Rock wool board processing management system and method based on data analysis

Abstract

The invention relates to the field of rock wool board processing, and particularly discloses a rock wool board processing management system and method based on data analysis, and the method comprises the steps: firstly obtaining a rock wool board cleaning surface image collected by a camera and brush pressure values of a rock wool board surface cleaning device at a plurality of preset time points, and then carrying out the deep learning technology, feature extraction and correlation analysis are carried out on the rock wool board and the rock wool board, and finally a classification result is obtained through a classifier to judge whether the brush pressure value of the rock wool board surface cleaning device needs to be adjusted or not, so that the cleanliness of the surface of the rock wool board is accurately evaluated, and it is ensured that the rock wool board reaches the cleaning standard before film covering.

Description

Technical Field

[0001] This application relates to the field of rock wool board processing, and more specifically, to a data analysis-based rock wool board processing management system and method. Background Technology

[0002] Rock wool board, also known as rock wool insulation and decorative board, is an inorganic fiber material made from natural rocks such as basalt through high-temperature melting and fiberization. This type of board is lightweight, has a low thermal conductivity, excellent heat absorption performance, and fire resistance, and is widely used in construction, industrial equipment, and transportation.

[0003] During the production process, rock wool boards often require a coating treatment to improve their durability and aesthetics. However, dust and debris adhere to the surface of the rock wool boards, necessitating a surface cleaning device. This cleaning process is based on a preset program and cannot be customized to suit specific conditions, which can easily affect the overall quality of the coating and production efficiency.

[0004] Therefore, a data analysis-based management system and method for rock wool board processing is desired. Summary of the Invention

[0005] To address the aforementioned technical problems, this application is proposed. Embodiments of this application provide a data analysis-based rock wool board processing management system and method. First, it acquires images of the cleaned surface of the rock wool board captured by a camera and brush pressure values ​​of a rock wool board surface cleaning device at multiple predetermined time points. Then, it utilizes deep learning technology to extract features and perform correlation analysis on the two images. Finally, it uses a classifier to obtain classification results to determine whether the brush pressure values ​​of the rock wool board surface cleaning device need adjustment, thereby accurately assessing the cleanliness of the rock wool board surface and ensuring that the rock wool board meets the cleanliness standards before lamination.

[0006] According to one aspect of this application, a data analysis-based rock wool board processing management system is provided, comprising:

[0007] The rock wool board processing data acquisition module is used to acquire images of the cleaned surface of the rock wool board captured by the camera and the brush pressure values ​​of the rock wool board surface cleaning device at multiple predetermined time points;

[0008] The rock wool board processing data extraction module is used to extract the rock wool board cleaning surface feature vector and the cleaning device brush pressure feature vector from the rock wool board cleaning surface image captured by the camera and the brush pressure values ​​of the rock wool board surface cleaning device at multiple predetermined time points.

[0009] The brush pressure value adjustment judgment module is used to determine whether the brush pressure value of the rock wool board surface cleaning device needs to be adjusted based on the feature vector of the rock wool board cleaning surface and the feature vector of the brush pressure of the cleaning device.

[0010] According to another aspect of this application, a data analysis-based method for managing the processing of rock wool boards is provided, comprising:

[0011] Acquire images of the cleaned surface of the rock wool board captured by the camera and the brush pressure values ​​of the rock wool board surface cleaning device at multiple predetermined time points;

[0012] The feature vectors of the cleaned surface of the rock wool board and the feature vectors of the brush pressure of the cleaning device are extracted from the images of the cleaned surface of the rock wool board captured by the camera and the brush pressure values ​​of the rock wool board surface cleaning device at multiple predetermined time points.

[0013] Based on the feature vector of the rock wool board cleaning surface and the feature vector of the brush pressure of the cleaning device, it is determined whether the brush pressure value of the rock wool board surface cleaning device needs to be adjusted.

[0014] Compared with existing technologies, this application provides a data analysis-based rock wool board processing management system and method. First, it acquires images of the clean surface of the rock wool board captured by a camera and the brush pressure values ​​of the rock wool board surface cleaning device at multiple predetermined time points. Then, it uses deep learning technology to extract features and perform correlation analysis on the two. Finally, it uses a classifier to obtain classification results to determine whether the brush pressure value of the rock wool board surface cleaning device needs to be adjusted, thereby accurately assessing the cleanliness of the rock wool board surface and ensuring that the rock wool board meets the cleanliness standards before lamination. Attached Figure Description

[0015] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0016] Figure 1 This is a block diagram of a data analysis-based rock wool board processing and management system according to an embodiment of this application.

[0017] Figure 2 This is a block diagram of the rock wool board processing data extraction module in the data analysis-based rock wool board processing management system according to an embodiment of this application.

[0018] Figure 3 This is a block diagram of the rock wool board cleaning surface feature extraction unit in the data analysis-based rock wool board processing management system according to an embodiment of this application.

[0019] Figure 4This is a block diagram of the pressure feature extraction unit of the cleaning device in the data analysis-based rock wool board processing management system according to an embodiment of this application.

[0020] Figure 5 This is a block diagram of the brush pressure value adjustment judgment module in the data analysis-based rock wool board processing management system according to an embodiment of this application.

[0021] Figure 6 This is a flowchart of a data analysis-based rock wool board processing management method according to an embodiment of this application. Detailed Implementation

[0022] Various exemplary embodiments, features, and aspects of this application will now be described in detail with reference to the accompanying drawings. The same reference numerals in the drawings denote elements that have the same or similar functions. Although various aspects of the embodiments are shown in the drawings, they are not necessarily drawn to scale unless specifically indicated otherwise.

[0023] The term “exemplary” as used herein means “serving as an example, embodiment, or illustration.” Any embodiment illustrated herein as “exemplary” is not necessarily to be construed as superior to or better than other embodiments.

[0024] Furthermore, to better illustrate this application, numerous specific details are provided in the following detailed description. Those skilled in the art should understand that this application can be implemented without certain specific details. In some instances, methods, means, components, and circuits well-known to those skilled in the art have not been described in detail in order to highlight the main points of this application.

[0025] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0026] Figure 1 This is a schematic block diagram of a data analysis-based rock wool board processing and management system according to an embodiment of this application. Figure 1As shown, the data analysis-based rock wool board processing management system 100 according to an embodiment of this application includes: a rock wool board processing data acquisition module 110, used to acquire images of the cleaned surface of the rock wool board captured by a camera and brush pressure values ​​of a rock wool board surface cleaning device at multiple predetermined time points; a rock wool board processing data extraction module 120, used to extract a rock wool board cleaned surface feature vector and a cleaning device brush pressure feature vector from the rock wool board cleaned surface images captured by the camera and the brush pressure values ​​of the rock wool board surface cleaning device at multiple predetermined time points; and a brush pressure value adjustment judgment module 130, used to determine whether the brush pressure value of the rock wool board surface cleaning device needs to be adjusted based on the rock wool board cleaned surface feature vector and the cleaning device brush pressure feature vector.

[0027] In the aforementioned data analysis-based rock wool board processing management system 100, the rock wool board processing data acquisition module 110 is used to acquire images of the cleaned surface of the rock wool board captured by a camera and brush pressure values ​​at multiple predetermined time points of the rock wool board surface cleaning device. It should be understood that rock wool board, also known as rock wool insulation decorative board, is an inorganic fiber material made from natural rocks such as basalt that are melted and fiberized at high temperatures. This type of board is widely used in construction, industrial equipment, and transportation due to its lightweight, low thermal conductivity, excellent heat absorption, and fire resistance. During the production process, rock wool boards typically require a coating process to enhance their durability and appearance. However, because dust and debris often accumulate on the surface of rock wool boards, a specialized cleaning device is needed. Current cleaning processes are based on fixed procedures and cannot be adjusted according to actual needs, which may affect the quality of the coating and production efficiency. Therefore, in the technical solution of this application, by acquiring images of the cleaned surface of the rock wool board captured by a camera and the brush pressure values ​​of the rock wool board surface cleaning device at multiple predetermined time points, and combining this with deep learning technology, it is determined whether the brush pressure values ​​of the rock wool board surface cleaning device need to be adjusted, so as to accurately assess the cleanliness of the rock wool board and ensure that it meets the required cleaning standards before lamination.

[0028] Specifically, acquiring images of the cleaned rock wool board surface and brush pressure values ​​captured by a camera is a crucial step in ensuring the processing quality of the rock wool board. The camera images provide detailed visual data of the cleaned rock wool board surface, enabling the system to analyze surface cleanliness and uniformity. Simultaneously, recording brush pressure values ​​at multiple time points allows for monitoring the performance and pressure variations of the cleaning device, which is essential for ensuring the brush remains stable and effective throughout the cleaning process. Comprehensive analysis of image and pressure data allows the system to dynamically adjust cleaning parameters during processing, thereby improving the overall quality and production efficiency of the rock wool board.

[0029] In the aforementioned data analysis-based rock wool board processing management system 100, the rock wool board processing data extraction module 120 is used to extract the rock wool board cleaning surface feature vector and the cleaning device brush pressure feature vector from the rock wool board cleaning surface image captured by the camera and the brush pressure values ​​of the rock wool board surface cleaning device at multiple predetermined time points. This enables comprehensive monitoring and optimization of the rock wool board cleaning process, thereby improving the control precision of the production process, reducing the need for manual intervention, and enhancing production efficiency and product quality.

[0030] Figure 2 This is a schematic block diagram of the rock wool board processing data extraction module in a data analysis-based rock wool board processing management system according to an embodiment of this application. Figure 2 As shown, in a specific embodiment of this application, the rock wool board processing data extraction module 120 includes: a rock wool board cleaning surface feature extraction unit 121, used to extract features from the rock wool board cleaning surface image captured by the camera to obtain the rock wool board cleaning surface feature vector; and a cleaning device pressure feature extraction unit 122, used to extract features from the brush pressure values ​​of the rock wool board surface cleaning device at multiple predetermined time points to obtain the cleaning device brush pressure feature vector.

[0031] As can be understood, image data provides detailed visual information about the surface of rock wool boards, including surface condition, dirt distribution, and cleanliness uniformity. Directly analyzing this image data is not only time-consuming but also difficult to process; therefore, feature extraction is necessary to transform the image data into feature vectors. The feature extraction process uses advanced image processing and computer vision techniques to extract a set of descriptive numerical features from the original images. These feature vectors quantify surface cleanliness, including dirt distribution, cleanliness unevenness, and other potential surface problems. Using these feature vectors, the system can systematically analyze the cleanliness of each rock wool board and determine whether it meets predetermined quality standards. This enables the system to perform efficient quality inspection and automated adjustments without manual intervention. Feature vectorization not only improves data processing speed but also enhances the system's sensitivity and accuracy in the cleaning process. Through this method, the cleanliness of rock wool boards can be objectively and quantitatively assessed, helping to promptly identify potential quality problems during production.

[0032] Furthermore, extracting features from the brush pressure values ​​of the rock wool board surface cleaning device at multiple predetermined time points to obtain the brush pressure feature vector is a crucial step in ensuring the cleaning effect of the rock wool board and optimizing the cleaning process. The brush pressure value directly affects the cleaning performance of the cleaning device; appropriate pressure can effectively remove dirt from the surface of the rock wool board, while excessively high or low pressure may lead to incomplete cleaning or damage to the rock wool board. By extracting features from the brush pressure values, time-series data can be transformed into representative feature vectors, which reflect the working status of the cleaning device at different time points. This feature extraction transforms the raw, complex pressure data into a simpler form with higher information density, enabling the system to perform analysis and decision-making more efficiently.

[0033] Figure 3 This is a block diagram of a rock wool board cleaning surface feature extraction unit in a data analysis-based rock wool board processing management system according to an embodiment of this application. Figure 3 As shown, in a specific embodiment of this application, the rock wool board cleaning surface feature extraction unit 121 includes: a cleaning surface feature encoding subunit 1211, used to perform feature encoding on the rock wool board cleaning surface image captured by the camera to obtain a rock wool board cleaning surface feature matrix; and a cleaning surface feature expansion subunit 1212, used to expand the rock wool board cleaning surface feature matrix into the rock wool board cleaning surface feature vector.

[0034] It is understandable that image data itself contains rich visual information, but analyzing raw images directly is not only complex but also computationally expensive. Feature encoding transforms this image data into a structured feature matrix, enabling the system to process and analyze image information more efficiently. Feature encoding extracts key features from images, such as texture, color, edges, shape, and region distribution, and converts this information into digital form, forming a feature matrix. This matrix compresses and summarizes the important information in the image, retaining features crucial for assessing the cleaning status while removing redundant and irrelevant information. The feature-matrixed data can be more easily integrated and compared with other data sources, such as brush pressure values ​​or production parameters. This comprehensive analysis provides a complete understanding of the rock wool board cleaning process, supporting more refined quality control and optimization strategies.

[0035] Furthermore, expanding the feature matrix of the rock wool board cleaning surface into feature vectors simplifies data representation and improves analysis efficiency. Feature matrices, when capturing complex cleaning state information in images, are typically presented as multidimensional arrays, where each element represents a different feature or attribute. While such matrices can comprehensively describe various surface details, directly processing such high-dimensional data in practical applications can lead to high computational complexity and slow processing speed. By expanding the feature matrix into feature vectors, these multidimensional data can be converted into one-dimensional arrays. This one-dimensional representation retains all the key information extracted from the feature matrix but is organized in a more compact form, enabling the model to perform calculations and decisions faster. The feature vector expansion process involves arranging each element of the matrix into a long vector in a specific order. This transformation linearizes all dimensions of the features, facilitating data input into various machine learning algorithms or data processing tools. Feature vectorized data can be more easily compared and combined with other data sources (such as brush pressure feature vectors) for comprehensive analysis and optimization. Feature vectorization helps reduce computational resource consumption and improves the efficiency of model training and inference.

[0036] In a specific embodiment of this application, the clean surface feature encoding subunit 1211 includes: passing the rock wool board clean surface image captured by the camera through a rock wool board clean surface image denoising module to obtain a rock wool board clean surface pixel enhancement image; and passing the rock wool board clean surface pixel enhancement image through a rock wool board clean surface feature extractor based on a pyramid network to obtain the rock wool board clean surface feature matrix.

[0037] It is understandable that in practical applications, the image acquisition process is often affected by ambient light, camera quality, and other interfering factors, leading to problems such as noise, blurriness, and uneven lighting in the image. This noise and poor image quality can severely impact image analysis results, especially when precise assessment of the cleanliness of rock wool board surfaces is required. The main function of the image denoising module is to remove these interfering factors from the original image, improving image clarity and contrast, making the cleanliness features of the rock wool board more prominent. By applying denoising algorithms such as mean filtering, Gaussian filtering, and median filtering, random noise and blurred areas in the image can be effectively reduced, making the surface details of the rock wool board more apparent. This clearer image enhances pixel contrast and detail, making minor defects such as dirt and uneven cleanliness easier to detect and identify. The denoised image is more suitable for subsequent image processing and analysis steps. Techniques such as feature extraction, pattern recognition, and image classification are sensitive to noise, and images with less noise provide more stable and reliable features, thereby improving the accuracy of analysis algorithms. Specifically, the image of the clean surface of the rock wool board captured by the camera is input into the encoder of the image noise reduction module for the clean surface of the rock wool board, wherein the encoder uses a convolutional layer to perform explicit spatial encoding on the image of the clean surface of the rock wool board captured by the camera to obtain image features; and the image features are input into the decoder of the image noise reduction module for the clean surface of the rock wool board, wherein the decoder uses a deconvolutional layer to perform deconvolution processing on the image features to obtain a pixel-enhanced image of the clean surface of the rock wool board.

[0038] Furthermore, the pixel-enhanced image of the cleaned surface of the rock wool board is processed by a feature extractor based on a pyramid network to obtain a feature matrix of the cleaned surface of the rock wool board. This is to efficiently extract important features from the image at multiple scales and transform these features into a structured data form, thereby supporting accurate cleaning status analysis and optimization. A pyramid network is a widely used technique in image processing that captures detailed information at different levels by constructing image pyramids at multiple scales. Pyramid networks can handle multi-scale information in images, solving the problem that single-scale feature extraction cannot fully capture details. In images of the cleaned surface of rock wool boards, dirt and cleanliness may exhibit different characteristics at different spatial scales. For example, some defects may be obvious at a larger scale, while subtle cleaning unevenness needs to be identified at a smaller scale. By progressively scaling the image, the pyramid network can capture these detailed information at different scales, thus forming a multi-level, multi-scale feature representation. By integrating these feature maps, a structured feature matrix is ​​finally formed, which fully reflects the details and features at various scales in the image. The feature matrix not only retains all the important information extracted from the original image, but also presents it in a form suitable for computation and analysis. Specifically, a shallow feature matrix is ​​extracted from the i-th layer of the pyramid network-based rock wool board cleaning surface feature extractor, where the i-th layer is the first to sixth layers of the pyramid network-based rock wool board cleaning surface feature extractor; a deep feature matrix is ​​extracted from the j-th layer of the pyramid network-based rock wool board cleaning surface feature extractor, where the ratio between the j-th layer and the i-th layer is greater than or equal to 5; and a shallow and deep feature fusion module is used to fuse the shallow feature matrix and the deep feature matrix to obtain the rock wool board cleaning surface feature matrix.

[0039] Figure 4 This is a block diagram of the pressure feature extraction unit of the cleaning device in a data analysis-based rock wool board processing management system according to an embodiment of this application. Figure 4 As shown, in a specific embodiment of this application, the cleaning device pressure feature extraction unit 122 includes: a brush pressure feature encoding subunit 1221, used to encode the brush pressure values ​​of the rock wool board surface cleaning device at multiple predetermined time points to obtain a cleaning device brush pressure feature map; and a brush pressure feature pooling subunit 1222, used to perform maximum pooling on the cleaning device brush pressure feature map to obtain the cleaning device brush pressure feature vector.

[0040] It is understandable that brush pressure is a critical parameter in the cleaning process of rock wool boards, as it directly affects the cleaning effect and the surface quality of the rock wool boards. Insufficient brush pressure may lead to incomplete cleaning, while excessive pressure may damage the surface of the rock wool boards. Therefore, accurately measuring and analyzing changes in brush pressure is essential for optimizing the cleaning process. The feature encoding process digitizes brush pressure data at different time points to form a unified feature map. These pressure values ​​are transformed into structured information that systematically reflects how brush pressure changes throughout the cleaning process. Through feature encoding, this time-series data can be transformed into a feature map with spatial and temporal dimensions, helping to reveal the dynamic impact of brush pressure at different time points and its effect on cleaning performance.

[0041] Furthermore, max pooling is applied to the feature map of the cleaning device brush pressure to obtain the feature vector of the cleaning device brush pressure. This is to enhance the expressive power of the data by simplifying and extracting key features, while reducing computational complexity. It should be understood that the brush pressure feature map contains brush pressure data at different time points, which is converted into a multi-dimensional matrix through feature encoding. Although these matrices can record the pressure distribution and changes in detail, directly using them for subsequent processing and analysis may face problems of heavy computational burden and excessively high data dimensionality. Max pooling is an effective dimensionality reduction technique that compresses data by extracting the maximum value from each pooling region of the feature map while retaining the most significant feature information. Specifically, the max pooling operation divides the feature map into several small regions (such as 2x2 windows) and then extracts the maximum value from each region. This process reduces the spatial dimensionality of the data while retaining the most important information within each region. The advantage of this is that it can eliminate the impact of small local changes on the overall feature extraction, ensuring that the most significant feature information is retained, thereby enhancing the representativeness and robustness of the features. Max pooling helps improve a model's generalization ability because it reduces the risk of overfitting to specific local features by decreasing the detail and complexity of the data.

[0042] In a specific embodiment of this application, the brush pressure feature encoding subunit 1222 includes: arranging the brush pressure values ​​of the rock wool board surface cleaning device at multiple predetermined time points in a two-dimensional manner to obtain a cleaning device brush pressure input matrix; and passing the cleaning device brush pressure input matrix through a cleaning device brush pressure convolution encoding model to obtain the cleaning device brush pressure feature map.

[0043] It's understandable that in the actual operation of cleaning devices, the pressure values ​​of the brushes are typically recorded at different points in time, providing information about the dynamic changes in the cleaning process. However, while time-series data itself reflects changes in brush pressure, many analytical methods and algorithms may require processing it in a specific structure. For example, many machine learning models and image processing techniques typically expect data to be presented in matrix form, where each dimension represents a specific feature or dimension. Arranging the time-series brush pressure values ​​in a two-dimensional fashion effectively organizes the pressure data from these time points into a matrix according to specific rules. This two-dimensional arrangement can form a matrix with row and column structures based on the order of the time points and the specific pressure value at each point. This structured input matrix not only visually displays the changes in pressure data over time but also captures the overall trend and local details of pressure changes. This matrix format allows the data to be directly used in various data processing techniques, such as image processing and convolutional neural networks (CNNs), which can effectively extract features and patterns from the matrix.

[0044] Furthermore, the cleaning device brush pressure input matrix is ​​processed through a cleaning device brush pressure convolutional coding model to obtain a cleaning device brush pressure feature map. This is to automatically extract and encode valuable feature information from the input matrix for deeper analysis and optimization. This process leverages the advantages of convolutional neural network (CNN) technology to efficiently process and interpret complex data structures. The cleaning device brush pressure input matrix is ​​formed by arranging the brush pressure values ​​at multiple time points in a two-dimensional manner, which can display the dynamic characteristics of pressure changes in a structured way. However, directly extracting useful information from such a two-dimensional matrix can be complex because it involves multi-dimensional data interaction and pattern recognition problems. Convolutional coding models have demonstrated powerful capabilities in this regard. Convolutional coding models (usually based on convolutional neural networks) can automatically extract local features from the input matrix through convolution operations. This operation involves sliding the convolution kernel (or filter) on the input matrix, performing weighted summation on local regions, and then passing the result to the next layer. Through this process, the convolutional coding model can capture local patterns in the input matrix, such as feature regions or trends of pressure changes. Each convolutional kernel focuses on detecting a specific type of feature in the matrix, thereby progressively extracting more complex and abstract features at multiple levels. After convolutional processing, the model generates one or more feature maps. Each feature map represents the response of the input matrix under the convolutional kernel operation, displaying the spatial distribution of different features. The feature maps not only reflect the local features of brush pressure but also reveal important patterns throughout the cleaning process, such as the uniformity of pressure distribution or local anomalies. This information is crucial for understanding and optimizing cleaning performance. Specifically, each layer of the cleaning device brush pressure convolutional coding model performs convolution processing, mean pooling based on the local feature matrix, and nonlinear activation processing on the input data during the forward pass of the layer to output the cleaning device brush pressure feature map from the last layer of the cleaning device brush pressure convolutional coding model, where the input of the cleaning device brush pressure convolutional coding model is the cleaning device brush pressure input matrix.

[0045] In the aforementioned data analysis-based rock wool board processing management system 100, the brush pressure adjustment judgment module 130 is used to determine whether the brush pressure value of the rock wool board surface cleaning device needs adjustment based on the rock wool board cleaning surface feature vector and the cleaning device brush pressure feature vector. It should be understood that the rock wool board cleaning surface feature vector represents the actual cleaning effect on the rock wool board surface during the cleaning process. This feature vector is typically obtained by scanning and analyzing the cleaned surface, and includes information such as the uniformity of the cleaned area, residue distribution, and surface quality. The cleaning surface feature vector can reveal the effectiveness and degree of cleaning of the current cleaning process, such as whether there are areas that have not been cleaned or whether there are signs of excessive wear. On the other hand, the cleaning device brush pressure feature vector reflects the pressure distribution of the brush during the cleaning process. Changes in brush pressure directly affect the cleaning effect; appropriate pressure ensures uniform cleaning of the surface, while excessively high or low pressure may lead to incomplete cleaning or damage to the surface. By combining these two feature vectors, it is possible to dynamically adjust the brush pressure of the cleaning device, thereby optimizing the cleaning effect.

[0046] Figure 5 This is a schematic block diagram of a brush pressure value adjustment judgment module in a data analysis-based rock wool board processing management system according to an embodiment of this application. In a specific embodiment of this application, the brush pressure value adjustment judgment module 130 includes: a rock wool board feature fusion unit 131, used to fuse the feature vector of the clean surface of the rock wool board and the feature vector of the brush pressure of the cleaning device to obtain a surface cleaning device pressure adjustment judgment feature vector; a rock wool board feature optimization unit 132, used to perform shift compensation correction based on nonlinear transformation on the feature vector of the cleaning device pressure adjustment judgment to obtain an optimized feature vector of the cleaning device pressure adjustment; and a brush pressure adjustment judgment classification unit 133, used to pass the optimized feature vector of the surface cleaning device pressure adjustment judgment through a classifier to obtain a classification result, the classification result being used to determine whether the brush pressure value of the rock wool board surface cleaning device needs to be adjusted.

[0047] It is understandable that fusing these two feature vectors to form a feature vector for judging pressure adjustment in surface cleaning devices allows for more comprehensive and precise adjustment decisions. The fusion process, by combining cleaning effectiveness with changes in brush pressure, reveals the relationship between the two. The advantage of this fusion strategy lies in its ability to integrate multifaceted information, providing a comprehensive perspective for optimizing the cleaning process. By comprehensively considering both cleaning effectiveness and brush pressure, it is possible to more accurately determine whether the current settings need adjustment, thereby achieving fine-tuning of the cleaning device and ensuring optimal cleaning results and long-term equipment operation.

[0048] Specifically, in the technical solution of this application, the features of the rock wool board image and the brush pressure are obtained from different data sources. This may lead to different types of noise in the surface cleaning device pressure adjustment judgment feature vector after feature extraction and feature fusion. The feature extraction stage is complex and relies on multiple steps, including image denoising, feature extraction, and feature vector expansion. Each of these steps may introduce small errors or interference, resulting in local features in the surface cleaning device pressure adjustment judgment feature vector potentially having input perturbations. Considering that the classifier operates based on the input surface cleaning device pressure adjustment judgment feature vector, if the surface cleaning device pressure adjustment judgment feature vector becomes unstable due to perturbations during data processing, the classifier may suffer from insufficient classification accuracy. Therefore, in the technical solution of this application, the surface cleaning device pressure adjustment judgment feature vector is subjected to shift compensation correction based on nonlinear transformation to obtain an optimized surface cleaning device pressure adjustment judgment feature vector.

[0049] The process of performing shift compensation correction based on nonlinear transformation on the pressure adjustment judgment feature vector of the cleaning device to obtain an optimized pressure adjustment judgment feature vector includes: determining the first and second pressure adjustment judgment weight matrices of the classifier before and after each iteration update; performing matrix multiplication of the first and second pressure adjustment judgment weight matrices with the pressure adjustment judgment feature vector of the cleaning device to obtain the first and second pressure adjustment judgment classification modulation feature vectors of the cleaning device; and calculating the positional difference between the first and second pressure adjustment judgment classification modulation feature vectors of the cleaning device to obtain the pressure adjustment judgment shift information representation. The following steps are taken: Calculate the F-norm of the position-mean vector between the first and second cleaning device pressure adjustment judgment classification modulation feature vectors as a shift compensation scaling factor for cleaning device pressure adjustment judgment; linearly scale the cleaning device pressure adjustment judgment shift information representation vector using the shift compensation scaling factor to obtain a scaled cleaning device pressure adjustment judgment shift information representation vector; input the scaled cleaning device pressure adjustment judgment shift information representation vector into a Sigmoid activation function to obtain a cleaning device pressure adjustment judgment backpropagation shift compensation representation vector; calculate the position-mean multiplication of the cleaning device pressure adjustment judgment backpropagation shift compensation representation vector with the cleaning device pressure adjustment judgment feature vector to obtain the optimized cleaning device pressure adjustment judgment feature vector.

[0050] The optimization steps are specifically represented as follows:

[0051]

[0052] Wherein, V represents the feature vector for judging the pressure adjustment of the cleaning device, S represents the shift compensation scaling factor for judging the pressure adjustment of the cleaning device, M2 represents the weight matrix for judging the pressure adjustment of the second cleaning device, M1 represents the weight matrix for judging the pressure adjustment of the first cleaning device, sigmoid represents the activation function, and ⊙ represents positional dot product. Indicates matrix multiplication. Indicates division by position difference. This indicates addition by position, ||·|| F Let V' represent the Frobenius norm, and let V' represent the optimized cleaning device pressure adjustment judgment feature vector.

[0053] In the technical solution of this application, to enhance the robustness of the classifier to local perturbations in the input feature vector, a shift compensation correction based on nonlinear transformation is performed on the feature vector for judging the pressure adjustment of the cleaning device. The key to this shift compensation correction lies in capturing information about the changes in the weight matrix during training and using this information to generate a shift information representation vector for judging the pressure adjustment of the cleaning device. This is then used to generate a backward propagation shift compensation representation vector for judging the pressure adjustment of the cleaning device through scaling and a nonlinear transformation (Sigmoid function). In this way, the feature vector for judging the pressure adjustment of the cleaning device is described by using the distribution differences of the classifier in different classification scenarios during training. This not only quantifies the displacement of the feature vector caused by changes in the weight matrix but also generates a compensation vector based on the displacement information. This compensation vector adjusts the weights of each element in the feature vector according to the degree of perturbation, thereby reducing the impact of perturbation on the classification result. This significantly improves the robustness of the classifier to local perturbations in the input data, enhances the stability and generalization ability of the classifier, and by reducing the impact of local perturbations on the classification results, the classifier can still maintain high classification accuracy when facing noisy or variable data.

[0054] Furthermore, the optimized surface cleaning device pressure adjustment judgment feature vector integrates the surface cleaning effect features of the rock wool board and the brush pressure features, providing a comprehensive data foundation for optimizing the cleaning effect. By inputting the optimized surface cleaning device pressure adjustment judgment feature vector into a classifier, automated analysis of the cleaning process can be achieved. The classifier is a machine learning algorithm that, after training, can classify the input optimized surface cleaning device pressure adjustment judgment feature vector, determining which category its corresponding cleaning state belongs to, such as "pressure adjustment needed" or "pressure setting appropriate." The main advantage of using a classifier lies in its ability to handle complex data patterns and relationships. By training on large amounts of historical data, the classifier can learn the potential relationships and influencing factors between various features, thereby accurately classifying new data in practical applications. This automated process not only improves the speed of decision-making but also reduces the possibility of human error. The classifier can process large amounts of data and provide timely feedback based on the latest feature vectors, ensuring that the cleaning process remains in optimal condition in each operating cycle.

[0055] In summary, this application first acquires images of the cleaned surface of the rock wool board captured by a camera and brush pressure values ​​of the rock wool board surface cleaning device at multiple predetermined time points. Then, deep learning technology is used to extract features and perform correlation analysis on the two. Finally, a classifier is used to obtain classification results to determine whether the brush pressure value of the rock wool board surface cleaning device needs to be adjusted, thereby accurately assessing the cleanliness of the rock wool board surface and ensuring that the rock wool board meets the cleanliness standards before coating.

[0056] As described above, the data analysis-based rock wool board processing management system 100 according to the embodiments of this application can be implemented in various terminal devices. In one example, the data analysis-based rock wool board processing management system 100 can be integrated into the terminal device as a software module and / or a hardware module. For example, the data analysis-based rock wool board processing management system 100 can be a software module in the operating system of the terminal device, or it can be an application developed for the terminal device; of course, the data analysis-based rock wool board processing management system 100 can also be one of many hardware modules of the terminal device.

[0057] Alternatively, in another example, the data analysis-based rock wool board processing management system 100 and the terminal device can also be separate devices, and the data analysis-based rock wool board processing management system 100 can be connected to the terminal device via wired and / or wireless networks, and transmit interactive information in accordance with an agreed data format.

[0058] Figure 6 This is a flowchart of a data analysis-based rock wool board processing management method according to an embodiment of this application. Figure 6 As shown, the rock wool board processing management method based on data analysis according to an embodiment of this application includes: S110, acquiring images of the cleaned surface of the rock wool board captured by a camera and brush pressure values ​​of a rock wool board surface cleaning device at multiple predetermined time points; S120, extracting a rock wool board cleaned surface feature vector and a cleaning device brush pressure feature vector from the rock wool board cleaned surface images captured by the camera and the brush pressure values ​​of the rock wool board surface cleaning device at multiple predetermined time points; S130, determining whether the brush pressure value of the rock wool board surface cleaning device needs to be adjusted based on the rock wool board cleaned surface feature vector and the cleaning device brush pressure feature vector.

[0059] Here, those skilled in the art will understand that the specific operations of each step in the above-described data analysis-based rock wool board processing management method have been referenced above. Figures 1 to 5 The data analysis-based rock wool board processing management system is described in detail here, and therefore, its repeated description will be omitted.

[0060] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0061] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0062] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0063] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0064] Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be embraced within the invention. No appended diagram markings in the claims should be construed as limiting the scope of the claims.

[0065] Furthermore, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices recited in a system claim may also be implemented by a single unit or device through software or hardware. The term "second class" is used to indicate names and does not indicate any specific order.

[0066] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit of the technical solutions of the present invention.

Claims

1. A data analysis-based rock wool board processing management system, characterized by, The method comprises the following steps: A rock wool board processing data acquisition module is used to acquire a rock wool board clean surface image collected by a camera and a plurality of predetermined time point brush pressure values of a rock wool board surface cleaning device; A rock wool board processing data extraction module is used to extract a rock wool board clean surface feature vector and a cleaning device brush pressure feature vector from the rock wool board clean surface image collected by the camera and the plurality of predetermined time point brush pressure values of the rock wool board surface cleaning device; A brush pressure value adjustment judgment module is used to judge whether the rock wool board surface cleaning device brush pressure value needs to be adjusted based on the rock wool board clean surface feature vector and the cleaning device brush pressure feature vector.

2. The data analysis based rock wool board processing management system according to claim 1, characterized in that, The rock wool board processing data extraction module comprises: A rock wool board clean surface feature extraction unit is used to perform feature extraction on the rock wool board clean surface image collected by the camera to obtain the rock wool board clean surface feature vector; A cleaning device pressure feature extraction unit is used to perform feature extraction on the plurality of predetermined time point brush pressure values of the rock wool board surface cleaning device to obtain the cleaning device brush pressure feature vector.

3. The data analysis based rock wool board processing management system according to claim 2, characterized in that, The rock wool board clean surface feature extraction unit comprises: A clean surface feature coding subunit is used to perform feature coding on the rock wool board clean surface image collected by the camera to obtain a rock wool board clean surface feature matrix; A clean surface feature unfolding subunit is used to unfold the rock wool board clean surface feature matrix into the rock wool board clean surface feature vector.

4. The data analysis based rock wool board processing management system according to claim 3, characterized in that, The clean surface feature coding subunit comprises: The rock wool board clean surface image collected by the camera is passed through a rock wool board clean surface image denoising module to obtain a rock wool board clean surface pixel enhancement image; The rock wool board clean surface pixel enhancement image is passed through a pyramid network-based rock wool board clean surface feature extractor to obtain the rock wool board clean surface feature matrix.

5. The data analysis based rock wool board processing management system according to claim 4, characterized in that, The cleaning device pressure feature extraction unit comprises: A brush pressure feature coding subunit is used to perform feature coding on the plurality of predetermined time point brush pressure values of the rock wool board surface cleaning device to obtain a cleaning device brush pressure feature map; A brush pressure feature pooling subunit is used to perform maximum value pooling on the cleaning device brush pressure feature map to obtain the cleaning device brush pressure feature vector.

6. The data analysis based rock wool board processing management system according to claim 5, characterized in that, The brush pressure feature coding subunit comprises: The plurality of predetermined time point brush pressure values of the rock wool board surface cleaning device are arranged in two dimensions to obtain a cleaning device brush pressure input matrix; The cleaning device brush pressure input matrix is passed through a cleaning device brush pressure convolution coding model to obtain the cleaning device brush pressure feature map.

7. The data analysis based rock wool board processing management system according to claim 6, characterized in that, The brush pressure value adjustment judgment module comprises: A rock wool board feature fusion unit is used to fuse the rock wool board clean surface feature vector and the cleaning device brush pressure feature vector to obtain a surface cleaning device pressure adjustment judgment feature vector; A rock wool board feature optimization unit is used to perform nonlinear transformation-based shift compensation correction on the cleaning device pressure adjustment judgment feature vector to obtain an optimized cleaning device pressure adjustment judgment feature vector; The brush pressure adjustment judgment classification unit is configured to input the optimized surface cleaning device pressure adjustment judgment feature vector into a classifier to obtain a classification result, and the classification result is used to determine whether the rock wool board surface cleaning device brush pressure value needs to be adjusted.

8. The data analysis based rock wool board processing management system according to claim 7, characterized in that, The rock wool board feature optimization unit comprises: determining the first cleaning device pressure adjustment judgment weight matrix and the second cleaning device pressure adjustment judgment weight matrix before and after each iteration update of the classifier; performing matrix multiplication of the first cleaning device pressure adjustment judgment weight matrix and the second cleaning device pressure adjustment judgment weight matrix with the cleaning device pressure adjustment judgment feature vector to obtain a first cleaning device pressure adjustment judgment classification modulation feature vector and a second cleaning device pressure adjustment judgment classification modulation feature vector; calculating the position difference between the first cleaning device pressure adjustment judgment classification modulation feature vector and the second cleaning device pressure adjustment judgment classification modulation feature vector to obtain a cleaning device pressure adjustment judgment shift information representation vector; calculating the F-norm of the position mean vector between the first cleaning device pressure adjustment judgment classification modulation feature vector and the second cleaning device pressure adjustment judgment classification modulation feature vector as a cleaning device pressure adjustment judgment shift compensation scaling factor; performing linear scaling on the cleaning device pressure adjustment judgment shift information representation vector by using the cleaning device pressure adjustment judgment shift compensation scaling factor to obtain a scaled cleaning device pressure adjustment judgment shift information representation vector, and inputting the scaled cleaning device pressure adjustment judgment shift information representation vector into a Sigmoid activation function to obtain a cleaning device pressure adjustment judgment back propagation shift compensation representation vector; calculating the position point multiplication of the cleaning device pressure adjustment judgment back propagation shift compensation representation vector and the cleaning device pressure adjustment judgment feature vector to obtain the optimized cleaning device pressure adjustment judgment feature vector.

9. A data analysis-based rock wool board processing management method, characterized by, It comprises: obtaining rock wool board clean surface images collected by a camera and brush pressure values of a rock wool board surface cleaning device at multiple predetermined time points; extracting rock wool board clean surface feature vectors and cleaning device brush pressure feature vectors from the rock wool board clean surface images collected by the camera and the brush pressure values of the rock wool board surface cleaning device at multiple predetermined time points; based on the rock wool board clean surface feature vectors and the cleaning device brush pressure feature vectors, determining whether the rock wool board surface cleaning device brush pressure value needs to be adjusted.

10. The data analysis-based rock wool board processing management method of claim 9, wherein, extracting rock wool board clean surface feature vectors and cleaning device brush pressure feature vectors from the rock wool board clean surface images collected by the camera and the brush pressure values of the rock wool board surface cleaning device at multiple predetermined time points, comprising: performing feature extraction on the rock wool board clean surface images collected by the camera to obtain the rock wool board clean surface feature vectors; performing feature extraction on the brush pressure values of the rock wool board surface cleaning device at multiple predetermined time points to obtain the cleaning device brush pressure feature vectors.

Images