An intelligent laser engraving detection system for an electric engraver

Abstract

The present invention discloses an intelligent laser engraving detection system for an electric engraving machine, which relates to the field of intelligent detection technology and realizes real-time monitoring of the operating status of the electric engraving machine, fault prediction and parameter optimization. The system obtains multi-source data such as operation, environment, imposition software and proofing effect through a data acquisition module, adopts a Transformer model to analyze long-term dependencies, and improves the accuracy of fault prediction. The optimization control module adjusts the engraving speed and vibration frequency according to the prediction results, prolongs the life of the engraving needle and thus reduces the white scraping phenomenon during printing. The image recognition module uses a convolutional neural network (CNN) to evaluate the quality of the engraving dots and feeds back to the optimization control module to form a closed-loop control. In addition, the system has an adaptive optimization function, realizes autonomous learning through a knowledge base and an evaluation and screening module, and continuously improves performance. The present invention significantly improves the operating efficiency, engraving quality and equipment life of the electric engraving machine, while reducing maintenance costs and resource consumption.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent detection, and particularly relates to an intelligent laser engraving detection system for an electric engraver. BACKGROUND

[0002] During the engraving process of the electric engraver, small dots are formed at the text entry point due to problems such as needle wear and unreasonable engraving parameter settings, and the edge is serrated, which reduces the ink content during printing, causes white scraping, and affects production efficiency. The traditional detection method relies on manual experience, is low in efficiency and low in accuracy. In recent years, artificial intelligence technology has gradually increased in the field of industrial detection. Through the introduction of image processing and deep learning technologies, partial automation of the engraving process has been achieved.

[0003] Chinese patent CN110175603B discloses an engraving text recognition method and system and a storage medium. The method includes: collecting the image of the measured object by a binocular camera, generating a depth map by convolution and deconvolution operation, extracting the text area by a region recommendation network, and performing text recognition based on a bidirectional long short-term memory network to obtain the recognition result of the engraved text. This technology combines binocular stereo vision and deep learning technology to improve the significance and efficiency of text recognition, reduces the cost and improves the reliability without manual detection, and can be widely applied to the field of text recognition. This method mainly focuses on the recognition of engraved text and does not involve the monitoring or fault prediction of the running state of the electric engraver, and cannot solve the actual fault problems such as needle wear and entry point scraping. In addition, its application is limited to text recognition and lacks real-time monitoring and optimization functions for engraving quality.

[0004] Chinese patent application CN116226758A discloses an engraving machine abnormality detection method based on improved particle filtering and convolutional network. The method includes offline training and online abnormality detection. In the offline training stage, running data is collected, an improved particle filtering algorithm and an expectation maximization algorithm are introduced, an accurate mathematical model is established, and a one-dimensional residual convolutional neural network is trained to realize data classification. In the online detection stage, the model estimates the noise sequence and judges the data anomaly. This method can mine the data correlation in the time dimension and improve the accuracy of abnormality detection. This technology only focuses on the abnormality detection of the running data of the engraving machine and cannot predict and prevent specific fault types such as needle wear or white scraping phenomenon. It also does not provide real-time parameter optimization function to reduce faults. In addition, its offline training mathematical model may not be able to adapt to the dynamic changes of the running state of the electric engraver and lacks the ability of self-adaptive learning and optimization.

[0005] The above design improves the character recognition efficiency through image processing and deep learning technology, or improves the abnormal detection accuracy through mathematical modeling, but still has certain limitations, such as: lack of fault prediction and prevention: the existing technology cannot effectively predict specific faults such as needle wear and white scratch of the cutting edge, and cannot take preventive measures, lack of real-time optimization function: the existing method cannot realize real-time adjustment of the engraving parameters, and cannot optimize the equipment performance according to the running state, insufficient self-adaptation: relying on static model or offline training, it is difficult to cope with dynamic changes in the operation of the electric engraving machine, incomplete quality control: lack of real-time monitoring and closed-loop feedback mechanism for engraving quality, and unable to ensure the effect of continuous optimization. SUMMARY

[0006] The purpose of the present application is to overcome the shortcomings of the prior art and provide an intelligent laser engraving detection system for an electric engraving machine to solve the above problems.

[0007] The purpose of the present application is achieved by the following technical scheme: an intelligent laser engraving detection system for an electric engraving machine, comprising:

[0008] A data acquisition module is used to acquire multi-source data of the electric engraving machine in real time, and the multi-source data includes electric engraving machine operation data, layout software data, environment data and sample effect data;

[0009] A data preprocessing module is used to clean, convert and compress the collected multi-source data to generate a format suitable for machine learning model processing;

[0010] A data encoding module is used to encode the preprocessed multi-source data into branch data chains, data chain strips and data columns in time sequence, wherein the branch data chain is the time sequence of each data source, the data chain strip is the combination of multiple branch data chains in time sequence, and the data column is the parallel combination of multiple branch data chains at the same time point;

[0011] A high-dimensional vector conversion module converts the data chain strip into a high-dimensional vector using a Transformer model, captures the long-term dependence relationship of the time step through a self-attention mechanism, and encodes the data column into a high-dimensional vector through a fully connected layer;

[0012] An influence correlation calculation module calculates the cosine similarity of the high-dimensional vector to quantify the influence correlation between different data sources;

[0013] A fault prediction module adds a prediction head based on the output layer of the Transformer model, including a regression layer for needle wear prediction and a classification layer or a regression layer for white scratch prediction of the cutting edge, respectively outputting the wear degree and the white scratch probability or degree;

[0014] An optimization control module adjusts the engraving speed and vibration frequency of the electric engraver in real time according to the fault prediction results, so as to prolong the service life of the needle and reduce the white scratch phenomenon.

[0015] An image recognition module analyzes the dot image by using a convolutional neural network (CNN), evaluates the dot quality, and feeds back the evaluation results to the optimization control module to realize closed-loop control.

[0016] An optimization process record module is used to record the input data, prediction results, optimization actions and feedback results of each optimization process.

[0017] An evaluation and screening module evaluates the optimization effect of each optimization process by using a defined evaluation index or reinforcement learning reward function, and screens out the optimization process with better optimization effect.

[0018] A knowledge base module is used to store the key features and corresponding actions of the screened optimization process, and to match the current data with the patterns in the knowledge base in real time during the work process, and automatically call the optimization strategy with the highest matching degree.

[0019] The electric engraver operation data collected by the data acquisition module includes the service life of the needle, the number of engraved lines, the engraving time, the vibration state and frequency of the needle, and the composition data of the needle material.

[0020] The data preprocessing module compresses the TIFF image to ensure that the text is not distorted.

[0021] The Transformer model in the high-dimensional vector conversion module includes multiple encoding layers and self-attention heads, which are used to generate time step vector representations of the data chain.

[0022] The needle wear prediction regression layer and the white scratch prediction classification layer or regression layer in the fault prediction module are trained and updated regularly based on historical data to improve prediction accuracy.

[0023] When the needle wear exceeds the preset threshold, the optimization control module automatically reduces the engraving speed, and adjusts the vibration frequency when the white scratch risk is predicted.

[0024] The image recognition module analyzes the dot size and edge sawtooth by using CNN, judges whether the ink content is sufficient during use, and inputs the judgment result as feedback to the optimization control module.

[0025] The optimization process record module stores each optimization process in a relational database, and the record content includes the data chain, data chain, data column, prediction result, adjusted parameter and image recognition evaluation result.

[0026] The evaluation and screening module uses the dot quality improvement amplitude or fault reduction rate as an evaluation index to screen out the optimization process with the best optimization effect.

[0027] The knowledge base module automatically calls the optimization strategy with the highest matching degree by calculating the similarity of the current data and the stored optimization process, realizes self-adaptive learning and optimization.

[0028] The beneficial effects of the present application are:

[0029] 1. The system acquires real-time multi-source data of the electric engraving machine through the data acquisition module, including operation data, environment data, layout software data and sample effect data, uses the Transformer model for high-dimensional vector conversion and self-attention mechanism analysis, can capture the long-term dependence between data, thereby significantly improving the accuracy of fault prediction, compared with the traditional maintenance method relying on artificial experience, the present application can identify problems such as stylus wear and white scratch, the prediction error rate is greatly reduced, the optimization control module adjusts the engraving speed and vibration frequency in real time according to the fault prediction result, effectively prolongs the service life of the stylus and reduces the white scratch phenomenon, at the same time, the image recognition module uses convolutional neural network (CNN) to analyze the engraving dot image, evaluates the dot quality and feeds back to the optimization control module, forms a closed loop control, ensures the effective implementation of the optimization measures, the data coding module encodes the multi-source data into branch data chain, data chain and data column in time sequence, realizes the structured processing and feature extraction of data through high-dimensional vector conversion and influence correlation calculation, which not only improves the input quality of the machine learning model, but also improves the training efficiency.

[0030] 2. Through predictive maintenance, the present application can reduce the occurrence of sudden failures and emergency repairs, significantly reduce the maintenance cost of enterprises, at the same time, prolong the service life of the stylus reduces the replacement frequency and consumable expenditure, saves the operating expenses for enterprises, the system reduces manual intervention through automatic adjustment of operation parameters, makes the electric engraving machine run continuously in the best state, which not only improves the production efficiency, but also improves the equipment utilization rate and reduces the resource waste.

[0031] 3. By optimizing operation parameters and prolonging equipment life, the present application reduces resource consumption and waste production, conforms to the trend of green manufacturing and sustainable development, and makes contributions to environmental protection, the system reduces unexpected downtime and emergency repairs caused by faults, improves the safety of operating personnel, and reduces the potential risks in the working environment.

[0032] 4. The present application applies Transformer model and CNN technology to the electric engraving machine detection field, realizes intelligent analysis and self-adaptive optimization of multi-source data, through the knowledge base module and the evaluation and screening module, the system can learn and optimize independently, continuously improve performance. BRIEF DESCRIPTION OF DRAWINGS

[0033] Figure 1 System interaction of the present application Figure 1 ;

[0034] Figure 2 It is a timing diagram of the present invention;

[0035] Figure 3 System interaction for the present invention Figure 2 . DETAILED DESCRIPTION

[0036] The technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only 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 skilled in the art without creative work are within the scope of protection of the present invention.

[0037] It is to be noted that the directions of "left", "right", "up", "down", "front", "back", "inside" and "outside" in the following schemes are all relative directions and are not listed here one by one.

[0038] Example 1:

[0039] like Figures 1 to 3 As shown, the following is a detailed description of Example 1, covering the specific implementation method, working process and beneficial effects. This example is intended to demonstrate the basic functional implementation of an intelligent laser engraving detection system for an electric engraving machine, including the coordinated operation of modules such as data acquisition, preprocessing, fault prediction and optimization control to achieve the goals of fault prediction and parameter optimization.

[0040] This embodiment provides an intelligent laser engraving detection system for an electric engraving machine. The system consists of multiple modules. The specific implementation of each module is as follows:

[0041] The data acquisition module is responsible for real-time collection of multi-source data from the electric engraving machine to comprehensively monitor the equipment's operating status and environmental conditions. Specific collection content includes:

[0042] Electric engraving machine operation data: Collect the engraving needle usage time (reflecting the cumulative working time of the engraving needle), the number of engraving lines (such as 160, 180, 200, 230, 270, 330, 410, 540 and other different specifications), engraving time (the duration of a single task), the engraving needle vibration state and frequency (monitoring the dynamic characteristics of the engraving needle through a vibration sensor) and the engraving needle material composition data (recording the material characteristics of the engraving needle through material analysis) through sensors and equipment interfaces.

[0043] Imposition software data: Extract imposition parameters and layout information from imposition software (such as ARTCOM, Engraving make (CEM), Collage, HelioDisk) and analyze their impact on engraving quality.

[0044] Environmental Data: Collect temperature and humidity within the workshop through temperature and humidity sensors to assess potential environmental factors affecting equipment operation.

[0045] Proofing Effect Data: Obtain the proofing results of the formal version through the proofing machine, such as hue deviation or dot quality data, for subsequent quality assessment.

[0046] The data preprocessing module processes the collected multi-source data to generate a format suitable for machine learning models, including the following steps:

[0047] Data cleaning: Fill in missing values (e.g., use interpolation method to fill in missing points of needle vibration frequency), remove or smooth abnormal values (e.g., remove mutation peaks in needle usage time).

[0048] Data conversion: Convert raw data to standardized format, such as one-hot encoding for categorical data (e.g., needle material type), and normalization for continuous data (e.g., vibration frequency).

[0049] Image compression: Compress TIFF images generated by the layout software using lossless compression algorithms (e.g., LZW), ensuring that text and details in the image are not distorted, while reducing storage and transmission resource occupation.

[0050] Fault prediction module based on Transformer model to realize fault prediction, the specific implementation is as follows:

[0051] Model architecture: Use Transformer model to process preprocessed multi-source data, use self-attention mechanism to capture long-term dependencies between data, and generate high-dimensional vector representation.

[0052] Prediction head design: Add two prediction heads to the output layer of the Transformer model:

[0053] Needle wear prediction: Output needle wear degree through regression layer, represented as a continuous value from 0 to 1 (0 represents no wear, 1 represents complete wear).

[0054] Incut white prediction: Output the probability or degree of white phenomenon through classification layer or regression layer, such as classifying white severity into three levels of slight, moderate, and severe, or outputting white risk value from 0 to 1.

[0055] Model training and updating: Use historical data (including multi-source data and corresponding fault records) to train the model, and regularly (e.g., monthly) update the model parameters using newly collected data to adapt to changes in equipment operation state and improve prediction accuracy.

[0056] The optimization control module adjusts the operation parameters of the electric engraver in real time according to the fault prediction results. The specific strategies include:

[0057] Optimization of needle wear: when the predicted needle wear value exceeds the preset threshold (e.g. 0.8), the system automatically reduces the engraving speed, for example from 100% to 80%, to slow down the wear speed of the needle.

[0058] Optimization of white scraping risk: when the predicted white scraping probability exceeds the preset threshold (e.g. 0.5), the system adjusts the vibration frequency of the needle, for example from 500Hz to 450Hz, to reduce the occurrence of white scraping.

[0059] Parameter execution: the adjusted parameters are applied in real time through the control interface of the electric engraver, ensuring that the optimization measures take effect immediately.

[0060] Working process

[0061] The running process of the system is a closed-loop real-time monitoring and optimization process, and the specific steps are as follows:

[0062] Data collection:

[0063] During the operation of the electric engraver, the data collection module collects operation data (needle usage time, engraving line number, etc.), layout software data, environmental data and sample effect data in real time through sensors and software interfaces, forming a multi-source data stream.

[0064] Data preprocessing:

[0065] The collected multi-source data is sent to the data preprocessing module for cleaning (filling missing values, removing outliers), conversion (encoding and normalization) and compression (TIFF image lossless compression), generating clean and uniform input data.

[0066] Data cleaning (interpolation method): to fill in missing values, the linear interpolation formula is used:

[0067]

[0068] Data normalization (Min-Max normalization): normalize the data using the following formula:

[0069]

[0070] Image compression (lossless compression): compress the TIFF images generated by the layout software, using the LZW algorithm, which generates lossless compressed files using a dictionary.

[0071] Fault prediction:

[0072] The pre-processed data is input into the fault prediction module, the Transformer model analyzes the data and generates a high-dimensional vector, and outputs the stylus wear degree and the probability or degree of white scratch as the fault prediction result through the prediction head.

[0073] Self-attention mechanism of Transformer model: The self-attention mechanism is used to capture long-term dependencies between data, and the calculation formula is as follows:

[0074]

[0075] Optimization control:

[0076] The optimization control module receives the fault prediction result and judges whether to trigger the adjustment condition:

[0077] If the stylus wear exceeds the preset threshold, the engraving speed is reduced.

[0078] If the white scratch risk exceeds the preset threshold, the vibration frequency is adjusted.

[0079] The adjusted parameters are fed back to the engraving machine through the control system to realize real-time optimization.

[0080] Loop execution:

[0081] The system continuously collects new data and repeats the above process to form a closed-loop control of dynamic monitoring and optimization.

[0082] Stylus wear prediction (regression layer output): The regression layer is used to predict the stylus wear, and the output formula is:

[0083]

[0084] White scratch prediction at the entry point (classification layer output): The classification layer is used to judge the white scratch risk, and the softmax function is used:

[0085]

[0086] Engraving speed adjustment: When the stylus wear prediction exceeds the threshold, the engraving speed is reduced, and the following formula can be used:

[0087]

[0088] Vibration frequency adjustment: When the white scratch risk prediction exceeds the threshold, the vibration frequency is adjusted to:

[0089]

[0090] Closed-loop control and dynamic optimization:

[0091] The system continuously runs, constantly optimizing control parameters through real-time collected new data and failure prediction results. For each prediction update, the objective function of the optimization control module can be expressed as:

[0092]

[0093] Dynamic parameter adjustment and optimization feedback:

[0094] In actual execution, the adjusted parameters will affect the operation of the equipment through the feedback mechanism. This process can be achieved through the formula of the feedback control system:

[0095]

[0096] Through the integration of multi-source data by the Transformer model and accurate prediction, the system can identify the stylus wear and white scratch risk in advance, reduce unexpected downtime caused by failures, and improve equipment reliability.

[0097] According to the wear prediction, the engraving speed is dynamically reduced to slow down the stylus wear rate and prolong its service life, thereby reducing the replacement frequency and related costs.

[0098] By adjusting the vibration frequency, the incidence of white scratch at the cutting edge is effectively reduced, ensuring the quality of engraved dots and improving the visual effect and market competitiveness of printed products.

[0099] The automatic parameter adjustment of the system reduces the need for manual intervention, enabling the electrographic engraver to run in the best state continuously and significantly improving production efficiency.

[0100] Through predictive maintenance, the occurrence of sudden failures and emergency repairs is reduced, the maintenance cost and resource waste of the enterprise are reduced, and long-term stable operation is ensured.

[0101] This embodiment realizes intelligent failure prediction and operation optimization of the electrographic engraver through the cooperative work of the data acquisition module, data preprocessing module, failure prediction module, and optimization control module. The system pays attention to detail processing (such as lossless compression of TIFF images) and dynamic adjustment (such as real-time optimization of speed and frequency) in specific implementation.

[0102] Embodiment 2:

[0103] For example, Figures 1 to 3As shown, the present embodiment further realizes the detailed functions of the data encoding module, high-dimensional vector conversion module, influence correlation calculation module, fault prediction module, and related optimization module based on the embodiment 1. Through these modules, the system can efficiently encode the multi-source data of the electric engraving machine into high-dimensional vectors, quantify the influence correlation between data sources, and combine image recognition and knowledge base optimization to realize accurate fault prediction and intelligent parameter adjustment, thereby improving the operation efficiency and engraving quality of the electric engraving machine.

[0104] The data acquisition module acquires the multi-source data of the electric engraving machine in real time, including:

[0105] Electric engraving machine operation data: such as the service life of the carving needle, engraving speed, vibration frequency, etc.

[0106] Rip software data: such as the number of engraving lines, dot design parameters, etc.

[0107] Environmental data: such as temperature, humidity, etc.

[0108] Proofing effect data: such as the preliminary evaluation results of dot quality.

[0109] The preprocessing module cleans (such as removing noise), converts (such as normalization processing), and compresses (such as dimensionality reduction or feature extraction) the collected multi-source data, generating a format suitable for machine learning model processing, for example, normalizing the service life of the carving needle and the environmental temperature to the range of 0 to 1, facilitating subsequent encoding and modeling.

[0110] The data encoding module encodes the preprocessed multi-source data in time sequence into the following three forms:

[0111] Branch data chain: each data source forms a time series, for example, the branch data chain of the service life of the carving needle is represented as [T1, T2, T3,...], where T1 is the service life of the carving needle at a certain time.

[0112] Data chain: multiple branch data chains are combined in time sequence into a complete sequence, such as [T1, L1, E1, T2, L2, E2,...], where L represents the number of engraving lines, and E represents the environmental temperature.

[0113] Data column: multiple source data at the same time point are combined side by side, such as the data column at time t1 is (T1, L1, E1).

[0114] This encoding method preserves the time sequence characteristics of the data (data chain) and the multi-source correlation (data column), providing structured input for high-dimensional vector conversion.

[0115] The high-dimensional vector conversion module uses the Transformer model to process the data chain and data column, generating a high-dimensional vector representation:

[0116] Model Architecture: The Transformer model contains multiple encoding layers and self-attention heads. The self-attention mechanism captures long-term dependencies in the data chain by calculating weights between time steps, such as analyzing the correlation between engraving needle usage duration and the number of engraved lines at different time points.

[0117] Data Chain Processing: The data chain is input into the Transformer model to generate high-dimensional vector representations for each time step, containing temporal context information.

[0118] Data Column Processing: The data columns are encoded into high-dimensional vectors through fully connected layers to capture joint features of multi-source data at the same time.

[0119] The final output high-dimensional vector combines temporal and multi-source characteristics, providing a rich information foundation for subsequent analysis.

[0120] The Influence Correlation Calculation Module quantifies the influence correlation between different data sources by calculating the cosine similarity of high-dimensional vectors:

[0121] Calculation Method: For two data sources (such as engraving needle usage duration and environmental temperature), extract their high-dimensional vectors and calculate the cosine similarity, the result is between -1 and 1, the closer to 1, the higher the correlation.

[0122] Application Example: If the similarity between the engraving needle usage duration and the environmental temperature is high, it indicates that temperature has a significant impact on the engraving needle wear, which can be used to optimize the weight of the prediction model.

[0123] Fault Prediction Module: Add a prediction head to the output layer of the Transformer model to realize two types of fault prediction:

[0124] Engraving Needle Wear Prediction: Output the wear degree (0 to 1 continuous value, 0 represents no wear, 1 represents complete wear) through the regression layer.

[0125] Entry Knife Scratching Prediction: Output the scratching severity (such as none, mild, moderate, severe) through the classification layer or the scratching risk value (0 to 1) through the regression layer.

[0126] Training and Updating: Train the model using historical data and update the parameters regularly (such as every month) to adapt to changes in device status.

[0127] Optimization Control Module: Adjust the parameters of the electric engraving machine in real-time based on the fault prediction results:

[0128] If the predicted engraving needle wear is high, reduce the engraving speed to prolong the life of the engraving needle.

[0129] If the predicted scratching risk is high, adjust the vibration frequency to reduce the scratching phenomenon.

[0130] The image recognition module uses a convolutional neural network (CNN) to analyze the engraving dot image, assesses the dot quality (such as uniformity, edge sharpness), and feeds the results back to the optimization control module, forming a closed-loop control. For example, if the dot edge is blurred, it prompts to adjust the vibration frequency.

[0131] The optimization process record module records the input data (such as high-dimensional vectors), prediction results (such as loss degree), optimization actions (such as reducing speed), and feedback results (such as dot quality improvement) of each optimization.

[0132] The evaluation and screening module evaluates the optimization effect through evaluation indicators (such as prediction accuracy, dot quality improvement amplitude) or reinforcement learning reward functions, and selects the optimization process with the best effect.

[0133] The knowledge base module stores the key features (such as data patterns under high loss conditions) and corresponding actions (such as reducing speed) of the selected optimization process, and matches the current data in real time during work to call the optimal strategy.

[0134] Working process

[0135] Data collection and preprocessing

[0136] The system collects data such as the service life of the engraving needle, the number of engraved lines, and the environmental temperature in real time, and generates structured data after cleaning and normalization.

[0137] Data encoding

[0138] Data is encoded into branch data chains (such as [T1, T2, T3,...]), data chains (such as [T1, L1, E1, T2, L2, E2,...]), and data columns (such as (T1, L1, E1)).

[0139] Branch data chain (time series encoding): Branch data chain is represented as time series, which can use the following dynamic encoding function:

[0140]

[0141] Data chain (multi-source data sequence encoding): Joint encoding of time series of multiple data sources can be defined as follows:

[0142]

[0143] Data column (parallel combination): The multi-source data at the same time is combined into a data column, defined as:

[0144]

[0145] High-dimensional vector conversion

[0146] Data chain is processed by Transformer model, self-attention mechanism captures temporal dependence, and generates high-dimensional vector.

[0147] Data column is encoded by fully connected layer, generating multi-source feature vector.

[0148] Data chain processing (Transformer model application): Apply Transformer model to data chain to generate high-dimensional vector representation at each time step, assuming each data source goes through self-attention mechanism to get high-dimensional vector H at each time step i , which can be generated by the following formula:

[0149]

[0150] Data column processing (fully connected layer encoding): Encode data column to convert multi-source features at each time into high-dimensional vector, defined as:

[0151]

[0152] Impact correlation calculation

[0153] Calculate the cosine similarity of high-dimensional vectors, for example, the similarity between needle use time and carving line number is 0.85, indicating a high correlation between the two.

[0154] Cosine similarity calculation: used to quantify the correlation between different data sources, for two data sources (such as needle use time T and environmental temperature E), their cosine similarity can be calculated by the following formula:

[0155]

[0156] Assuming the similarity between needle use time and environmental temperature is high (for example, 0.85), it indicates that temperature has a significant impact on needle wear, and the model should increase the weight of temperature data.

[0157] Fault prediction

[0158] Transformer output layer predicts needle wear (e.g. 0.75, close to threshold) and scuffing risk (e.g. 0.6, moderate risk).

[0159] Needle wear prediction (regression layer output): use regression layer to predict needle wear, output formula is:

[0160]

[0161] Knife entry whiteout prediction (classification layer output): Classify or regress the whiteout risk, using a softmax function to output probabilities.

[0162]

[0163] Parameter optimization

[0164] The optimization control module reduces the engraving speed and adjusts the vibration frequency according to the prediction results, while the image recognition module analyzes the dot quality and feeds back the adjustment effect.

[0165] Parameter optimization (speed and frequency adjustment): Adjust parameters through fault prediction results. If the loss prediction is high, reduce the engraving speed. Assume the current speed is v current , the optimized speed is:

[0166]

[0167] Dot quality assessment: Use a convolutional neural network (CNN) to analyze the engraved dot image and generate an assessment result. Assume I dot is the dot image, and the evaluation function can be expressed as:

[0168]

[0169] Recording and optimization

[0170] The optimization process record module stores the data of this operation, the evaluation and selection module evaluates the effect, and the knowledge base module updates the optimization strategy.

[0171] Optimization process record: Record the input data, prediction results, optimization actions and feedback results of each optimization. The optimization process can be implemented through the following record function:

[0172]

[0173] Knowledge base update and call: Store the key features of the optimization process and the corresponding actions, and call the optimal strategy in real time. The update process can be represented by the following function:

[0174]

[0175] Continuous operation

[0176] The system cycles through the above steps to achieve dynamic monitoring and optimization.

[0177] The data encoding module and high-dimensional vector conversion module capture time series and multi-source characteristics through data chains, data chain links and Transformer models, improving data utilization efficiency.

[0178] The cosine similarity quantifies the correlation between data sources, providing a scientific basis for fault prediction and parameter adjustment.

[0179] The Transformer model combines a prediction head and a periodic update mechanism, significantly improving the accuracy of needle loss and whiteboard prediction.

[0180] The optimization control module and image recognition module form a closed-loop control, adjusting parameters in real time to extend the life of the needle and improve the quality of engraving.

[0181] The knowledge base module learns from historical data and real-time matching, automatically invoking optimal strategies to enhance the system's intelligent level.

[0182] Example 2 realizes efficient processing and intelligent optimization of multi-source data of the electric engraving machine through the collaborative work of data encoding, high-dimensional vector conversion, influence correlation calculation, and fault prediction modules.

[0183] Example 3:

[0184] As shown in Figures 1 to 3 Example 1 (data collection, preprocessing, fault prediction, and optimization control) and Example 2 (data encoding, high-dimensional vector conversion, and influence correlation calculation), the image recognition module, optimization process recording module, evaluation and screening module, and knowledge base module are further implemented. Through these modules, the system can evaluate the quality of the engraved dots in real time, record the optimization process, screen high-quality optimization strategies, and achieve adaptive learning through the knowledge base, thereby optimizing the operating parameters of the electric engraving machine and enhancing the intelligent level.

[0185] The image recognition module is based on convolutional neural network (CNN) technology and is used to analyze the images of engraved dots, evaluate the quality of the dots, and feed the results back to the optimization control module. The implementation details are as follows:

[0186] Dot image acquisition: After the electric engraving machine completes engraving, a high-resolution camera captures the dot image in real time and transmits the image to the image recognition module.

[0187] Image preprocessing: The collected image is subjected to grayscale processing to reduce noise interference, denoising through Gaussian filtering, and edge enhancement technology to highlight the dot outline and details.

[0188] CNN model architecture: A multi-layer CNN model is constructed, including convolution layers (extracting dot features such as size and edge shape), pooling layers (reducing computational complexity), and fully connected layers (outputting quality evaluation results). The model learns dot features such as dot size uniformity, edge jaggedness, and shape consistency through training.

[0189] Dot quality evaluation: The model outputs quantitative indicators, including:

[0190] Uniformity of dot size (measured by calculating the standard deviation, the smaller the better);

[0191] Edge jaggy degree (the fewer the better);

[0192] Sufficient ink content (judged by dot saturation, for example, insufficient saturation may indicate insufficient ink content during printing).

[0193] Feedback mechanism: feedback the evaluation results (such as dot quality score or ink content deficiency prompt) to the optimization control module, for example, if there are too many edge jaggies, it may trigger adjustment of the vibration frequency; if the ink content is insufficient, it may prompt adjustment of the engraving speed.

[0194] The optimization process record module is responsible for recording detailed data of each optimization, providing support for subsequent analysis and learning, and the specific implementation is as follows:

[0195] Record content: including:

[0196] Branch data chain: such as time series data of the service life of the engraving needle;

[0197] Data chain: combination of multi-source data (such as running data and environmental data) in chronological order;

[0198] Data column: combination of multi-source data at the same time;

[0199] Prediction results: such as the output of the fault prediction module, the wear value of the engraving needle and the risk of whiteout;

[0200] Adjustment parameters: such as specific changes in engraving speed and vibration frequency;

[0201] Image recognition evaluation results: such as dot quality score.

[0202] Storage method: use a relational database (such as MySQL) to store optimization records, design an "optimization record table" containing fields such as "timestamp", "branch data chain", "prediction results", "adjustment parameters", "dot quality score", etc., for easy query and analysis.

[0203] The evaluation and screening module analyzes the optimization effect through evaluation indicators and selects high-quality optimization processes, and the specific implementation is as follows:

[0204] Evaluation indicators:

[0205] Dot quality improvement amplitude: calculate the difference between the dot quality scores before and after adjustment;

[0206] Fault reduction rate: such as the percentage reduction in whiteout phenomenon after adjustment.

[0207] Reinforcement Learning Mechanism (Optional): Adopt reinforcement learning algorithm, treat the optimized action as action, and the improvement of mesh quality as reward. Through trial and error, learn the optimal strategy.

[0208] Screening Criteria: Set threshold (e.g. mesh quality improvement amplitude exceeds 10% or scratch reduction rate exceeds 15%), screen the optimization processes that meet the conditions, and mark them as "high-quality optimization".

[0209] Knowledge Base Module stores high-quality optimization processes and supports adaptive optimization, which is implemented as follows:

[0210] Knowledge Base Construction: Store the key features (e.g. data patterns under high loss conditions) and corresponding actions (e.g. reduce engraving speed by 10%) of the screened high-quality optimization processes in the knowledge base.

[0211] Similarity Calculation: In the running process, use Euclidean distance or cosine similarity to calculate the similarity between the current data and the optimization processes in the knowledge base.

[0212] Automatic Call: When the similarity exceeds the preset threshold (e.g. 0.9), automatically call the optimization strategy with the highest matching degree and apply it to the current running to achieve adaptive optimization.

[0213] Working Process

[0214] Data Acquisition Module: Real-time acquisition of engraving machine running data, environmental data, etc. After data preprocessing module cleaning, conversion and compression, data encoding module generates data chain, data chain and data column, and then high-dimensional vector conversion module generates input features.

[0215] Fault Prediction and Preliminary Optimization

[0216] Fault Prediction Module: Based on the Transformer model, output the degree of stylus loss and scratch risk. Optimization control module adjusts the engraving speed or vibration frequency according to the prediction results, prolongs the stylus life and reduces the scratch.

[0217] Image Recognition Evaluation

[0218] After engraving, the image recognition module analyzes the mesh image and outputs the mesh quality score and ink content evaluation. If the quality is not up to standard (e.g. too many edge sawteeth), the result is fed back to the optimization control module to trigger further adjustment.

[0219] Optimization Record

[0220] Optimization Process Record Module: Stores the input data, prediction results, adjustment parameters and mesh quality score of this optimization to a relational database.

[0221] Evaluation and Screening

[0222] The evaluation and screening module calculates the optimization effect (such as the improvement range of mesh quality), and if it exceeds the threshold (such as 10%), it is marked as high-quality optimization.

[0223] Knowledge base updating and calling

[0224] The high-quality optimization process is stored in the knowledge base module. In subsequent operation, the system calculates the similarity between the current data and the knowledge base mode in real time, and automatically calls the optimization strategy with the highest matching degree.

[0225] Continuous optimization

[0226] Through recording, evaluation and learning, the system forms an adaptive optimization closed loop to continuously improve the operation effect.

[0227] The image recognition module analyzes the mesh size, edge sawtooth and ink content through CNN to objectively quantify the engraving quality, improving the precision and reliability of quality control.

[0228] The image recognition result is fed back to the optimization control module in real time, forming a closed loop of parameter adjustment and quality evaluation to ensure effective implementation of optimization measures.

[0229] The optimization process record module stores the detailed data of each optimization in a relational database, providing a rich data basis for subsequent improvement and problem tracing.

[0230] The knowledge base module realizes adaptive learning of the system by similarity matching and automatic calling of optimization strategies, reduces manual intervention, and improves the intelligent level.

[0231] The evaluation and screening module screens high-quality optimization processes and stores them in the knowledge base, ensuring that the system continuously accumulates experience and continuously improves the operation efficiency and engraving quality of the electric engraver.

[0232] In embodiment 3, through the cooperative work of the image recognition module, the optimization process record module, the evaluation and screening module and the knowledge base module, real-time evaluation of engraving quality and adaptive optimization of operation parameters are realized. On the basis of embodiments 1 and 2, this embodiment adds closed loop control and intelligent learning ability, improving the intelligence and practicality of the electric engraver detection system.

[0233] The above is only the preferred embodiment of the present application, it should be understood that the present application is not limited to the form disclosed herein, should not be regarded as excluding other embodiments, and can be used in various other combinations, modifications and environments, and can be modified by the above teaching or related art or knowledge within the scope of the concept described herein, and the modification and change made by the person skilled in the art without departing from the spirit and scope of the present application shall be within the protection scope of the claims attached to the present application.

Claims

1. An intelligent laser engraving detection system for an electric engraving machine, characterized in that: include: Data acquisition module, used to collect multi-source data of the electric engraving machine in real time; The data preprocessing module is used to clean, transform, and compress the collected multi-source data to generate a format suitable for machine learning model processing; The data encoding module is used to encode the pre-processed multi-source data into branch data chains, data chains and data columns in chronological order; The high-dimensional vector conversion module uses the Transformer model to convert data chains into high-dimensional vectors, captures long-term dependencies of time steps through the self-attention mechanism, and encodes data columns into high-dimensional vectors through a fully connected layer; The impact correlation calculation module quantifies the impact correlation between different data sources by calculating the cosine similarity of high-dimensional vectors; A fault prediction module adds a prediction head based on the output layer of the Transformer model; the fault prediction module includes a regression layer for predicting needle loss and a classification layer or regression layer for predicting blade whitening, which output the degree of loss and the probability or degree of whitening respectively. The Transformer model in the high-dimensional vector conversion module includes multiple encoding layers and self-attention heads for generating a time step vector representation of the data chain; Optimize the control module and adjust the engraving speed and vibration frequency of the electric engraving machine in real time according to the fault prediction results to extend the life of the engraving needle and reduce the whitening phenomenon. An image recognition module uses a convolutional neural network (CNN) to analyze the engraved dot image, evaluate the dot quality, and feed the evaluation results back to the optimization control module to achieve closed-loop control. The image recognition module uses CNN to analyze the dot size and edge jaggedness to determine whether the ink content is sufficient during use, and the judgment results are fed back to the optimization control module as feedback; It also includes an optimization process recording module, an evaluation and screening module, and a knowledge base module. The evaluation and screening module evaluates the effect of each optimization through a defined evaluation index or reinforcement learning reward function, and screens out the optimization process with better optimization effect. The optimization process recording module is used to record the input data, prediction results, optimization actions, and feedback results of each optimization process. The needle loss prediction regression layer and the blade whitening prediction classification layer or regression layer in the fault prediction module are trained with historical data and updated regularly to improve prediction accuracy. The knowledge base module is used to store the key features and corresponding actions of the screened optimization process, and to match the current data with the patterns in the knowledge base in real time during the working process, and automatically call the optimization strategy with the highest matching degree; the multi-source data includes the operation data of the electric engraving machine, the imposition software data, the environmental data and the proofing effect data; the operation data of the electric engraving machine collected by the data acquisition module includes the usage time of the engraving needle, the number of engraving lines, the engraving time, the vibration state and frequency of the engraving needle, and the composition data of the engraving needle material.

2. The intelligent laser engraving detection system for an electric engraving machine according to claim 1, characterized in that: The branch data chain is a time series of each data source, the data chain is a combination of multiple branch data chains in time sequence, and the data column is a parallel combination of data from multiple branch data chains at the same time point. The data preprocessing module compresses the TIFF image to ensure that the text is not distorted.

3. The intelligent laser engraving detection system for an electric engraving machine according to claim 1, characterized in that: The optimization control module automatically reduces the engraving speed when predicting that the engraving needle loss exceeds a preset threshold, and adjusts the vibration frequency when predicting the risk of whitening.

4. The intelligent laser engraving detection system for an electric engraving machine according to claim 1, characterized in that: The optimization process recording module stores each optimization process in a relational database, and the record content includes branch data chains, data chains, data columns, prediction results, adjusted parameters and image recognition evaluation results.

5. The intelligent laser engraving detection system for an electric engraving machine according to claim 1, characterized in that: The evaluation and screening module uses the improvement of the quality of the network points or the failure reduction rate as an evaluation indicator to screen out the optimization process with the best optimization effect.

6. The intelligent laser engraving detection system for an electric engraving machine according to claim 1, characterized in that: The knowledge base module automatically calls the optimization strategy with the highest matching degree by calculating the similarity between the current data and the storage optimization process, thereby achieving adaptive learning and optimization.

Images