Intelligent ink drop control method and system for digital printing press and digital printing press

Abstract

The present invention relates to the technical field of digital printing presses, and in particular to an intelligent ink drop control method, system and digital printing press of a digital printing press. The method of the present invention includes real-time data acquisition, multi-dimensional parameter modeling, real-time ink drop behavior prediction, adaptive inkjet parameter adjustment, closed-loop feedback control, adaptive matching of medium characteristics, intelligent multi-region optimization and environmental factor compensation. Ink drop behavior and environmental data are collected in real time by high-speed cameras and multiple sensors, and a nonlinear model is constructed using deep learning technology to generate a dynamic parameter optimization curve. The method of the present invention significantly improves printing quality and efficiency, and is applicable to a variety of media and complex scenes.

Description

Technical Field

[0001] The present invention relates to the technical field of digital printing presses, and particularly to an intelligent ink droplet control method, system and digital printing press for a digital printing press. Background Art

[0002] Digital printing technology has developed rapidly in recent years. With its advantages of no need for plate making, on-demand printing and high efficiency and flexibility, it has been widely used in the fields of commercial printing, packaging printing and industrial applications. During the inkjet process of traditional digital printing equipment, the speed, shape and landing position of ink droplets are key factors affecting printing quality. However, due to the dynamic changes in the printing environment, medium characteristics and equipment performance, it is difficult for the existing technology to ensure the stability and accuracy of ink droplet behavior, which poses a severe challenge to high-precision and high-consistency printing.

[0003] Traditional solutions of digital printing technology mainly focus on inkjet print head control, fixed parameter setting and simple feedback adjustment mechanisms. These technologies have laid the foundation for the initial popularization of digital printing equipment, but still have significant limitations when facing complex environments, diverse media and high-precision requirements.

[0004] 1. High environmental sensitivity: Environmental factors such as temperature, humidity, static electricity and equipment vibration will significantly affect the flight trajectory and landing point of ink droplets, resulting in problems such as blurred image edges and uneven colors. Traditional inkjet devices usually adopt fixed parameter settings and cannot adapt to real-time environmental changes.

[0005] 2. Poor medium adaptability: Different printing media (such as paper, plastic, fabric and metal surfaces) have different ink absorbency, roughness and conductivity. Traditional systems often require manual adjustment of inkjet parameters, which is both time-consuming and difficult to ensure optimal parameters.

[0006] 3. Insufficient dynamic adjustment ability: Inkjet parameters (such as voltage, pulse frequency and ejection angle) are usually set before printing, and there is a lack of real-time feedback and dynamic adjustment mechanism during the actual inkjet process, resulting in limited accuracy and efficiency.

[0007] 4. Nonlinear complexity not fully resolved: The relationship between ink droplet behavior and environment, medium and inkjet parameters is highly nonlinear and dynamic. The existing technology lacks effective modeling and optimization means and is difficult to comprehensively improve printing quality.

[0008] To solve the above problems, some advanced digital printing technologies attempt to introduce sensor networks and simple feedback mechanisms to adjust inkjet parameters by real-time monitoring of the environment and medium state. However, these technologies are usually based on static models, unable to capture the complex nonlinear relationships between multi-dimensional parameters, and also unable to make full use of the dynamic information in the time series. Therefore, the improvement of printing accuracy and efficiency is still limited. Summary of the Invention

[0009] To solve the above technical problems, the object of the present invention is to provide an intelligent ink droplet control method for a digital printing press. This method constructs a non-linear multi-dimensional model, integrates environmental parameters, medium characteristics, and inkjet parameters, realizes multi-module linkage, dynamically predicts the flight behavior of ink droplets, and updates the inkjet parameters in real time through high-frequency control signals. This method combines an innovative non-linear model of time series analysis and deep learning, and can adapt to environmental changes and medium differences in real time. At the same time, using a high-speed vision monitoring system and an environmental sensor network, it dynamically collects inkjet behavior and environmental state data, generates an optimization curve, and guides the adjustment of inkjet parameters. Through this method, the printing accuracy, efficiency, and stability can be significantly improved, meeting the high-quality printing requirements under complex environments and multiple media.

[0010] To achieve the above object, the present invention adopts the following technical solutions:

[0011] An intelligent ink droplet control method for a digital printing press, characterized in that the method comprises the following steps:

[0012] 1) Real-time data collection: Collect data on the shape, speed, landing position, and color distribution of ink droplets during the inkjet process, as well as environmental parameters, through a high-speed camera and environmental sensors. The environmental parameters include temperature, humidity, static electricity, and the surface roughness of the medium.

[0013] 2) Multi-dimensional parameter modeling: Use deep learning technology to construct a non-linear model between ink droplet behavior, environmental factors, medium characteristics, and inkjet parameters, and generate a dynamic parameter optimization curve.

[0014] 3) Real-time ink droplet behavior prediction: Predict the flight trajectory deviation, deformation of ink droplets, and their impact on the target printing quality based on the time series analysis algorithm.

[0015] 4) Adaptive inkjet parameter adjustment: Dynamically adjust the voltage, pulse frequency, ejection time, and nozzle opening and closing state of the inkjet head according to the ink droplet behavior prediction results to optimize the size, speed, and ejection angle of the ink droplets.

[0016] 5) Closed-loop feedback control: Generate a control signal by real-time monitoring of the ink droplet landing point deviation data, and update the inkjet parameters at a frequency of at least 5000 times per second.

[0017] Preferably, the multi-dimensional parameter modeling uses a combined model of a convolutional neural network (CNN) and a recurrent neural network (RNN) to analyze both the spatial behavior characteristics of ink droplets and the dynamic changes in the time series.

[0018] Preferably, the non-linear model uses multi-layer non-linear transformation to integrate environmental parameters, medium characteristics, and inkjet parameters into a dynamic prediction framework:

[0019] ;

[0020] where: y(t): the behavior of the ink droplet includes velocity, size, and landing deviation; Φ: a dynamic multi-layer non-linear function implemented by a neural network; ϵ(t): unmodeled error.

[0021] The model structure combines the following multi-level non-linear calculations:

[0022] ;

[0023] X = [E, M, P]: input vector, including environmental parameter E, medium property M, and inkjet parameter P;

[0024] W 1 , W 2 , W 3 : weight matrices optimized by training data;

[0025] b 1 , b 2 , b 3 : bias vectors;

[0026] σ(·): activation function;

[0027] The input parameter X is processed as follows:

[0028] Environmental parameter E(t) = [T, H, S, V]: including temperature and humidity, static electricity, and vibration;

[0029] Medium property M(t) = [A, R, C]: ink absorbency, roughness, and conductivity;

[0030] Inkjet parameter P(t) = [V s , F p , θ]: nozzle voltage, pulse frequency, and ejection angle;

[0031] Decomposed by the deep feature extraction layer into:

[0032] Environmental module: ;

[0033] Medium module: ;

[0034] Inkjet module: ;

[0035] Finally, combine the outputs of these modules:

[0036] ;

[0037] The output passes through the final layer:

[0038] ;

[0039] Combine with the dynamic behavior of the time series and add time-related parameters:

[0040] ;

[0041] Wherein α i is the time decay coefficient, which is used to weight the influence of historical data on the current decision.

[0042] Preferably, the method further includes analyzing the ink absorption, roughness, and conductivity of the medium surface through a medium recognition module, and dynamically loading an optimized inkjet parameter solution for the corresponding medium.

[0043] Preferably, the medium recognition module automatically identifies the medium type and loads the optimal inkjet parameters through a high-resolution scanning device and material library matching technology.

[0044] Preferably, the method further includes dividing the target image into multiple regions by using a region segmentation algorithm, and adjusting the inkjet parameters for each region separately.

[0045] Preferably, the method further includes dynamically adjusting the inkjet parameters in combination with real-time environmental data to compensate for inkjet deviations caused by changes in temperature, humidity, or static electricity.

[0046] Preferably, the environmental factor compensation includes adjusting the electric field distribution of the nozzle according to the data of the static electricity sensor to reduce the deviation of the ink droplets during the spraying process.

[0047] Furthermore, the present invention also provides an intelligent ink droplet control system for a digital printing press, which implements the method described above, including:

[0048] 1) High-speed vision monitoring unit: used to collect real-time data on the position, shape, and color distribution of ink droplets;

[0049] 2) Environmental perception module: including temperature and humidity sensors, static electricity sensors, and medium roughness detection devices;

[0050] 3) Deep learning module: used to analyze the collected data and generate optimized inkjet parameters;

[0051] 4) Inkjet control unit: connected to the deep learning module, used to dynamically adjust the voltage, pulse frequency, and spraying time of the inkjet head;

[0052] 5) Closed-loop feedback control module: generates a control signal according to real-time monitoring data;

[0053] 6) Medium recognition module: used to analyze and match the characteristic parameters of different media.

[0054] Furthermore, the present invention also provides a digital printing press, comprising:

[0055] 1) The intelligent ink droplet control system described above;

[0056] 2) A media transport module: used to transport various media including paper, plastic, fabric, and metal surfaces;

[0057] 3) An ink system: supporting UV-curable ink and water-based ink, and having an automatic cleaning function;

[0058] 4) A control panel: used to set and monitor printing parameters in real time.

[0059] Due to the adoption of the above technical solutions, the present invention realizes the intelligence and high efficiency of the digital printing press in the dynamic adjustment of multi-dimensional parameters by introducing deep learning modeling, non-linear dynamic optimization, and closed-loop control technologies. The specific technical effects include the following points:

[0060] 1. Improve printing quality:

[0061] The present invention realizes the dynamic optimization of ink droplet behavior by collecting data on the shape, speed, landing position, and color distribution of ink droplets in real time and combining deep learning modeling. At the same time, the accuracy of the ink droplet landing point is improved, effectively reducing image blurring and color unevenness caused by offset and deformation. Further, the present invention adopts a region segmentation algorithm to independently adjust the inkjet parameters for different regions of the target image, ensuring the printing quality uniformity and detail expressiveness of complex images and multi-level structures.

[0062] 2. Enhance environmental adaptability:

[0063] The present invention collects data on temperature, humidity, static electricity, and media surface roughness in real time through an environmental perception module and dynamically adjusts the inkjet parameters, significantly reducing the impact of environmental fluctuations on printing quality. At the same time, based on the static electricity sensor data, the electric field distribution of the nozzle is adjusted, effectively reducing the offset phenomenon of ink droplets during the spraying process.

[0064] 3. Improve multi-media compatibility

[0065] The present invention automatically analyzes the ink absorption, roughness, and conductivity of the media through a media recognition module and dynamically loads the optimal inkjet parameters, enabling it to adapt to various media such as paper, plastic, fabric, and metal. It solves the problem that traditional printing presses need to be manually adjusted frequently during the media type switching process. Further, a high-resolution scanning device combined with a material library matching technology provides exclusive optimization solutions for different media, improving the wide compatibility and operation convenience of the printing press.

[0066] 4. Improve printing efficiency

[0067] The present invention utilizes closed-loop feedback control with a parameter update frequency of at least 5000 times per second to ensure that the inkjet parameters are always in the optimal state during the high-speed inkjet process. This reduces the downtime for adjustment and production interruptions, and improves the equipment utilization rate. At the same time, by combining time series analysis technology, it can predict in real time the deviation and deformation of the ink droplet flight trajectory, thereby optimizing the inkjet parameters in advance and shortening the response time for parameter adjustment.

[0068] 5. Resource conservation

[0069] The present invention dynamically adjusts the ink droplet size and ejection angle, reduces the phenomenon of ink overflow and repeated ejection, and improves the ink utilization rate by at least 30%. The high-precision inkjet control and area-level optimization significantly reduce the scrap rate caused by printing deviation, saving the media cost.

[0070] 6. Improve system stability and reliability

[0071] The deep learning module of the present invention continuously optimizes the non-linear model by analyzing the historical data of the environment, media, and inkjet parameters, improving the long-term stability and intelligent level of the system. Whether in complex environments (such as high humidity, high static electricity) or multi-media scenarios, the present invention can maintain stable performance and reduce the risk of downtime or failure caused by environmental changes.

[0072] In summary, the present invention integrates real-time data acquisition, deep learning modeling, dynamic parameter optimization, and closed-loop control technologies, providing a highly intelligent solution for digital printing presses. It can adapt to a variety of complex environments and media, improve printing quality, efficiency, and resource utilization rate, while reducing the need for manual intervention and significantly enhancing the overall performance and market competitiveness of digital printing equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 is a flowchart of the method of the present invention.

[0074] Figure 2 is a dynamic curve graph of the fluctuations of temperature (red dashed line) and humidity (blue dotted line) over time.

[0075] Figure 3 is a dynamic curve graph of voltage (green solid line) and frequency (orange solid line) based on the optimization model. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0076] Next, in combination with the embodiments of the present invention, the technical solutions in the embodiments will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0077] AsFigure 1 An intelligent ink droplet control method for a digital printing press, the method comprising the following steps:

[0078] 1) Real-time data acquisition: Collect data on the shape, speed, landing position, and color distribution of ink droplets during the inkjet process, as well as environmental parameters, through a high-speed camera and environmental sensors. The environmental parameters include temperature, humidity, static electricity, and the surface roughness of the medium.

[0079] 2) Multidimensional parameter modeling: Use deep learning technology to construct a non-linear model between ink droplet behavior, environmental factors, medium characteristics, and inkjet parameters, and generate a dynamic parameter optimization curve.

[0080] 3) Real-time ink droplet behavior prediction: Predict the flight trajectory deviation, deformation of ink droplets, and their impact on the target printing quality based on time series analysis algorithms.

[0081] 4) Adaptive inkjet parameter adjustment: Dynamically adjust the voltage, pulse frequency, ejection time, and nozzle opening / closing state of the inkjet head according to the ink droplet behavior prediction results to optimize the size, speed, and ejection angle of the ink droplets.

[0082] 5) Closed-loop feedback control: Generate a control signal by real-time monitoring of ink droplet landing point deviation data, and update the inkjet parameters at a frequency of at least 5000 times per second.

[0083] 6) Adaptive matching of medium characteristics: Analyze the ink absorption, roughness, and conductivity of the medium surface through a medium identification module, and dynamically load the inkjet parameter optimization scheme corresponding to the medium.

[0084] 7) Intelligent multi-region optimization: Use a region segmentation algorithm to divide the target image into multiple regions, and adjust the inkjet parameters for each region separately.

[0085] 8) Environmental factor compensation: Dynamically adjust the inkjet parameters in combination with real-time environmental data to compensate for inkjet deviations caused by changes in temperature, humidity, or static electricity.

[0086] I. Real-time data acquisition

[0087] In order to accurately capture the ink droplet behavior and environmental changes during the inkjet process, real-time data acquisition needs to be implemented in combination with a high-speed camera, environmental sensors, and data preprocessing technology.

[0088] 1. High-speed camera collects ink droplet data

[0089] 1.1 Camera selection:

[0090] Use an industrial-grade high-speed camera with a high frame rate (≥5000 FPS) to ensure continuous imaging of ink droplets during flight. The resolution selection should meet the requirements for ink droplet detail detection (such as ≥1920×1080 pixels).

[0091] 1.2 Data acquisition process:

[0092] 1.2.1 Shape detection:

[0093] Extract the contour features of the ink droplets through an image segmentation algorithm. Apply edge detection methods (such as the Canny algorithm) to obtain the shape of the ink droplets.

[0094] 1.2.2 Velocity calculation:

[0095] Calculate the flying speed based on the position change of the ink droplets in consecutive frames:

[0096] .

[0097] 1.2.3 Landing position recording:

[0098] Use the image coordinates to convert to actual space coordinates and record the landing point of the ink droplet: (x actual , y actual ).

[0099] 1.2.4 Color distribution analysis:

[0100] Extract the color channel values of the ink droplet image (such as RGB or LAB color space), and statistically analyze the color uniformity and saturation.

[0101] 1.2.5 Synchronization processing:

[0102] Ensure the time synchronization between the camera and the inkjet action to avoid data offset.

[0103] 2. Environmental sensors collect environmental parameters

[0104] 2.1 Sensor selection:

[0105] Temperature and humidity sensor: Measure the temperature T(t) and humidity H(t) of the inkjet environment. Select a high-precision sensor (such as a temperature error of ±0.1°C and a humidity error of ±1%).

[0106] Electrostatic sensor: Measure the electrostatic intensity S(t) in the inkjet environment, using a non-contact electrostatic sensor.

[0107] Roughness detection sensor: Use a laser or optical sensor to measure the micro-roughness RRR of the medium surface.

[0108] 2.2 Data acquisition process:

[0109] Temperature and humidity: Collect temperature and humidity data T(t) and H(t) once per second and record the change trend.

[0110] Electrostatic: Measure the electrostatic field intensity S(t) near the inkjet path and the medium surface.

[0111] Roughness: The laser sensor scans the surface of the medium and calculates the roughness parameter R, using the root mean square height (RMS) index:

[0112] 。

[0113] 2.3 Data fusion:

[0114] Synchronously fuse the data of each sensor to form a unified input of environmental parameters.

[0115] 3. Data preprocessing

[0116] 3.1 Denoising processing:

[0117] Apply Gaussian filtering or median filtering to the images collected by the camera to remove noise. Use Kalman filtering for the sensor data to smooth the changes in environmental parameters.

[0118] 3.2 Normalization:

[0119] Normalize data with different dimensions to a unified range (such as [0, 1]):

[0120] 。

[0121] 3.3 Time synchronization:

[0122] Use a unified clock to align the time of the camera and sensor data to ensure data consistency.

[0123] II. Multidimensional parameter modeling

[0124] The multidimensional parameter modeling of the present invention combines a convolutional neural network (CNN) and a recurrent neural network (RNN) to establish a non-linear relationship between ink droplet behavior, environmental factors, medium characteristics, and inkjet parameters, and then generates a dynamic parameter optimization curve. Figure 2 The fluctuations of temperature (red dashed line) and humidity (blue dotted line) over time are given, simulating the changes under complex environmental conditions. Figure 3 The voltage (green solid line) and frequency (orange solid line) are given to be dynamically adjusted based on the optimization model to adapt to real-time environmental changes and inkjet requirements.

[0125] 1. Non-linear model

[0126] The non-linear model uses multi-layer non-linear transformation to integrate environmental parameters, medium characteristics, and inkjet parameters into a dynamic prediction framework:

[0127] ;

[0128] where: y(t): the behavior of the ink droplet includes velocity, size, and landing deviation; Φ: a dynamic multi-layer non-linear function implemented by a neural network; ϵ(t): unmodeled error.

[0129] The model structure combines the following multi-level non-linear calculations:

[0130] ;

[0131] X = [E, M, P]: the input vector, including the environmental parameter E, the medium property M, and the inkjet parameter P; W 1 , W 2 , W 3 : weight matrices, optimized by training data; b 1 , b 2 , b 3 : bias vectors; σ(·): activation function.

[0132] The input parameter X is processed as follows:

[0133] The environmental parameter E(t) = [T, H, S, V]: including temperature and humidity, static electricity, vibration;

[0134] The medium property M(t) = [A, R, C]: ink absorbency, roughness, conductivity;

[0135] The inkjet parameter P(t) = [V s , F p , θ]: nozzle voltage, pulse frequency, ejection angle;

[0136] It is decomposed by the deep feature extraction layer into:

[0137] Environmental module: ;

[0138] Medium module: ;

[0139] Inkjet module: ;

[0140] Finally, the outputs of these modules are combined:

[0141] ;

[0142] The output passes through the final layer:

[0143] ;

[0144] Combined with the time series dynamic behavior, time-related parameters are added:

[0145] ;

[0146] whereα i is the time decay coefficient, which is used to weight the influence of historical data on the current decision.

[0147] III. Real-time Ink Drop Behavior Prediction

[0148] The real-time ink drop behavior prediction of the present invention uses a time series analysis algorithm to predict the flight trajectory deviation, deformation of the ink drop and its impact on printing quality. The long short-term memory network (LSTM) is used to process the time series data. The following are the specific implementation steps:

[0149] 1. LSTM Model Structure

[0150] Input layer: Input the time series data X of the sliding window seq .

[0151] LSTM hidden layer: Capture the long-term and short-term dependencies in the time series:

[0152] ;

[0153] h t : Hidden state; c t : Cell state.

[0154] Output layer: Output the ink drop trajectory deviation Δx, Δy and deformation ΔS:

[0155] .

[0156] 2. Prediction Steps

[0157] (1) Prediction of Flight Trajectory Deviation

[0158] Predict the trajectory deviation of the ink drop during flight:

[0159] Model input: ;

[0160] Model output: .

[0161] (2) Prediction of Ink Drop Deformation

[0162] Predict the deformation of the ink drop caused by environmental and speed changes during flight:

[0163] Input: X seq；

[0164] Output: .

[0165] IV. Adaptive Inkjet Parameter Adjustment

[0166] Adaptive inkjet parameter adjustment analyzes the prediction results of droplet behavior and dynamically optimizes the parameters of the inkjet head (voltage, pulse frequency, ejection time, and nozzle opening / closing state) to ensure precise control of droplet size, velocity, and ejection angle. The following are the specific implementation methods:

[0167] 1. Input and analysis of prediction results

[0168] (1) Input data

[0169] Obtain the following prediction results from the real-time droplet behavior prediction module:

[0170] ;

[0171] Δx: Deviation of the droplet flight trajectory; ΔS(t): Degree of droplet deformation.

[0172] Meanwhile, the environmental and medium data collected in real time:

[0173] ;

[0174] Environmental parameters E(t)=[T(t),H(t),S(t)]; Medium characteristics M(t)=[A,R,C].

[0175] (2) Deviation definition

[0176] Landing deviation:

[0177] ;

[0178] Target behavior adjustment goal: Reduce Δd and minimize ΔS(t).

[0179] 2. Rules and formulas for dynamic parameter adjustment

[0180] (1) Nozzle voltage Vs

[0181] Adjust the voltage to optimize the droplet volume and reduce deformation.

[0182] Adjustment rule:

[0183]

[0184] K v : Adjustment coefficient (determined according to the medium characteristics); If the droplet is too large (ΔS>0), reduce V s ; If the droplet is too small (ΔS<0), increase V s .

[0185] (2) Pulse frequency Fp

[0186] Adjust the frequency to compensate for the trajectory deviation.

[0187] Adjustment rules:

[0188]

[0189] K f : Frequency adjustment coefficient; increase the frequency to improve the stability of ink droplets and reduce deviation.

[0190] (3) Jetting time t s

[0191] Optimize the momentum to correct the trajectory.

[0192] Adjustment rules:

[0193] t s (t)=t s (t−1)+K t (Δx+Δy)

[0194] K t : Time adjustment coefficient; dynamically adjust the jetting time according to the deviation direction.

[0195] (4) Nozzle opening and closing state

[0196] The nozzle opening and closing state determines the initial angle and shape of the ink droplets; dynamically control the nozzle opening and closing angle to correct the flight trajectory.

[0197] Adjustment rules:

[0198]

[0199] K θ : Angle adjustment coefficient; correct the jetting direction to reduce deviation.

[0200] 3. Parameter adjustment algorithm

[0201] Based on the input prediction data and real-time acquisition data, execute the following adaptive parameter adjustment process:

[0202] Step 1: Calculate the deviation

[0203] Calculate the trajectory deviation Δd and deformation ΔS according to the prediction result.

[0204] Judge the magnitude of the deviation:

[0205] If Δd > threshold, perform deviation compensation; if ΔS > threshold, optimize the deformation.

[0206] Step 2: Determine the adjustment amount

[0207] Calculate the adjustment parameters according to the magnitude of the deviation:

[0208]

[0209] The adjustment amount of each parameter is calculated by the above rules.

[0210] Step 3: Update parameters in real time

[0211] Adjust the parameters of the inkjet head in real time:

[0212] Voltage: V s (t + 1)=V s (t)+ΔV s ;

[0213] Frequency: F p (t + 1)=F p (t)+ΔF p ;

[0214] Time: t s (t + 1)=t s (t)+Δt s ;

[0215] Angle: θ(t + 1)=θ(t)+Δθ ;

[0216] V. Closed-loop feedback control

[0217] The closed-loop feedback control generates control signals to dynamically adjust the inkjet parameters by monitoring the deviation of the ink drop landing point in real time, ensuring the stability and accuracy of the ink drop behavior. The following is the specific implementation method:

[0218] 1. Data acquisition and deviation calculation

[0219] (1) Collect data in real time

[0220] Collect the following data through a high-speed camera and sensors:

[0221] Landing coordinates: The actual landing coordinates of the ink drop (x actual , y actual );

[0222] Target coordinates: The ideal landing coordinates of the ink drop (x target , y target ).

[0223] (2) Deviation calculation

[0224] Calculate the deviation between the actual landing point and the target landing point of the ink drop:

[0225] Δdx = x actual −x target , Δdy = y actual −ytarget

[0226]

[0227] Δdx, Δdy: Deviations in the X-axis and Y-axis directions; Δd: Overall deviation for subsequent control.

[0228] 2. Control Signal Generation

[0229] (1) PID Controller Design

[0230] Use a proportional-integral-derivative (PID) controller to generate a control signal for adjusting inkjet parameters.

[0231] The output of the PID controller is:

[0232]

[0233] u(t): Control signal (adjustment amount); e(t) = Δd(t): Current deviation; K p , K i , K d : Proportional, integral, and derivative gains.

[0234] Independently control Δdx and Δdy respectively:

[0235]

[0236]

[0237] 3. Parameter Adjustment

[0238] (1) Adjust the nozzle voltage V s

[0239] The voltage affects the droplet size, and the adjustment formula is:

[0240] ;

[0241] K v : Voltage adjustment coefficient.

[0242] (2) Adjust the pulse frequency F p

[0243] The pulse frequency affects the droplet velocity, and the adjustment formula is:

[0244]

[0245] K f : Frequency adjustment coefficient.

[0246] (3) Adjust the injection angle θ

[0247] The angle controls the landing direction, and the adjustment formula is:

[0248]

[0249] K θ : Angle adjustment coefficient.

[0250] (4) Dynamic jetting time t s

[0251] The jetting time affects the momentum of the ink droplet, and the adjustment formula is:

[0252]

[0253] K t : Time adjustment coefficient.

[0254] 4. High-frequency control update

[0255] (1) Control update frequency

[0256] The operating frequency of the closed-loop controller is at least 5000 times per second; each time the latest deviation data is collected, the control signal is generated in real time and the inkjet parameters are updated.

[0257] (2) Filtering processing

[0258] To avoid the interference of sensor noise on the control signal, filtering technology is used to smooth the input data:

[0259] Kalman filter:

[0260]

[0261] z t : Measured value; K t : Filtering gain.

[0262] VI. Adaptive matching of media characteristics

[0263] The adaptive matching of media characteristics analyzes the surface characteristics of the media (such as ink absorbency, roughness, conductivity, etc.) through the media identification module, and dynamically loads the optimal inkjet parameter scheme to achieve fast adaptation and efficient printing of various media. The following are the specific implementation methods:

[0264] 1. Data acquisition

[0265] (1) Media characteristic acquisition

[0266] The key characteristics of the media are detected in real time through a variety of sensors:

[0267] Ink absorbency A: Use a high-precision ink diffusion sensor or image analysis device to measure the diffusion rate and diffusion area of the ink droplet on the media surface. Measurement formula: A = diffusion area / ink droplet volume;

[0268] Surface roughness R: Use a laser displacement sensor or an optical microscope to scan the microscopic morphology of the medium surface and calculate the roughness parameters.

[0269] Typical indicators:

[0270] ;

[0271] z i : Surface height value; : Average surface height.

[0272] Conductivity C: Use a contact or non-contact conductivity sensor to detect the conductivity of the medium and evaluate the impact of static electricity on the inkjet behavior. Measurement unit: (S / m).

[0273] (2) Data acquisition synchronization

[0274] Synchronously collect the medium characteristic data [A, R, C] and the environmental data [T, H, S] to form complete input features.

[0275] 2. Medium identification

[0276] (1) Material classification

[0277] Classify the medium characteristic data through the medium identification module:

[0278] Use the labeled data to train a classification model (such as Support Vector Machine SVM or Convolutional Neural Network CNN).

[0279] Classify the medium into specific types (such as paper, plastic, fabric, metal, etc.) according to parameters such as ink absorbency A, roughness R, and conductivity C.

[0280] (2) Classification model formula

[0281] The classification model is Φ: medium type = Φ(A, R, C);

[0282] Φ is a non-linear mapping function based on deep learning.

[0283] (3) Data matching

[0284] According to the identified medium type, search for the corresponding optimized parameter solution in the material database:

[0285] P optimal = DB(Φ(A, R, C));

[0286] DB: Material database; P optimal : Optimized inkjet parameter solution.

[0287] 3. Dynamic Loading of Inkjet Parameters

[0288] (1) Parameter Mapping Rules

[0289] Map the optimization solutions in the material database to the inkjet parameters:

[0290] Nozzle voltage V s : V s = f 1 (A);

[0291] Adjust the voltage according to the ink absorption to control the droplet size.

[0292] Pulse frequency F p : F p = f 2 (R);

[0293] Adjust the frequency according to the roughness to optimize the droplet stability.

[0294] Jetting angle θ: θ = f 3 (C);

[0295] Adjust the jetting direction according to the conductivity to reduce electrostatic interference.

[0296] (2) Dynamic Parameter Adjustment Formula

[0297] Dynamically update the inkjet parameters according to the collected real-time data:

[0298] P(t) = P baseline + ΔP(A, R, C);

[0299] P baseline : Default inkjet parameters; ΔP(A, R, C): Parameter correction amount calculated based on the medium characteristics.

[0300] 4. Adaptive Matching Algorithm

[0301] (1) Data Processing

[0302] Collect input features: X = [A, R, C, T, H, S]

[0303] (2) Matching Algorithm

[0304] Based on the input features, dynamically select the optimization solution:

[0305] The preset rules select the inkjet parameters according to the input features; use the trained deep learning model to predict the parameter adjustment amount.

[0306] (3) Output

[0307] Generate the optimized inkjet parameters P optimal , including: nozzle voltage V s; Pulse frequency F p ; Jetting time t s ; Jetting angle θ.

[0308] VII. Intelligent multi - region optimization

[0309] Intelligent multi - region optimization divides the target image into multiple regions through a region segmentation algorithm and adjusts the ink - jet parameters separately based on the region characteristics to ensure the details and overall quality of complex image printing. The following is the specific implementation method:

[0310] 1. Data pre - processing

[0311] (1) Input data

[0312] Target image: The resolution of the input image I is (W, H), and the pixel values represent color or grayscale.

[0313] Segmentation target: Divide multiple regions {R1, R2, …, Rn} according to image characteristics (such as brightness, texture, color, etc.).

[0314] (2) Normalize the pixel values of the input image to ensure that the pixel value range is [0, 1][0, 1][0, 1].

[0315] 2. Region segmentation algorithm

[0316] (1) Segmentation basis

[0317] Segment the image according to the following characteristics:

[0318] Color: Based on pixel color (such as RGB, LAB color space).

[0319] Brightness: Based on the grayscale value distribution.

[0320] Texture: Based on local features (such as gradient or wavelet transform).

[0321] Complexity: The richness of regional details.

[0322] (2) Algorithm

[0323] Select a suitable segmentation algorithm to divide the image into multiple regions.

[0324] 2.1 Threshold - based segmentation

[0325] Compare the pixel values of the image with one or more thresholds to divide regions:

[0326]

[0327] T i : Threshold, determined according to the image histogram or Otsu algorithm.

[0328] 2.2 K-means Clustering

[0329] Cluster the pixel values and assign each pixel to the nearest cluster center.

[0330] Clustering formula:

[0331] ;

[0332] k: The number of divided regions; μ i : The mean value of each region.

[0333] 2.3 Superpixel Segmentation

[0334] Use the SLIC (Simple Linear Iterative Clustering) algorithm to generate superpixels.

[0335] Superpixel region merging formula:

[0336] ;

[0337] c: Color distance weight; s: Spatial distance weight.

[0338] 3. Region Feature Analysis

[0339] (1) Color and Brightness Features

[0340] Calculate the color mean and brightness mean of each region:

[0341] ;

[0342] I c (x,y): Pixel color value; I l (x,y): Pixel brightness value.

[0343] (2) Texture Features

[0344] Extract the texture features of each region (such as gradient magnitude or wavelet energy):

[0345]

[0346] T i : The average texture feature of the region.

[0347] (3) Complexity Features

[0348] Define the detail complexity of the region:

[0349]

[0350] Var: The variance of pixel values within the region.

[0351] 4. Inkjet Parameter Optimization

[0352] (1) Inkjet Parameter Mapping

[0353] Dynamically adjust inkjet parameters according to regional characteristics:

[0354] Nozzle voltage V s :

[0355] Adjust the droplet size according to the average color and brightness: V s =f 1 (C i ,L i )

[0356] Pulse frequency F p :

[0357] Adjust the droplet velocity according to the texture characteristics: F p =f 2 (T i )

[0358] Jetting time t s :

[0359] Adjust the droplet momentum according to the complexity: t s =f 3 (D i )

[0360] (2) Dynamically Load Parameters

[0361] Apply the optimized parameters to the corresponding area:

[0362] P i ={V s ,F p ,t s ,θ},∀R i .

[0363] 5. Multi - area Inkjet Execution

[0364] (1) Area Priority Sorting

[0365] Sort according to the regional complexity Di, and give priority to processing areas with complex details:

[0366] P riority (R i )=D i

[0367] (2) Step - by - step Inkjet Execution

[0368] Complete the inkjet tasks for each area in order of priority to ensure that details are prioritized.

[0369] VIII. Environmental Factor Compensation

[0370] Environmental factor compensation dynamically adjusts the inkjet parameters (voltage, pulse frequency, ejection time, and ejection angle) of the inkjet head by real-time monitoring of parameters such as temperature, humidity, and static electricity, so as to reduce the impact of environmental changes on the behavior of ink droplets. The following are the specific implementation methods:

[0371] 1. Environmental Data Acquisition and Processing

[0372] (1) Environmental Data Acquisition

[0373] The following environmental parameters are obtained in real time through sensors:

[0374] Temperature T(t): The environmental temperature is collected through a temperature sensor (unit: °C);

[0375] Humidity H(t): The environmental humidity is collected through a humidity sensor (unit: %RH);

[0376] Static electricity S(t): The electrostatic field strength of the inkjet path and the medium surface is measured through an electrostatic sensor (unit: kV / m).

[0377] (2) Data Preprocessing

[0378] The collected data is denoised and normalized:

[0379] Denoising: Use a Kalman filter to smooth the sensor noise.

[0380] Normalization: Scale the data to a unified range [0,1].

[0381] (3) Deviation Calculation

[0382] Calculate the deviation required for compensation according to the changes in environmental parameters:

[0383] ΔP = f(T, H, S)

[0384] ΔP: The correction value of the inkjet parameters caused by environmental changes.

[0385] 2. Compensation Model

[0386] (1) Parameter Relationship Model

[0387] Build the relationship between environmental factors and inkjet behavior through experiments or data training:

[0388] Temperature T(t): An increase in temperature will reduce the viscosity of the ink droplets and increase the flight speed.

[0389] Compensation formula:

[0390]

[0391] K T : Temperature compensation coefficient; T 0 : Reference temperature.

[0392] Humidity H(t): An increase in humidity will increase the diffusibility of ink droplets.

[0393] Compensation formula:

[0394]

[0395] K H : Humidity compensation coefficient; H0: Reference humidity.

[0396] Electrostatic S(t): An increase in static electricity will cause ink droplet deviation.

[0397] Compensation formula:

[0398]

[0399] K S : Electrostatic compensation coefficient.

[0400] (2) Comprehensive compensation formula

[0401] Calculate and correct the inkjet parameters by comprehensively considering temperature, humidity, and static electricity:

[0402] .

[0403] 3. Real-time compensation process

[0404] (1) Data acquisition and analysis

[0405] Obtain the temperature T(t), humidity H(t), and static electricity S(t) in real time.

[0406] Calculate the deviation between the current environmental parameters and the reference values: ΔT = T(t) − T 0 , ΔH = H(t) − H 0 , ΔS = S(t) − S 0 ;

[0407] (2) Generate compensation signal

[0408] Calculate the compensation signal based on the deviation:

[0409] ΔP = [ΔV s , ΔF p , Δt s , Δθ]

[0410] ΔV s : Voltage adjustment amount; ΔF p : Pulse frequency adjustment amount; Δt s : Jetting time adjustment amount; Δθ: Jetting angle adjustment amount.

[0411] (3) Real-time parameter adjustment

[0412] Apply the compensation signal to the inkjet head:

[0413] Voltage adjustment: V s (t + 1)=V s (t)+ΔV s ;

[0414] Frequency adjustment: F p (t + 1)=F p (t)+ΔF p ;

[0415] Time adjustment: t s (t + 1)=t s (t)+Δt s ;

[0416] Angle adjustment: θ(t + 1)=θ(t)+Δθ.

[0417] The following combines a practical application scenario to elaborate in detail on the implementation process of the specific embodiments of the present invention for high-precision printing of a complex image in a high-humidity environment.

[0418] 1. Example background

[0419] 1.1 Environmental conditions:

[0420] Temperature: 30 °C; Humidity: 75%; Electrostatic intensity: 2 kV / m 2 .

[0421] 1.2 Medium characteristics:

[0422] Medium type: Smooth plastic sheet; Ink absorption: A = 0.3; Surface roughness: R = 0.2 μm; Conductivity: C = 0.1 S / m.

[0423] 1.3 Target image:

[0424] Resolution: 300 dpi.

[0425] Characteristics: The image includes a highlight area (such as the white part), a high-texture area (the part with complex details), and a medium-brightness area (the background part).

[0426] 2. Specific implementation steps

[0427] Step 1: Real-time data acquisition

[0428] Ink droplet behavior data acquisition: Use a high-speed camera to capture the shape, speed, and landing position of ink droplets in real time.

[0429] Analyze the shape of the ink droplet: Use the Canny algorithm to extract the contour of the ink droplet.

[0430] Calculate the ink droplet velocity: v(t)=Δx / Δtv(t); Assume the displacement between frames is 0.2 mm and the time interval is 0.001 s, then: v(t)=0.2 / 0.001 = 200 mm / s;

[0431] Collect environmental parameters:

[0432] Temperature T(t)=30°C, humidity H(t)=75%, static electricity intensity S(t)=2 kV / m; The laser sensor measures the surface roughness of the medium R = 0.2 μm.

[0433] Step 2: Multidimensional parameter modeling

[0434] Construct input data:

[0435] X(t)=[T(t),H(t),S(t),A,R,C,Vs,Fp,θ]

[0436] Deep learning model:

[0437] The convolutional neural network extracts the medium characteristics and environmental characteristics; LSTM processes the inkjet parameters and time series data.

[0438] Predict the behavior of the ink droplet:

[0439] Landing point offset: Δx(t)=0.1 mm, Δy(t)=0.2 mm; Ink droplet deformation: ΔS(t)=0.05 mm.

[0440] Step 3: Real-time ink droplet behavior prediction

[0441] Predict the trajectory offset according to the LSTM model:

[0442]

[0443] Predict the deformation value: ΔS(t)=0.05 mm.

[0444] Step 4: Adaptive inkjet parameter adjustment

[0445] Adjust the nozzle voltage: The ink droplet deformation ΔS = 0.05 mm, adjust the voltage: V s (t + 1)=V s (t)+K v *ΔS;

[0446] Assume K v =10 V / mm, then: ΔV s =10*0.05 = 0.5 V;

[0447] Pulse frequency adjustment: offset Δd = 0.223 mm, adjustment frequency: F p (t + 1) = F p (t) − K f *Δd;

[0448] Assume K f = 50 kHz / mm, then: ΔF p = 50 ⋅ 0.223 = 11.15 kHz

[0449] Jet angle adjustment: offset direction: Δθ = −K θ *arctan(Δy / Δx);

[0450] Assume K θ = 5 ° / rad, then: Δθ = −5*arctan(0.2 / 0.1) = −5*1.107 = −5.54°.

[0451] Step 5: Closed-loop feedback control

[0452] Real-time monitoring of the landing point deviation: deviation value: Δd = 0.223 mm

[0453] PID control signal generation:

[0454]

[0455] Proportional gain K p = 2, integral gain K i = 0.5, derivative gain K d = 1, error e(t) = 0.223e, then:

[0456]

[0457] Dynamically calculate the control signal for adjusting the inkjet parameters.

[0458] Step 6: Adaptive matching of media characteristics

[0459] Detect media characteristics: measured ink absorption A = 0.3, roughness R = 0.2 μm, conductivity C = 0.1 S / m.

[0460] Material library matching: query the material library:

[0461] P optimal = DB(A, R, C);

[0462] Obtain the optimized parameter solution V s = 15 V, F p = 20 kHz, θ = 2 ° .

[0463] Step 7: Intelligent Multi-region Optimization

[0464] Region Segmentation: Use the K-means algorithm to divide the image into three regions:

[0465] Highlight Region R 1 : White background;

[0466] High-texture Region R 2 : Image details;

[0467] Medium-brightness Region R 3 : Background pattern.

[0468] Dynamic Inkjet Adjustment:

[0469] Highlight Region R 1 : Reduce voltage and increase frequency;

[0470] High-texture Region R 2 : Increase jetting time and voltage;

[0471] Medium-brightness Region R 3 : Keep default parameters.

[0472] Step 8: Environmental Factor Compensation

[0473] Temperature and Humidity Compensation: When the humidity increases to H(t) = 75%, adjust the voltage: V s (t + 1)=V s (t)+K H *(H(t)−H 0 );

[0474] Assume K H =0.1, then: ΔV s =0.1*(75−50)=2.5 V;

[0475] Electrostatic Compensation: When the electrostatic intensity is S(t)=2 kV / m, adjust the jetting angle: θ(t + 1)=θ(t)−K S *S(t);

[0476] Assume K S =1° / kV / m, then: Δθ=﹣1*2=﹣2°.

[0477] The foregoing is a description of embodiments of the present invention. Through the above description of the disclosed embodiments, those skilled in the art can implement or use the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the broadest scope consistent with the principles and novel features disclosed herein.

Claims

1. An intelligent ink drop control method for a digital printing press, characterized in that: The method comprises the following steps: 1) Real-time data collection: The shape, speed, landing point location and color distribution data of ink droplets during the inkjet process, as well as environmental parameters, are collected through high-speed cameras and environmental sensors. The environmental parameters include temperature, humidity, static electricity and medium surface roughness; 2) Multi-dimensional parameter modeling: Use deep learning technology to build a nonlinear model between ink droplet behavior and environmental factors, media characteristics and inkjet parameters, and generate a dynamic parameter optimization curve; 3) Real-time ink droplet behavior prediction: Based on the time series analysis algorithm, the flight trajectory deviation and deformation of the ink droplets and their impact on the target printing quality are predicted; 4) Adaptive inkjet parameter adjustment: According to the prediction results of ink droplet behavior, the voltage, pulse frequency, injection time and nozzle opening and closing state of the inkjet head are dynamically adjusted to optimize the size, speed and injection angle of the ink droplets; 5) Closed-loop feedback control: Generate control signals by real-time monitoring of ink droplet landing point deviation data, and update inkjet parameters at a frequency of at least 5,000 times per second; In step 2), the nonlinear model uses multi-layer nonlinear transformation to integrate environmental parameters, media characteristics and inkjet parameters into a dynamic prediction framework: ； in, y(t): ink drop behavior including speed, size, and drop point deviation; Φ: dynamic multi-layer nonlinear function, implemented by neural network; ϵ(t): unmodeled error; The model structure combines the following multi-level nonlinear calculations: ； X=[E, M, P]: input vector, including environmental parameters E, medium characteristics M and inkjet parameters P; W1, W2, W3: weight matrix, optimized by training data; b1, b2, b3: bias vector; σ(·): activation function; The input parameter X is processed as follows: Environmental parameters E(t)=[T, H, S, V]: including temperature, humidity, static electricity, and vibration; Medium properties M(t)=[A, R, C]: ink absorption, roughness, conductivity; Inkjet parameters P(t)=[V s , F p ,θ]: nozzle voltage, pulse frequency, injection angle; Decomposed into deep feature extraction layers: Environment module: ; Media Module: ; Inkjet module: ; Finally the outputs of these modules are combined: ； The output goes through the final layer: ； Combined with the dynamic behavior of time series, time-related parameters are added: ； in α i is the time decay coefficient, which is used to weight the impact of historical data on current decisions.

2. The intelligent ink drop control method of a digital printing press according to claim 1, characterized in that: The multi-dimensional parameter modeling utilizes a joint model of a convolutional neural network (CNN) and a recurrent neural network (RNN) to analyze both the spatial behavior characteristics of ink droplets and the dynamic changes in the time series.

3. The intelligent ink drop control method of a digital printing press according to claim 1, characterized in that: The method also includes analyzing the ink absorption, roughness and conductivity of the medium surface through a medium identification module, and dynamically loading an inkjet parameter optimization solution for the corresponding medium.

4. The intelligent ink drop control method of a digital printing press according to claim 3, characterized in that: The media identification module automatically identifies the media type and loads the optimal inkjet parameters through high-resolution scanning equipment and material library matching technology.

5. The intelligent ink drop control method of a digital printing press according to claim 1, characterized in that: The method also includes dividing the target image into a plurality of regions by using a region segmentation algorithm, and adjusting the inkjet parameters for each region respectively.

6. The intelligent ink drop control method of a digital printing press according to claim 1, characterized in that: The method also includes dynamically adjusting inkjet parameters in combination with real-time environmental data to compensate for inkjet deviations caused by changes in temperature, humidity or static electricity.

7. The intelligent ink drop control method of a digital printing press according to claim 6, characterized in that: Environmental factor compensation includes adjusting the electric field distribution of the nozzle according to the data of the electrostatic sensor to reduce the deviation of the ink droplet during the ejection process.

8. An intelligent ink drop control system for a digital printing press, characterized in that: The system implements the method described in any one of claims 1 to 7, including: 1) High-speed visual monitoring unit: used to collect ink drop position, shape and color distribution data in real time; 2) Environmental perception module: including temperature and humidity sensors, electrostatic sensors and medium roughness detection devices; 3) Deep learning module: used to analyze the collected data and generate optimized inkjet parameters; 4) Inkjet control unit: connected to the deep learning module, used to dynamically adjust the voltage, pulse frequency and injection time of the inkjet head; 5) Closed-loop feedback control module: generates control signals based on real-time monitoring data; 6) Medium identification module: used to analyze and match the characteristic parameters of different media.

9. A digital printing press, characterized in that: include: 1) The intelligent ink drop control system according to claim 8; 2) Media transport module: used to transport a variety of media including paper, plastic, cloth and metal surfaces; 3) Ink system: supports UV curing ink and water-based ink, and has automatic cleaning function; 4) Control panel: used to set and monitor printing parameters in real time.

Images