Printing communication optical power prediction method

Abstract

The invention provides a printing communication optical power prediction method, which belongs to the technical field of optical power prediction, and comprises the following steps of: 1, acquiring various data of optical power in a printing communication system according to a running log of a printing system, and preprocessing the data to acquire characteristics of the printing communication optical power; 2, combining the characteristics of the printing communication optical power to obtain main factors influencing the optical power; 3, determining a generation mechanism of optical power and a transmission process of an optical signal according to main factors, and constructing a deep learning model based on automation; 4, training the deep learning model, and evaluating and optimizing the model; and step 5, deploying the model to an actual printing communication system to realize real-time prediction of the optical power. The problems that manual feature selection is needed and is greatly influenced by environmental factors, and meanwhile, a prediction model possibly has over-fitting, so that a prediction result is inaccurate, and the printing efficiency and the printing image quality are reduced are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of optical power prediction, and in particular to a method for predicting optical power of printed communications. Background Art

[0002] With the continuous development of printed communication technology, accurate prediction of printed communication optical power becomes more and more important.

[0003] However, traditional printing communication optical power prediction methods require a large number of actual tests, which are costly and time-consuming. At the same time, manual feature selection is required and is greatly affected by environmental factors. At the same time, the prediction model may have overfitting problems, resulting in inaccurate prediction results, optical signal interruption or poor transmission effect due to insufficient or excessive optical power, reducing printing efficiency and printed image quality.

[0004] Therefore, the present invention proposes a method for predicting optical power of printed communication. Summary of the invention

[0005] The present invention provides a method for predicting optical power for printed communications, which is used to solve the problems in the prior art that a large number of actual tests are required, which is costly and time-consuming, and that manual feature selection is required and is greatly affected by environmental factors. At the same time, the prediction model may have an overfitting problem, resulting in inaccurate prediction results, optical signal interruption or poor transmission effect due to insufficient or excessive optical power, thereby reducing printing efficiency and the quality of printed images.

[0006] In one aspect, the present invention provides a method for predicting optical power of printed communication, comprising: Step 1: Acquire various data of optical power in the printing communication system according to the operation log of the printing system, pre-process the data, and acquire the characteristics of the printing communication optical power according to the pre-processing results; Step 2: By combining the characteristics of the printed communication optical power, the main factors affecting the optical power are obtained; Step 3: Determine the generation mechanism of optical power and the transmission process of optical signals according to the main factors, and construct an automated deep learning model according to the generation mechanism and the transmission process of optical signals; Step 4: Train the deep learning model, evaluate the model using cross-validation technology, and optimize the model based on the evaluation results; Step 5: Deploy the model to the actual printed communication system to achieve real-time prediction of optical power.

[0007] A method for predicting optical power of printed communication provided by the present invention further includes: Determine and lock the frequency of the printing laser of the printing system, and introduce part of the printing laser that maintains the same frequency into the quantum frequency standard; The average intensity of the printed laser is determined using the optical frequency shift produced by a quantum frequency standard; Acquire irradiance data of a light-emitting component that generates a printing laser based on the average light intensity, and generate an irradiance point value sequence of the light-emitting component according to the irradiance data; Determine the optical power gain based on the irradiance point value sequence, and determine the correlation matrix between the optical power adjustment amount and the change amount of the voltage parameter of the light emitting component according to the optical power gain; Determining corresponding adjustment parameters of the optical power according to the correlation matrix and the adjustment difference of the voltage parameters of the light emitting component; Determine the light source delivery rate interval of the light emitting component according to the corresponding adjustment difference, and determine the performance requirement parameters based on the light source delivery rate interval; Determine the light supply gain of the light emitting component according to the performance requirement parameters, and determine the light compensation gain according to the preset light supply compensation relationship and the light supply gain; The optical attenuator and optical gain device are set according to the optical compensation gain to adjust the light to realize optical power detection.

[0008] According to a method for predicting optical power of printing communication provided by the present invention, various data of optical power in the printing communication system are obtained according to the operation log of the printing system, the data are preprocessed, and the characteristics of the optical power of printing communication are obtained according to the preprocessing result, including: A recording method for obtaining optical power data based on the operation log of the printing system; Searching for various data related to optical power in the operation log according to the recording method of the optical power data; Analyze the various data, and obtain the variation law of the optical power of the printing communication system according to the analysis result; According to the change rules, the potential structure and relationship in the data are determined based on a machine learning algorithm, and the characteristics of the printed communication optical power are obtained according to the potential structure and relationship, and engineering processing is performed.

[0009] According to a method for predicting optical power of printed communication provided by the present invention, by combining the characteristics of the optical power of printed communication, the main factors affecting the optical power are obtained, including: By calculating the correlation and weight between the characteristics of the printed communication optical power based on a statistical analysis method; The features of the printed communication optical power are combined according to the correlation degree and weight between the features to obtain a feature set; The main factors affecting the optical power are determined based on the implicit variables according to the feature set.

[0010] According to a method for predicting optical power of printed communication provided by the present invention, the generation mechanism of optical power and the transmission process of optical signals are determined according to the main factors, and an automated deep learning model is constructed according to the generation mechanism and the transmission process of optical signals, including: Determine the influence degree on optical power and their interaction according to the main factors; Obtain the type and working mode of the light source, and determine the generation mechanism of light power based on the degree of influence of factors on light power; Obtain the absorption and scattering of the optical signal and determine the transmission process of the optical signal based on the wavelength and refractive index; An automation-based deep learning model is constructed based on the generation mechanism and transmission process of the optical signal.

[0011] According to a method for predicting optical power of printed communication provided by the present invention, the type and working mode of the light source are obtained, and the generation mechanism of the optical power is determined based on the influence of factors on the optical power, including: Obtaining the type of the light source, and obtaining the corresponding power output and efficiency according to the type of the light source; determining an operating mode of the light source according to the power output and the efficiency; The current magnitudes of different light sources are obtained according to the working mode, and the generation mechanism of the optical power is determined according to the influence degree of the factors on the optical power based on the current magnitudes.

[0012] According to a method for predicting optical power of printed communication provided by the present invention, the deep learning model is trained, the model is evaluated using a cross-validation technique, and the model is optimized according to the evaluation result, including: Obtain an optical signal training set, train a deep learning model using the optical signal training set, and obtain a training model; Obtain the task characteristics of the deep learning model and select the cross-validation method based on the task characteristics; The training model is evaluated on the optical signal validation set through the cross-validation method, and the accuracy index of the model is calculated; Adjust the parameters or hyperparameters of the training model according to the accuracy index to achieve model optimization; Cross-validation methods include K-fold cross-validation, cross-validation round method, and Stratified K-fold cross-validation; Precision indicators include: accuracy, F1 score, mean square error, and determination coefficient.

[0013] According to a method for predicting optical power of printed communication provided by the present invention, a model is deployed in an actual printed communication system to realize real-time prediction of optical power, including: Convert the optimized model into an executable file and configure the deep learning framework in the print communication system and set the corresponding parameters and options; Importing the weights of the executable file into a deep learning framework; The model is loaded and compiled according to the API of the deep learning framework, and the prediction of printed communication optical power is realized based on the loaded and compiled model.

[0014] Compared with the prior art, the present invention has the following beneficial effects: By obtaining and combining various optical power data in the printing communication system, the characteristics of optical power are obtained and combined, the influencing factors are obtained, a deep learning model is constructed and trained and deployed, and real-time prediction of optical power is realized without the need for a large number of actual tests. At the same time, the feature selection is more accurate and not easily affected by environmental factors. At the same time, the prediction model does not have the problem of overfitting, resulting in accurate prediction results, avoiding optical signal interruption or poor transmission effect caused by insufficient or excessive optical power, and improving printing efficiency and printed image quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0016] Figure 1 It is a flow chart of a method for predicting optical power of printed communication provided by an embodiment of the present invention; Figure 2 It is a schematic diagram of a flow chart for preprocessing various data of optical power in a printing communication system and obtaining characteristics of the printing communication optical power provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] Embodiment 1: The embodiment of the present invention provides a method for predicting optical power of printed communication, such as Figure 1 As shown, the method mainly includes the following steps: Step 1: Acquire various data of optical power in the printing communication system according to the operation log of the printing system, pre-process the data, and acquire the characteristics of the printing communication optical power according to the pre-processing results; Step 2: By combining the characteristics of the printed communication optical power, the main factors affecting the optical power are obtained; Step 3: Determine the generation mechanism of optical power and the transmission process of optical signals according to the main factors, and construct an automated deep learning model according to the generation mechanism and the transmission process of optical signals; Step 4: Train the deep learning model, evaluate the model using cross-validation technology, and optimize the model based on the evaluation results; Step 5: Deploy the model to the actual printed communication system to achieve real-time prediction of optical power.

[0019] In this embodiment, the operation log of the printing system refers to a document that records various events, states, and operations during the operation of the printing system.

[0020] In this embodiment, the optical power data is recorded in a manner including: direct measurement, indirect calculation or sensor reading.

[0021] In this embodiment, the characteristics of the printing communication optical power may be: light source type, wavelength range, irradiation angle, and ambient temperature.

[0022] In this embodiment, the main factors affecting the optical power may be: the material, wavelength, temperature and pressure of the light source.

[0023] In this embodiment, the optical power generation mechanism includes: Current-driven type: The optical power is mainly driven by the current of the light source, such as LED lights, lasers, etc.

[0024] Luminous flux control type: The optical power is mainly controlled by the luminous flux of the light source, for example, by adjusting the reflector or lens of the light source to change the luminous flux.

[0025] Thermal radiation type: The light power is mainly generated by the thermal radiation of the light source, such as incandescent lamps, fluorescent lamps, etc.

[0026] In this embodiment, the transmission process of the optical signal refers to the process of light propagating in a medium.

[0027] In this embodiment, cross validation is a technique used to evaluate the performance of a model by dividing the dataset into multiple parts and splitting it into two groups, then using one of the groups to train the model and testing it on the other group to evaluate the generalization ability of the model.

[0028] In this embodiment, optimization may include: adjusting the structure of the model and modifying the hyperparameters of the model.

[0029] The beneficial effects of the above technical solution are: obtaining and combining the characteristics of optical power through various optical power data in the printing communication system, obtaining influencing factors, building a deep learning model and training and deploying it, and realizing real-time prediction of optical power without the need for a large number of actual tests. At the same time, the feature selection is relatively accurate and not easily affected by environmental factors. At the same time, the prediction model does not have the problem of overfitting, resulting in accurate prediction results, avoiding optical signal interruption or poor transmission effect caused by insufficient or excessive optical power, and improving printing efficiency and the quality of printed images.

[0030] Embodiment 2: Based on Example 1, the embodiment of the present invention further includes: Determine and lock the frequency of the printing laser of the printing system, and introduce part of the printing laser that maintains the same frequency into the quantum frequency standard; The average intensity of the printed laser is determined using the optical frequency shift produced by a quantum frequency standard; Acquire irradiance data of a light-emitting component that generates a printing laser based on the average light intensity, and generate an irradiance point value sequence of the light-emitting component according to the irradiance data; Determine the optical power gain based on the irradiance point value sequence, and determine the correlation matrix between the optical power adjustment amount and the change amount of the voltage parameter of the light emitting component according to the optical power gain; Determining corresponding adjustment parameters of the optical power according to the correlation matrix and the adjustment difference of the voltage parameters of the light emitting component; Determine the light source delivery rate interval of the light emitting component according to the corresponding adjustment difference, and determine the performance requirement parameters based on the light source delivery rate interval; Determine the light supply gain of the light emitting component according to the performance requirement parameters, and determine the light compensation gain according to the preset light supply compensation relationship and the light supply gain; The optical attenuator and optical gain device are set according to the optical compensation gain to adjust the light to realize optical power detection.

[0031] In this embodiment, the frequency of the printing laser refers to the number of vibrations per second of the laser beam used in the printing process.

[0032] In this embodiment, the quantum frequency standard is a method for defining and measuring the frequency of light based on the internal energy level transitions of atoms or molecules.

[0033] In this embodiment, the frequency shift of light produced by the quantum frequency standard refers to the difference between the observed frequency of light and the rest frequency of an object when the frequency of light of an object is measured by a standard called a quantum frequency standard.

[0034] In this embodiment, the average light intensity of the printing laser refers to the average power density of the laser beam output during the printing process.

[0035] In this embodiment, the irradiance data of the light-emitting component refers to the intensity and uniformity of the light generated by the printing laser under specific conditions.

[0036] In this embodiment, the irradiance point value sequence of the light-emitting component refers to a point value sequence formed by the change of the radiation intensity of the printing laser over time within a given time interval. This point value sequence is usually plotted with the time axis as the horizontal axis and the irradiance value as the vertical axis.

[0037] In this embodiment, the optical power gain refers to the ratio of the total optical power of light to the optical power of incident light when light passes through the optical element.

[0038] In this embodiment, the correlation matrix between the optical power adjustment amount and the voltage parameter change amount of the light-emitting component is a set of data tables that describe the linear relationship between them, wherein each row corresponds to a specific voltage parameter change amount, and each column corresponds to a specific optical power adjustment amount. The correlation matrix can be used to evaluate the strength and direction of the relationship between the optical power adjustment amount and the voltage parameter change amount.

[0039] In this embodiment, the corresponding adjustment parameters of the optical power refer to some parameters that need to be changed in order to control or adjust the output power of the laser under certain conditions, such as materials and temperature.

[0040] In this embodiment, the light source transmission rate interval of the light emitting component refers to the rated operating current range of the light emitting component.

[0041] In this embodiment, the light source delivery rate refers to the ratio of the luminous flux passing through the light-emitting component to the current passing through the component.

[0042] In this embodiment, the light supply gain of the light emitting component refers to the ratio between the actual light power emitted and the theoretical value when a certain voltage is applied to the light emitting component to cause it to emit light of a specific wavelength under a certain voltage.

[0043] In this embodiment, the optical compensation gain means that when light is irradiated onto a certain material, the absorption and emission characteristics of the material make the degree of absorption and emission of light by the material equal, so that the material looks like it is receiving and emitting the same light.

[0044] The beneficial effect of the above technical solution is: setting the optical attenuator and the optical gain device according to the optical compensation gain to adjust the light to achieve optical power detection can make the optical power detection more accurately reflect the changes in the actual optical signal and avoid measurement errors caused by changes in light source intensity.

[0045] Embodiment 3: Based on Example 2, the embodiment of the present invention obtains various data of optical power in the printing communication system according to the operation log of the printing system, pre-processes the data, and obtains the characteristics of the printing communication optical power according to the pre-processing results, such as Figure 2 As shown, including: S01: A recording method for obtaining optical power data according to an operation log of a printing system; S02: searching for various data related to optical power in the operation log according to the recording method of the optical power data; S03: Analyze the various data, and obtain the variation law of the optical power of the printing communication system according to the analysis result; S04: Determine the potential structure and relationship in the data based on the machine learning algorithm according to the change law, obtain the characteristics of the printed communication optical power according to the potential structure and relationship, and perform engineering processing.

[0046] In this embodiment, the operation log of the printing system refers to a document that records various events, states, and operations during the operation of the printing system.

[0047] In this embodiment, the optical power data is recorded in a manner including: direct measurement, indirect calculation or sensor reading.

[0048] In this embodiment, the various data related to optical power refer to a series of parameters related to optical power measurement and monitoring in an optical communication system, such as: Optical power: Indicates the optical power value output by the light source.

[0049] Optical power fluctuation: Indicates the difference between the measured optical power value and its average value.

[0050] Optical power distortion: Indicates the degree of optical power attenuation caused by factors such as loss and reflection during the transmission of optical signals.

[0051] Wavelength: represents the length of a complete electromagnetic spectrum line segment in the propagation of light.

[0052] In this embodiment, in the printing communication system, the variation pattern of optical power is usually related to the length and loss of the optical fiber. When light passes through the optical fiber, a certain loss will occur, which will cause the intensity of the optical signal to gradually weaken. In addition, different types of optical fibers and light sources may also affect the optical power. The optical power in the printing communication system will gradually decrease with the increase of distance. This is because the propagation speed of light in the optical fiber is very fast, but when entering the optical fiber interface, the propagation direction of the light will also change due to the change in the refractive index of the medium, resulting in the loss of optical power.

[0053] In this embodiment, the characteristics of the printing communication optical power may be: light source type, wavelength range, irradiation angle, and ambient temperature.

[0054] In this embodiment, the engineering processing may be: dimensionality reduction, smoothing, and feature enhancement.

[0055] In this embodiment, the data related to optical power may include the magnitude of optical power, fluctuation range, and frequency distribution. There is a certain correlation between these data. For example, the magnitude and fluctuation range of optical power may be affected by parameters such as the type, wavelength, and current of the light source; the frequency distribution of optical power may be affected by parameters such as the material, structure, and length of the optical fiber.

[0056] The beneficial effects of the above technical solution are: obtaining various data on optical power in the printing communication system according to the operation log of the printing system, and preprocessing the data can ensure the quality and accuracy of the optical power data. At the same time, obtaining the characteristics of the printing communication optical power according to the preprocessing results can reduce the complexity of feature selection and improve the quality of the features, and at the same time, provide a basis for the subsequent construction of the model.

[0057] Embodiment 4: Based on Example 3, the embodiment of the present invention combines the characteristics of the printed communication optical power to obtain the main factors affecting the optical power, including: By calculating the correlation and weight between the characteristics of the printed communication optical power based on a statistical analysis method; The features of the printed communication optical power are combined according to the correlation degree and weight between the features to obtain a feature set; The main factors affecting the optical power are determined based on the implicit variables according to the feature set.

[0058] In this embodiment, the statistical analysis method may be: correlation analysis, principal component analysis.

[0059] In this embodiment, in the printed communication system, implicit variables refer to those variables that are not directly measured but have a significant impact on the performance of other variables, such as material properties, environmental conditions, and light source characteristics.

[0060] In this embodiment, the main factors affecting the optical power may be: the material, wavelength, temperature and pressure of the light source.

[0061] The beneficial effects of the above technical solution are: combining the features of printed communication optical power according to the degree of correlation and weight of different features to obtain a feature set, and obtaining the factors affecting the optical power based on implicit variables, which can better understand the prediction results of printed communication optical power and formulate more effective measures to reduce or eliminate the impact of these factors, thereby improving printing efficiency and printing quality.

[0062] Embodiment 5: Based on Example 4, the embodiment of the present invention determines the generation mechanism of optical power and the transmission process of optical signals according to the main factors, and constructs an automated deep learning model according to the generation mechanism and the transmission process of optical signals, including: Determine the influence degree on optical power and their interaction according to the main factors; Obtain the type and working mode of the light source, and determine the generation mechanism of light power based on the degree of influence of factors on light power; Obtain the absorption and scattering of the optical signal and determine the transmission process of the optical signal based on the wavelength and refractive index; An automation-based deep learning model is constructed based on the generation mechanism and transmission process of the optical signal.

[0063] In this embodiment, in the printed communication system, the main factors affecting the optical power include: Fiber type: Different types of optical fibers have different effects on optical power. Single-mode fiber has higher optical power transmission performance than multi-mode fiber.

[0064] Length: The optical power will gradually attenuate while passing through the optical fiber, so measures need to be taken to compensate for this attenuation during long-distance transmission, such as using optical amplifiers, optical repeaters and other equipment.

[0065] Loss: In optical fiber, some losses will be generated during the propagation of light, such as absorption loss, scattering loss and bending loss.

[0066] In this embodiment, the interaction between the main factors can be: Fiber type and loss: Loss will affect the choice of fiber type. For example, if a longer transmission distance is required, you need to choose a fiber type that can withstand greater loss. At the same time, the choice of fiber type will also affect the size of the loss. For example, single-mode fiber has lower loss than multi-mode fiber.

[0067] Length and loss: The longer the length, the smaller the optical power. At the same time, loss will also affect the optical power. When the optical signal is transmitted in a long-distance optical fiber, more optical signals will be lost, resulting in a decrease in optical power.

[0068] Light source and optical fiber: The higher the power of the light source, the stronger the optical signal generated. However, the power of the light source cannot be too high, otherwise it will produce too high a light flux, resulting in distortion of the optical signal. At the same time, the optical fiber will also affect the optical signal. For example, the loss, dispersion, and reflection of the optical fiber will affect the transmission of the optical signal.

[0069] In this embodiment, the type of light source refers to the wavelength range of light emitted by the light source, including: LED lamp, HID lamp, Xenon lamp, halogen lamp.

[0070] In this embodiment, the working mode of the light source is, for example: LED lamp generates light through semiconductor materials, while HID lamp generates light through gas discharge. Xenon lamp and halogen lamp are also generated through discharge, but their discharge process is different from that of HID lamp.

[0071] In this embodiment, the optical power generation mechanism includes: Current-driven type: The optical power is mainly driven by the current of the light source, such as LED lights, lasers, etc.

[0072] Luminous flux control type: The optical power is mainly controlled by the luminous flux of the light source, for example, by adjusting the reflector or lens of the light source to change the luminous flux.

[0073] Thermal radiation type: The light power is mainly generated by the thermal radiation of the light source, such as incandescent lamps, fluorescent lamps, etc.

[0074] In this embodiment, the transmission process of the optical signal refers to the process of light propagating in a medium.

[0075] The beneficial effects of the above technical solution are: determining the generation mechanism of optical power and the transmission process of optical signals according to influencing factors, and constructing an automated deep learning model based on the generation mechanism and the transmission process of optical signals, which is closer to the real physical mechanism and has higher accuracy and reliability. At the same time, the model structure and parameters can be adaptively adjusted to adapt to different application scenarios and needs, thereby improving the generalization ability of the model and enhancing the accuracy of prediction.

[0076] Embodiment 6: Based on Example 5, the embodiment of the present invention obtains the type and working mode of the light source, and determines the generation mechanism of the optical power based on the influence degree of the factors on the optical power, including: Obtaining the type of the light source, and obtaining the corresponding power output and efficiency according to the type of the light source; determining an operating mode of the light source according to the power output and the efficiency; The current magnitudes of different light sources are obtained according to the working mode, and the generation mechanism of the optical power is determined according to the influence degree of the factors on the optical power based on the current magnitudes.

[0077] In this embodiment, the type of light source refers to the wavelength range of light emitted by the light source, including: LED lamp, HID lamp, Xenon lamp, halogen lamp.

[0078] In this embodiment, obtaining the corresponding power output according to the type of light source refers to calculating the power input required to achieve a specific brightness or illumination level according to the type or specification of the light source.

[0079] In this embodiment, the power output of the light source refers to the amount of energy emitted by the light source per second.

[0080] In this embodiment, the efficiency of the light source refers to the ability of the light source to convert electrical energy into light energy.

[0081] In this embodiment, the working mode of the light source is, for example: LED lamp generates light through semiconductor materials, while HID lamp generates light through gas discharge. Xenon lamp and halogen lamp are also generated through discharge, but their discharge process is different from that of HID lamp.

[0082] In this embodiment, the optical power generation mechanism includes: Current-driven type: The optical power is mainly driven by the current of the light source, such as LED lights, lasers, etc.

[0083] Luminous flux control type: The optical power is mainly controlled by the luminous flux of the light source, for example, by adjusting the reflector or lens of the light source to change the luminous flux.

[0084] Thermal radiation type: The light power is mainly generated by the thermal radiation of the light source, such as incandescent lamps, fluorescent lamps, etc.

[0085] The beneficial effects of the above technical solution are: determining the working mode of the light source according to the type of light source, thereby determining the optical power generation mechanism according to the current size of different light sources, and being able to automatically adjust the optical power generation mechanism according to the actual environment, making the system more stable and reliable, and avoiding unstable situations.

[0086] Embodiment 7: Based on Example 6, the embodiment of the present invention trains the deep learning model, uses cross-validation technology to evaluate the model, and optimizes the model according to the evaluation results, including: Obtain an optical signal training set, train a deep learning model using the optical signal training set, and obtain a training model; Obtain the task characteristics of the deep learning model and select the cross-validation method based on the task characteristics; The training model is evaluated on the optical signal validation set through the cross-validation method, and the accuracy index of the model is calculated; Adjust the parameters or hyperparameters of the training model according to the accuracy index to achieve model optimization; Cross-validation methods include K-fold cross-validation, cross-validation round method, and Stratified K-fold cross-validation; Precision indicators include: accuracy, F1 score, mean square error, and determination coefficient.

[0087] In this embodiment, the optical signal training set refers to a data set used to train a model, which contains a large amount of optical signal data.

[0088] In this embodiment, the task characteristics of the deep learning model refer to the type of problem that the model is designed to solve.

[0089] In this embodiment, the cross-validation method includes K-fold cross-validation, cross-validation round method, and Stratified K-fold cross-validation.

[0090] In this embodiment, the accuracy indicators of the model include: accuracy, F1 score, mean square error, and determination coefficient.

[0091] In this embodiment, optimization may include: adjusting the structure of the model and modifying the hyperparameters of the model.

[0092] The beneficial effects of the above technical solution are: training the deep learning model, using cross-validation technology to evaluate the model, and optimizing the model according to the evaluation results, which can improve the accuracy and generalization ability of the model and ensure the accuracy of the prediction results.

[0093] Embodiment 8: Based on Example 7, the embodiment of the present invention deploys the model into an actual printing communication system to achieve real-time prediction of optical power, including: Convert the optimized model into an executable file and configure the deep learning framework in the print communication system and set the corresponding parameters and options; Importing the weights of the executable file into a deep learning framework; The model is loaded and compiled according to the API of the deep learning framework, and the prediction of printed communication optical power is realized based on the loaded and compiled model.

[0094] In this embodiment, converting a model into an executable file refers to the process of converting a computer model or a mathematical model into an executable file that can be run on a computer, such as pytorch.

[0095] In this embodiment, the deep learning framework is a software framework for implementing deep learning, which provides many tools and functions to help developers build deep neural networks and perform training and reasoning.

[0096] In this embodiment, the parameters and options of the deep learning framework include: data input format, model structure, activation function and loss function.

[0097] In this embodiment, the weight of the executable file generally refers to the execution permission or security level of the executable file.

[0098] In this embodiment, the API of the deep learning framework refers to a set of programming interfaces and tools that can be used to build, train, and deploy deep learning models.

[0099] The beneficial effects of the above technical solution are: by converting the model into an executable file and importing it into a deep learning framework for loading and compiling, the prediction of printed communication optical power can be realized, thereby improving the accuracy and efficiency of the prediction, avoiding the cost and time of redeveloping and training the model in the field of printed communication optical power, and at the same time, improving the versatility of the model.

[0100] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0101] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for predicting optical power of printed communication, characterized in that: include: Step 1: Acquire various data of optical power in the printing communication system according to the operation log of the printing system, pre-process the data, and acquire the characteristics of the printing communication optical power according to the pre-processing results; Step 2: By combining the characteristics of the printed communication optical power, the main factors affecting the optical power are obtained; Step 3: Determine the generation mechanism of optical power and the transmission process of optical signals according to the main factors, and construct an automated deep learning model according to the generation mechanism and the transmission process of optical signals; Step 4: Train the deep learning model, evaluate the model using cross-validation technology, and optimize the model based on the evaluation results; Step 5: Deploy the model to the actual printed communication system to achieve real-time prediction of optical power.

2. The method for predicting optical power of printed communication according to claim 1, characterized in that: Also includes: Determine and lock the frequency of the printing laser of the printing system, and introduce part of the printing laser that maintains the same frequency into the quantum frequency standard; The average intensity of the printed laser is determined using the optical frequency shift produced by a quantum frequency standard; Acquire irradiance data of a light-emitting component that generates a printing laser based on the average light intensity, and generate an irradiance point value sequence of the light-emitting component according to the irradiance data; Determine the optical power gain based on the irradiance point value sequence, and determine the correlation matrix between the optical power adjustment amount and the change amount of the voltage parameter of the light emitting component according to the optical power gain; Determining corresponding adjustment parameters of the optical power according to the correlation matrix and the adjustment difference of the voltage parameters of the light emitting component; Determine the light source delivery rate interval of the light emitting component according to the corresponding adjustment difference, and determine the performance requirement parameters based on the light source delivery rate interval; Determine the light supply gain of the light emitting component according to the performance requirement parameters, and determine the light compensation gain according to the preset light supply compensation relationship and the light supply gain; The optical attenuator and optical gain device are set according to the optical compensation gain to adjust the light to realize optical power detection.

3. The method for predicting optical power of printed communication according to claim 1, characterized in that: According to the operation log of the printing system, various data of the optical power in the printing communication system are obtained, the data are preprocessed, and the characteristics of the printing communication optical power are obtained according to the preprocessing results, including: A recording method for obtaining optical power data based on the operation log of the printing system; Searching for various data related to optical power in the operation log according to the recording method of the optical power data; Analyze the various data, and obtain the variation law of the optical power of the printing communication system according to the analysis result; According to the change rules, the potential structure and relationship in the data are determined based on a machine learning algorithm, and the characteristics of the printed communication optical power are obtained according to the potential structure and relationship, and engineering processing is performed.

4. The method for predicting optical power of printed communication according to claim 1, characterized in that: By combining the characteristics of the printed communication optical power, the main factors affecting the optical power are obtained, including: By calculating the correlation and weight between the characteristics of the printed communication optical power based on a statistical analysis method; The features of the printed communication optical power are combined according to the correlation degree and weight between the features to obtain a feature set; The main factors affecting the optical power are determined based on the implicit variables according to the feature set.

5. The method for predicting optical power of printed communication according to claim 1, characterized in that: Determine the generation mechanism of optical power and the transmission process of optical signals according to the main factors, and build an automated deep learning model according to the generation mechanism and the transmission process of optical signals, including: Determine the influence degree on optical power and their interaction according to the main factors; Obtain the type and working mode of the light source, and determine the generation mechanism of light power based on the degree of influence of factors on light power; Obtain the absorption and scattering of the optical signal and determine the transmission process of the optical signal based on the wavelength and refractive index; An automation-based deep learning model is constructed based on the generation mechanism and transmission process of the optical signal.

6. The method for predicting optical power of printed communication according to claim 1, characterized in that: Obtain the type and working mode of the light source, and determine the mechanism of light power generation based on the degree of influence of factors on light power, including: Obtaining the type of the light source, and obtaining the corresponding power output and efficiency according to the type of the light source; determining an operating mode of the light source according to the power output and the efficiency; The current magnitudes of different light sources are obtained according to the working mode, and the generation mechanism of the optical power is determined according to the influence degree of the factors on the optical power based on the current magnitudes.

7. The method for predicting optical power of printed communication according to claim 1, characterized in that: The deep learning model is trained, the model is evaluated using cross-validation technology, and the model is optimized according to the evaluation results, including: Obtain an optical signal training set, train a deep learning model using the optical signal training set, and obtain a training model; Obtain the task characteristics of the deep learning model and select the cross-validation method based on the task characteristics; The training model is evaluated on the optical signal validation set through the cross-validation method, and the accuracy index of the model is calculated; Adjust the parameters or hyperparameters of the training model according to the accuracy index to achieve model optimization; Cross-validation methods include K-fold cross-validation, cross-validation round method, and Stratified K-fold cross-validation; Precision indicators include: accuracy, F1 score, mean square error, and determination coefficient.

8. The method for predicting optical power of printed communication according to claim 1, characterized in that: The model is deployed in actual printed communication systems to achieve real-time prediction of optical power, including: Convert the optimized model into an executable file and configure the deep learning framework in the print communication system and set the corresponding parameters and options; Importing the weights of the executable file into a deep learning framework; The model is loaded and compiled according to the API of the deep learning framework, and the prediction of printed communication optical power is realized based on the loaded and compiled model.

Images