Intelligent control system and method for rock wool production line

Abstract

The invention relates to the field of intelligent production control, and provides an intelligent control system and method for a rock wool production line, which adopts a data analysis method based on artificial intelligence to analyze thermal infrared images of a cupola furnace at a plurality of time points so as to understand melting change characteristics of raw materials. And meanwhile, the feeding speed data of the feeding system at multiple time points are analyzed to obtain feeding speed time sequence characteristics, then the melting change characteristics of the raw materials and the feeding speed time sequence characteristics are fused and optimized, and finally the adjustment direction of the feeding speed is obtained through a classifier. Thus, by accurately controlling the feeding speed, it can be ensured that the raw materials reach the optimal state in the melting process, and therefore the quality and consistency of products are improved.

Description

Technical Field

[0001] The present application relates to the field of intelligent production control, and more specifically, to an intelligent control system and method for a rock wool production line. Background Art

[0002] Rock wool originated in Hawaii, and the rock wool production line simulates the natural process of Hawaiian volcanic eruption. Rock wool products are made of high-quality basalt and dolomite. After melting under high temperature, they are centrifuged into fibers by centrifuges at high speed. At the same time, a certain amount of binder, dust-proof oil and water-repellent are sprayed into them. Finally, they are collected by cotton collectors, cured, dried and cut to make rock wool products of different specifications and uses.

[0003] The process flow of the rock wool production line mainly includes the feeding system, cupola system, centrifugal fiber forming system, binder system, forming, pleating and pressurizing system, curing and drying system, longitudinal and transverse cutting system and stacking and packaging system. The feeding system accurately weighs the raw materials of rock wool production, basalt, slag and coke according to the formula ratio and feeds them into the feeding device, and then the feeding device feeds the mixed raw materials into the cupola; and the melting process of the mixed raw materials is realized in the cupola. When the feeding speed of the feeding system is too fast, it will cause the raw materials to accumulate in the cupola, affecting the melting effect and even causing failure; when the feeding speed is too slow, it will reduce production efficiency.

[0004] Therefore, an optimized control solution for rock wool production line is needed. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present application provides an intelligent control system and method for a rock wool production line.

[0006] According to one aspect of the present application, an intelligent control system for a rock wool production line is provided, comprising:

[0007] A rock wool production-related data acquisition module, used to obtain thermal infrared images of the cupola at multiple predetermined time points within a predetermined time period and the feeding speed values ​​of the feeding system at the multiple predetermined time points;

[0008] A rock wool production related data encoding module, used for performing data joint analysis on the thermal infrared images of the cupola at the plurality of predetermined time points and the feeding speed values ​​of the feeding system at the plurality of predetermined time points to obtain a raw material melting response change characteristic vector;

[0009] The feeding speed adjustment result generating module is used to obtain the adjustment direction of the feeding speed value of the feeding system at the current time point based on the raw material melting response change characteristic vector.

[0010] According to another aspect of the present application, there is provided an intelligent control method for a rock wool production line, comprising:

[0011] Acquire thermal infrared images of the cupola at a plurality of predetermined time points within a predetermined time period and feeding speed values ​​of the feeding system at the plurality of predetermined time points;

[0012] Performing data joint analysis on the thermal infrared images of the cupola at the plurality of predetermined time points and the feeding speed values ​​of the feeding system at the plurality of predetermined time points to obtain a characteristic vector of the raw material melting response change;

[0013] Based on the raw material melting response change characteristic vector, the adjustment direction of the feeding speed value of the feeding system at the current time point is obtained.

[0014] This application has significant technical effects due to the adoption of the above technical solutions:

[0015] The intelligent control system and method for rock wool production line provided by the present application adopts a data analysis method based on artificial intelligence, analyzes the thermal infrared images of the cupola at multiple time points to understand the melting change characteristics of the raw materials, and analyzes the feeding speed data of the feeding system at multiple time points to obtain the feeding speed timing characteristics, then fuses and optimizes the melting change characteristics of the raw materials and the feeding speed timing characteristics, and finally obtains the adjustment direction of the feeding speed through a classifier. In this way, by accurately controlling the feeding speed, it can be ensured that the raw materials reach the optimal state during the melting process, thereby improving the quality and consistency of the product. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used 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 accompanying drawings, the same reference numerals generally represent the same components or steps.

[0017] Figure 1 The figure is a flow chart of an intelligent control method for a rock wool production line according to an embodiment of the present application.

[0018] Figure 2 In the intelligent control method for a rock wool production line according to an embodiment of the present application, a flow chart is provided in which a data joint analysis is performed on the thermal infrared images of the blast furnace at the multiple predetermined time points and the feeding speed values ​​of the feeding system at the multiple predetermined time points to obtain a characteristic vector of the raw material melting response change.

[0019] Figure 3The present invention is a flow chart of performing image content feature extraction on the thermal infrared images of the cupola at the plurality of predetermined time points in the intelligent control method for a rock wool production line according to an embodiment of the present application to obtain a feature vector emphasizing the change in the melting state of the raw material.

[0020] Figure 4 The present invention is a flow chart of performing time series feature extraction on the feeding speed values ​​of the feeding system at the plurality of predetermined time points in the intelligent control method for a rock wool production line according to an embodiment of the present application to obtain a feeding speed time series feature vector.

[0021] Figure 5 The present invention is a flow chart of obtaining the adjustment direction of the feeding speed value of the feeding system at the current time point based on the characteristic vector of the raw material melting response change in the intelligent control method for the rock wool production line according to the embodiment of the present application.

[0022] Figure 6 The system block diagram of an intelligent control system for a rock wool production line according to an embodiment of the present application. DETAILED DESCRIPTION

[0023] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0024] Rock wool is made of high-quality basalt and dolomite. After melting at high temperature, it is centrifuged into fibers by a centrifuge, and then a binder, dust-proof oil and water-repellent agent are added. It is collected by a cotton collector, solidified, dried and cut into products of different specifications and uses. The process flow of the rock wool production line includes feeding, cupola, centrifugal fiberization, binder, forming, pleating and pressurizing, solidification and drying, longitudinal and transverse cutting and stacking packaging system. The feeding system accurately weighs raw materials such as basalt, slag and coke according to the formula ratio and feeds them into the feeding device, and then enters the cupola to melt the raw materials. If the feeding speed is too fast, the raw materials will accumulate in the cupola, affecting the melting effect or even failure, and if it is too slow, it will reduce production efficiency. Therefore, an intelligent control solution for rock wool production lines is expected.

[0025] In recent years, deep learning and neural networks have been widely used in computer vision, natural language processing, text signal processing and other fields. In addition, deep learning and neural networks have also shown a level close to or even beyond that of humans in image classification, object detection, semantic segmentation, text translation and other fields. The development of deep learning and neural networks provides new solutions and solutions for the intelligent control of rock wool production lines.

[0026] Figure 1 FIG. 1 is a flow chart of an intelligent control method for a rock wool production line according to an embodiment of the present application. Figure 1As shown, the intelligent control method for a rock wool production line according to an embodiment of the present application includes: S110, acquiring thermal infrared images of the cupola at multiple predetermined time points within a predetermined time period and the feeding speed values ​​of the feeding system at the multiple predetermined time points; S120, performing a joint data analysis on the thermal infrared images of the cupola at the multiple predetermined time points and the feeding speed values ​​of the feeding system at the multiple predetermined time points to obtain a raw material melting response change characteristic vector; S130, obtaining the adjustment direction of the feeding speed value of the feeding system at the current time point based on the raw material melting response change characteristic vector.

[0027] In step S110, the thermal infrared images of the cupola at multiple predetermined time points within a predetermined time period and the feeding speed values ​​of the feeding system at the multiple predetermined time points are obtained. It should be understood that the thermal infrared image of the cupola is captured by a thermal infrared camera, which mainly reflects the surface temperature distribution of objects in the furnace (mainly raw materials and flames). Specifically, different brightness in the thermal infrared image represents different temperature levels. Through the change of brightness, the temperature difference in different areas of the cupola can be understood, thereby understanding the melting state of the raw materials. The feeding speed value of the feeding system refers to the speed at which the feeding system delivers raw materials into the cupola. By analyzing the feeding speed values ​​at multiple time points, the changing trend of the feeding speed can be obtained. In general, the thermal infrared image provides real-time visual feedback of the melting state of the raw materials in the cupola, while the feeding speed value reflects the operating state of the feeding system. By combining the two, the melting effect of the raw materials can be monitored in real time, and the feeding speed can be dynamically adjusted according to the change of the melting state.

[0028] In step S120, data joint analysis is performed on the thermal infrared images of the cupola at the plurality of predetermined time points and the feeding speed values ​​of the feeding system at the plurality of predetermined time points to obtain a characteristic vector of the raw material melting response change. Specifically, Figure 2 The present invention is a flow chart of performing data joint analysis on the thermal infrared images of the cupola at the plurality of predetermined time points and the feeding speed values ​​of the feeding system at the plurality of predetermined time points in the intelligent control method for the rock wool production line according to the embodiment of the present application to obtain the characteristic vector of the raw material melting response change. Figure 2As shown, a data joint analysis is performed on the thermal infrared images of the cupola at the multiple predetermined time points and the feeding speed values ​​of the feeding system at the multiple predetermined time points to obtain a raw material melting response change characteristic vector, including: S121, image content feature extraction is performed on the thermal infrared images of the cupola at the multiple predetermined time points to obtain a raw material melting state change emphasis feature vector; S122, time series feature extraction is performed on the feeding speed values ​​of the feeding system at the multiple predetermined time points to obtain a feeding speed time series feature vector; S123, the feeding speed time series feature vector and the raw material melting state change emphasis feature vector are merged to obtain the raw material melting response change feature vector.

[0029] More specifically, Figure 3 The present invention is a flow chart of extracting image content features from the thermal infrared images of the cupola at the plurality of predetermined time points in the intelligent control method for the rock wool production line according to the embodiment of the present application to obtain a feature vector emphasizing the change in the melting state of the raw material. Figure 3 As shown, image content feature extraction is performed on the thermal infrared images of the blast furnace at the multiple predetermined time points to obtain a raw material melting state change emphasis feature vector, including: S1211, the thermal infrared images of the blast furnace at the multiple predetermined time points are respectively passed through a thermal infrared image encoder based on a spatial attention mechanism to obtain a plurality of raw material melting state feature matrices; S1212, the plurality of raw material melting state feature matrices are aggregated along the channel into a raw material melting state change feature map; S1213, the raw material melting state change feature map is passed through a melting state feature emphasizer based on a channel attention mechanism to obtain a raw material melting state change emphasis feature map; S1214, mean pooling processing is performed on each feature matrix along the channel dimension of the raw material melting state change emphasis feature map to obtain the raw material melting state change emphasis feature vector.

[0030] The thermal infrared image of the cupola contains a lot of temperature distribution information, but not all areas of information are equally important for judging the melting state of the raw materials. In order to automatically focus on the most important areas in the image, that is, those areas that best reflect the changes in the melting state of the raw materials, it is necessary to pass the thermal infrared images of the cupola at the multiple predetermined time points through a thermal infrared image encoder based on a spatial attention mechanism to obtain multiple raw material melting state feature matrices. The thermal infrared image encoder based on the spatial attention mechanism in this application is a convolutional neural network model using a spatial attention mechanism. The spatial attention mechanism can reduce the amount of computation without significantly reducing performance by focusing on key areas in the thermal infrared image and reducing attention to non-important areas.

[0031] In an embodiment of the present application, the thermal infrared images of the cupola at the multiple predetermined time points are respectively passed through a thermal infrared image encoder based on a spatial attention mechanism to obtain multiple raw material melting state feature matrices. One possible implementation method may be: use each layer of the thermal infrared image encoder to perform the following on the input data in the forward pass of the layer: convolve the input data based on a convolution kernel to obtain a convolution feature map; pass the convolution feature map through a spatial attention module to obtain a spatial attention score matrix; multiply the spatial attention score matrix and each feature matrix of the convolution feature map along the channel dimension by the spatial attention feature map at each position point; perform pooling processing on the spatial attention feature map along the channel dimension to obtain a pooling feature map; perform nonlinear activation on the pooling feature map to obtain an activation feature map; wherein the input of the first layer of the thermal infrared image encoder based on the spatial attention mechanism is the thermal infrared images of the cupola at the multiple predetermined time points, and the output of the last layer of the thermal infrared image encoder based on the spatial attention mechanism is the multiple raw material melting state feature matrices.

[0032] Each raw material melting state feature matrix is ​​extracted from the thermal infrared images at different time points, and each of them reflects the melting state of the raw material at that time point. In order to obtain a feature map containing time dynamic information, this feature map not only contains the static melting state information at each time point, but also implies the trend and pattern of the melting state changing over time. It is necessary to aggregate the multiple raw material melting state feature matrices along the channel into a raw material melting state change feature map. Aggregating multiple feature matrices into one feature map helps the model better capture the changing laws of the raw material melting state, so that the future melting state can be predicted more accurately to adjust the feeding speed.

[0033] In the characteristic diagram of the change in the melting state of the raw material, different channels may represent the melting state characteristics at different time points, but not all channel characteristics are equally important for the final judgment of the melting state of the raw material. In order to emphasize those channel characteristics that are more critical to the judgment of the melting state and suppress those relatively unimportant characteristics, it is necessary to pass the characteristic diagram of the change in the melting state of the raw material through a melting state feature emphasizer based on a channel attention mechanism to obtain a raw material melting state change emphasized feature diagram. The melting state feature emphasizer based on the channel attention mechanism in this application is a convolutional neural network model using a channel attention mechanism. It should be understood that through the channel attention mechanism, the model can dynamically adjust the contribution of different channel features, thereby more accurately capturing the changing trend of the raw material melting state.

[0034] In an embodiment of the present application, a possible implementation method of passing the raw material melting state change feature map through a melting state feature emphasizer based on a channel attention mechanism to obtain a raw material melting state change emphasized feature map can be: each layer of the melting state feature emphasizer performs the following on the input data in the forward transfer of the layer: convolution processing on the input data based on a two-dimensional convolution kernel to generate a convolution feature map; pooling processing on the convolution feature map to generate a pooling feature map; activation processing on the pooling feature map to generate an activation feature map; calculating the quotient of the eigenvalue mean of the feature matrix corresponding to each channel in the activation feature map and the sum of the eigenvalue mean of the feature matrix corresponding to all channels as the weighting coefficient of the feature matrix corresponding to each channel; weighting the feature matrix of each channel with the weighting coefficient of each channel in the activation feature map to generate a channel attention feature map; wherein the input of the first layer of the melting state feature emphasizer based on the channel attention mechanism is the raw material melting state change feature map, and the output of the last layer of the melting state feature emphasizer based on the channel attention mechanism is the raw material melting state change emphasized feature map.

[0035] In order to make the raw material melting state characteristics more concise and clear, and to better reflect the overall state change of the raw material melting, the technical solution of the present application needs to perform mean pooling processing on the characteristic map of the raw material melting state change emphasis along the channel dimension of each characteristic matrix to obtain the raw material melting state change emphasis feature vector. Mean pooling can effectively reduce the dimension of the raw material melting state change emphasis feature map by calculating the mean of each channel of the feature map, while retaining important information.

[0036] More specifically, Figure 4 The present invention is a flow chart of extracting the time series characteristics of the feeding speed values ​​of the feeding system at the plurality of predetermined time points in the intelligent control method for the rock wool production line according to the embodiment of the present application to obtain the time series characteristic vector of the feeding speed. Figure 4 As shown, the feeding speed values ​​of the feeding system at the multiple predetermined time points are subjected to timing feature extraction to obtain a feeding speed timing feature vector, including: S1221, arranging the feeding speed values ​​of the feeding system at the multiple predetermined time points according to the time dimension to obtain a feeding speed timing input vector; S1222, passing the feeding speed timing input vector through a feeding speed timing encoder to obtain the feeding speed timing feature vector.

[0037] The feeding speed of the rock wool production line changes with the changes of time, raw material status, equipment status and other factors. In order to clearly grasp the changing trend and dynamic characteristics of the feeding speed, it is necessary to arrange the feeding speed values ​​of the feeding system at the multiple predetermined time points according to the time dimension to obtain the feeding speed time series input vector. Compared with the data at a single time point, the time series data of the feeding speed contains more contextual information and historical dependencies. Using this information as input can help to more accurately understand the operating status and changing trend of the rock wool production line, thereby improving the accuracy and reliability of control decisions.

[0038] In order to extract useful feature information from the original feeding speed time series data, it is necessary to pass the feeding speed time series input vector through a feeding speed time series encoder to obtain a feeding speed time series feature vector. The feeding speed time series encoder in this application is a convolutional neural network model using a one-dimensional convolution kernel. An important feature of the feeding speed time series data is its internal time dependency, that is, the feeding speed data at the current time point may be affected by the feeding speed data at the previous time point. The feeding speed time series encoder can capture this time dependency through its internal structure, thereby more accurately understanding the inherent laws of the time series data.

[0039] In an embodiment of the present application, one possible implementation method of passing the feeding speed time series input vector through a feeding speed time series encoder to obtain a feeding speed time series feature vector may be: using a one-dimensional convolution layer of the feeding speed time series encoder to perform one-dimensional convolution encoding on the feeding speed time series input vector using the following one-dimensional convolution encoding formula to extract high-dimensional implicit correlation features between feature values ​​at each position in the input vector; wherein the one-dimensional convolution encoding formula is:

[0040]

[0041] Among them, a is the width of the convolution kernel in the x direction, F(a) is the convolution kernel parameter vector, G(xa) is the local vector matrix operated with the convolution kernel function, w is the size of the convolution kernel, X represents the feeding speed timing input vector, and Cov(X) represents the one-dimensional convolution encoding of the feeding speed timing input vector.

[0042] The time series characteristic vector of the feeding speed reflects the speed change and its internal law of the feeding system at different time points; the characteristic vector emphasizing the change of the raw material melting state captures the state change information of the raw material during the melting process; in order to comprehensively reflect the interaction and response relationship between the feeding speed change and the raw material melting state, and provide comprehensive information for more accurate control, it is necessary to fuse the feeding speed time series characteristic vector and the raw material melting state change emphasis characteristic vector to obtain the raw material melting response change characteristic vector. Through vector fusion, a control model for the response between the feeding speed and the raw material melting state can be constructed, comprehensively considering the mutual influence between multiple factors, thereby improving the control accuracy and response speed.

[0043] Figure 5 The present invention is a flow chart of obtaining the adjustment direction of the feeding speed value of the feeding system at the current time point based on the characteristic vector of the raw material melting response change in the intelligent control method for the rock wool production line according to the embodiment of the present application. Figure 5 As shown, based on the raw material melting response change characteristic vector, the adjustment direction of the feeding speed value of the feeding system at the current time point is obtained, including: S131, the raw material melting response change characteristic vector is subjected to entropy reduction adjustment based on the information uncertainty change measurement to obtain an optimized raw material melting response change characteristic vector; S132, the optimized raw material melting response change characteristic vector is passed through a feeding speed adjustment result generator based on a classifier to obtain a classification result, and the classification result is that the feeding speed value of the feeding system at the current time point should be increased, decreased or kept unchanged.

[0044] In particular, considering that when processing the thermal infrared image data of a blast furnace, images at multiple time points may show similar patterns in certain features, which will lead to the extraction of some repeated information. At the same time, when the interval between the collected data points is too small, there will be repeated features in the feeding speed time series features obtained using the feeding speed time series encoder for the feeding speed value data of the feeding system at multiple time points, which will lead to feature redundancy in the raw material melting response change feature vector obtained by fusion. When there is feature redundancy in the raw material melting response change feature vector, the classifier needs to process more feature information. When the classifier screens and judges among more features, this will cause the performance of the classifier to deteriorate, and it is difficult to accurately determine the feeding speed adjustment direction of the feeding system at the current time point. Therefore, in the technical solution of the present application, it is necessary to perform entropy reduction adjustment on the raw material melting response change feature vector based on the measurement of information uncertainty changes to obtain an optimized raw material melting response change feature vector.

[0045] Specifically, in an embodiment of the present application, the raw material melting response change characteristic vector is subjected to entropy reduction adjustment based on information uncertainty change measurement to obtain an optimized raw material melting response change characteristic vector, including: extracting each row vector of the feeding speed prediction parameter space to obtain a set of feeding speed prediction parameter row vectors; calculating the feeding speed class probability entropy degradation simulation value between the raw material melting response change characteristic vector and each feeding speed prediction parameter row vector in the set of feeding speed prediction parameter row vectors to obtain a set of feeding speed class probability entropy degradation simulation values; calculating the square root of the sum of squares of the feeding speed class probability entropy degradation simulation values ​​to obtain a feeding speed entropy degradation modulation value; based on the feeding speed entropy degradation modulation value, performing characteristic distribution modulation on the raw material melting response change characteristic vector to obtain the optimized raw material melting response change characteristic vector.

[0046] In an embodiment of the present application, another expression of performing entropy reduction adjustment based on information uncertainty change measurement on the raw material melting response change characteristic vector to obtain an optimized raw material melting response change characteristic vector can be: processing the raw material melting response change characteristic vector with the following formula to obtain an optimized raw material melting response change characteristic vector; wherein the formula is:

[0047] P={p1,p2,...,p j ,...,p n}

[0048] D j =Softmax(M m v c +H m p j +b m )

[0049]

[0050] Among them, P represents the set of row vectors of feeding speed prediction parameters, p1, p2, p j 、p n represents the first, second, jth and nth feeding speed prediction parameter row vectors in the set of feeding speed prediction parameter row vectors, v c represents the characteristic vector of the raw material melting response change, Softmax represents the normalized exponential function, M m represents the first weight matrix, H m represents the second weight matrix, b m represents the bias vector, D j represents the jth probability entropy degradation simulation value of the feeding speed class in the set of probability entropy degradation simulation values ​​of the feeding speed class, t represents the feeding speed entropy degradation modulation value, v c⊙t represents the characteristic vector v of the raw material melting response change c Each eigenvalue of the t-th power, represents the ith eigenvalue of the raw material melting response change eigenvector, α represents a predetermined hyperparameter, represents addition by position, and V′ represents the optimized raw material melting response change characteristic vector.

[0051] In the technical solution of the present application, the raw material melting response change feature vector has feature redundancy, which leads to the problem that when it is classified and judged by the classifier, there will be a class squeeze between categories relative to the class label of the classifier, which affects the classification iteration effect. Based on this, in the technical solution of the present application, the raw material melting response change feature vector is subjected to entropy reduction adjustment based on the measurement of information uncertainty change. First, each row vector is extracted from the classifier's feeding speed prediction parameter space to form a row vector set. Then, the feeding speed class probability entropy degradation simulation value between the raw material melting response change feature vector and each row vector in the above row vector set is calculated to measure the uncertainty or information entropy change of the raw material melting response change feature vector under given parameters. Then, all the feeding speed class probability entropy degradation simulation values ​​are squared and summed, and then the square root is taken to obtain a comprehensive feeding speed entropy degradation modulation value, which reflects the degree of entropy change of the entire system when processing the feature vector. Finally, the raw material melting response change characteristic vector is adjusted according to the entropy degradation modulation value of the feeding speed obtained above. Specifically, the attribute set of the raw material melting response change characteristic vector is represented in a power series based on the entropy degradation modulation value, and each attribute value of the raw material melting response change attribute vector is added as a set aggregated bounded deviation represented by an aggregate radius, so as to perform attribute selection and sparse optimization on the raw material melting response change attribute vector under a unified entropy degradation modulation value restriction standard, so as to ensure that the attribute vector contains sufficient data but is not too cumbersome so as to introduce unnecessary interference or redundancy.

[0052] Finally, the optimized raw material melting response change feature vector is passed through the classifier-based feeding speed adjustment result generator to obtain a classification result, and the classification result is that the feeding speed value of the feeding system at the current time point should be increased, decreased or unchanged. It should be understood that the optimized raw material melting response change feature vector contains rich production information. By inputting the optimized raw material melting response change feature vector into the classifier, the corresponding feeding speed adjustment suggestion can be automatically output, that is, the classification result of the feeding speed adjustment direction at the current time point is generated. When the classifier determines that the current feeding speed should be increased, the feeding system will increase the feeding speed accordingly; when it is judged that it should be reduced, the feeding speed will be reduced; if it is judged to remain unchanged, the current speed will be maintained. By optimizing the adjustment strategy of the feeding speed, unnecessary waste and loss can be reduced, and it helps to maintain the stable operation and efficient production of the production line.

[0053] In an embodiment of the present application, a possible implementation method of passing the optimized raw material melting response change feature vector through a classifier-based feeding speed adjustment result generator to obtain a classification result that the feeding speed value of the feeding system at the current time point should be increased, decreased or remained unchanged can be: using the fully connected layer of the classifier to fully connect encode the optimized raw material melting response change feature vector to obtain a fully connected encoded feature vector of the raw material melting response change; inputting the fully connected encoded feature vector of the raw material melting response change into the Softmax classification function of the classifier to obtain the probability value of the optimized raw material melting response change feature vector belonging to each classification label, the classification label including a value for indicating that the feeding speed value of the feeding system at the current time point should be increased, a value for indicating that the feeding speed value of the feeding system at the current time point should be decreased, and a value for indicating that the feeding speed value of the feeding system at the current time point should remain unchanged; and determining the classification label corresponding to the largest of the probability values ​​as the classification result.

[0054] In summary, the intelligent control method for the rock wool production line based on the embodiment of the present application is explained, which adopts a data analysis method based on artificial intelligence, analyzes the thermal infrared images of the cupola at multiple time points to understand the melting change characteristics of the raw materials, and analyzes the feeding speed data of the feeding system at multiple time points to obtain the feeding speed timing characteristics, and then fuses and optimizes the melting change characteristics of the raw materials and the feeding speed timing characteristics, and finally obtains the adjustment direction of the feeding speed through the classifier. In this way, by accurately controlling the feeding speed, it can be ensured that the raw materials reach the optimal state during the melting process, thereby improving the quality and consistency of the product.

[0055] Figure 6 : is a system block diagram of an intelligent control system for a rock wool production line according to an embodiment of the present application. Figure 6As shown, according to an embodiment of the present application, an intelligent control system 100 for a rock wool production line includes: a rock wool production-related data acquisition module 110, which is used to obtain thermal infrared images of the cupola at multiple predetermined time points within a predetermined time period and the feeding speed values ​​of the feeding system at the multiple predetermined time points; a rock wool production-related data encoding module 120, which is used to perform data joint analysis on the thermal infrared images of the cupola at the multiple predetermined time points and the feeding speed values ​​of the feeding system at the multiple predetermined time points to obtain a raw material melting response change characteristic vector; a feeding speed adjustment result generation module 130, which is used to obtain an adjustment direction of the feeding speed value of the feeding system at the current time point based on the raw material melting response change characteristic vector.

[0056] Here, those skilled in the art will appreciate that the specific functions and operations of the various units and modules in the intelligent control system 100 for the rock wool production line have been described in detail above. Figures 1 to 5 The description of the intelligent control method for a rock wool production line has been introduced in detail, and therefore, its repeated description will be omitted.

[0057] In summary, the intelligent control system 100 for the rock wool production line based on the embodiment of the present application is explained, which adopts a data analysis method based on artificial intelligence, analyzes the thermal infrared images of the cupola at multiple time points to understand the melting change characteristics of the raw materials, and analyzes the feeding speed data of the feeding system at multiple time points to obtain the feeding speed timing characteristics, and then fuses and optimizes the melting change characteristics of the raw materials and the feeding speed timing characteristics, and finally obtains the adjustment direction of the feeding speed through the classifier. In this way, by accurately controlling the feeding speed, it can be ensured that the raw materials reach the best state during the melting process, thereby improving the quality and consistency of the product.

[0058] As described above, the intelligent control system 100 for a rock wool production line according to an embodiment of the present application can be implemented in various wireless terminals, such as a server for intelligent control of a rock wool production line. In one example, the intelligent control system 100 for a rock wool production line according to an embodiment of the present application can be integrated into a wireless terminal as a software module and / or a hardware module. For example, the intelligent control system 100 for a rock wool production line can be a software module in the operating system of the wireless terminal, or can be an application developed for the wireless terminal; of course, the intelligent control system 100 for a rock wool production line can also be one of the many hardware modules of the wireless terminal.

[0059] Alternatively, in another example, the intelligent control system 100 for the rock wool production line and the wireless terminal may also be separate devices, and the intelligent control system 100 for the rock wool production line may be connected to the wireless terminal via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0060] Finally, the present application also relates to a computer-readable storage medium for implementing an intelligent control method for a rock wool production line according to one or more embodiments of the present application. The computer-readable storage medium mentioned herein includes various types of computer storage media, which may be any available medium that can be accessed by a general-purpose or special-purpose computer. For example, a computer-readable storage medium may include RAM, ROM, EPROM, E2PROM, registers, hard disks, removable disks, CD-ROMs or other optical disk storage, disk storage or other magnetic storage devices, or any other temporary or non-temporary medium that can be used to carry or store desired program code units in the form of instructions or data structures and can be accessed by a general-purpose or special-purpose computer, or a general-purpose or special-purpose processor. The description of the computer-readable storage medium according to the present application can refer to the explanation of the intelligent control method for a rock wool production line according to the present application, and no further description will be given.

Claims

1. An intelligent control method for a rock wool production line, characterized in that: include: Acquire thermal infrared images of the cupola at a plurality of predetermined time points within a predetermined time period and feeding speed values ​​of the feeding system at the plurality of predetermined time points; Performing data joint analysis on the thermal infrared images of the cupola at the plurality of predetermined time points and the feeding speed values ​​of the feeding system at the plurality of predetermined time points to obtain a characteristic vector of the raw material melting response change; Based on the raw material melting response change characteristic vector, the adjustment direction of the feeding speed value of the feeding system at the current time point is obtained.

2. The intelligent control method for a rock wool production line according to claim 1, characterized in that: Performing data joint analysis on the thermal infrared images of the cupola at the plurality of predetermined time points and the feeding speed values ​​of the feeding system at the plurality of predetermined time points to obtain a raw material melting response change characteristic vector, including: Extracting image content features from the thermal infrared images of the cupola at the plurality of predetermined time points to obtain a raw material melting state change emphasized feature vector; Extracting time series characteristics of the feeding speed values ​​of the feeding system at the plurality of predetermined time points to obtain a feeding speed time series characteristic vector; The material feeding speed time series feature vector and the material melting state change emphasis feature vector are fused to obtain the material melting response change feature vector.

3. The intelligent control method for a rock wool production line according to claim 2, characterized in that: Extracting image content features from the thermal infrared images of the cupola at the plurality of predetermined time points to obtain a raw material melting state change emphasis feature vector includes: The thermal infrared images of the cupola at the plurality of predetermined time points are respectively passed through a thermal infrared image encoder based on a spatial attention mechanism to obtain a plurality of raw material melting state feature matrices; Aggregating the plurality of raw material melting state characteristic matrices along the channel into a raw material melting state change characteristic diagram; The raw material melting state change feature map is passed through a melting state feature emphasizer based on a channel attention mechanism to obtain a raw material melting state change emphasized feature map; The raw material melting state change emphasis feature map is subjected to mean pooling processing of each feature matrix along the channel dimension to obtain the raw material melting state change emphasis feature vector.

4. The intelligent control method for a rock wool production line according to claim 3, characterized in that: Extracting the time series characteristics of the feeding speed values ​​of the feeding system at the plurality of predetermined time points to obtain a feeding speed time series characteristic vector includes: Arrange the feeding speed values ​​of the feeding system at the plurality of predetermined time points according to the time dimension to obtain a feeding speed time sequence input vector; The feeding speed timing input vector is passed through a feeding speed timing encoder to obtain the feeding speed timing characteristic vector.

5. The intelligent control method for a rock wool production line according to claim 4, characterized in that: The thermal infrared image encoder based on the spatial attention mechanism is a convolutional neural network model using the spatial attention mechanism, the melting state feature emphasizer based on the channel attention mechanism is a convolutional neural network model using the channel attention mechanism, and the feeding speed timing encoder is a convolutional neural network model using a one-dimensional convolution kernel.

6. The intelligent control method for a rock wool production line according to claim 5, characterized in that: Based on the raw material melting response change characteristic vector, the adjustment direction of the feeding speed value of the feeding system at the current time point is obtained, including: Performing entropy reduction adjustment on the raw material melting response change characteristic vector based on information uncertainty change measurement to obtain an optimized raw material melting response change characteristic vector; The optimized raw material melting response change feature vector is passed through a classifier-based feeding speed adjustment result generator to obtain a classification result, and the classification result is whether the feeding speed value of the feeding system at the current time point should be increased, decreased or kept unchanged.

7. The intelligent control method for a rock wool production line according to claim 6, characterized in that: The raw material melting response change characteristic vector is subjected to entropy reduction adjustment based on information uncertainty change measurement to obtain an optimized raw material melting response change characteristic vector, including: Extracting each row vector of the feeding speed prediction parameter space to obtain a set of feeding speed prediction parameter row vectors; Calculating a feeding speed class probability entropy degradation simulation value between the raw material melting response change characteristic vector and each feeding speed prediction parameter row vector in the set of feeding speed prediction parameter row vectors to obtain a set of feeding speed class probability entropy degradation simulation values; Calculating the square root of the sum of squares of the feeding speed class probability entropy degradation simulation values ​​to obtain a feeding speed entropy degradation modulation value; Based on the entropy degradation modulation value of the feeding speed, the characteristic distribution modulation is performed on the raw material melting response change characteristic vector to obtain the optimized raw material melting response change characteristic vector.

8. An intelligent control system for a rock wool production line, characterized in that: include: A rock wool production-related data acquisition module, used to obtain thermal infrared images of the cupola at multiple predetermined time points within a predetermined time period and the feeding speed values ​​of the feeding system at the multiple predetermined time points; A rock wool production related data encoding module, used for performing data joint analysis on the thermal infrared images of the cupola at the plurality of predetermined time points and the feeding speed values ​​of the feeding system at the plurality of predetermined time points to obtain a raw material melting response change characteristic vector; The feeding speed adjustment result generating module is used to obtain the adjustment direction of the feeding speed value of the feeding system at the current time point based on the raw material melting response change characteristic vector.

9. The intelligent control system for a rock wool production line according to claim 8, characterized in that: The rock wool production related data coding module includes: A cupola thermal infrared image encoding unit, used for extracting image content features from the thermal infrared images of the cupola at the plurality of predetermined time points to obtain a raw material melting state change emphasized feature vector; A feeding speed time series feature extraction unit, used for performing time series feature extraction on the feeding speed values ​​of the feeding system at the plurality of predetermined time points to obtain a feeding speed time series feature vector; The material feeding speed time series feature vector and the material melting state change emphasis feature vector are fused to obtain the material melting response change feature vector.

10. The intelligent control system for a rock wool production line according to claim 9, characterized in that: The feeding speed adjustment result generating module comprises: A raw material melting response change characteristic optimization unit, used for performing entropy reduction adjustment on the raw material melting response change characteristic vector based on information uncertainty change measurement to obtain an optimized raw material melting response change characteristic vector; The optimized raw material melting response change characteristic analysis unit is used to pass the optimized raw material melting response change characteristic vector through the classifier-based feeding speed adjustment result generator to obtain a classification result, and the classification result is that the feeding speed value of the feeding system at the current time point should be increased, decreased or remained unchanged.

Images