a diagonal matrix of L1*L1, w1 = w(1,1), w2 = w(2,1),..., wn = w(n,1), and L1 = w(n1,m); S4: using the weighted least squares method, the quadratic polynomial model obtained in S3 is fitted with the weights assigned in S2 to calculate the initial solution of the coefficient vector of the quadratic polynomial model; based on the initial solution, the residual error is calculated based on the feature matrix, the observation vector and the weight matrix, and the abnormal data points are identified and removed again according to the standard deviation of the residual error; the steps of fitting and removing abnormal data points are repeated until the standard deviation of the residual error converges or no new abnormal data points are identified, so as to obtain the optimal solution of the coefficient vector; S5: write the optimal solution of the coefficient vector obtained into the non-volatile memory of the optical module, which is used for real-time calibration and calculation of optical power when the optical module is working.

2. The optical module DDMI optical power calibration method according to claim 1, wherein, In the S1, the test equipment comprises a test host, a to-be-tested optical module, an optical power meter, a temperature control box, an optical switch, an adjustable optical attenuator and a light source, and S1 specifically comprises: S11: initialize the test equipment, wherein: ， ， Define a list of m calibration temperature points. express The j-th calibration temperature point in the data, j=1,2,...,m; defining a list of n calibration optical power points, denoting the i-th calibration optical power point in the list, i = 1, 2,..., n; M defines the number of consecutive measurements under each (i, j) condition, and M≥3; S12: set multiple calibration temperature points and multiple calibration optical power points, and the setting process comprises an outer loop and an inner loop: Outer loop: traverse the calibration temperature points , j = 1, 2, …, m; control the temperature control box to set the environment temperature to , wait for the environment temperature to be stable; Inner loop: at current ambient temperature , traverse the calibration optical power point , i = 1, 2,..., n; then control the adjustable optical attenuator or light source to adjust the input optical power to the target value , measure M times in succession, and record the true optical power P of the M times of measurement synchronously m (i, j, k) and the module raw DDMI value R raw (i, j, k), where k = 1, 2,..., M; Real-time computation of each light power true value P m Mean and variance of (i,j,k): ； ； ； ； wherein, represents the mean value of the i-th calibration optical power true value measured M times at the temperature represents the mean value of the i-th module original DDMI value measured M times at the temperature represents the mean value of the i-th calibration optical power true value measured M times at the temperature represents the mean value of the i-th module original DDMI value measured M times at the temperature represents the i-th light power true value measurement M times after P at temperature T, the i-th module original DDMI value measurement M times after R m the variance of (i,j,k). represents the i-th light power true value measurement M times after P at temperature T, the i-th module original DDMI value measurement M times after R raw the variance of (i,j,k).

3. The optical module DDMI optical power calibration method according to claim 2, wherein, In S2, for each calibration optical power point If the variance of the M measurements is greater than a preset threshold, it is determined that the data point is unstable, and re-measurement or elimination of abnormal data points is performed; wherein the list of calibration optical power points after re-measurement or elimination of abnormal data points is .

4. The optical module DDMI optical power calibration method of claim 1, wherein, In the S4, the solution formula of the weighted least squares method is: ； The initial solution of the available coefficient vector β first : ； The calculation method of the residual error ε is: ； The standard deviation calculation formula of the residual error is: ； wherein denotes the standard deviation of the residuals, ε ι denotes the residual of the i-th data point; L1is the total number of valid data points in the current iteration, i = 1, 2,..., L1; p is the number of parameters in the model.

5. The optical module DDMI optical power calibration method according to claim 4, wherein, The condition for judging that a data point is an abnormal data point through the residual error is: ； If an abnormal point is detected, the elements corresponding to the abnormal point in the feature matrix, the observation vector and the weight matrix are removed, and new feature matrix, observation vector and weight matrix are formed, and data amount checking is performed, if the number of remaining data points L1 is insufficient, the process is terminated.

6. The optical module DDMI optical power calibration method according to claim 5, wherein, After the loop is terminated, the optimal solution of the coefficient vector is obtained: ； where β final denotes the optimal solution of the coefficient vector; T denotes the matrix transpose.

Images