High-power laser aging on-line test system and method

Abstract

The invention relates to the technical field of aging testing of lasers, in particular to an online aging testing system and method for a high-power laser, and the method comprises the steps: collecting the illumination intensity data of a processed laser through an illumination intensity sensor; the method specifically comprises the following steps: dynamically attenuating the output power of a laser to avoid measurement distortion caused by thermal saturation of an attenuation sheet; constructing a spectrum adaptation model, and expanding the spectral response range of the illumination intensity sensor based on an inherent spectral response edge signal and cross-band relevance of the sensor; acquiring external temperature data of the laser through a temperature sensor to obtain a heat conductivity coefficient, and calculating internal temperature data of the laser according to the heat conductivity coefficient and the external temperature data; preprocessing the acquired illumination intensity data and internal temperature data to form a laser parameter data set, and transmitting the laser parameter data set to an upper computer; and the light attenuation process of the laser is obtained by analyzing data in the laser parameter data set.

Description

Technical Field

[0001] The present invention relates to the technical field of laser aging testing, and in particular to a high-power laser aging online testing system and method. Background Art

[0002] As a light source, the stability of the output optical power of the laser in the field of optical fiber communication directly affects the communication quality. Optical attenuation will cause the optical power to drop, which in turn affects the transmission distance of the signal and the signal-to-noise ratio of the receiving end. Therefore, it is necessary to monitor and measure the optical attenuation of the laser to ensure its optical power output within the normal working range and to ensure the stable operation of the communication system. By measuring the optical attenuation of the laser, we can understand its performance change trend during use, so as to evaluate its reliability and predict its life. For example, the optical performance of a COS laser tube may gradually decline during long-term use, resulting in a decrease in output optical power and changes in spectral characteristics. The aging test is designed to simulate the actual use conditions and accelerate the aging process of the laser tube so as to obtain information on its performance degradation in a short time, thereby evaluating its reliability and life.

[0003] The existing technology usually uses an attenuation plate to attenuate the laser energy. This single fixed attenuation method has poor stability, especially when the attenuation plate is saturated, resulting in measurement distortion. Summary of the invention

[0004] The present invention expands the spectral response range of the light intensity sensor through an algorithm, considers the nonlinear relationship between the actual spectrum and the channel response, introduces nonlinear terms in the relationship function, and infers the actual light intensity data when the attenuation sheet is saturated.

[0005] The technical solution proposed by the present invention is: a high-power laser aging online testing method, the method comprising:

[0006] The light intensity data of the laser after processing is collected by the light intensity sensor; specifically, the output power of the laser is dynamically attenuated to avoid measurement distortion caused by thermal saturation of the attenuation plate; a spectral adaptation model is constructed to expand the spectral response range of the light intensity sensor based on the edge signal and cross-band correlation of the sensor's inherent spectral response, thereby obtaining light intensity data beyond the original spectral response range of the light intensity sensor;

[0007] The external temperature data of the laser is collected by a temperature sensor to obtain the thermal conductivity, and the internal temperature data of the laser is calculated using the thermal conductivity and the external temperature data;

[0008] After preprocessing the acquired light intensity data and internal temperature data, a laser parameter data set is formed and transmitted to the host computer;

[0009] By analyzing the data in the laser parameter data set, the laser light decay process is obtained.

[0010] Preferably, the dynamically attenuating the output power of the laser to avoid measurement distortion caused by thermal saturation of the attenuation sheet includes:

[0011] Control the output power of the laser through laser pulse width modulation;

[0012] Construct a dynamic energy adjustment matrix to dynamically adjust the laser energy output by the laser;

[0013] Specifically include the following steps:

[0014] Assume that the thermal saturation influence matrix is The diagonal elements in the thermal saturation influence matrix represent the influence of thermal saturation of thermal attenuation sheets corresponding to different laser channels on laser energy attenuation, and the non-diagonal elements represent the thermal saturation coupling influence between different laser channels, where n represents the number of laser channels;

[0015] Then, when the attenuation plate reaches thermal saturation, the laser energy after passing through the attenuation plate is:

[0016] in, They represent the laser energy after being affected by the thermal saturation of the attenuator and the initial laser energy respectively;

[0017] Building a dynamic adjustment matrix Among them, d mm It indicates the energy adjustment ratio of the mth laser in the mth time slice during the laser pulse width adjustment process, where m indicates the number of time slices;

[0018] The adjusted energy is: Wherein, P0 represents the laser initial pulse width adjustment matrix.

[0019] Preferably, the constructing of the spectral adaptation model, based on the edge signal and cross-band correlation of the inherent spectral response of the sensor, expands the spectral response range of the light intensity sensor, including:

[0020] Construct spectral correlation matrix;

[0021] Constructing a relational function for spectral inversion to calculate and obtain spectral values ​​that exceed the corresponding range of the sensor; the spectral adaptation model includes a spectral correlation matrix and a relational function;

[0022] The constructing of the spectral correlation matrix comprises:

[0023] Get the number of channels and the central wavelength of each channel to form a wavelength sequence λ = (λ1, λ2, ..., λ n );

[0024] Determine the range of the target spectrum to be inverted, wherein the range of the target spectrum includes a spectrum lower than the lowest central wavelength Δλ of the laser channel and higher than the highest central wavelength Δλ of the laser channel;

[0025] Obtain known standard spectral data with a wide spectral coverage from the database, obtain spectral values ​​at the central wavelength of each channel and multiple wavelengths within the target spectral range, and set the sampling number k. The value of the standard spectral data at the corresponding wavelength is S = [s1, s2, ..., s k ] T ;

[0026] Calculate the correlation coefficient between the light intensity sensor corresponding to each channel and the spectral value at the sampling wavelength Among them, R it represents the response value of the i-th channel in the t-th time, is the mean of the response values ​​of the ith channel, Represents the mean of the spectral values ​​at the jth sampling wavelength;

[0027] With the correlation coefficient as the element, the spectral correlation matrix A = [ρ ij ].

[0028] Preferably, the construction of the relationship function for spectral inversion to calculate and obtain spectral values ​​beyond the corresponding range of the sensor includes:

[0029] Assume the target spectrum is X = [x1, x2, ..., x l ] T , where x l Indicates the spectral value at the lth wavelength in the target spectral range;

[0030] Establish the relationship function AX = B, where B = [b1, b2, b i , …, b u ] T , where a ij is the element in the spectral correlation matrix, s j Represents the value of the standard spectrum data at the jth sampling wavelength;

[0031] Using the least squares method to solve the relationship function, the solution obtained is: X = (A T A) T B;

[0032] If the nonlinear relationship between the actual spectrum and the channel response is considered, a nonlinear term is introduced into the relationship function;

[0033] By fitting the relationship function with a polynomial, we can obtain the relationship function in the form of a quadratic polynomial:

[0034] Among them, c j0 , c ji , c ijk represents the unknown coefficient, r i Represents the response value of the i-th channel;

[0035] The value of the unknown coefficient is calculated by using the known standard spectrum data and the corresponding channel response value of the sensor using the curve fitting algorithm;

[0036] Substitute the known edge spectral response value of the light intensity sensor into the relationship function to calculate the spectral value beyond the response range of the light intensity sensor.

[0037] Preferably, the constructing of the spectral adaptation model, based on the edge signal and cross-band correlation of the inherent spectral response of the sensor, expands the spectral response range of the light intensity sensor, and further includes:

[0038] The spectral adaptation model is constructed through convolutional neural network, which inputs low-spectral data and outputs high-spectral data; specifically, it includes:

[0039] Obtain edge wavelength data in the spectral response range of the light intensity sensor, that is, the lowest response wavelength and the highest response wavelength in a band, to form a low-spectrum data set X LOW ;

[0040] Acquire wavelength data beyond the preset spectral response range of the light intensity sensor to form a hyperspectral dataset X HIG ;

[0041] Normalize the data in the low-spectral dataset and the high-spectral dataset to form a sample dataset;

[0042] The sample data set is used to train the convolutional neural network, and the trained convolutional neural network is used to output the predicted hyperspectral data, that is, the extended spectral response range of the light intensity sensor in a band is [λ i,min -Δλ′,λ i,max +Δλ′], where λ i,min , i,max represent respectively the minimum wavelength and the maximum wavelength to which the light intensity sensor responds in the i-th band, wherein the band includes the visible light band, the infrared band and the ultraviolet band;

[0043] The convolutional neural network includes an input layer, a hidden layer and an output layer. The number of neurons in the input layer matches the number of data in the low-spectrum data set, and low-spectrum data of one band is received through neurons; the number of neurons in the output layer matches the number of data in the high-spectrum data set;

[0044] The neural network is optimized using the Adam optimizer.

[0045] Preferably, collecting external temperature data of the laser through a temperature sensor to obtain thermal conductivity, and calculating and obtaining internal temperature data of the laser using the thermal conductivity and external temperature data, comprises:

[0046] Construct a three-dimensional transient thermal model and use the inverse heat conduction algorithm to derive the three-dimensional transient temperature field inside the laser. The specific steps are as follows:

[0047] Acquire the laser structure data and the structure data of all the internal components from the database; construct a three-dimensional model of the laser according to the acquired laser structure data and the structure data of all the internal components;

[0048] Construct the governing equations for three-dimensional transient heat conduction:

[0049] Where α′ is the thermal diffusion coefficient;

[0050] The three-dimensional model of the laser is discretized by the finite difference method, and the object is divided into multiple three-dimensional units, each of which has a size of ΔX, ΔY, and ΔZ; ΔT represents the length of the time step;

[0051] Use the central difference method to discretize time and space to obtain:

[0052]

[0053] in, represents the temperature at the position (I, J, K) at the wth time step;

[0054] Set boundary conditions; forward solve the three-dimensional transient heat conduction equation based on the surface temperature and boundary conditions measured by the temperature sensor to obtain the temperature distribution inside the object;

[0055] Calculating the sensitivity of the temperature field to the parameters of the three-dimensional transient thermal model, wherein the parameters of the three-dimensional transient thermal model include the laser surface material information and boundary conditions;

[0056] Using the surface temperature and sensitivity of the laser measured by the temperature sensor, the model parameters are estimated by the least square method or the gradient descent method;

[0057] Using the estimated model parameters, the three-dimensional transient heat conduction equation is solved to obtain the temperature distribution inside the laser, that is, the temperature field distribution.

[0058] Preferably, the step of acquiring the laser light attenuation process by analyzing the data in the laser parameter data set includes:

[0059] Extract light intensity data, arrange the measured light data in chronological order, and form a light-time variation curve; the light attenuation trend of the laser is reflected through the light-time variation curve;

[0060] Establish a relationship model between various materials and chips inside the laser tube and the light attenuation of the laser tube, and set up a correction matrix;

[0061] Based on the correction matrix and the aging factor of the corresponding material, analyze the materials and chips that have the greatest impact on light decay; specifically, the following steps are included:

[0062] Assume that the laser tube contains z different materials or chips, represented by M1, M2, ..., M k express;

[0063] Construct the material aging factor vector Among them, μ i Indicates material or chip M i The aging factor, the larger the value of the aging factor, the greater the impact of the aging degree of the material or chip on the performance of the laser tube;

[0064] Establish the relationship between laser output power and aging factor in, represents the output power vector of the laser tube, Where u represents the number of power measurement points, that is, the number of laser tubes; p lu Represents the output power of the laser tube measured at the uth measurement point; represents the original output power, that is, the theoretical output power when not affected by material aging; C represents the coefficient matrix of factors affecting the output power due to material aging factors;

[0065] Constructing the Correction Matrix Where I represents the unit matrix, k′ represents the adjustable coefficient, which is used to adjust the correction amplitude; the correction matrix satisfies the following relationship:

[0066] When the output power of the laser tube is corrected to 100%, it corresponds to the theoretical output power have

[0067] That is, the corrected value is:

[0068]

[0069] Observe the coefficients of the items related to different materials in the above formula, and the absolute values ​​of the coefficients reflect the effect of the aging of the corresponding materials on the laser light attenuation.

[0070] Preferably, it includes constructing an edge computing resource calling matrix, calling edge computing resources according to the computing demand, and realizing real-time temperature-light intensity coupling analysis on the local FPGA, which specifically includes the following steps:

[0071] Obtain edge computing resource data, the edge computing data including CPU resources, memory resources, and storage resources;

[0072] Obtain the CPU resource usage, memory usage, and storage resource usage of the host computer;

[0073] Constructing edge computing resource call matrix where d 11 ,d 12 ,d 13 Respectively represent the call intensity of CPU, memory and storage resources when the amount of computation is low; d 21 ,d 22 ,d 23 Indicates the intensity of the computation on CPU, memory, and storage resources; d 31 ,d 32 ,d 33 Indicates the intensity of high computing workload on CPU, memory and storage resources;

[0074] The CPU resource occupancy rate, memory occupancy rate and storage resource occupancy rate of the host computer are used to determine the computational workload of the received computing task and to allocate the CPU, memory and storage resources in the edge computing device according to the computational workload;

[0075] The monitoring module in the FPGA or the interaction with the operating system monitors the utilization rate of the CPU, memory and storage resources and the progress of task execution.

[0076] A high-power laser aging online test system, the test system includes multiple acquisition modules, an acquisition control unit, a cooling unit, a host computer and a cabinet, the acquisition module, the acquisition control unit and the cooling unit are all arranged in the cabinet, the acquisition control unit is connected to the host computer data; the cooling unit is used to reduce the internal temperature of the cabinet, the acquisition module includes a laser lamp holder, a light intensity sensor, an attenuation sheet and a temperature sensor, and the light intensity sensor is connected to the acquisition control unit. In this embodiment, the cooling unit adopts water cooling to adjust the temperature.

[0077] A computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement the visual recognition-based cooking unattended reminder method.

[0078] Beneficial effects of the present invention:

[0079] 1. The present invention adjusts the laser intensity through dynamic attenuation of laser pulse width modulation, realizes closed-loop control of optical power on the basis of existing fixed attenuation, and can avoid measurement distortion caused by thermal saturation of attenuation plates in traditional solutions.

[0080] 2. The present invention establishes a matrix that can reflect the spectral correlation between the spectra of each channel, and constructs a relationship function to realize the inversion of the spectrum that exceeds the response range of the light intensity sensor through the known sensor edge spectral response, thereby expanding its response range without replacing the sensor, and solving the problem that the actual light intensity cannot be accurately measured when the attenuation plate is saturated. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] Figure 1 The present invention is a flowchart of a high-power laser aging online testing method. DETAILED DESCRIPTION

[0082] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art can think of other obvious variations. The basic principles of the present invention defined in the following description can be applied to other embodiments, variations, improvements, equivalents, and other technical solutions that do not deviate from the spirit and scope of the present invention.

[0083] It is to be understood that the term "one" should be understood as "at least one" or "one or more", that is, in one embodiment, the number of an element may be one, while in another embodiment, the number of the element may be multiple, and the term "one" should not be understood as a limitation on the quantity.

[0084] Embodiment 1:

[0085] refer to Figure 1 The technical solution provided by the present invention is: a high-power laser aging online testing method, comprising the following steps:

[0086] Step 1. Collect processed light intensity data of the laser through a light intensity sensor; specifically, the following steps are performed: dynamically attenuating the output power of the laser to avoid measurement distortion caused by thermal saturation of the attenuation plate; constructing a spectral adaptation model to expand the spectral response range of the light intensity sensor based on the edge signals and cross-band correlation of the sensor's inherent spectral response, thereby obtaining light intensity data beyond the original spectral response range of the light intensity sensor.

[0087] The output power of the laser is dynamically attenuated to avoid measurement distortion caused by thermal saturation of the attenuator, including:

[0088] Step 1.1.1, controlling the output power of the laser by laser pulse width modulation;

[0089] Step 1.1.2, construct a dynamic energy adjustment matrix to dynamically adjust the laser energy output by the laser. Based on the existing fixed attenuation through the attenuator, the closed-loop control of the optical power can be realized to avoid the measurement distortion caused by the thermal saturation of the attenuator.

[0090] Step 1.1.2 includes the following steps:

[0091] Assume that the thermal saturation influence matrix is The diagonal elements in the thermal saturation influence matrix represent the influence of thermal saturation of thermal attenuation sheets corresponding to different laser channels on laser energy attenuation, and the non-diagonal elements represent the thermal saturation coupling influence between different laser channels, where n represents the number of laser channels;

[0092] Then, when the attenuation plate reaches thermal saturation, the laser energy after passing through the attenuation plate is:

[0093] in, They represent the laser energy after being affected by the thermal saturation of the attenuator and the initial laser energy respectively;

[0094] Building a dynamic adjustment matrix Among them, d mm It indicates the energy adjustment ratio of the mth laser in the mth time slice during the laser pulse width adjustment process, where m indicates the number of time slices;

[0095] The adjusted energy is: Wherein, P0 represents the laser initial pulse width adjustment matrix.

[0096] A spectral adaptation model is constructed to expand the spectral response range of the light intensity sensor based on the edge signals and cross-band correlation of the sensor's inherent spectral response.

[0097] For example, if the target range is 0.4μm-0.8μm, and the sensor channel range is 0.5μm-0.7μm, how to expand the channel range of the light intensity sensor without replacing the light intensity sensor. This can be achieved by the following steps:

[0098] Step 1.2.1, construct the spectral correlation matrix;

[0099] Step 1.2.2, construct a relationship function for spectral inversion, and calculate and obtain spectral values ​​that exceed the corresponding range of the sensor; the spectral adaptation model includes a spectral correlation matrix and a relationship function.

[0100] Among them, constructing a spectral correlation matrix and establishing cross-band correlation includes the following steps:

[0101] Get the number of channels and the central wavelength of each channel to form a wavelength sequence λ = (λ1, λ2, ..., λ n ); The so-called channel refers to the measurable band of the light intensity sensor, including the visible light band, near infrared band, ultraviolet band and far infrared band, and each band corresponds to a channel. For example, for a multispectral sensor with 5 channels, the central wavelengths are 0.45μm, 0.55μm, 0.65μm, 0.85μm and 1.2μm, respectively, and the wavelength range covers the visible light to near infrared band.

[0102] Determine the range of the target spectrum to be inverted, wherein the range of the target spectrum includes spectra that are a certain wavelength Δλ lower than the lowest center wavelength of the laser channel and a certain wavelength Δλ higher than the highest center wavelength of the laser channel; that is, the corresponding range of the Δλ wavelength is extended based on the edge response signal of the spectrum of the light intensity sensor.

[0103] Obtain known standard spectral data with a wide spectral coverage from the database, obtain spectral values ​​at the central wavelength of each channel and multiple wavelengths within the target spectral range, and set the sampling number k. The value of the standard spectral data at the corresponding wavelength is S = [s1, s2, ..., s k ] T ;

[0104] Calculate the correlation coefficient between the light intensity sensor corresponding to each channel and the spectral value at the sampling wavelength Among them, R it represents the response value of the ith channel in the tth time, R i is the mean of the response values ​​of the ith channel, Represents the mean of the spectral values ​​at the jth sampling wavelength;

[0105] With the correlation coefficient as the element, the spectral correlation matrix A = [ρ ij ].

[0106] Among them, building a relationship function for spectral inversion and calculating and obtaining spectral values ​​beyond the corresponding range of the sensor include the following steps:

[0107] Assume the target spectrum is X = [x1, x2, ..., x l ] T , where x l Indicates the spectral value at the lth wavelength in the target spectral range;

[0108] Establish the relationship function AX = B, where B = [b1, b2, b i , …, b u ] T , where a ijis the element in the spectral correlation matrix, s j Represents the value of the standard spectrum data at the jth sampling wavelength;

[0109] Using the least squares method to solve the relationship function, the solution obtained is: X = (A T A) T B;

[0110] If the nonlinear relationship between the actual spectrum and the channel response is considered, a nonlinear term is introduced into the relationship function;

[0111] By fitting the relationship function with a polynomial, we can obtain the relationship function in the form of a quadratic polynomial:

[0112] Among them, c j0 , c ji , c ijk represents the unknown coefficient, r i Represents the response value of the i-th channel;

[0113] The value of the unknown coefficient is calculated by using the known standard spectrum data and the corresponding channel response value of the sensor using the curve fitting algorithm;

[0114] Substitute the known edge spectral response value of the light intensity sensor into the relationship function to calculate the spectral value beyond the response range of the light intensity sensor.

[0115] Step 2: Collect the external temperature data of the laser through the temperature sensor to obtain the thermal conductivity, and use the thermal conductivity and external temperature data to calculate the internal temperature data of the laser. Specifically, the following steps are included:

[0116] Construct a three-dimensional transient thermal model and use the inverse heat conduction algorithm to derive the three-dimensional transient temperature field inside the laser. The specific steps are as follows:

[0117] Acquire the laser structure data and the structure data of all the internal components from the database; construct a three-dimensional model of the laser according to the acquired laser structure data and the structure data of all the internal components;

[0118] Construct the governing equations for three-dimensional transient heat conduction:

[0119] Where α′ is the thermal diffusion coefficient;

[0120] The three-dimensional model of the laser is discretized by the finite difference method, and the object is divided into multiple three-dimensional units, each of which has a size of ΔX, ΔY, and ΔZ; ΔT represents the length of the time step;

[0121] Use the central difference method to discretize time and space to obtain:

[0122]

[0123] in, represents the temperature at the position (I, J, K) at the wth time step;

[0124] Set boundary conditions; forward solve the three-dimensional transient heat conduction equation based on the surface temperature and boundary conditions measured by the temperature sensor to obtain the temperature distribution inside the object;

[0125] Calculating the sensitivity of the temperature field to the parameters of the three-dimensional transient thermal model, wherein the parameters of the three-dimensional transient thermal model include the laser surface material information and boundary conditions;

[0126] Using the surface temperature and sensitivity of the laser measured by the temperature sensor, the model parameters are estimated by the least square method or the gradient descent method;

[0127] Using the estimated model parameters, the three-dimensional transient heat conduction equation is solved to obtain the temperature distribution inside the laser, that is, the temperature field distribution.

[0128] Step 3: After preprocessing the acquired light intensity data and internal temperature data, a laser parameter data set is formed and transmitted to the host computer;

[0129] Step 4: Obtain the laser light decay process by analyzing the data in the laser parameter data set. This includes the following steps:

[0130] Extract light intensity data, arrange the measured light data in chronological order, and form a light-time variation curve; the light attenuation trend of the laser is reflected through the light-time variation curve;

[0131] Establish a relationship model between various materials and chips inside the laser tube and the light attenuation of the laser tube, and set up a correction matrix;

[0132] Based on the correction matrix and the aging factor of the corresponding material, analyze the materials and chips that have the greatest impact on light decay; specifically, the following steps are included:

[0133] Assume that the laser tube contains z different materials or chips, represented by M1, M2, ..., M k express;

[0134] Construct the material aging factor vector Among them, μ i Indicates material or chip M i The aging factor, the larger the value of the aging factor, the greater the impact of the aging degree of the material or chip on the performance of the laser tube;

[0135] Establish the relationship between laser output power and aging factor in, represents the output power vector of the laser tube, Where u represents the number of power measurement points, that is, the number of laser tubes; p lu Represents the output power of the laser tube measured at the uth measurement point; represents the original output power, that is, the theoretical output power when not affected by material aging; C represents the coefficient matrix of factors affecting the output power due to material aging factors;

[0136] Constructing the Correction Matrix Where I represents the unit matrix, k′ represents the adjustable coefficient, which is used to adjust the correction amplitude; the correction matrix satisfies the following relationship:

[0137] When the output power of the laser tube is corrected to 100%, it corresponds to the theoretical output power have

[0138] That is, the corrected value is:

[0139]

[0140] Observe the coefficients of the items related to different materials in the above formula, and the absolute values ​​of the coefficients reflect the effect of the aging of the corresponding materials on the laser light attenuation.

[0141] For example, if a laser tube is made of two materials, there are 3 measurement points;

[0142] The aging factor vector is

[0143] Assume k′ = 0.1,

[0144] The correction matrix is:

[0145]

[0146] From the above results, we can see that the coefficient related to the first material is the largest, so it can be judged that the aging of the first material has a greater impact on light decay. Improvements can be made to the corresponding materials to increase the life of the laser.

[0147] Embodiment 2:

[0148] The difference between the illumination intensity of the second embodiment and the first embodiment is that:

[0149] 1. When expanding the spectral response range of the light intensity sensor, the sampling convolutional neural network constructs a spectral adaptation model, which specifically includes the following steps:

[0150] Obtain edge wavelength data in the spectral response range of the light intensity sensor, that is, the lowest response wavelength and the highest response wavelength in a band, to form a low-spectrum data set X LOW ;

[0151] Acquire wavelength data beyond the preset spectral response range of the light intensity sensor to form a hyperspectral dataset X HIG ;

[0152] Normalize the data in the low-spectral dataset and the high-spectral dataset to form a sample dataset;

[0153] The sample data set is used to train the convolutional neural network, and the trained convolutional neural network is used to output the predicted hyperspectral data, that is, the extended spectral response range of the light intensity sensor in a band is [λ i,min -Δλ′,λ i,max +Δλ′], where λ i,min , i,max represent respectively the minimum wavelength and the maximum wavelength to which the light intensity sensor responds in the i-th band, wherein the band includes the visible light band, the infrared band and the ultraviolet band;

[0154] The convolutional neural network includes an input layer, a hidden layer and an output layer. The number of neurons in the input layer matches the number of data in the low-spectrum data set, and low-spectrum data of one band is received through neurons; the number of neurons in the output layer matches the number of data in the high-spectrum data set;

[0155] The neural network is optimized using the Adam optimizer.

[0156] 2. Based on the four steps in Example 1, a step of improving technical efficiency is added to reduce the impact of network bandwidth on computing efficiency during online measurement. Specifically:

[0157] Construct an edge computing resource call matrix, call edge computing resources according to the computing requirements, and implement real-time temperature-light intensity coupling analysis on the local FPGA. The specific steps include:

[0158] Obtain edge computing resource data, the edge computing data including CPU resources, memory resources, and storage resources;

[0159] Obtain the CPU resource usage, memory usage, and storage resource usage of the host computer;

[0160] Constructing edge computing resource call matrix where d 11 d 12 d 13Respectively represent the call intensity of CPU, memory and storage resources when the amount of computation is low; d 21 ,d 22 ,d 23 Indicates the intensity of the computation on CPU, memory, and storage resources; d 31 ,d 32 ,d 33 Indicates the intensity of high computing workload on CPU, memory and storage resources;

[0161] The CPU resource occupancy rate, memory occupancy rate and storage resource occupancy rate of the host computer are used to determine the computational workload of the received computing task and to allocate the CPU, memory and storage resources in the edge computing device according to the computational workload;

[0162] The monitoring module in the FPGA or the interaction with the operating system monitors the utilization rate of the CPU, memory and storage resources and the progress of task execution.

[0163] In the embodiments disclosed by the present invention, the processes described above with reference to the flowchart can be implemented as a computer software program. The embodiments disclosed by the present invention include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part, and / or installed from a removable medium. When the computer program is executed by the central processing unit (CPU), the above functions defined in the method of the present invention are executed. It should be noted that the above-mentioned computer-readable medium of the present invention can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media can include, but are not limited to: an electrical connection with one or more wire segments, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device, or device. In the present invention, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in combination with an instruction execution system, device, or device. The program code contained on the computer-readable medium may be transmitted using any suitable medium, including but not limited to: wireless segments, wire segments, optical cables, RF, etc., or any suitable combination of the above.

[0164] The flowchart in the accompanying drawings illustrates the possible architecture, functions and operations of the methods according to various embodiments of the present invention. In this regard, each box in the flowchart or block diagram may represent a module, a program segment or a part of a code, and the module, a program segment or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box may also occur in a different order from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they may sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or the flowchart, and the combination of boxes in the block diagram and / or the flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0165] Those skilled in the art should understand that the embodiments of the present invention described above and shown in the accompanying drawings are only examples and do not limit the present invention. The objectives of the present invention have been fully and effectively achieved. The functions and structural principles of the present invention have been demonstrated and explained in the embodiments. Without departing from the principles, the implementation methods of the present invention may be subject to any changes or modifications.

Claims

1. A high-power laser aging online testing method, characterized in that: The method comprises: The light intensity data of the laser after processing is collected by the light intensity sensor; specifically, the output power of the laser is dynamically attenuated to avoid measurement distortion caused by thermal saturation of the attenuation plate; a spectral adaptation model is constructed to expand the spectral response range of the light intensity sensor based on the edge signal and cross-band correlation of the sensor's inherent spectral response, thereby obtaining light intensity data beyond the original spectral response range of the light intensity sensor; The external temperature data of the laser is collected by a temperature sensor to obtain the thermal conductivity, and the internal temperature data of the laser is calculated using the thermal conductivity and the external temperature data; After preprocessing the acquired light intensity data and internal temperature data, a laser parameter data set is formed and transmitted to the host computer; By analyzing the data in the laser parameter data set, the laser light decay process is obtained.

2. A high-power laser aging online testing method according to claim 1, characterized in that: The method of dynamically attenuating the output power of the laser to avoid measurement distortion caused by thermal saturation of the attenuation sheet includes: Control the output power of the laser through laser pulse width modulation; Construct a dynamic energy adjustment matrix to dynamically adjust the laser energy output by the laser; Specifically include the following steps: Assume that the thermal saturation influence matrix is The diagonal elements in the thermal saturation influence matrix represent the influence of thermal saturation of thermal attenuation sheets corresponding to different laser channels on laser energy attenuation, and the non-diagonal elements represent the thermal saturation coupling influence between different laser channels, where n represents the number of laser channels; Then, when the attenuation plate reaches thermal saturation, the laser energy after passing through the attenuation plate is: in, They represent the laser energy after being affected by the thermal saturation of the attenuator and the initial laser energy respectively; Building a dynamic adjustment matrix Among them, d mm It indicates the energy adjustment ratio of the mth laser in the mth time slice during the laser pulse width adjustment process, where m indicates the number of time slices; The adjusted energy is: Wherein, P0 represents the laser initial pulse width adjustment matrix.

3. A high-power laser aging online testing method according to claim 2, characterized in that: The spectral adaptation model is constructed to expand the spectral response range of the light intensity sensor based on the edge signal and cross-band correlation of the inherent spectral response of the sensor, including: Construct spectral correlation matrix; Constructing a relational function for spectral inversion to calculate and obtain spectral values ​​that exceed the corresponding range of the sensor; the spectral adaptation model includes a spectral correlation matrix and a relational function; The constructing of the spectral correlation matrix comprises: Get the number of channels and the central wavelength of each channel to form a wavelength sequence λ = (λ1, λ2, ..., λ n ); Determine the range of the target spectrum to be inverted, wherein the range of the target spectrum includes a spectrum lower than the lowest central wavelength Δλ of the laser channel and higher than the highest central wavelength Δλ of the laser channel; Obtain known standard spectral data with a wide spectral coverage from the database, obtain spectral values ​​at the central wavelength of each channel and multiple wavelengths within the target spectral range, and set the sampling number k. The value of the standard spectral data at the corresponding wavelength is S = [s1, s2, ..., s k ] T ; Calculate the correlation coefficient between the light intensity sensor corresponding to each channel and the spectral value at the sampling wavelength Among them, R it represents the response value of the i-th channel in the t-th time, is the mean of the response values ​​of the ith channel, Represents the mean of the spectral values ​​at the jth sampling wavelength; With the correlation coefficient as the element, the spectral correlation matrix A = [ρ ij ].

4. A high-power laser aging online testing method according to claim 3, characterized in that: The constructed relationship function is used for spectral inversion, and the spectral value exceeding the corresponding range of the sensor is calculated and obtained, including: Assume the target spectrum is X = [x1, x2, ..., x l ] T , where x l Indicates the spectral value at the lth wavelength in the target spectral range; Establish the relationship function AX = B, where B = [b1, b2, b i , …, b u ] T , where a ij is the element in the spectral correlation matrix, s j Represents the value of the standard spectrum data at the jth sampling wavelength; Using the least squares method to solve the relationship function, the solution obtained is: X = (A T A) T B; If the nonlinear relationship between the actual spectrum and the channel response is considered, a nonlinear term is introduced into the relationship function; By fitting the relationship function with a polynomial, we can obtain the relationship function in the form of a quadratic polynomial: Among them, c j0 , c ji , c ijk represents the unknown coefficient, r i Represents the response value of the i-th channel; The value of the unknown coefficient is calculated by using the known standard spectrum data and the corresponding channel response value of the sensor using the curve fitting algorithm; Substitute the known edge spectral response value of the light intensity sensor into the relationship function to calculate the spectral value beyond the response range of the light intensity sensor.

5. A high-power laser aging online testing method according to claim 4, characterized in that: The spectral adaptation model is constructed to expand the spectral response range of the light intensity sensor based on the edge signal and cross-band correlation of the inherent spectral response of the sensor, and further includes: The spectral adaptation model is constructed through convolutional neural network, which inputs low-spectral data and outputs high-spectral data; specifically, it includes: Obtain edge wavelength data in the spectral response range of the light intensity sensor, that is, the lowest response wavelength and the highest response wavelength in a band, to form a low-spectrum data set X LOW ; Acquire wavelength data beyond the preset spectral response range of the light intensity sensor to form a hyperspectral dataset X HIG ; Normalize the data in the low-spectral dataset and the high-spectral dataset to form a sample dataset; The sample data set is used to train the convolutional neural network, and the trained convolutional neural network is used to output the predicted hyperspectral data, that is, the extended spectral response range of the light intensity sensor in a band is [λ i,min -Δλ′,λ i,max +Δλ′], where λ i,min , i,max represent respectively the minimum wavelength and the maximum wavelength to which the light intensity sensor responds in the i-th band, wherein the band includes the visible light band, the infrared band and the ultraviolet band; The convolutional neural network includes an input layer, a hidden layer and an output layer. The number of neurons in the input layer matches the number of data in the low-spectrum data set, and low-spectrum data of one band is received through neurons; the number of neurons in the output layer matches the number of data in the high-spectrum data set; The neural network is optimized using the Adam optimizer.

6. A high-power laser aging online testing method according to claim 5, characterized in that: The method of collecting external temperature data of the laser by a temperature sensor to obtain thermal conductivity, and calculating and obtaining internal temperature data of the laser by using the thermal conductivity and external temperature data, includes: Construct a three-dimensional transient thermal model and use the inverse heat conduction algorithm to derive the three-dimensional transient temperature field inside the laser. The specific steps are as follows: Acquire the laser structure data and the structure data of all the internal components from the database; construct a three-dimensional model of the laser according to the acquired laser structure data and the structure data of all the internal components; Construct the governing equations for three-dimensional transient heat conduction: Where α′ is the thermal diffusion coefficient; The three-dimensional model of the laser is discretized by the finite difference method, and the object is divided into multiple three-dimensional units, each of which has a size of ΔX, ΔY, and ΔZ; ΔT represents the length of the time step; Use the central difference method to discretize time and space to obtain: in, represents the temperature at the position (I, J, K) at the wth time step; Set boundary conditions; forward solve the three-dimensional transient heat conduction equation based on the surface temperature and boundary conditions measured by the temperature sensor to obtain the temperature distribution inside the object; Calculating the sensitivity of the temperature field to the parameters of the three-dimensional transient thermal model, wherein the parameters of the three-dimensional transient thermal model include the laser surface material information and boundary conditions; Using the surface temperature and sensitivity of the laser measured by the temperature sensor, the model parameters are estimated by the least square method or the gradient descent method; Using the estimated model parameters, the three-dimensional transient heat conduction equation is solved to obtain the temperature distribution inside the laser, that is, the temperature field distribution.

7. A high-power laser aging online testing method according to claim 6, characterized in that: The step of obtaining the laser light attenuation process by analyzing the data in the laser parameter data set includes: Extract light intensity data, arrange the measured light data in chronological order, and form a light-time variation curve; the light attenuation trend of the laser is reflected through the light-time variation curve; Establish a relationship model between various materials and chips inside the laser tube and the light attenuation of the laser tube, and set up a correction matrix; Based on the correction matrix and the aging factor of the corresponding material, analyze the materials and chips that have the greatest impact on light decay; specifically, the following steps are included: Assume that the laser tube contains z different materials or chips, represented by M1, M2, ..., M k express; Construct the material aging factor vector Among them, μ i Indicates material or chip M i The aging factor, the larger the value of the aging factor, the greater the impact of the aging degree of the material or chip on the performance of the laser tube; Establish the relationship between laser output power and aging factor in, represents the output power vector of the laser tube, Where u represents the number of power measurement points, that is, the number of laser tubes; p lu Represents the output power of the laser tube measured at the uth measurement point; represents the original output power, that is, the theoretical output power when not affected by material aging; C represents the coefficient matrix of factors affecting the output power due to material aging factors; Constructing the Correction Matrix Where I represents the unit matrix, k′ represents the adjustable coefficient, which is used to adjust the correction amplitude; the correction matrix satisfies the following relationship: When the output power of the laser tube is corrected to 100%, it corresponds to the theoretical output power have That is, the corrected value is: Observe the coefficients of the items related to different materials in the above formula, and the absolute values ​​of the coefficients reflect the effect of the aging of the corresponding materials on the laser light attenuation.

8. A high-power laser aging online testing method according to claim 7, characterized in that: It includes building an edge computing resource call matrix, calling edge computing resources according to the computing requirements, and implementing real-time temperature-light intensity coupling analysis on the local FPGA. The specific steps include: Obtain edge computing resource data, the edge computing data including CPU resources, memory resources, and storage resources; Obtain the CPU resource usage, memory usage, and storage resource usage of the host computer; Constructing edge computing resource call matrix where d 11 d 12 d 13 Respectively represent the call intensity of CPU, memory and storage resources when the amount of computation is low; d 21 d 22 d 23 Indicates the intensity of the computation on CPU, memory, and storage resources; d 31 d 32 d 33 Indicates the intensity of high computing workload on CPU, memory and storage resources; The CPU resource occupancy rate, memory occupancy rate and storage resource occupancy rate of the host computer are used to determine the computational workload of the received computing task and to allocate the CPU, memory and storage resources in the edge computing device according to the computational workload; The monitoring module in the FPGA or the interaction with the operating system monitors the utilization rate of the CPU, memory and storage resources and the progress of task execution.

9. A high-power laser aging online test system, the test system includes multiple acquisition modules, an acquisition control unit, a cooling unit, a host computer and a cabinet, characterized in that: The acquisition module, acquisition control unit and cooling unit are all arranged in the cabinet, and the acquisition control unit is connected to the host computer data; the cooling unit is used to reduce the internal temperature of the cabinet, the acquisition module includes a laser lamp holder, a light intensity sensor, an attenuation plate and a temperature sensor, and the light intensity sensor is connected to the acquisition control unit.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement the visual recognition-based unattended cooking reminder method described in any one of claims 1 to 8.

Images