Intelligent control system and method for building material production process

Abstract

The invention relates to the technical field of intelligent control, and more particularly discloses an intelligent control system and method for a building material production process, which are combined with a speed sensor and a video acquisition device to realize the capabilities of real-time monitoring and automatic stirring speed adjustment in the light building material production process. The feature vectors in the stirring process are extracted, the stirring speed data are fused, and adjustment suggestions are obtained through classifier processing, so that the stirring uniformity is ensured, the production efficiency is improved, and the manual operation cost is reduced.

Description

Technical Field

[0001] The present application relates to the field of intelligent control technology, and more specifically, to an intelligent control system and method for a building material production process. Background Art

[0002] Lightweight material is a new type of composite material. It uses alkali-resistant glass fiber as reinforcement, sulfoaluminate low-alkalinity cement as binder and suitable aggregate as base material. It is a new type of inorganic composite material made through spraying, vertical mold casting, extrusion, slurry flow and other processes to replace gravel, sand, etc., which can greatly reduce weight.

[0003] Lightweight building materials need to be mixed evenly when used. If they are not mixed evenly, raw powder clumps and particles will appear, resulting in poor material quality. Existing mixing agitators have low mixing efficiency and require manual hand-held operation, which is time-consuming and labor-intensive to use.

[0004] Therefore, an intelligent control system and method for a building material production process is desired. Summary of the Invention

[0005] The present application is proposed to solve the above technical problems. The embodiments of the present application provide an intelligent control system and method for a building material production process, which utilizes a speed sensor and a video monitoring device to achieve intelligent monitoring and automatic adjustment of the mixing speed during the production of lightweight building materials, thereby improving production efficiency and quality.

[0006] Accordingly, according to one aspect of the present application, there is provided an intelligent control system for a building material production process, comprising:

[0007] The building material production data acquisition module is used to collect stirring speed values at multiple predetermined time points through a speed sensor and to collect monitoring videos of the stirring process using a video acquisition device;

[0008] a building material production data processing module, configured to extract a mixing process variation feature vector from the monitoring video of the mixing process, and extract a mixing speed feature vector from the mixing speed values at the plurality of predetermined time points;

[0009] A building material production data fusion module is used to fuse the mixing process change feature vector and the mixing speed feature vector to obtain an optimized mixing process correlation feature vector;

[0010] The building material production data analysis module is used to pass the optimized mixing process associated feature vector through a classifier to obtain a classification result, and the result is used to indicate whether the mixing speed at the current time point should be increased or decreased.

[0011] According to another aspect of the present application, there is also provided an intelligent control method for a building material production process, which includes:

[0012] The stirring speed values at a plurality of predetermined time points are collected by a speed sensor and the monitoring video of the stirring process is collected by a video acquisition device;

[0013] Extracting a stirring process change feature vector from the monitoring video of the stirring process, and extracting a stirring speed feature vector from the stirring speed values at the plurality of predetermined time points;

[0014] fusing the stirring process change feature vector and the stirring speed feature vector to obtain an optimized stirring process correlation feature vector;

[0015] The optimized stirring process associated feature vector is passed through a classifier to obtain a classification result, and the result is used to indicate whether the stirring speed at the current time point should be increased or decreased.

[0016] Compared to existing technologies, this application provides an intelligent control system and method for building material production processes. Combining a speed sensor and video acquisition device, this system enables real-time monitoring of the lightweight building material production process and automatic adjustment of mixing speed. By extracting feature vectors from the mixing process, integrating mixing speed data, and processing it with a classifier, it generates adjustment recommendations, thereby ensuring mixing uniformity, improving production efficiency, and reducing manual operation costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The above and other purposes, features, and advantages of the present application will become more apparent through a more detailed description of the embodiments of the present application in conjunction with the accompanying drawings. The accompanying drawings are intended to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0018] Figure 1 Schematic diagram of a block diagram of an intelligent control system for a building material production process according to an embodiment of the present application.

[0019] Figure 2 This is a block diagram of a building material production data processing module in an intelligent control system for a building material production process according to an embodiment of the present application.

[0020] Figure 3 Schematic diagram of a block diagram of a monitoring video extraction unit in an intelligent control system for a building material production process according to an embodiment of the present application.

[0021] Figure 4 Flowchart of an intelligent control method for a building material production process according to an embodiment of the present application. DETAILED DESCRIPTION

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

[0023] The word “exemplary” is used exclusively herein to mean “serving as an example, example, or illustration.” Any embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments.

[0024] In addition, numerous specific details are provided in the following detailed description to better illustrate the present application. Those skilled in the art will appreciate that the present application can be practiced without certain specific details. In some instances, methods, means, components, and circuits well known to those skilled in the art are not described in detail in order to highlight the main purpose of the present 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 the technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. Throughout the description of this application, "plurality" means two or more, unless otherwise specifically defined.

[0026] Figure 1 The figure shows a block diagram of an intelligent control system for a building material production process according to an embodiment of the present application. Figure 1 As shown, the intelligent control system 100 for the building material production process according to the embodiment of the present application includes: a building material production data acquisition module 110, which is used to collect stirring speed values at multiple predetermined time points through a speed sensor and to collect monitoring videos of the stirring process using a video acquisition device; a building material production data processing module 120, which is used to extract stirring process change feature vectors from the monitoring videos of the stirring process, and to extract stirring speed feature vectors from the stirring speed values at the multiple predetermined time points; a building material production data fusion module 130, which is used to fuse the stirring process change feature vector and the stirring speed feature vector to obtain an optimized stirring process associated feature vector; a building material production data analysis module 140, which is used to pass the optimized stirring process associated feature vector through a classifier to obtain a classification result, and the result is used to indicate whether the stirring speed at the current time point should be increased or decreased.

[0027] In an embodiment of the present application, the building material production data acquisition module 110 is configured to collect stirring speed values at multiple predetermined time points using a speed sensor and to capture monitoring videos of the stirring process using a video acquisition device. It should be understood that the collection of stirring speed values provides real-time stirring status information, as stirring speed is one of the key parameters that directly influences the stirring process. Real-time stirring speed data collected by the speed sensor can provide information on the current operating status of the stirrer, speed variations, and any abnormalities that may occur during the stirring process. This data can help the monitoring system promptly identify problems and implement appropriate control measures to ensure the stability and quality of the stirring process. Monitoring videos provide more intuitive and comprehensive information about the stirring process. Through video monitoring, the operating status of the stirrer, the mixing of the stirred materials, and any problems such as clumping or uneven mixing can be observed. Video data provides more detailed information, helping analysts fully understand various aspects of the stirring process, including visual features and motion changes. The combined use of stirring speed values and monitoring video data allows for a more comprehensive and accurate analysis and understanding of the stirring process. By combining these two data sources, multi-dimensional monitoring and analysis of the stirring process can be achieved, allowing for more precise adjustment of the stirring speed to optimize the stirring effect. Through technical means such as machine learning models, these data sources can be used to automatically adjust and optimize the stirring speed, thereby improving production efficiency and product quality.

[0028] In an embodiment of the present application, the building material production data processing module 120 is configured to extract a mixing process variation feature vector from a surveillance video of the mixing process, and to extract a mixing speed feature vector from the mixing speed values at the plurality of predetermined time points. It should be understood that surveillance videos typically contain a large amount of information, and extracting feature vectors can convert video information into a more compact representation, helping the system better understand and analyze changes in the mixing process. By extracting feature vectors, machine learning algorithms can be applied for pattern recognition and anomaly detection, thereby promptly identifying anomalies and problems in the mixing process. Feature vector extraction can provide a foundation for subsequent automated analysis and decision-making, helping the system achieve more intelligent mixing process control. By extracting the mixing speed feature vector, the operating status of the agitator at different time points can be analyzed, including speed variation trends and fluctuations, helping to understand the dynamic characteristics of the mixing process. Mixing speed is a critical parameter that affects mixing performance and product quality. Extracting the speed feature vector can help the system achieve precise control and optimization of the mixing speed, thereby improving product quality and production efficiency. Correlating the mixing speed feature vector with other process parameters can reveal the relationship between different parameters in the mixing process, helping the system to more comprehensively understand the entire mixing process.

[0029] Specifically, in one embodiment of the present application, Figure 2FIG2 is a block diagram of a building material production data processing module in an intelligent control system for a building material production process according to an embodiment of the present application. Figure 2 As shown, in the intelligent control system 100 of the above-mentioned building material production process, the building material production data processing module 120 includes: a monitoring video extraction unit 121, which is used to extract key frames from the monitoring video of the mixing process and then obtain the mixing process change feature vector through convolution encoding; a mixing speed extraction unit 122, which is used to obtain the mixing speed feature vector by passing the mixing speed values at the multiple predetermined time points through a temporal encoder including a one-dimensional convolution layer.

[0030] Accordingly, in a specific example of the present application, the monitoring video extraction unit 121 is used to extract key frames from the monitoring video of the stirring process and then perform convolution coding to obtain the stirring process change feature vector. It should be understood that extracting key frames from the video can help reduce the amount of data, extract the most representative and information-rich frames in the video, avoid the problem of processing redundant information, and retain key timing information. Convolutional neural networks (CNNs) perform well in the field of image processing and can effectively learn image features. By inputting key frames into CNN for convolution coding, high-level features in the image can be extracted, including texture, shape, structure and other information, so as to better represent the changes in the stirring process. The feature vector obtained by convolution coding can better capture the spatial information and changes in the video, and convert complex image information into a compact, high-dimensional feature representation, which facilitates subsequent data analysis and processing. The feature vectors extracted based on CNN can be applied to tasks such as pattern recognition, classification and anomaly detection. These feature vectors can be used as input to help the system better understand the characteristics and laws of the stirring process and realize automated analysis and decision-making. After converting video data into feature vectors, it can be combined with other data sources (such as stirring speed data) for comprehensive analysis, enabling multi-dimensional monitoring and control of the stirring process. This allows for more accurate regulation of the stirring process and optimization of production efficiency and product quality. Therefore, by extracting key frames from the surveillance video and generating feature vectors through convolutional coding, key features in the video information can be effectively extracted, providing a richer and more effective data representation for the analysis and control of the stirring process, thereby achieving more accurate stirring process monitoring and optimization.

[0031] further, Figure 3 FIG2 is a block diagram of a monitoring video extraction unit in an intelligent control system for a building material production process according to an embodiment of the present application. Figure 3As shown, in the building material production data processing module 120 of the intelligent control system 100 of the above-mentioned building material production process, the monitoring video extraction unit 121 includes: a key frame extraction subunit 1211, which is used to extract multiple mixing process monitoring key frames from the monitoring video of the mixing process; a deep and shallow fusion subunit 1212, which is used to pass the multiple mixing process monitoring key frames respectively through a first convolutional neural network model including a deep and shallow fusion module to obtain multiple mixing process monitoring feature matrices; a three-dimensional convolution subunit 1213, which is used to aggregate the multiple mixing process monitoring feature matrices into a three-dimensional mixing process feature tensor along the time dimension and then obtain the mixing process change feature vector through a second convolutional neural network model using a three-dimensional convolution kernel.

[0032] Specifically, the keyframe extraction subunit 1211 is configured to extract multiple mixing process monitoring keyframes from the monitoring video of the mixing process. It should be understood that extracting multiple keyframes can more comprehensively represent the different stages and changes of the entire mixing process. The information captured by each keyframe may vary, providing a more comprehensive perspective for understanding the characteristics and evolution of the mixing process. Different keyframes can capture the state and changes at different time points during the mixing process. These keyframes can help analysts or systems better understand the dynamic characteristics of the mixing process, including speed changes, material mixing level, and so on. By comparing keyframes at different time points, anomalies or problems in the mixing process can be more easily detected. For example, sudden unusual behavior or state changes may be more easily identified in keyframes. Extracting multiple keyframes provides richer data information, helping to build a more accurate and comprehensive mixing process model. These keyframes can be used as training data for machine learning models, improving the model's accuracy and generalization capabilities. Compared to using the entire video data, extracting keyframes can effectively compress the data volume, reducing storage and processing costs. This allows for more efficient subsequent data analysis and processing.

[0033] Specifically, the deep-shallow fusion subunit 1212 is configured to separately pass the multiple stirring process monitoring keyframes through a first convolutional neural network model including a deep-shallow fusion module to obtain multiple stirring process monitoring feature matrices. It should be understood that the first convolutional neural network model has excellent performance in image processing and can effectively learn image features. By inputting each keyframe separately into the first convolutional neural network model including a deep-shallow fusion module, high-level features from each keyframe, including texture, shape, structure, and other information, can be extracted, thereby better representing the stirring process monitoring information. The deep-shallow fusion module can help integrate feature information from different levels, combining the advantages of shallow and deep features. This improves the representational and generalization capabilities of features, making the extracted features richer and more meaningful. By processing each keyframe separately, the unique information contained in each keyframe is preserved. This allows for a more comprehensive capture of the diversity and variability of the stirring process, resulting in a richer and more accurate feature representation. By separately inputting multiple keyframes into the first convolutional neural network model including a deep-shallow fusion module for processing, data can be processed in parallel, improving processing efficiency and speed. This allows for faster generation of multiple mixing process monitoring feature matrices to support real-time monitoring and analysis. By processing multiple keyframes and generating multiple mixing process monitoring feature matrices, the model's generalization and adaptability can be improved. These feature matrices can be used to train the model, leading to a better understanding of the characteristics and patterns of the mixing process.

[0034] Correspondingly, the deep-shallow fusion subunit includes: a shallow feature extraction secondary subunit for extracting a shallow feature map from the i-th layer of the first convolutional neural network model, where j is greater than or equal to 1 and less than or equal to 6; a deep feature extraction secondary subunit for extracting a deep feature map from the j-th layer of the first convolutional neural network model, where the ratio between the j-th layer and the i-th layer is greater than or equal to 5 and less than or equal to 10; a feature map fusion secondary subunit for using the deep-shallow feature fusion module of the first convolutional neural network model to fuse the shallow feature map and the deep feature map to obtain a fused feature map; a feature map dimensionality reduction secondary subunit for performing global pooling on the fused feature map along the channel dimension to obtain the multiple stirring process monitoring feature matrices.

[0035] Specifically, the three-dimensional convolution subunit 1213 is configured to aggregate the multiple stirring process monitoring feature matrices along the time dimension into a three-dimensional stirring process feature tensor, and then use a second convolutional neural network model with a three-dimensional convolution kernel to obtain the stirring process change feature vector. It should be understood that by aggregating multiple monitoring feature matrices along the time dimension, monitoring information at different time points can be integrated to capture changes in the stirring process over time. This allows for a more comprehensive understanding of the dynamic characteristics and evolution of the stirring process. The second convolutional neural network model using a three-dimensional convolution kernel can extract features in the spatiotemporal domain, better capturing the temporal and spatial information of the stirring process. This results in a more representative and rich representation of the stirring process features. The three-dimensional convolution operation helps capture local correlations and temporal patterns between different time points. This allows for better identification of important time periods and change patterns in the stirring process, improving the effectiveness and expressiveness of feature extraction. The parameter sharing mechanism in the second convolutional neural network model can effectively reduce the number of model parameters and improve the model's generalization capability. By sharing the convolution kernel and operating on the entire stirring process feature tensor, data information can be more effectively utilized and the risk of overfitting can be reduced. The three-dimensional convolution operation can process spatiotemporal information in parallel, improving computational efficiency and processing speed. This allows for faster acquisition of the stirring process change feature vector, supporting real-time monitoring and analysis requirements. Therefore, aggregating multiple stirring process monitoring feature matrices along the time dimension into a three-dimensional stirring process feature tensor and using a second convolutional neural network model with a three-dimensional convolution kernel to obtain the stirring process change feature vector helps to capture the temporal variation characteristics of the stirring process more comprehensively and effectively, providing a richer and more accurate data basis for further analysis, monitoring, and optimization. This method can better understand the dynamic characteristics of the stirring process and provide strong support for the control and optimization of the stirring process.

[0036] Correspondingly, the three-dimensional convolution subunit includes: a stirring process encoding secondary subunit, which is used to use the second convolutional neural network model to perform three-dimensional convolution encoding on the three-dimensional stirring process feature tensor to obtain a stirring process change feature map; and a stirring process dimensionality reduction secondary subunit, which is used to perform global mean pooling on each feature matrix along the channel dimension of the stirring process change feature map to obtain the stirring process change feature vector.

[0037] Furthermore, the stirring process encoding secondary subunit is used to perform three-dimensional convolution encoding on the three-dimensional stirring process feature tensor using the second convolutional neural network model to obtain a stirring process change feature map. It should be understood that the three-dimensional convolution operation can extract features in both time and space dimensions, better capturing the temporal and spatial correlations during the stirring process. This can effectively characterize the changing patterns and dynamic characteristics of the stirring process. Through three-dimensional convolutional encoding, the second convolutional neural network model can effectively model the temporal information of the stirring process and capture the correlations and changes between different time points. This helps to more comprehensively understand the dynamic characteristics and evolution process of the stirring process. Three-dimensional convolutional encoding can map the three-dimensional stirring process feature tensor into a higher-level feature representation, extracting important features and patterns in the stirring process. This can obtain a more representative and meaningful stirring process change feature map. The second convolutional neural network model can take into account the information of surrounding pixels when performing the convolution operation, thereby better understanding the relationship and contextual information between different regions in the stirring process. This helps to improve the expressiveness and accuracy of the features. The parameter sharing mechanism in the second convolutional neural network model can reduce the number of model parameters and improve the generalization ability of the model. By sharing convolution kernels and operating on the three-dimensional stirring process feature tensor, data information can be more effectively utilized and the risk of overfitting can be reduced. Therefore, using a second convolutional neural network model to perform three-dimensional convolution encoding on the three-dimensional stirring process feature tensor to obtain a stirring process variation feature map can more comprehensively and effectively capture the spatiotemporal variation characteristics of the stirring process, providing a richer and more accurate data foundation for further analysis, monitoring, and optimization. This approach helps to better understand the dynamic characteristics of the stirring process and provides strong support for its control and optimization.

[0038] Furthermore, the stirring process dimensionality reduction secondary subunit is configured to perform global mean pooling on each feature matrix along the channel dimension of the stirring process change feature map to obtain the stirring process change feature vector. It should be understood that global mean pooling can integrate the spatial information in each feature matrix into a single value, thereby reducing the data dimensionality. This helps reduce the length of the feature vector, simplify the model structure, and reduce computational complexity. Mean pooling can preserve important information in the feature map because it considers the average of the entire feature map rather than local information. This helps capture global features and important patterns in the stirring process change feature map. Global mean pooling helps reduce the risk of model overfitting by reducing the number of parameters and model complexity. This helps improve the model's generalization ability and enable it to better adapt to new data. The mean pooling operation is translation invariant to a certain extent, meaning it is robust to translations of the input feature map. This helps the model better handle positional changes and translations in the stirring process change feature map. Global mean pooling is a simple and effective operation that can accelerate the computational process and reduce computational cost. This helps improve the efficiency of model training and inference, especially when processing large-scale data.

[0039] Accordingly, in a specific example of the present application, the stirring speed extraction unit 122 is used to pass the stirring speed values at the multiple predetermined time points through a temporal encoder including a one-dimensional convolutional layer to obtain the stirring speed feature vector. It should be understood that the one-dimensional convolutional layer can effectively extract features from time series data and capture patterns and associations between different time points. Through the temporal encoder, the temporal characteristics in the stirring speed data can be better understood, thereby extracting meaningful feature vectors. The one-dimensional convolutional layer can learn local features, that is, identify important patterns and features within a local range. This helps to capture local changes and regularities in the stirring speed data and improve the expressiveness of features. Compared with the fully connected layer, the one-dimensional convolutional layer has the advantage of parameter sharing, which can reduce the amount of model parameters and improve the generalization ability of the model. This helps to avoid overfitting and improve the model's adaptability to new data. The one-dimensional convolutional layer has translation invariance to a certain extent and can identify patterns between different time points without being affected by time offset. This helps the model better understand the time information in the stirring speed data. The stirring speed feature vector extracted by the time series encoder can better represent the important features and patterns of stirring speed data, providing more effective input for subsequent analysis, prediction or decision-making.

[0040] Specifically, the stirring speed extraction unit includes: a discrete data structuring subunit, configured to arrange the stirring speed values at the plurality of predetermined time points into a stirring speed input vector according to the time dimension; and a fully connected encoding subunit, configured to use the fully connected layer of the temporal encoder to perform fully connected encoding on the stirring speed input vector using the following formula to extract high-dimensional implicit features of the rotation speed value at each position in the stirring speed input vector, wherein the formula is: Where X is the stirring speed input vector, Y is the output vector, W is the weight matrix, and B is the bias vector. represents matrix multiplication; a one-dimensional convolutional encoding subunit, configured to use the one-dimensional convolutional layer of the temporal encoder to perform one-dimensional convolutional encoding on the stirring speed input vector using the following formula to extract high-dimensional implicit correlation features of the correlation between the rotational speed values at each position in the stirring speed input vector; wherein the formula is:

[0041]

[0042] Wherein, a is the width of the convolution kernel in the x direction, F is the convolution kernel parameter vector, G is the local vector matrix operated with the convolution kernel function, w is the size of the convolution kernel, X represents the stirring speed input vector, and Cov(X) represents the one-dimensional convolution encoding of the stirring speed input vector.

[0043] In an embodiment of the present application, the building material production data fusion module 130 is configured to fuse the mixing process variation feature vector and the mixing speed feature vector to obtain an optimized mixing process correlation feature vector. It should be understood that the characteristics of the mixing process include not only the spatial variation characteristics of the mixing process, but also the time series characteristics of the mixing speed. By fusing these two different types of feature vectors, spatial and temporal information can be comprehensively considered to more comprehensively describe the characteristics of the mixing process. Fusion of feature vectors from different sources can improve the expressive power of features, making the final feature vector more representative and discriminative. This helps improve the model's understanding and modeling capabilities of the mixing process. Fusion of multiple features can provide more information, help reduce the risk of model overfitting, and improve the model's generalization ability. Fusion of features can help the model better adapt to different mixing process data. Fusion of different types of features can provide more comprehensive information, helping to improve the model's predictive performance. Combining the mixing process variation characteristics and the mixing speed characteristics can more accurately predict the development and outcome of the mixing process. The characteristics of the mixing process are influenced by multiple factors, including spatial structure, time series, and speed. Fusion of different feature vectors can comprehensively consider these factors to better understand and describe the correlation characteristics of the mixing process.

[0044] Specifically, in one embodiment of the present application, the building material production data fusion module 130 includes: a building material production data fusion unit, used to perform weighted fusion of the mixing process change feature vector and the mixing speed feature vector to obtain a mixing process associated feature vector; a building material production data modulation unit, used to perform eigenspace regression constrained kernel projection modulation on the mixing process associated feature vector to obtain an optimized mixing process associated feature vector.

[0045] While advanced deep learning models are employed to convert data from diverse sources (such as video surveillance and speed sensor data) into feature representations, these models may focus more on extracting the most discriminative features from large amounts of data, rather than fully capturing the subtle structural information within these features. For example, while the first convolutional neural network model effectively extracts visual features from keyframes and combines information from different levels through a deep and shallow fusion module, it may not fully resolve all complex patterns associated with stirring state changes, particularly those involving dynamic changes over long time spans or requiring comprehensive understanding across multiple feature dimensions. Furthermore, during feature vector fusion, simply combining the stirring process change feature vector from the video data with the stirring speed feature vector from the speed sensor data can result in the loss of important structural information implicit in the original data. This is because direct feature concatenation or simple weighted summation operations fail to fully consider the intrinsic connections and interaction mechanisms between the two types of features. Such processing can overlook details crucial for accurately determining stirring speed adjustments, thereby compromising the accuracy and reliability of the final decision. Therefore, the present application further performs eigenspace regression-constrained kernel projection modulation on the stirring process associated feature vector to obtain an optimized stirring process associated feature vector.

[0046] More specifically, the building material production data modulation unit is used to: first, construct a pixel topology correlation matrix of the mixing process correlation feature vector, which is expressed as follows:

[0047]

[0048] Wherein, V represents the associated characteristic vector of the stirring process, v i and v j Respectively represent the eigenvalues of the i-th and j-th positions of the associated eigenvector of the stirring process, d(v i ,v j ) indicates calculating the Euclidean distance, D i,j Represents the eigenvalue of the (i, j) position of the pixel topological correlation matrix.

[0049] That is, by constructing a pixel topological correlation matrix, the spatial-temporal coupling strength between visual texture changes and speed fluctuations during the stirring process is explicitly quantified. By strengthening the weight distribution of high-correlation units (such as particle aggregation areas and speed drop nodes) in the pixel topological correlation matrix, redundant noise signals are suppressed, so that the subsequent classifier can identify the critical triggering conditions for stirring speed adjustment based on the internal dependency structure of the features, thereby improving the control system's prediction accuracy and real-time response to the trend of stirring uniformity degradation.

[0050] Secondly, the kernel space feature of the pixel topology correlation matrix is mined through a deep convolution layer to obtain the dynamic response matrix of the kernel space associated with the stirring process, which is expressed as follows:

[0051] M=Conv(D)

[0052] Wherein, D represents the pixel topological association matrix, Conv represents the convolutional layer, and M represents the stirring process association kernel space dynamic response matrix.

[0053] Specifically, the convolution kernel is used to learn the abstract dynamic laws of feature associations within the kernel space. The local association topology (e.g., the progressive response relationship between adjacent feature nodes) and global long-range dependencies (e.g., the lag effect between the vortex at the edge of the agitator and the velocity value at the center) in the pixel topology association matrix are uniformly encoded into a task-adaptive dynamic response pattern through nonlinear convolution operations. This generates a pixel topology association matrix, and by mining the causal transmission laws implicit in the feature associations (e.g., the temporal correlation between the change in the particle distribution gradient in the video and the abnormal velocity sensor reading), a dynamic response function is established between the multi-dimensional data of the mixing process, achieving closed-loop control of nonlinear dynamic anomalies in the lightweight material mixing process.

[0054] Then, the pixel topology correlation matrix is expanded to obtain a set of coding vectors of the associated intrinsic components of the stirring process, which can be expressed as follows:

[0055]

[0056] Where T represents the transpose of the vector, Λ represents the diagonal matrix, λ1 and λ m denote the first and mth eigenvalues of the diagonal matrix respectively, U denotes the set of coding vectors of the associated eigencomponents of the stirring process, x1, x2, x m They represent the first, second and mth stirring process associated eigencomponent encoding vectors respectively.

[0057] Specifically, based on the principle of orthogonal basis vector generation from spectral decomposition, the high-dimensional pixel topological correlation matrix is mapped into a low-dimensional orthogonal eigenmode space, thereby explicitly separating the physical coupling between material distribution and mechanical motion during the mixing process. The resulting set of encoding vectors for the intrinsic components of the mixing process can, through the linear combination reconstruction capability of the orthogonal basis vectors, compress the multidimensional correlation information of the mixing process into interpretable physical patterns, thereby blocking the evolution path of inhomogeneity in the early stages of lightweight material mixing.

[0058] Next, each stirring process-associated intrinsic component coding vector in the set of stirring process-associated characteristic intrinsic component coding vectors is input into the attention-focused feature weight modulation network to obtain a set of enhanced stirring process-associated intrinsic component coding vectors, which is expressed as follows:

[0059] Y=Transformer{[x1,x2,…,x m ]}=[y1,y2,…,y m ]

[0060] Among them, Transformer represents a sequence model based on the self-attention mechanism, Y represents the set of encoding vectors of the associated intrinsic components of the reinforcement mixing process, y1, y2, y m They represent the encoding vectors of the first, second and mth enhanced stirring process associated eigencomponents respectively.

[0061] That is, the cross-modal contextual relationship between the eigencomponent encoding vectors of the associated features of the stirring process is mined through the self-attention mechanism, and the task-relevance weights of the eigencomponent encoding vectors of the associated features of each stirring process under the real-time stirring state are dynamically calculated, thereby converting the abstract mathematical orthogonal basis into dynamic control knowledge with physical interpretability, so that the subsequent classifier can accurately capture the early signs of phase change in the stirring state based on the attention-weighted eigenspace, avoiding material waste caused by delayed manual intervention.

[0062] Then, each enhanced stirring process associated intrinsic component coding vector in the set of the enhanced stirring process associated intrinsic component coding vectors is dot-product-projected with the stirring process associated kernel space dynamic response matrix to obtain a set of stirring process associated intrinsic component kernel constraint coding vectors, which is expressed as follows:

[0063]

[0064] in, represents matrix multiplication, S represents the characteristic scale of the dynamic response matrix of the stirring process correlation kernel space, y i represents the encoding vector of the intrinsic component associated with the i-th enhanced stirring process, L represents the length of the encoding vector of the intrinsic component associated with the enhanced stirring process, z irepresents the kernel constraint encoding vector of the associated eigencomponent of the i-th stirring process.

[0065] That is, a cross-domain interaction field between the global structural pattern and the local dynamic response is constructed through projection operation, so that the orthogonal basis vectors of the coding vector of the associated eigencomponent of the enhanced stirring process are nonlinearly modulated by the task-related correlation path in the dynamic response matrix of the associated kernel space of the stirring process, thereby integrating the physical conservation law of lightweight material stirring and the data-driven dynamic anomaly propagation model into a unified feature representation, thereby generating the kernel constraint coding vector of the associated eigencomponent of the stirring process.

[0066] Finally, the set of kernel constraint encoding vectors associated with the stirring process is aggregated to obtain the optimized stirring process associated eigenvalue vector, which is expressed as follows:

[0067] V'=Concat{z1,z2,…,z m}

[0068] Among them, Concat represents the cascade function, z1, z2, z m They represent the first, second and mth stirring process associated eigencomponent kernel constraint encoding vectors respectively, and V' represents the optimized stirring process associated eigenvector.

[0069] Specifically, through the mathematical operation of feature cascade fusion, the physical conservation patterns in the orthogonal eigenspace are cross-domain coupled with the anomaly propagation patterns in the dynamic response of the core space, constructing a multi-scale joint feature representation that reflects the macroscopic stability constraints of the mixing system while capturing microscopic dynamic anomalies. Effectively, the generated optimized mixing process-associated feature vectors, through a joint embedding space formed by multi-component fusion, encode the implicit phase transition critical conditions during the lightweight material mixing process as classifiable explicit feature boundaries. This enables subsequent classifiers to accurately identify the transition inflection point from steady-state to unsteady-state mixing based on the composite gradient changes of the fused features, thereby triggering multi-parameter coordinated speed control instructions during the latent period before the formation of raw dough.

[0070] In an embodiment of the present application, the building material production data analysis module 140 is configured to pass the optimized mixing process-associated feature vector through a classifier to obtain a classification result, which indicates whether the mixing speed at the current time point should be increased or decreased. It should be understood that mapping the mixing process-associated feature vector to the classification result through a classifier can provide decision support for determining whether the mixing speed at the current time point should be increased or decreased. Based on the classifier's results, automated adjustment of the mixing speed can be implemented, reducing manual intervention and improving production efficiency. Based on the classification results, the system can automatically adjust the mixing speed, making the mixing process more stable and efficient. The mixing process-associated feature vector is processed in real time to quickly determine whether the mixing speed should be increased or decreased. This real-time nature ensures timely and effective adjustment of the mixing process, avoiding problems or wasted resources. Using a classifier to determine the adjustment direction for the mixing speed at the current time point can help optimize the mixing process, improving product quality and production efficiency. Adjusting the mixing speed based on the classification results helps achieve better mixing and reaction effects. Adjusting the mixing speed based on the classification results of the mixing process-associated feature vector implements a data-driven decision-making process. This method allows for more objective and accurate decisions on mixing speed adjustment based on data characteristics. Therefore, the mixing process-associated feature vectors are passed through a classifier to obtain a classification result, which is then used to indicate whether the mixing speed at the current time point should be increased or decreased. This can provide decision support, automated adjustment, real-time performance, optimized mixing effects, and data-driven decision-making, helping to improve the efficiency and quality of the mixing process. This method can make the mixing process more intelligent and precise, providing better support and guarantee for industrial production.

[0071] Accordingly, in one embodiment of the present application, the building material production data analysis module 140 is configured to: use the classifier to process the mixing process associated feature vector using the following formula to obtain the classification result;

[0072] Wherein, the formula is: softmax{(W n ,B n ):…:(W1,B1)|X}, where W1 to W n is the weight matrix, B1 to B n is the bias vector, X is the associated feature vector of the stirring process, softmax represents the softmax function, and O represents the classification result.

[0073] In summary, the intelligent control system and method for the building material production process described in the embodiments of this application, combined with a speed sensor and video acquisition device, enables real-time monitoring of the lightweight building material production process and automatic adjustment of the mixing speed. By extracting the characteristic vector of the mixing process, integrating the mixing speed data, and processing it with a classifier, adjustment recommendations are generated, thereby ensuring mixing uniformity, improving production efficiency, and reducing manual operation costs.

[0074] As described above, the intelligent control system 100 for the building material production process according to the embodiments of the present application can be implemented in various terminal devices, such as a server of the intelligent control system for the building material production process. In one example, the intelligent control system 100 for the building material production process can be integrated into the terminal device as a software module and / or a hardware module. For example, the intelligent control system 100 for the building material production process can be a software module in the operating system of the terminal device, or can be an application developed for the terminal device; of course, the intelligent control system 100 for the building material production process can also be one of the many hardware modules of the terminal device.

[0075] Alternatively, in another example, the intelligent control system 100 of the building material production process and the terminal device may also be separate devices, and the intelligent control system 100 of the building material production process may be connected to the terminal device via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0076] Figure 4 Flowchart of the intelligent control method for the building material production process according to the embodiment of the present application. Figure 4 As shown, the intelligent control method for the building material production process according to the embodiment of the present application includes the following steps: S110, collecting stirring speed values at multiple predetermined time points through a speed sensor and collecting monitoring videos of the stirring process using a video acquisition device; S120, extracting a stirring process change feature vector from the monitoring video of the stirring process, and extracting a stirring speed feature vector from the stirring speed values at the multiple predetermined time points; S130, fusing the stirring process change feature vector and the stirring speed feature vector to obtain an optimized stirring process associated feature vector; S140, passing the optimized stirring process associated feature vector through a classifier to obtain a classification result, and the result is used to indicate whether the stirring speed at the current time point should be increased or decreased.

[0077] Here, those skilled in the art will appreciate that the specific operations of each step in the intelligent control method for the production process of building materials have been described in the above reference. Figures 1 to 3 The description of the intelligent control system for the building material production process has been introduced in detail, and therefore, its repeated description will be omitted.

[0078] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0079] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0080] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0081] The units described as separate components may or may not be physically separate, and the components shown as units 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 these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0082] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0083] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0084] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. An intelligent control system for a building material production process, characterized in that: include: The building material production data acquisition module is used to collect stirring speed values at multiple predetermined time points through a speed sensor and to collect monitoring videos of the stirring process using a video acquisition device; a building material production data processing module, configured to extract a mixing process variation feature vector from the monitoring video of the mixing process, and extract a mixing speed feature vector from the mixing speed values at the plurality of predetermined time points; A building material production data fusion module is used to fuse the mixing process change feature vector and the mixing speed feature vector to obtain an optimized mixing process correlation feature vector; The building material production data analysis module is used to pass the optimized mixing process associated feature vector through a classifier to obtain a classification result, and the result is used to indicate whether the mixing speed at the current time point should be increased or decreased.

2. The intelligent control system for the building material production process according to claim 1, characterized in that: The building material production data processing module includes: A monitoring video extraction unit, configured to extract key frames from the monitoring video of the stirring process and then perform convolution coding to obtain a characteristic vector of a change in the stirring process; The stirring speed extraction unit is configured to pass the stirring speed values at the plurality of predetermined time points through a temporal encoder comprising a one-dimensional convolutional layer to obtain the stirring speed feature vector.

3. The intelligent control system for the building material production process according to claim 2, characterized in that: The surveillance video extraction unit includes: a key frame extraction subunit, configured to extract a plurality of stirring process monitoring key frames from the monitoring video of the stirring process; a deep-shallow fusion subunit, configured to pass the plurality of stirring process monitoring key frames through a first convolutional neural network model including a deep-shallow fusion module to obtain a plurality of stirring process monitoring feature matrices; A three-dimensional convolution subunit is used to aggregate the multiple stirring process monitoring feature matrices into a three-dimensional stirring process feature tensor along the time dimension and then obtain the stirring process change feature vector by using a second convolutional neural network model with a three-dimensional convolution kernel.

4. The intelligent control system for the building material production process according to claim 3, characterized in that: The deep and shallow fusion subunit includes: a shallow feature extraction secondary subunit, configured to extract a shallow feature map from the i-th layer of the first convolutional neural network model, where j is greater than or equal to 1 and less than or equal to 6; a deep feature extraction secondary subunit, configured to extract a deep feature map from the jth layer of the first convolutional neural network model, wherein a ratio between the jth layer and the ith layer is greater than or equal to 5 and less than or equal to 10; A feature map fusion secondary subunit, configured to fuse the shallow feature map and the deep feature map using the deep and shallow feature fusion module of the first convolutional neural network model to obtain a fused feature map; The feature map dimensionality reduction secondary subunit is used to perform global pooling along the channel dimension on the fused feature map to obtain the multiple stirring process monitoring feature matrices.

5. The intelligent control system for the building material production process according to claim 4, characterized in that: The three-dimensional convolution subunit includes: a stirring process encoding secondary subunit, configured to perform three-dimensional convolution encoding on the three-dimensional stirring process feature tensor using the second convolutional neural network model to obtain a stirring process change feature map; The stirring process dimensionality reduction secondary subunit is used to perform global mean pooling on each feature matrix along the channel dimension of the stirring process change feature map to obtain the stirring process change feature vector.

6. The intelligent control system for the building material production process according to claim 5, characterized in that: The stirring speed extraction unit comprises: a discrete data structuring subunit, configured to arrange the stirring speed values at the plurality of predetermined time points into a stirring speed input vector according to a time dimension; A fully connected encoding subunit is configured to use the fully connected layer of the temporal encoder to perform fully connected encoding on the stirring speed input vector using the following formula to extract high-dimensional implicit features of the rotation speed value at each position in the stirring speed input vector, wherein the formula is: Where X is the stirring speed input vector, Y is the output vector, W is the weight matrix, and B is the bias vector. Represents matrix multiplication; A one-dimensional convolutional encoding subunit is configured to use the one-dimensional convolutional layer of the temporal encoder to perform one-dimensional convolutional encoding on the stirring speed input vector using the following formula to extract a high-dimensional implicit correlation feature of the correlation between the rotation speed values at each position in the stirring speed input vector; wherein the formula is: Wherein, a is the width of the convolution kernel in the x direction, F is the convolution kernel parameter vector, G is the local vector matrix operated with the convolution kernel function, w is the size of the convolution kernel, X represents the stirring speed input vector, and Cov(X) represents the one-dimensional convolution encoding of the stirring speed input vector.

7. The intelligent control system for the building material production process according to claim 6, characterized in that: The building material production data fusion module includes: A building material production data fusion unit is used to perform weighted fusion on the mixing process change feature vector and the mixing speed feature vector to obtain a mixing process correlation feature vector; The building material production data modulation unit is used to perform eigenspace regression constrained kernel projection modulation on the mixing process associated feature vector to obtain an optimized mixing process associated feature vector.

8. The intelligent control system for the building material production process according to claim 7, characterized in that: The building material production data modulation unit is used to: Constructing a pixel topological correlation matrix of the stirring process correlation feature vector; Performing kernel space feature mining on the pixel topology correlation matrix through a deep convolution layer to obtain a stirring process correlation kernel space dynamic response matrix; Expanding the pixel topology correlation matrix to obtain a set of coding vectors of associated intrinsic components of the stirring process; Inputting each stirring process associated intrinsic component encoding vector in the set of stirring process associated feature intrinsic component encoding vectors into an attention focused feature weight modulation network to obtain a set of enhanced stirring process associated intrinsic component encoding vectors; Performing dot product projection on each enhanced stirring process associated intrinsic component coding vector in the set of the enhanced stirring process associated intrinsic component coding vectors and the stirring process associated kernel space dynamic response matrix to obtain a set of stirring process associated intrinsic component kernel constraint coding vectors; The set of the stirring process associated eigencomponent kernel constraint encoding vectors is aggregated to obtain the optimized stirring process associated eigenvector.

9. The intelligent control system for the building material production process according to claim 8, characterized in that: The building material production data analysis module is used to: use the classifier to process the optimized mixing process associated feature vector using the following formula to obtain the classification result; Wherein, the formula is: softmax{(W n ,B n ):…:(W1,B1)|X}, where W1 to W n is the weight matrix, B1 to B n is the bias vector, X is the associated feature vector of the optimized stirring process, softmax represents the softmax function, and O represents the classification result.

10. An intelligent control method for a building material production process, characterized in that: include: The stirring speed values at a plurality of predetermined time points are collected by a speed sensor and the monitoring video of the stirring process is collected by a video acquisition device; Extracting a stirring process change feature vector from the monitoring video of the stirring process, and extracting a stirring speed feature vector from the stirring speed values at the plurality of predetermined time points; fusing the stirring process change feature vector and the stirring speed feature vector to obtain an optimized stirring process correlation feature vector; The optimized stirring process associated feature vector is passed through a classifier to obtain a classification result, and the result is used to indicate whether the stirring speed at the current time point should be increased or decreased.

Images