Waveform debugging method and device, storage medium and computer program product

Abstract

The invention discloses a waveform debugging method and device, a storage medium and a computer program product, and relates to the technical field of ink-jet printing, and the waveform debugging method comprises the steps: obtaining target ink droplet form data and a parameter prediction model of specified ink-jet printing equipment, wherein the parameter prediction model is a machine learning model for describing a mapping relationship between the waveform parameter of the specified ink-jet printing equipment and the ink droplet form; and inputting the target ink droplet form data into a parameter prediction model to obtain a target waveform parameter, matched with the target ink droplet form data, of the specified ink-jet printing equipment. The technical problem that the existing waveform debugging process is easily influenced by subjective factors, so that the debugging stability is poor is solved.

Description

Technical Field

[0001] The present application relates to the field of inkjet printing technology, and in particular to a waveform debugging method, device, storage medium and computer program product. Background Art

[0002] In the field of OLED (Organic Light-Emitting Diode) inkjet printing, precise control of ink droplets is crucial to the image quality and device performance of OLED display devices.

[0003] At present, common waveform debugging methods mainly rely on manual debugging. These methods usually require technicians to manually set and adjust the waveform parameters of inkjet printers, such as voltage, rise time, duration, fall time, etc. Then the quality of the waveform parameters is judged by observing the changes in the ink droplet morphology, and then a set of waveform parameters with better ink droplet morphology is selected. However, this waveform debugging method is too dependent on the personal experience and subjective judgment of technicians. The debugging process is easily affected by subjective factors, resulting in poor debugging stability.

[0004] The above contents are only used to assist in understanding the technical solution of the present application and do not constitute an admission that the above contents are prior art. Summary of the invention

[0005] The main purpose of this application is to provide a waveform debugging method, device, storage medium and computer program product, aiming to solve the technical problem that the existing waveform debugging process is easily affected by subjective factors, resulting in poor debugging stability.

[0006] To achieve the above purpose, the present application proposes a waveform debugging method, which includes: Acquire target ink droplet morphology data and a parameter prediction model of a specified inkjet printing device, wherein the parameter prediction model is a machine learning model that describes a mapping relationship between waveform parameters of the specified inkjet printing device and ink droplet morphology; Inputting the target ink droplet morphology data into a parameter prediction model to obtain target waveform parameters that match the target ink droplet morphology data for the designated inkjet printing device; Before the step of obtaining the parameter prediction model of the specified inkjet printing device, it includes: Generate multiple groups of sample waveform parameter combinations; Controlling the designated inkjet printing device to perform inkjet printing operations according to each of the sample waveform parameter combinations, and collecting ink droplet morphology data under each of the sample waveform parameter combinations as a training sample set; A parameter prediction model is obtained by training based on the training sample set; The step of collecting ink droplet morphology data under each of the sample waveform parameter combinations as a training sample set includes: Collecting ink droplet morphology images and ink droplet morphology parameters under each of the sample waveform parameter combinations as ink droplet morphology data; Calculating the numerical distribution characteristic value of the ink droplet morphology parameter, and constructing a parameter feature based on the numerical distribution characteristic value corresponding to the ink droplet morphology parameter; Extracting features from the ink droplet morphology image to obtain ink droplet image features; The parameter features and the ink drop image features are concatenated to obtain an input feature matrix; An output parameter matrix is ​​constructed based on the sample waveform parameter combination; The input feature matrix and the corresponding output parameter matrix are used as training samples, and each of the training samples is used as a training sample set.

[0007] In one embodiment, the step of obtaining a parameter prediction model based on the training sample set includes: Marking the input feature matrix whose numerical distribution feature value is within the predetermined distribution range as having good quality, and marking the input feature matrix whose numerical distribution feature value is not within the predetermined distribution range as having poor quality; Assigning a first weight to an input feature matrix marked as having good quality, and assigning a second weight to an input feature matrix marked as having poor quality, wherein the first weight is greater than the second weight; The input feature matrix and the output parameter matrix after weight allocation are input into a predetermined initial regression model for training until the predetermined initial regression model converges to obtain a parameter prediction model.

[0008] In one embodiment, after the step of inputting the target ink droplet morphology data into a parameter prediction model to obtain target waveform parameters matching the specified inkjet printing device with the target ink droplet morphology data, the waveform debugging method further includes: Controlling the designated inkjet printing device to perform inkjet printing according to the target waveform parameters to obtain a real-time ink droplet morphology, and comparing the real-time ink droplet morphology with the target ink droplet morphology data to obtain an ink droplet error; After the ink drop error is not less than a predetermined error threshold, a parameter adjustment amount is calculated according to the ink drop error, and the target waveform parameters are updated according to the parameter adjustment amount until the ink drop error is less than the predetermined error threshold, thereby obtaining updated target waveform parameters.

[0009] In one embodiment, after the step of updating the target waveform parameters according to the parameter adjustment amount until the ink drop error is less than a predetermined error threshold to obtain updated target waveform parameters, the method includes: Using the updated target waveform parameters and corresponding target ink droplet morphology data as new training samples; The parameter prediction model is iteratively updated based on the new training samples to obtain a new parameter prediction model.

[0010] In one embodiment, the step of iteratively updating the parameter prediction model based on the new training sample to obtain a new parameter prediction model includes: Acquire the current training sample set of the parameter prediction model, and use the new training sample and the current training sample set as a new training sample set; Assigning a third weight to the input feature matrix in the new training sample, and assigning a fourth weight to the input feature matrix in the current training sample of the current training sample set, wherein the third weight is greater than the fourth weight; The new training sample set after weight allocation is input into the predetermined initial regression model for training until the predetermined initial regression model converges to obtain a new parameter prediction model.

[0011] In addition, to achieve the above-mentioned purpose, the present application also proposes a waveform debugging device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the waveform debugging method described above.

[0012] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by the processor, the steps of the waveform debugging method described above are implemented.

[0013] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the steps of the waveform debugging method described above are implemented.

[0014] One or more technical solutions proposed in this application have at least the following technical effects: The present application obtains the target ink droplet morphology data and the parameter prediction model of the specified inkjet printing device, wherein the parameter prediction model is a machine learning model that describes the mapping relationship between the waveform parameters of the specified inkjet printing device and the ink droplet morphology. Then the target ink droplet morphology data is input into the parameter prediction model to obtain the target waveform parameters that match the target inkjet printing device and the target ink droplet morphology data. Therefore, the present application constructs a parameter prediction model that describes the mapping relationship between the waveform parameters of the specified inkjet printing device and the ink droplet morphology in advance, so that in the scenario where the ink droplet morphology needs to be specified, the target waveform parameters that match the target ink droplet morphology data can be predicted by the parameter prediction model, so that the specified inkjet printing device can spray ink droplets of the corresponding shape of the target ink droplet morphology data. The present application abandons the traditional manual debugging method and instead uses a machine learning model to describe the mapping relationship between the waveform parameters of the specified inkjet printing device and the ink droplet morphology, reducing the interference of subjective factors of technicians, effectively improving the stability of waveform debugging, and also simultaneously improving the debugging efficiency. In addition, since the target ink droplet morphology data in the present application can be changed according to specific needs, the present application is not only able to output the "optimal waveform parameters" under a certain fixed evaluation standard, but can output target waveform parameters that are adapted to the target ink droplet morphology data, thereby enabling the inkjet printing device under the waveform debugging method of the present application to be applied to a wider range of scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0016] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0017] Figure 1 A flowchart diagram of the first embodiment of the waveform debugging method of the present application is provided; Figure 2 A flowchart diagram of the second embodiment of the waveform debugging method of the present application is provided; Figure 3 A schematic diagram of a waveform debugging scenario involved in an embodiment of the present application; Figure 4 A flowchart diagram of the third embodiment of the waveform debugging method of the present application is provided; Figure 5 It is a structural schematic diagram of the waveform debugging device of the hardware operating environment involved in the waveform debugging method in the embodiment of the present application.

[0018] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0019] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0020] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0021] The main solution of the embodiment of the present application is: to obtain target ink droplet morphology data and a parameter prediction model of a specified inkjet printing device, wherein the parameter prediction model is a machine learning model that describes the mapping relationship between the waveform parameters of the specified inkjet printing device and the ink droplet morphology; the target ink droplet morphology data is input into the parameter prediction model to obtain the target waveform parameters of the specified inkjet printing device that match the target ink droplet morphology data.

[0022] In this embodiment, for ease of description, the waveform debugging device is used as the execution subject for explanation below.

[0023] Since the common waveform debugging methods in the prior art mainly rely on manual debugging, these methods usually require technicians to manually set and adjust the waveform parameters of the inkjet printer, such as voltage, rise time, duration, fall time, etc. Then the quality of the waveform parameters is judged by observing the changes in the ink droplet morphology, and then a set of waveform parameters with better ink droplet morphology is selected. However, this waveform debugging method is too dependent on the personal experience and subjective judgment of the technicians, and the debugging process is easily affected by subjective factors, resulting in poor debugging stability.

[0024] The present application provides a solution, by constructing a parameter prediction model that describes the mapping relationship between the waveform parameters of a specified inkjet printing device and the ink droplet morphology in advance, so that in the scenario where the ink droplet morphology needs to be specified, the target waveform parameters that match the target ink droplet morphology data can be predicted through the parameter prediction model, so that the specified inkjet printing device can eject ink droplets of the corresponding morphology of the target ink droplet morphology data. The present application abandons the traditional manual debugging method, and instead adopts a machine learning model to describe the mapping relationship between the waveform parameters of the specified inkjet printing device and the ink droplet morphology, reducing the interference of subjective factors of technicians, effectively improving the stability of waveform debugging, and also simultaneously improving the debugging efficiency. In addition, since the target ink droplet morphology data in the present application can be changed according to specific needs, the present application is not only able to output the "optimal waveform parameters" under a certain fixed evaluation standard, but can output the target waveform parameters adapted to it according to the different target ink droplet morphology data, so that the inkjet printing device under the waveform debugging method of the present application can be applied to a wider range of scenarios.

[0025] Based on this, the present application embodiment provides a waveform debugging method, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the waveform debugging method of the present application.

[0026] In this embodiment, the waveform debugging method includes steps S10 to S20: Step S10, obtaining target ink droplet morphology data and a parameter prediction model of a specified inkjet printing device, wherein the parameter prediction model is a machine learning model that describes a mapping relationship between waveform parameters of the specified inkjet printing device and ink droplet morphology; It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a server and other electronic devices that can realize the above functions.

[0027] It should also be noted that the parameter prediction model is a machine learning model that describes the mapping relationship between the waveform parameters of the specified inkjet printing device and the ink droplet morphology. The machine learning model can be a support vector machine regression, a neural network model, a decision tree model, etc.

[0028] In addition, it should be noted that the target ink drop shape data is data describing the ink drop shape expected to be ejected by the specified inkjet printing device. The specified inkjet printing device is a device expected to perform an inkjet printing operation, and the specified inkjet printing device may be a complete inkjet printing device or a component of the inkjet printing device that performs an inkjet operation (such as a nozzle and a nozzle controller).

[0029] This embodiment can obtain the volume, speed, angle and other morphological parameters of the ink droplets that the user expects the specified inkjet printing device to eject, as well as the distribution characteristic values ​​of each morphological parameter (such as mean value, variance, standard deviation, etc.) as target morphological parameters, and obtain the target morphological image of the ink droplets that the user expects the specified inkjet printing device to eject (that is, the image of the ink droplet falling process at a certain moment, such as shape integrity, whether there are satellite droplets, etc.). Then the target morphological parameters and the target morphological image are used as target ink droplet morphological data. This embodiment obtains the ink droplet morphological data under each sample waveform parameter combination as a training sample set, and trains the obtained parameter prediction model. It can be understood that the parameter prediction model can be a machine learning model trained in advance, or it can be a machine learning model trained during waveform debugging.

[0030] Step S20, inputting the target ink droplet morphology data into a parameter prediction model to obtain target waveform parameters that match the target ink droplet morphology data for the designated inkjet printing device.

[0031] Since the parameter prediction model describes the mapping relationship between the waveform parameters and ink droplet morphology of the specified inkjet printing device, after the target ink droplet morphology data is input into the parameter prediction model, the parameter prediction model can output the target waveform parameters matching the target ink droplet morphology data through the mapping relationship between the waveform parameters and the ink droplet morphology. In addition, the ink droplets ejected by the specified inkjet printing device in the inkjet printing operation can meet the target ink droplet morphology data based on the target waveform parameters.

[0032] The first embodiment of the present application provides a waveform debugging method, by obtaining target ink droplet morphology data and a parameter prediction model of a specified inkjet printing device, wherein the parameter prediction model is a machine learning model that describes the mapping relationship between the waveform parameters of the specified inkjet printing device and the ink droplet morphology. Then the target ink droplet morphology data is input into the parameter prediction model to obtain the target waveform parameters that match the target inkjet printing device and the target ink droplet morphology data. Therefore, the present application constructs a parameter prediction model that describes the mapping relationship between the waveform parameters of the specified inkjet printing device and the ink droplet morphology in advance, so that in the scenario where the ink droplet morphology needs to be specified, the target waveform parameters that match the target ink droplet morphology data can be predicted by the parameter prediction model, so that the specified inkjet printing device can eject ink droplets of the corresponding morphology of the target ink droplet morphology data. The present application abandons the traditional manual debugging method, and instead uses a machine learning model to describe the mapping relationship between the waveform parameters of the specified inkjet printing device and the ink droplet morphology, reducing the interference of subjective factors of technicians, effectively improving the stability of waveform debugging, and also simultaneously improving the debugging efficiency. In addition, since the target ink droplet morphology data in the present application can be changed according to specific needs, the present embodiment is not only able to output the "optimal waveform parameters" under a certain fixed evaluation standard, but can output target waveform parameters adapted to the target ink droplet morphology data, thereby enabling the inkjet printing device under the waveform debugging method of the present embodiment to be applied to a wider range of scenarios.

[0033] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above-mentioned embodiment 1 can be referred to the above introduction, and will not be repeated in the following. Figure 2 Before step S10, the waveform debugging method further includes steps A10 to A30: Step A10, generating multiple groups of sample waveform parameter combinations; Step A20, controlling the designated inkjet printing device to perform inkjet printing operations according to each of the sample waveform parameter combinations, and collecting ink droplet morphology data under each of the sample waveform parameter combinations as a training sample set; Step A30: obtaining a parameter prediction model based on the training sample set.

[0034] It should be noted that the sample waveform parameter combination includes a group of waveform parameters for describing a voltage waveform for controlling ink droplet ejection, such as voltage magnitude, voltage rise time, voltage duration, and voltage fall time.

[0035] See also Figure 3 , Figure 3 Schematic diagram of the scene of collecting ink droplet morphology data involved in the embodiment of the present application. Figure 3As shown, the designated inkjet printing device may include a nozzle controller and a nozzle. In this embodiment, each sample waveform parameter combination is sent to the nozzle controller, so that the nozzle connected to the nozzle controller ejects ink droplets, and then the ejected ink droplets are observed by an inkjet observer to obtain ink droplet morphology data.

[0036] Regarding the generation of multiple sets of sample waveform parameter combinations, this embodiment can obtain each optional waveform parameter by selecting a value within the parameter range corresponding to each waveform parameter. Then, after arranging and combining each waveform parameter, multiple sets of sample waveform parameter combinations are obtained. Exemplarily, this embodiment can set multiple levels for waveform parameters (such as voltage, voltage rise time, voltage duration, voltage drop time, etc.) according to the parameter range corresponding to each waveform parameter to generate each sample waveform parameter combination. For example, a 2^4 design. 2 represents 2 levels (high and low) of each factor (i.e., waveform parameter), and 4 is the number of waveform parameters. This design can generate 16 sample waveform parameter combinations. In order to further improve the debugging accuracy on the basis of ensuring the debugging efficiency, this embodiment can also divide each waveform parameter into key factors (i.e., factors that have a greater impact on the ink droplet morphology, such as voltage, voltage duration) and non-key factors (i.e., factors that have a smaller impact on the ink droplet morphology, such as voltage rise time, voltage drop time), wherein the number of levels of the key factors is more than that of the non-key factors. Exemplarily, factor levels can be added to key factors (such as voltage and voltage duration) to achieve a more refined division of key factors. For example, the voltage magnitude can be expanded from two levels (15V and 20V) to multiple levels (15V, 17.5V, 20V), and the voltage duration can be expanded from two levels (0.1us and 10us) to more levels (0.1us, 5us, 10us). For non-critical factors (such as voltage rise time and voltage fall time), a smaller number of levels can be maintained to reduce the number of experiments and ensure debugging efficiency. In addition, the present embodiment can also perform computer simulation on each generated sample waveform parameter combination, predict the influence of different sample waveform parameters on ink droplet formation through physical models and mathematical calculations, analyze the ink movement inside the nozzle and the final morphology of the ejected ink droplets based on fluid dynamics or electromagnetic simulation models, and then select the sample waveform parameter combination for model training from each sample waveform parameter combination according to the simulation results. Alternatively, perform preliminary physical tests on the generated sample waveform parameter combination, and select the sample waveform parameter combination for model training from each sample waveform parameter combination according to the physical test results. For example, the generated sample waveform parameter combinations are input into a designated inkjet printing device, an inkjet test is performed on an actual glass substrate, the ink droplet morphology is observed, and some sample waveform parameter combinations with better ink droplet morphology are used as sample waveform parameter combinations for model training.

[0037] Regarding the collection of ink droplet morphology data under each of the sample waveform parameter combinations, this embodiment can realize the collection of ink droplet morphology data through a high-precision ink droplet observer. This embodiment can move the ink droplet observer to the ink droplet observation position of the specified inkjet printing device; select the i-th group of sample waveform parameter combinations from the set of each of the sample waveform parameter combinations, and configure it to the nozzle controller of the specified inkjet printing device, and set a fixed flash spray frequency and flash spray times in the nozzle controller so that the nozzle connected to the nozzle controller performs the corresponding inkjet printing operation. With the help of the ink droplet observer, the volume (pl), speed (m / s), angle (°) and other morphological parameters of the ink droplets ejected by the nozzle are continuously collected and the ink droplet morphology image under the corresponding morphological parameters is saved at the same time. The i-th group of sample waveform parameter combinations is associated with the collected ink droplet morphology parameters and ink droplet morphology image as ink droplet morphology data. Then, it is determined whether all sample waveform parameter combinations have completed the ink droplet morphology data acquisition operation. If all sample waveform parameter combinations have not completed the ink droplet morphology data acquisition operation, i=i+1 (the initial value of i is 1) can be set to continue the ink droplet morphology data acquisition operation for other sample waveform parameter combinations. If all sample waveform parameter combinations have completed the ink droplet morphology data acquisition operation, this embodiment can directly use the ink droplet morphology data under each of the sample waveform parameter combinations as a training sample set, or convert the sample waveform parameter combination in the ink droplet morphology data into an output parameter matrix, convert the ink droplet morphology parameters and ink droplet morphology image in the ink droplet morphology data into an input feature matrix, use the output parameter matrix and the input feature matrix as training samples, and use each of the training samples as a training sample set.

[0038] Furthermore, this embodiment can obtain a parameter prediction model based on the training sample set training. The following description of model training is carried out by taking the support vector machine regression model as the parameter prediction model. In this embodiment, the initialized support vector regression model is used for training, and the radial basis function kernel is selected as the kernel function of the support vector machine regression model so as to be able to handle complex nonlinear relationships. γ (gamma) is used as the kernel function hyperparameter, and the initial value is set to "scale". By default, γ is automatically adjusted according to the number of features and feature variance of the data; the hyperparameters of the support vector regression model (such as the kernel function hyperparameter γ and the regularization parameter C) are tuned using methods such as cross-validation and random search. The operation of the hyperparameter tuning process is as follows: determine the hyperparameters that need to be tuned and their value ranges. For example, for the regularization parameter C, a series of values ​​in the logarithmic space (such as 0.1, 1, 10, 100) can be tried; for the kernel function parameter γ, a series of values ​​(such as 0.001, 0.01, 0.1, 1) can also be tried; a part of the hyperparameter combination is randomly selected for testing. Compared with GridSearch, random search can find near-optimal solutions faster with the same computing resources. Determine the indicators used to evaluate model performance. For regression problems, indicators such as negative mean square error or coefficient of determination can be used as indicators to evaluate model performance. Use k-fold cross validation to evaluate the performance of each set of hyperparameter combinations. Divide the training sample set into k subsets, train the model on k-1 subsets, and verify it on the remaining subset. Repeat k times and take the average performance as the performance evaluation of the set of hyperparameters. Run random search, use cross validation to evaluate the model on each hyperparameter combination, and record the performance indicators of each combination. After obtaining the preliminary results of random search, further fine-tune according to the range of the best results. For example, if the best value of C is between 1 and 10, a more detailed grid search can be performed, such as [1, 2, 5, 10], and then use cross validation again on each hyperparameter combination to evaluate the model, and record the performance indicators of each combination. According to the average performance indicators of cross validation, find the best hyperparameter combination for training the final parameter prediction model.

[0039] In this embodiment, the training sample set can be divided into a training set and a test set, for example, 80% of the training samples in the training sample set are used as the training set for training, and 20% of the training samples in the training sample set are used as the test set for testing; the support vector machine regression model is trained using the training set so that the support vector machine regression model can effectively learn the mapping relationship between ink droplet morphology and waveform parameters; and then the test set is used to evaluate the performance of the trained support vector machine regression model (i.e., the accuracy, precision and recall rate of the support vector machine regression model); after the performance of the trained support vector machine regression model meets the predetermined performance requirements, the trained support vector machine regression model can be used as a parameter prediction model. In a feasible implementation, the step of collecting ink droplet morphology data under each of the sample waveform parameter combinations as a training sample set in step A20 may include steps B10 to B60: Step B10, collecting ink droplet morphology images and ink droplet morphology parameters under each of the sample waveform parameter combinations as ink droplet morphology data; Step B20, calculating the numerical distribution characteristic value of the ink droplet morphology parameter, and constructing a parameter feature based on the numerical distribution characteristic value corresponding to the ink droplet morphology parameter; Step B30, extracting features from the ink droplet morphology image to obtain ink droplet image features; Step B40, combining the parameter features and the ink drop image features to obtain an input feature matrix; Step B50, constructing an output parameter matrix based on the sample waveform parameter combination; Step B60: taking the input feature matrix and the corresponding output parameter matrix as training samples, and taking each of the training samples as a training sample set.

[0040] It should be noted that the numerical distribution characteristic value is a numerical value used to describe the distribution characteristics of the ink droplet morphology parameter, such as a mean value, a standard deviation, a variance, and the like.

[0041] After collecting the ink droplet morphology images and ink droplet morphology parameters under each of the sample waveform parameter combinations as ink droplet morphology data through the ink droplet observation instrument, the ink droplet morphology parameters corresponding to each group of sample waveform parameter combinations can be calculated to obtain the numerical distribution characteristic values ​​(such as mean and standard deviation) of each ink droplet morphology parameter under the sample waveform parameter combination, and then the numerical distribution characteristic values ​​corresponding to the ink droplet morphology parameters are converted into parameter features in the form of vectors or matrices, for example, the parameter features are 6-dimensional characteristic vectors, i.e., the mean and standard deviation corresponding to the volume, velocity, and angle of the ink droplet. Furthermore, the present embodiment can pre-process the ink droplet morphology images. Exemplarily, the preprocessing operations include grayscale processing, such as using OpenCV or other image processing libraries to grayscale the ink droplet morphology image to reduce computational complexity; noise reduction processing, such as using Gaussian filtering or median filtering to remove noise in the ink droplet morphology image to improve image quality; image enhancement processing, such as using contrast adjustment or edge detection methods to enhance the ink droplet morphology image to highlight the morphological features of the ink droplets; image cropping processing, such as cropping the ink droplet morphology image to remove useless areas in the ink droplet morphology image and only retain the content within the ink droplet area of ​​interest to reduce the amount of calculation and data redundancy. Then, this embodiment performs feature extraction on the preprocessed ink droplet morphology image. For example, a model for feature extraction (such as a convolutional neural network model) can be constructed, and the convolutional neural network is used to extract high-level features of the preprocessed ink droplet morphology image, wherein the high-level features include information such as the shape, edge, and texture of the ink droplet, and then the high-level features are subjected to dimensionality reduction processing to convert the high-dimensional data into low-dimensional feature representation to obtain ink droplet image features, thereby reducing the feature space dimension of the ink droplet image features, which helps to improve the training efficiency and generalization ability of the subsequent parameter prediction model. Then, this embodiment can splice the parameter features and the ink droplet image features to obtain an input feature matrix. For example, the 6-dimensional parameter features (the average value and standard deviation of the volume, velocity, and angle of the ink droplet) and the N-dimensional ink droplet image features are spliced ​​together to form a complete N+6-dimensional feature matrix vector (i.e., the input feature matrix), whereby the input feature matrix contains features describing the ink droplet morphology parameters and the ink droplet morphology image, reflecting the ejection effect of the ink droplets under different sample waveform parameter combinations. Furthermore, the input feature matrix can also be standardized to ensure that the analysis is performed on the same data scale.

[0042] Based on the sample waveform parameter combination, an output parameter matrix is ​​constructed, which is the target of model training and contains waveform parameter combinations corresponding to the input feature matrix. The dimension of the output parameter matrix is: m×p, where m represents the number of waveform parameter combinations and p represents the dimension of the waveform parameter (such as voltage magnitude, voltage rise time, voltage duration, voltage drop time, etc.). Therefore, this embodiment uses each input feature matrix and the corresponding output parameter matrix as training samples, and uses each of the training samples as a training sample set.

[0043] In a feasible implementation, step A30 may include steps C10 to C30: Step C10, marking the input feature matrix whose numerical distribution eigenvalues ​​are within the predetermined distribution range as having good quality, and marking the input feature matrix whose numerical distribution eigenvalues ​​are not within the predetermined distribution range as having poor quality; Step C20, assigning a first weight to the input feature matrix marked as good quality, and assigning a second weight to the input feature matrix marked as poor quality, wherein the first weight is greater than the second weight; Step C30, inputting the weighted input feature matrix and the output parameter matrix into a predetermined initial regression model for training until the predetermined initial regression model converges to obtain a parameter prediction model.

[0044] It should be noted that the predetermined distribution range is a pre-set range of ink droplet morphology parameters that characterize a better ink droplet distribution. For example, the predetermined distribution range of the volume mean is 1pl to 10pl, the predetermined distribution range of the velocity mean is 0.5m / s to 5m / s, and the predetermined distribution range of the angle mean is 80° to 95°.

[0045] In this embodiment, the input feature matrix whose numerical distribution eigenvalues ​​are within the predetermined distribution range can be marked as having good quality, and the input feature matrix whose numerical distribution eigenvalues ​​are not within the predetermined distribution range can be marked as having poor quality; a first weight is assigned to the input feature matrix marked as having good quality, and a second weight is assigned to the input feature matrix marked as having poor quality, wherein the first weight is greater than the second weight; and the input feature matrix and the output parameter matrix after weight assignment are input into a predetermined initial regression model for training until the predetermined initial regression model converges to obtain a parameter prediction model.

[0046] In this embodiment, by assigning a higher weight to the input feature matrix with better ink droplet distribution and a lower weight to the input feature matrix with poor ink droplet distribution, the trained parameter prediction model can be more inclined to output waveform parameters that can produce better ink droplet distribution.

[0047] In the second embodiment of the present application, the embodiment generates multiple sets of sample waveform parameter combinations; controls the specified inkjet printing device to perform inkjet printing operations according to each of the sample waveform parameter combinations, and collects ink droplet morphology data under each of the sample waveform parameter combinations as a training sample set; and obtains a parameter prediction model based on the training sample set. Thus, a parameter prediction model capable of describing the mapping relationship between the waveform parameters of the specified inkjet printing device and the ink droplet morphology is obtained through training.

[0048] Based on the first embodiment of the present application, in the third embodiment of the present application, the same or similar contents as those in the above-mentioned embodiment 1 can be referred to the above introduction, and will not be repeated in the following. Figure 4 After step S20, the waveform debugging method further includes steps S30 to S40: Step S30, controlling the designated inkjet printing device to perform inkjet printing according to the target waveform parameters to obtain a real-time ink droplet morphology, and comparing the real-time ink droplet morphology with the target ink droplet morphology data to obtain an ink droplet error; Step S40, after the ink drop error is not less than a predetermined error threshold, a parameter adjustment amount is calculated according to the ink drop error, and the target waveform parameters are updated according to the parameter adjustment amount until the ink drop error is less than the predetermined error threshold, thereby obtaining updated target waveform parameters.

[0049] It should be noted that the predetermined error threshold is a threshold that indicates a large error between the real-time ink droplet morphology data and the target ink droplet morphology data.

[0050] As the designated inkjet printing device is used, the real-time ink droplet morphology ejected by the designated inkjet printing device may not match the target ink droplet morphology data under the target waveform parameters output by the parameter prediction model due to the wear and tear of the device parts, replacement, ink component differences, etc. This embodiment can also control the designated inkjet printing device to perform inkjet printing operations according to the target waveform parameters to obtain real-time ink droplet morphology, and compare the real-time ink droplet morphology with the target ink droplet morphology data to obtain ink droplet errors. The ink droplet error may include the volume error, speed error, and angle error of the real-time ink droplet ejected. It can then be determined whether the ink droplet error is less than a predetermined error threshold. If the ink droplet error is less than the predetermined error threshold, the designated inkjet printing device is controlled to perform inkjet printing operations based on the target waveform parameters. After the ink droplet error is not less than the predetermined error threshold, the parameter adjustment amount is calculated based on the ink droplet error. Exemplarily, the parameter adjustment amount may be obtained using a PID algorithm, an incremental adjustment algorithm, a least squares method, a minimum likelihood estimation algorithm, and the like. Taking the incremental adjustment algorithm as an example, for each waveform parameter in the target waveform parameter, if the ink drop error is a positive value, the value of each waveform parameter is reduced; if the ink drop error is a negative value, the value of each waveform parameter is increased, wherein the parameter adjustment value is scaled according to the size of the ink drop error. Then, the target waveform parameter is updated according to the parameter adjustment amount. Further, it can be checked whether each waveform parameter in the updated target waveform parameter is within the predetermined operating range of the specified inkjet printing device (i.e., the parameter adjustment range that can be achieved by the specified inkjet printing device). If there is a waveform parameter in the target waveform parameter that exceeds the predetermined operating range, the waveform parameter is adjusted back to the nearest boundary value of the predetermined operating range. If there is no waveform parameter in the target waveform parameter that exceeds the predetermined operating range, step S30 can be performed until the ink drop error is less than the predetermined error threshold, and the updated target waveform parameter is obtained. Then, the specified inkjet printing device can be controlled to perform inkjet printing based on the updated target waveform parameter. Therefore, this embodiment can provide real-time feedback and adjust the target waveform parameters during the subsequent inkjet printing process, so that under the interference of other factors, the specified inkjet printing device can still spray ink droplets that meet the target ink droplet morphology data.

[0051] In a feasible implementation manner, step S40 may include steps S50 to S60: Step S50, using the updated target waveform parameters and the corresponding target ink droplet shape data as new training samples; Step S60, iteratively updating the parameter prediction model based on the new training sample to obtain a new parameter prediction model.

[0052] This embodiment can use the updated target waveform parameters and the corresponding target ink droplet morphology data as new training samples. Then obtain the current training sample set of the parameter prediction model, and use the new training samples and the current training sample set as new training sample sets. Then input the new training sample set into a predetermined initial regression model for training until the predetermined initial regression model converges to obtain a new parameter prediction model. As another example, this embodiment can also obtain the current training sample set of the parameter prediction model, and use the new training samples and the current training sample set as new training sample sets; assign a third weight to the input feature matrix in the new training sample, and assign a fourth weight to the input feature matrix in the current training sample of the current training sample set, wherein the third weight is greater than the fourth weight; input the new training sample set after weight assignment into the predetermined initial regression model for training until the predetermined initial regression model converges to obtain a new parameter prediction model.

[0053] In a feasible implementation, step S60 may include steps S61 to S63: Step S61, obtaining the current training sample set of the parameter prediction model, and using the new training sample and the current training sample set as a new training sample set; Step S62, assigning a third weight to the input feature matrix in the new training sample, and assigning a fourth weight to the input feature matrix in the current training sample of the current training sample set, wherein the third weight is greater than the fourth weight; Step S63, inputting the new training sample set after weight allocation into a predetermined initial regression model for training until the predetermined initial regression model converges to obtain a new parameter prediction model.

[0054] This embodiment can obtain the current training sample set of the parameter prediction model, merge the training sample and the current training sample set into a new training sample set, and then assign a third weight to the input feature matrix in the new training sample, and assign a fourth weight to the input feature matrix in the current training sample of the current training sample set, wherein the third weight is greater than the fourth weight. Then, the new training sample set after the weight assignment is input into the predetermined initial regression model for training until the predetermined initial regression model converges to obtain a new parameter prediction model. Thus, the new parameter prediction model can replace the original parameter test model. Therefore, this embodiment assigns a higher weight to the new training sample, so that the new parameter prediction model is more biased towards the new training sample, so that the new parameter prediction model matches the real-time status of the specified inkjet printing device, thereby further improving the prediction accuracy of the parameter prediction model for the target waveform parameters.

[0055] In the third embodiment of the present application, the specified inkjet printing device is controlled to perform inkjet printing according to the target waveform parameters to obtain the real-time ink droplet morphology, and the real-time ink droplet morphology is compared with the target ink droplet morphology data to obtain the ink droplet error; after the ink droplet error is not less than the predetermined error threshold, the parameter adjustment amount is calculated according to the ink droplet error, and the target waveform parameters are updated according to the parameter adjustment amount until the ink droplet error is less than the predetermined error threshold, thereby obtaining the updated target waveform parameters. Therefore, this embodiment can provide real-time feedback and adjust the target waveform parameters in the subsequent inkjet printing process, so that under the interference of other factors, the specified inkjet printing device can still spray ink droplets that meet the target ink droplet morphology data.

[0056] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the waveform debugging method of the present application. More simple transformations based on this technical concept are all within the scope of protection of the present application.

[0057] The present application provides a waveform debugging device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the waveform debugging method in the above-mentioned embodiment one.

[0058] Reference below Figure 5 , which shows a schematic diagram of the structure of a waveform debugging device suitable for implementing the embodiment of the present application. The waveform debugging device in the embodiment of the present application may include but is not limited to terminal devices such as laptop computers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions: tablet computers), desktop computers, servers, etc. Figure 5 The waveform debugging device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0059] like Figure 5As shown, the waveform debugging device may include a processing device 1001 (such as a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the waveform debugging device are also stored. The processing device 1001, ROM1002 and RAM1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the waveform debugging device to communicate with other devices wirelessly or by wire to exchange data. Although the figure shows a waveform debugging device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or have alternatively.

[0060] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0061] The waveform debugging device provided by the present application adopts the waveform debugging method in the above embodiment, which can solve the technical problem that the existing waveform debugging process is easily affected by subjective factors, resulting in poor debugging stability. Compared with the prior art, the beneficial effects of the waveform debugging device provided by the present application are the same as the beneficial effects of the waveform debugging method provided by the above embodiment, and the other technical features in the waveform debugging device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0062] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0063] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0064] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, and the computer-readable program instructions are used to execute the waveform debugging method in the above-mentioned embodiment.

[0065] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM: Random Access Memory), a read-only memory (ROM: Read Only Memory), an erasable programmable read-only memory (EPROM: Erasable Programmable Read Only Memory or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM: CD-Read Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency: Radio Frequency), etc., or any suitable combination of the above.

[0066] The computer-readable storage medium may be included in the waveform debugging device; or it may exist independently without being assembled into the waveform debugging device.

[0067] The above-mentioned computer-readable storage medium carries one or more programs. When the above-mentioned one or more programs are executed by the waveform debugging device, the waveform debugging device: obtains target ink droplet morphology data and a parameter prediction model of a specified inkjet printing device, wherein the parameter prediction model is a machine learning model that describes the mapping relationship between the waveform parameters and ink droplet morphology of the specified inkjet printing device; inputs the target ink droplet morphology data into the parameter prediction model to obtain the target waveform parameters of the specified inkjet printing device that match the target ink droplet morphology data.

[0068] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0069] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the 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 square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0070] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0071] The readable storage medium provided by the present application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned waveform debugging method, and can solve the technical problem that the existing waveform debugging process is easily affected by subjective factors, resulting in poor debugging stability. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided by the present application are the same as the beneficial effects of the waveform debugging method provided by the above-mentioned embodiment, and will not be repeated here.

[0072] The present application also provides a computer program product, including a computer program, which implements the steps of the waveform debugging method as described above when executed by a processor.

[0073] The computer program product provided by the present application can solve the technical problem that the existing waveform debugging process is easily affected by subjective factors, resulting in poor debugging stability. Compared with the prior art, the beneficial effects of the computer program product provided by the present application are the same as the beneficial effects of the waveform debugging method provided by the above embodiment, and will not be repeated here.

[0074] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A waveform debugging method, characterized in that: The waveform debugging method comprises: Acquire target ink droplet morphology data and a parameter prediction model of a specified inkjet printing device, wherein the parameter prediction model is a machine learning model that describes a mapping relationship between waveform parameters of the specified inkjet printing device and ink droplet morphology; Inputting the target ink droplet morphology data into a parameter prediction model to obtain target waveform parameters that match the target ink droplet morphology data for the designated inkjet printing device; Before the step of obtaining the parameter prediction model of the specified inkjet printing device, it includes: Generate multiple groups of sample waveform parameter combinations; Controlling the designated inkjet printing device to perform inkjet printing operations according to each of the sample waveform parameter combinations, and collecting ink droplet morphology data under each of the sample waveform parameter combinations as a training sample set; A parameter prediction model is obtained by training based on the training sample set; The step of collecting ink droplet morphology data under each of the sample waveform parameter combinations as a training sample set includes: Collecting ink droplet morphology images and ink droplet morphology parameters under each of the sample waveform parameter combinations as ink droplet morphology data; Calculating the numerical distribution characteristic value of the ink droplet morphology parameter, and constructing a parameter feature based on the numerical distribution characteristic value corresponding to the ink droplet morphology parameter; Extracting features from the ink droplet morphology image to obtain ink droplet image features; The parameter features and the ink drop image features are concatenated to obtain an input feature matrix; An output parameter matrix is ​​constructed based on the sample waveform parameter combination; The input feature matrix and the corresponding output parameter matrix are used as training samples, and each of the training samples is used as a training sample set.

2. The waveform debugging method according to claim 1, characterized in that: The step of obtaining a parameter prediction model based on the training sample set training comprises: Marking the input feature matrix whose numerical distribution feature value is within the predetermined distribution range as having good quality, and marking the input feature matrix whose numerical distribution feature value is not within the predetermined distribution range as having poor quality; Assigning a first weight to an input feature matrix marked as having good quality, and assigning a second weight to an input feature matrix marked as having poor quality, wherein the first weight is greater than the second weight; The input feature matrix and the output parameter matrix after weight allocation are input into a predetermined initial regression model for training until the predetermined initial regression model converges to obtain a parameter prediction model.

3. The waveform debugging method according to any one of claims 1 to 2, characterized in that: After the step of inputting the target ink droplet morphology data into a parameter prediction model to obtain target waveform parameters matching the specified inkjet printing device with the target ink droplet morphology data, the waveform debugging method further includes: Controlling the designated inkjet printing device to perform inkjet printing according to the target waveform parameters to obtain a real-time ink droplet morphology, and comparing the real-time ink droplet morphology with the target ink droplet morphology data to obtain an ink droplet error; After the ink drop error is not less than a predetermined error threshold, a parameter adjustment amount is calculated according to the ink drop error, and the target waveform parameters are updated according to the parameter adjustment amount until the ink drop error is less than the predetermined error threshold, thereby obtaining updated target waveform parameters.

4. The waveform debugging method according to claim 3, characterized in that: After the step of updating the target waveform parameters according to the parameter adjustment amount until the ink drop error is less than a predetermined error threshold to obtain updated target waveform parameters, the method includes: Using the updated target waveform parameters and corresponding target ink droplet morphology data as new training samples; The parameter prediction model is iteratively updated based on the new training samples to obtain a new parameter prediction model.

5. The waveform debugging method according to claim 4, characterized in that: The step of iteratively updating the parameter prediction model based on the new training sample to obtain a new parameter prediction model includes: Acquire the current training sample set of the parameter prediction model, and use the new training sample and the current training sample set as a new training sample set; Assigning a third weight to the input feature matrix in the new training sample, and assigning a fourth weight to the input feature matrix in the current training sample of the current training sample set, wherein the third weight is greater than the fourth weight; The new training sample set after weight allocation is input into the predetermined initial regression model for training until the predetermined initial regression model converges to obtain a new parameter prediction model.

6. A waveform debugging device, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the waveform debugging method according to any one of claims 1 to 5.

7. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the waveform debugging method according to any one of claims 1 to 5 are implemented.

8. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the steps of the waveform debugging method according to any one of claims 1 to 5 are implemented.

Images