Intelligent control system and method for waste tire cracking production line

Through artificial intelligence technology, the thermal distribution and air volume characteristics during the cracking process of waste tires are extracted and matched, and the cooling air volume is adjusted adaptively, solving the problem of inaccurate control in the existing technology and improving the intelligent control effect of cracking production of waste tires.

CN119046701AInactive Publication Date: 2024-11-29克拉玛依顺通环保科技有限责任公司

Patent Information

Application Number
CN202411529990.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-30
Publication Date
2024-11-29
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing technology cannot automatically adjust the cool air volume according to the real-time heat distribution during the cracking of waste tires, resulting in insufficient control and difficulty in dealing with changes in complex production environments. It is not possible to fully utilize the depth information of the heat distribution and cool air volume data, and ignore the temperature gradient and The timing of the cool air volume changes dynamically, affecting combustion efficiency and stability.

Method used

Using image analysis and data processing technology based on artificial intelligence, the semantic features of thermal distribution maps and wind volume are extracted through the hollow convolutional neural network model and Bi-LSTM model, and the cross-domain dynamic gated matching module is used to match features and adaptively adjust. The cool air volume of the high-temperature fan.

Benefits of technology

It realizes rapid response to heat distribution and changes in the volume of cool air, accurately controls the volume of cool air, and improves the intelligent control accuracy and efficiency of waste tire cracking production.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119046701A_ABST
    Figure CN119046701A_ABST
Patent Text Reader

Abstract

The invention provides an intelligent control system and method for a waste tire cracking production line, relates to the field of intelligent control, and adopts an image analysis and data processing technology based on artificial intelligence to perform semantic feature extraction and optimization on time sequence data of a thermal distribution diagram in a decomposition furnace. Time sequence correlation is carried out on the time sequence data of the air volume of the high-temperature fan, so that the air volume of the high-temperature fan at the next time point is adaptively controlled according to cross-modal dynamic matching characteristics between the optimized thermal distribution semantic characteristics and the air volume time sequence characteristics. Therefore, the thermal distribution and the air volume change can be quickly responded, so that the air volume can be accurately controlled, and more intelligent waste tire cracking production control is realized.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

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

[0002] Tire pyrolysis, also known as pyrolysis, refers to the high-temperature pyrolysis reaction of organic matter under oxygen-free conditions, which converts the rubber material in the tire into a variety of low-molecular-weight organic compounds, thereby facilitating the recycling of waste tires. By pyrolyzing waste rubber in a pyrolysis furnace, the flue gas generated in the process can be condensed and separated to obtain heavy oil, light oil, pyrolysis gas and a small amount of water vapor, and the remaining part can be made into pyrolysis carbon black. The pyrolysis of waste rubber is the main method for recycling waste tires.

[0003] Chinese patent CN107129168B discloses a method for suspended catalytic cracking and catalytic oxidation flameless combustion of waste tires, which mixes and homogenizes waste tires with liquid catalytic cracking oxidant, and then feeds them into the decomposition furnace of a dry cement production line. By increasing the air volume of the high-temperature fan, utilizing the air flow of the kiln system and the temperature field of the decomposition furnace, the suspended catalytic cracking and flameless combustion of the waste tires are achieved.

[0004] In the above-mentioned patent, the process of suspended catalytic cracking and flameless combustion of modified fuel by the bottom temperature field in the decomposition furnace is to gradually increase the amount of modified fuel, gradually reduce the amount of tail coal, and synchronously adjust the air volume of the system high-temperature fan to maintain it at a level higher than the equilibrium air volume of the kiln system. This adjustment mainly relies on artificially set thresholds and empirical adjustment strategies. However, this adjustment method cannot automatically adjust the air volume according to the real-time changing thermal distribution, resulting in inaccurate control and difficulty in coping with complex production environment changes, especially when the system is faced with sudden changes or requires precise control. It seems powerless. In addition, this adjustment method fails to fully utilize the deep information of thermal distribution and air volume data, resulting in the inability to accurately capture and respond to subtle changes in temperature and air volume during combustion. In other words, this method ignores the complex patterns and trends hidden in the data, such as temperature gradients, dynamic changes in air volume time series, etc. These features are crucial to optimizing combustion efficiency and stability.

[0005] Therefore, an optimized control scheme for a waste tire pyrolysis production line is desired. Summary of the invention

[0006] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides an intelligent control system and method for a waste tire pyrolysis production line.

[0007] According to one aspect of the present application, an intelligent control method for a waste tire cracking production line is provided, which includes: crushing waste tires to obtain waste tire rubber blocks; spraying liquid catalytic cracking oxidant on the surface of the waste tire rubber blocks, and performing homogenization treatment to obtain modified fuel; sending the modified fuel to a decomposition furnace, so as to perform suspended catalytic cracking and catalytic oxidation flameless combustion on the modified fuel through the bottom temperature field in the decomposition furnace to obtain a mineralizer raw material; wherein, sending the modified fuel to the decomposition furnace, so as to perform suspended catalytic cracking and catalytic oxidation flameless combustion on the modified fuel through the bottom temperature field in the decomposition furnace to obtain a mineralizer raw material, characterized in that it includes: obtaining a time series of a thermal distribution diagram in the decomposition furnace; obtaining a time series of an air volume of a high-temperature fan; extracting a time series of a thermal distribution diagram in the decomposition furnace through a thermal distribution semantic feature based on a void convolutional neural network model; The extractor is used to obtain the time series of the semantic feature vector of the thermal distribution map; the time series of the semantic feature vector of the thermal distribution map is passed through a feature sequence optimization module based on feature significance distribution to obtain the time series of the semantic feature vector of the thermal distribution map after selection and optimization; after the time series of the air volume is arranged as an air volume time series input vector according to the time dimension, the air volume time series input vector is passed through an air volume time series encoder based on a Bi-LSTM model to obtain an air volume time series association feature vector; the air volume time series association feature vector and the time series of the semantic feature vector of the thermal distribution map after selection and optimization are passed through a cross-domain dynamic gating matching module based on local key matrix association to obtain an air volume-thermal distribution cross-domain dynamic semantic matching feature vector; based on the air volume-thermal distribution cross-domain dynamic semantic matching feature vector, a control result is obtained, and the control result is used to indicate whether the air volume of the high-temperature fan at the next time point should increase, decrease or remain unchanged.

[0008] According to another aspect of the present application, an intelligent control system for a waste tire pyrolysis production line is provided, which includes: a thermal distribution map data acquisition module, used to obtain the time series of the thermal distribution map in the decomposition furnace; a high-temperature fan air volume data acquisition module, used to obtain the time series of the air volume of the high-temperature fan; a thermal distribution map semantic feature extraction module, used to pass the time series of the thermal distribution map in the decomposition furnace through a thermal distribution semantic feature extractor based on a void convolutional neural network model to obtain a time series of thermal distribution map semantic feature vectors; a thermal distribution map semantic feature selection and optimization module, used to pass the time series of the thermal distribution map semantic feature vectors through a feature sequence optimization module based on feature significance distribution to obtain a time series of thermal distribution map semantic feature vectors after selection and optimization; and an air volume time series correlation feature extraction module. An extraction module is used to arrange the time series of the air volume into an air volume time series input vector according to the time dimension, and then pass the air volume time series input vector through an air volume time series encoder based on a Bi-LSTM model to obtain an air volume time series associated feature vector; an air volume-thermal distribution cross-domain dynamic semantic matching feature generation module is used to pass the time series of the air volume time series associated feature vector and the selected optimized thermal distribution map semantic feature vector through a cross-domain dynamic gating matching module based on local key matrix association to obtain an air volume-thermal distribution cross-domain dynamic semantic matching feature vector; a high-temperature fan air volume adjustment result generation module is used to obtain a control result based on the air volume-thermal distribution cross-domain dynamic semantic matching feature vector, and the control result is used to indicate whether the air volume of the high-temperature fan at the next time point should increase, decrease or remain unchanged.

[0009] Compared with the prior art, the intelligent control system and method for the waste tire cracking production line provided by the present application uses artificial intelligence-based image analysis and data processing technology to extract and optimize the semantic features of the time series data of the thermal distribution diagram in the decomposition furnace, and to perform time series association on the time series data of the air volume of the high-temperature fan, so as to adaptively control the air volume of the high-temperature fan at the next time point according to the cross-modal dynamic matching features between the optimized thermal distribution semantic features and the air volume time series features. In this way, it is possible to quickly respond to changes in thermal distribution and air volume, so as to accurately control the air volume, thereby achieving more intelligent waste tire cracking production control. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, reference is now made to the following description in conjunction with the accompanying drawings.

[0011] In the attached picture: Figure 1 The present invention is a flowchart of an intelligent control method for a waste tire pyrolysis production line according to an embodiment of the present application.

[0012] Figure 2 In the intelligent control method for a waste tire pyrolysis production line according to an embodiment of the present application, the modified fuel is sent to a decomposition furnace, and the modified fuel is subjected to suspended catalytic cracking and catalytic oxidation flameless combustion through the bottom temperature field in the decomposition furnace to obtain a mineralizer raw material. DETAILED DESCRIPTION

[0013] 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.

[0014] Scrap tire pyrolysis is an environmentally friendly and efficient way of resource recovery. It converts waste tires into a variety of valuable by-products such as heavy oil, light oil, pyrolysis gas and pyrolysis carbon black by subjecting them to high temperature treatment in an oxygen-free environment. Chinese patent CN107129168B proposes a method of mixing waste tires with liquid catalytic pyrolysis oxidants and pyrolyzing them in a decomposition furnace of a dry cement production line. This method treats waste tires by adjusting the fuel dosage and the air volume of the high-temperature fan, but it has certain limitations because it relies on artificially set thresholds and empirical adjustment strategies and lacks the ability to automatically respond to real-time changes in thermal distribution. In addition, this method fails to fully utilize the deep information of thermal distribution and air volume data, and ignores key features such as temperature gradient and dynamic changes in air volume time series, which are of great significance for improving combustion efficiency and stability. Therefore, the method provided by this patent has limitations in accurately controlling and optimizing the combustion process, and further research and improvement are needed to improve its efficiency and adaptability.

[0015] Based on this, the present application provides an intelligent control method for a waste tire pyrolysis production line. Figure 1 FIG. 1 is a flow chart of an intelligent control method for a waste tire pyrolysis production line according to an embodiment of the present application. Figure 1 As shown, the intelligent control method for the waste tire cracking production line according to the embodiment of the present application includes: S110, crushing the waste tires to obtain waste tire rubber blocks; S120, spraying the liquid catalytic cracking oxidant on the surface of the waste tire rubber blocks, and homogenizing them to obtain modified fuel; S130, sending the modified fuel to a decomposition furnace to perform suspended catalytic cracking and catalytic oxidation flameless combustion on the modified fuel through the bottom temperature field in the decomposition furnace to obtain a mineralizer raw material.

[0016] Accordingly, in sending the modified fuel to the decomposition furnace, and subjecting the modified fuel to suspended catalytic cracking and catalytic oxidation flameless combustion through the bottom temperature field in the decomposition furnace to obtain the mineralizer raw material, the technical concept of the present application is to obtain the time series of the thermal distribution diagram in the decomposition furnace, and the time series of the air volume of the high-temperature fan, and use image analysis and data processing technology based on artificial intelligence to extract and optimize the semantic features of the thermal distribution diagram, and perform time series association on the air volume, so as to adaptively control the air volume of the high-temperature fan at the next time point according to the cross-modal dynamic matching characteristics between the optimized thermal distribution semantic features and the air volume time series features. In this way, it is possible to quickly respond to changes in thermal distribution and air volume, and at the same time more comprehensively understand the subtle dynamic changes of parameters in the combustion process, so as to accurately control the size of the air volume, ensure that the combustion process runs in the best state, and thus achieve more intelligent waste tire cracking production control.

[0017] Figure 2 The present invention is a flow chart of an intelligent control method for a waste tire pyrolysis production line according to an embodiment of the present application, wherein the modified fuel is sent to a calciner to perform suspended catalytic pyrolysis and catalytic oxidation flameless combustion on the modified fuel through a bottom temperature field in the calciner to obtain a mineralizer raw material. Figure 2 As shown, the modified fuel is sent to the decomposition furnace to perform suspended catalytic cracking and catalytic oxidation flameless combustion on the modified fuel through the bottom temperature field in the decomposition furnace to obtain a mineralizer raw material, including: S131, obtaining the time series of the thermodynamic distribution diagram in the decomposition furnace; S132, obtaining the time series of the air volume of the high-temperature fan; S133, passing the time series of the thermodynamic distribution diagram in the decomposition furnace through a thermodynamic distribution semantic feature extractor based on a void convolutional neural network model to obtain a time series of the thermodynamic distribution diagram semantic feature vector; S134, passing the time series of the thermodynamic distribution diagram semantic feature vector through a feature sequence optimization module based on feature significance distribution to obtain a time series of the thermodynamic distribution diagram semantic feature vector selected after optimization. time sequence; S135, after arranging the time series of the air volume into an air volume time series input vector according to the time dimension, the air volume time series input vector is passed through an air volume time series encoder based on a Bi-LSTM model to obtain an air volume time series associated feature vector; S136, the air volume time series associated feature vector and the time series of the selected optimized thermal distribution map semantic feature vector are passed through a cross-domain dynamic gating matching module based on local key matrix association to obtain an air volume-thermal distribution cross-domain dynamic semantic matching feature vector; S137, based on the air volume-thermal distribution cross-domain dynamic semantic matching feature vector, a control result is obtained, and the control result is used to indicate whether the air volume of the high-temperature fan at the next time point should increase, decrease or remain unchanged.

[0018] In step S131 and step S132, the time series of the thermodynamic distribution diagram in the decomposition furnace and the time series of the air volume of the high-temperature fan are obtained respectively. It should be understood that the thermodynamic distribution diagram in the decomposition furnace provides the temperature conditions of different areas in the decomposition furnace, and can monitor the temperature changes in the modified fuel cracking process in real time. By obtaining the time series data of the thermodynamic distribution diagram, it can help understand the temperature change trend of the decomposition furnace over time. The air volume directly affects the supply of oxygen in the decomposition furnace, and then affects the combustion efficiency and the progress of the cracking reaction. By obtaining the time series data of the air volume of the high-temperature fan, it can help understand the change trend of the air volume over time. By analyzing the two obtained time series data, the problems existing in the decomposition furnace can be identified in real time, and then the thermodynamic distribution and airflow velocity in the decomposition furnace can be optimized by adjusting the air volume of the high-temperature fan in real time, so as to improve the efficiency of the cracking reaction and the product quality. In particular, in a specific embodiment of the present application, the time series data of the thermodynamic distribution diagram in the decomposition furnace can be collected by an infrared thermal imager, and the time series data of the air volume of the high-temperature fan can be collected by a flow sensor.

[0019] In step S133, the time series of the thermal distribution diagram in the decomposition furnace is passed through a thermal distribution semantic feature extractor based on a hole convolutional neural network model to obtain a time series of semantic feature vectors of the thermal distribution diagram. Accordingly, considering that each diagram in the time series of the thermal distribution diagram contains semantic feature information about the thermal distribution, such as the temperature of different areas and the flow pattern of heat flow in the furnace, based on this, in the technical solution of the present application, the time series of the thermal distribution diagram in the decomposition furnace is passed through a thermal distribution semantic feature extractor based on a hole convolutional neural network model to capture and perceive the semantic features of the thermal distribution, and obtain a time series of semantic feature vectors of the thermal distribution diagram. Those skilled in the art should know that the hole convolutional neural network model uses a hole convolution kernel (i.e., holes are inserted between the elements of the convolution kernel), which enables the model to expand the receptive field without increasing the number of parameters, and then can capture the global features in the thermal distribution diagram in a larger range, so as to better understand the overall trend of temperature changes. In addition, the temperature information in the thermal distribution map usually has spatial correlation. When extracting features from the thermal distribution map, the atrous convolution can well preserve this spatial structure information, thereby improving the representation ability of the extracted semantic features of the thermal distribution map.

[0020] In step S134, the time series of the semantic feature vectors of the thermal distribution diagram is passed through a feature sequence optimization module based on feature significance distribution to obtain a time series of the semantic feature vectors of the thermal distribution diagram after selection and optimization. It should be understood that the importance of each thermal distribution semantic feature of the time series of the semantic feature vectors of the thermal distribution diagram is different in the entire time domain. Therefore, in order to help identify the periodic changes and frequency components in the thermal distribution diagram, so as to more centrally understand the periodic behavior of the thermal dynamics in the furnace, in the technical solution of the present application, the time series of the semantic feature vectors of the thermal distribution diagram is passed through a feature sequence optimization module based on feature significance distribution to obtain a time series of the semantic feature vectors of the thermal distribution diagram after selection and optimization. In particular, the feature sequence optimization module based on feature significance distribution is a method for feature extraction and optimization using energy spectrum analysis, and its core goal is to optimize the feature vector sequence by evaluating the performance of the feature in energy distribution. Specifically, first, the characteristic energy distribution spectrum vector of each semantic feature vector of the thermal distribution map is calculated to map each vector to the space of the energy distribution spectrum to more effectively represent and process the features, because the energy spectrum can capture the important information of the features in the frequency domain, and at the same time, it can more easily identify and analyze the patterns in the thermal distribution map, such as periodic changes or specific heat flow patterns, to obtain a sequence of the semantic feature energy distribution spectrum vectors of the thermal distribution map. Next, the mean vector of the sequence of energy distribution spectrum vectors is calculated as the center vector of the semantic energy spectrum distribution of the thermal distribution map to represent the average characteristics of all energy distribution spectrum vectors in the entire time series, reflecting the general characteristics of the entire sequence, which helps to evaluate the correlation or difference between each feature and the central feature. Then, the semantic difference between each energy distribution spectrum vector and the central vector is calculated to obtain a sequence of span factors of the semantic energy distribution spectrum of the thermal distribution map, that is, the energy distribution spectrum span factor represents the distance from the central vector and reflects the discreteness of the feature. Finally, each energy distribution spectrum span factor is compared and selected with the preset threshold to obtain the time series of the semantic feature vector of the thermal distribution map after the final selection that meets the conditions. In this way, the interference of noise and irrelevant information can be removed, and the most representative and distinctive features can be retained to provide support for the subsequent analysis of thermal changes in the decomposition furnace.

[0021] Specifically, in an embodiment of the present application, the time series of the semantic feature vectors of the thermal distribution map is passed through a feature sequence optimization module based on feature significance distribution to obtain a time series of the semantic feature vectors of the thermal distribution map after selection and optimization, including: calculating the feature energy distribution spectrum vectors of each thermal distribution map semantic feature vector in the time queue of the semantic feature vectors of the thermal distribution map to obtain a sequence of the semantic feature energy distribution spectrum vectors of the thermal distribution map; calculating the positional mean vector of the sequence of the semantic feature energy distribution spectrum vectors of the thermal distribution map as the center vector of the semantic energy spectrum distribution of the thermal distribution map; calculating the energy distribution spectrum span factor between the center vector of the semantic energy spectrum distribution of the thermal distribution map and each thermal distribution map semantic feature energy distribution spectrum vector in the sequence of the semantic feature energy distribution spectrum vector of the thermal distribution map to obtain a sequence of the semantic energy distribution spectrum span factors of the thermal distribution map; based on the comparison between each thermal distribution map semantic energy distribution spectrum span factor in the sequence of the semantic energy distribution spectrum span factors of the thermal distribution map and a preset threshold, determining the selected optimized thermal distribution map semantic A time series of feature vectors; wherein, the energy distribution spectrum span factors between the central vector of the semantic energy spectrum distribution of the thermal distribution map and each of the semantic feature energy distribution spectrum vectors of the thermal distribution map in the sequence of the semantic feature energy distribution spectrum vectors of the thermal distribution map are calculated to obtain a sequence of the semantic energy distribution spectrum span factors of the thermal distribution map, including: calculating the positional difference between the central vector of the semantic energy spectrum distribution of the thermal distribution map and the semantic feature energy distribution spectrum vector of the thermal distribution map to obtain the semantic differential feature vector of the thermal distribution map; calculating the covariance matrix between the central vector of the semantic energy spectrum distribution of the thermal distribution map and the semantic feature energy distribution spectrum vector of the thermal distribution map, and calculating the inverse of the covariance matrix to obtain the semantic inverse covariance matrix of the thermal distribution map; calculating the transposed vector of the semantic differential feature vector of the thermal distribution map and the product of the semantic inverse covariance matrix of the thermal distribution map and the semantic differential feature vector of the thermal distribution map to obtain the semantic distribution value of the thermal distribution map; calculating the square root of the semantic distribution value of the thermal distribution map to obtain the semantic energy distribution spectrum span factor of the thermal distribution map.

[0022] More specifically, in an embodiment of the present application, the characteristic energy distribution spectrum vector of each thermal distribution map semantic feature vector in the time queue of the thermal distribution map semantic feature vector is calculated to obtain a sequence of thermal distribution map semantic feature energy distribution spectrum vectors, including: respectively calculating the characteristic distribution energy synergy representation vector between each thermal distribution map semantic feature vector and other thermal distribution map semantic feature vectors in the time queue of the thermal distribution map semantic feature vector to obtain a sequence of multiple thermal distribution map semantic feature distribution energy synergy representation vectors; calculating the characteristic distribution energy synergy factor vector of each sequence of thermal distribution map semantic feature distribution energy synergy representation vectors in the sequence of multiple thermal distribution map semantic feature distribution energy synergy representation vectors to obtain a sequence of thermal distribution map semantic feature energy distribution spectrum vectors.

[0023] More specifically, in an embodiment of the present application, the feature distribution energy collaborative representation vectors between each of the thermal distribution map semantic feature vectors and other thermal distribution map semantic feature vectors in the time queue of the thermal distribution map semantic feature vector are calculated respectively to obtain a sequence of multiple thermal distribution map semantic feature distribution energy collaborative representation vectors, including: extracting a predetermined thermal distribution map semantic feature vector from the time queue of the thermal distribution map semantic feature vector; performing element-by-element maximum value extraction on the predetermined thermal distribution map semantic feature vector and each of the other thermal distribution map semantic feature vectors respectively to obtain a sequence of thermal distribution map semantic feature distribution energy collaborative representation vectors corresponding to the predetermined thermal distribution map semantic feature vector.

[0024] More specifically, in an embodiment of the present application, the feature distribution energy synergy factor vector of each sequence of the thermal distribution map semantic feature distribution energy synergy representation vectors in the sequence of the multiple thermal distribution map semantic feature distribution energy synergy representation vectors is calculated to obtain a sequence of the thermal distribution map semantic feature energy distribution spectrum vectors, including: respectively extracting the maximum value of each thermal distribution map semantic feature distribution energy synergy representation vector in the sequence of the thermal distribution map semantic feature distribution energy synergy representation vector to obtain a sequence of thermal distribution map semantic feature maximum values; respectively calculating the mean and variance of each thermal distribution map semantic feature distribution energy synergy representation vector to obtain a sequence of thermal distribution map semantic feature means and a sequence of thermal distribution map semantic feature variances; adding the sequence of thermal distribution map semantic feature variances and preset hyperparameters by position to obtain a sequence of thermal distribution map semantic feature first energy distribution coefficients; calculating the thermal distribution map semantic feature maximum The square of the difference between the maximum value of the thermal distribution map semantic feature and the mean value of the thermal distribution map semantic feature in each group corresponding to the sequence of large values ​​and the sequence of the mean values ​​of the thermal distribution map semantic features is obtained to obtain a sequence of thermal distribution map semantic difference values; the sequence of thermal distribution map semantic feature variances is multiplied by constant two by position to obtain a sequence of modulated thermal distribution map semantic feature variances; the thermal distribution map semantic difference values ​​and the sequence of modulated thermal distribution map semantic feature variances in each group corresponding to the thermal distribution map semantic difference values ​​and the modulated thermal distribution map semantic feature variances are added to the preset hyperparameter to obtain a sequence of second energy distribution coefficients of thermal distribution map semantic features; the sequence of first energy distribution coefficients of the thermal distribution map semantic features is divided by the sequence of second energy distribution coefficients of the thermal distribution map semantic features by position to obtain the thermal distribution map semantic feature energy distribution spectrum vector composed of multiple feature distribution energy synergy factors.

[0025] In the embodiment of the present application, specifically, the time series of the semantic feature vector of the heat distribution map is passed through a feature sequence optimization module based on feature significance distribution to obtain a time series of the semantic feature vector of the heat distribution map after selection optimization, including: passing the time series of the semantic feature vector of the heat distribution map through a feature sequence optimization module based on feature significance distribution, and processing it with the following selection optimization formula to obtain the time series of the semantic feature vector of the heat distribution map after selection optimization; wherein, the selection optimization formula is: .

[0026] in, and Respectively represent the first and The heat map semantic feature vector, It means taking the maximum value between two vectors element by element. For the and The heat map semantic feature distribution energy synergistic representation vector between the heat map semantic feature vectors, To extract the maximum value of a vector, and They are The mean and variance of To preset hyperparameters, is the first in the sequence of the energy distribution spectrum vector of the semantic feature of the thermal distribution graph The energy distribution spectrum vector of the semantic feature of the thermal distribution map The characteristic distribution energy synergy factor, For the The energy distribution spectrum vector of the semantic feature of the thermal distribution map The characteristic distribution energy synergy factor, A sequence of energy distribution spectrum vectors representing the semantic features of the thermal distribution graph, is the first in the sequence of the energy distribution spectrum vector of the semantic feature of the thermal distribution graph The energy distribution spectrum vector of the semantic feature of the thermal distribution map, is the number of vectors in the sequence of energy distribution spectrum vectors of the semantic feature of the thermal distribution graph, is the center vector of the semantic energy spectrum distribution of the thermal distribution graph, is the transpose of a vector, It is The covariance matrix between the energy distribution spectrum vector of the semantic feature of the thermal distribution map and the center vector of the semantic energy spectrum distribution of the thermal distribution map, For the The span factor of the thermal distribution map semantic energy distribution spectrum corresponding to the thermal distribution map semantic feature energy distribution spectrum vector, is the preset threshold, is the time series of the semantic feature vector of the heat distribution map after the optimization. Select the optimized semantic feature vector of the heat distribution map, is the time series of the semantic feature vector of the heat distribution map after the selection optimization.

[0027] In step S135, after the time series of the air pull volume is arranged according to the time dimension as an air pull volume time series input vector, the air pull volume time series input vector is passed through an air pull volume time series encoder based on a Bi-LSTM model to obtain an air pull volume time series correlation feature vector. Accordingly, considering that the time series of the air pull volume has time series fluctuations or periodic changes in the time dimension, therefore, in the technical solution of the present application, the time series of the air pull volume is arranged according to the time dimension as an air pull volume time series input vector. Next, considering that the air pull volume time series input vector has a time series correlation relationship, and the data at the current time point is affected by the previous and subsequent data, and the Bi-LSTM model can process past and future information at the same time, which is very important for understanding the long-term dependencies in time series data. Based on this, in the technical solution of the present application, the air pull volume time series input vector is passed through an air pull volume time series encoder based on a Bi-LSTM model to refine and capture the correlation between each time point, and obtain the air pull volume time series correlation feature vector.

[0028] In step S136, the time series of the airflow temporal association feature vector and the selected optimized thermal distribution map semantic feature vector are passed through a cross-domain dynamic gating matching module based on local key matrix association to obtain the airflow-thermal distribution cross-domain dynamic semantic matching feature vector. It should be understood that the airflow temporal association feature vector reflects the temporal dynamic feature information of the rate of gas flow in the furnace, which is a key factor affecting combustion efficiency and thermal distribution. The time series of the selected optimized thermal distribution map semantic feature vector provides a static snapshot of the temperature information in the furnace at different time points. The change in temperature will cause the adjustment of the airflow, so in order to more accurately control the size of the airflow based on the matching response between the two, in the technical solution of the present application, the time series of the airflow temporal association feature vector and the selected optimized thermal distribution map semantic feature vector are passed through a cross-domain dynamic gating matching module based on local key matrix association to obtain the airflow-thermal distribution cross-domain dynamic semantic matching feature vector. That is, the cross-domain dynamic gated matching module based on local key matrix association adopts a cross-domain attention mechanism based on the Transformer architecture, and performs self-attention operations through the interaction of query vectors, key vectors and value vectors to achieve the fusion and integration of information from different domains. This method allows the model to dynamically adjust its focus when processing features from different data sources, thereby more effectively combining and utilizing cross-domain information.

[0029] In detail, first, a linear transformation is performed on the time series of the semantic feature vector of the thermal distribution map after the selection and optimization so that the dimension of each feature remains consistent, and each linearly transformed vector is arranged into an initial key matrix of the thermal distribution. Next, each row vector in each initial key matrix of the thermal distribution and the time-series associated feature vector of the air pull volume are calculated to perform feature distribution difference calculation to obtain a set of energy coefficients of the difference in air pull volume-thermal distribution feature distribution, so as to quantify the difference between the semantic features of the thermal distribution map and the time-series associated features of the air pull volume using the time-series associated features of the air pull volume as a reference anchor point, and provide a difference measure for subsequent cross-domain information integration. In particular, since each row vector is calculated to be different from the time-series associated feature vector of the air pull volume, the set of feature distribution difference energy coefficients as a whole also reflects the local time-series association information between each row vector. Then, each distribution difference energy coefficient is compared with a predetermined threshold. For the case where the coefficient is less than the threshold, the mean vector of the time series correlation feature vector of the air volume and the corresponding row vector is used as the row vector in the key matrix of the updated thermal distribution. Here, the update operation can identify the thermal distribution features that are closely related to the change of the air volume, and strengthen these features to highlight their importance in the model. Then, based on the time series correlation feature vector of the air volume, the query vector and value vector are constructed, and each row vector in the key matrix of the updated thermal distribution is used as the key vector, which is input into the cross-domain attention gate query module based on the Transformer structure to obtain the sequence of cross-domain query attention vectors of the air volume-thermal distribution, so as to better capture and express the complex relationship between the air volume and the thermal distribution. Finally, the mean vector of the sequence of cross-domain query attention vectors is calculated as the cross-domain dynamic semantic matching feature vector of the air volume-thermal distribution. In this way, all query attention information in the thermal distribution sequence and between the air volume can be comprehensively considered, thereby providing help for the subsequent control of the air volume.

[0030] Specifically, in an embodiment of the present application, the time series of the airflow volume time series association feature vector and the selected optimized thermal distribution map semantic feature vector are passed through a cross-domain dynamic gating matching module based on local key matrix association to obtain the airflow volume-thermal distribution cross-domain dynamic semantic matching feature vector, including: linearly transforming each selected optimized thermal distribution map semantic feature vector in the time series of the selected optimized thermal distribution map semantic feature vector to obtain an initial key matrix of thermal distribution, wherein each thermal distribution row vector in the initial key matrix of thermal distribution is a linearly transformed semantic feature vector of the selected optimized thermal distribution map; calculating the relationship between the airflow volume time series association feature vector and each thermal distribution row vector in the initial key matrix of thermal distribution. The characteristic distribution difference energy coefficients between the distribution row vectors are obtained to obtain a set of the pull air volume-thermal distribution characteristic distribution difference energy coefficients; each pull air volume-thermal distribution characteristic distribution difference energy coefficient in the set of the pull air volume-thermal distribution characteristic distribution difference energy coefficients is compared with a predetermined threshold value, and in response to the pull air volume-thermal distribution characteristic distribution difference energy coefficient being less than the predetermined threshold value, the mean vector between the thermal distribution row vector corresponding to the pull air volume-thermal distribution characteristic distribution difference energy coefficient and the pull air volume time series associated characteristic vector is used as the updated thermal distribution row vector to obtain an updated thermal distribution key matrix; the pull air volume time series associated characteristic vector is processed using a value embedding matrix to obtain a pull air volume time series associated value vector ; The air volume time series associated feature vector is used as the query vector, each updated thermal distribution row vector in the updated thermal distribution key matrix is ​​used as the key vector, and the air volume time series associated value vector is input into the cross-domain attention gate query module based on the Transformer structure to obtain a sequence of air volume-thermal distribution cross-domain query attention vectors; The positional mean vector of the sequence of air volume-thermal distribution cross-domain query attention vectors is calculated to obtain the air volume-thermal distribution cross-domain dynamic semantic matching feature vector; wherein, the air volume time series associated feature vector is used as the query vector, each updated thermal distribution row vector in the updated thermal distribution key matrix is ​​used as the key vector, and the air volume time series associated value vector is input into the cross-domain attention gate query module based on the Transformer structure to obtain a sequence of air volume-thermal distribution cross-domain query attention vectors; A cross-domain attention gate query module of the rmer structure is used to obtain a sequence of cross-domain query attention vectors of airflow-heat distribution, including: calculating the product between the query vector and the transposed vector of the updated thermal distribution row vector to obtain a query-update thermal distribution association matrix, and then dividing the query-update thermal distribution association matrix by the square root of the length of the updated thermal distribution row vector by position to obtain a query-update thermal distribution modulation matrix; inputting the query-update thermal distribution modulation matrix into the softmax function to obtain a query-update thermal distribution activation matrix; multiplying the query-update thermal distribution activation matrix by the airflow temporal association value vector to obtain the cross-domain query attention vector of the airflow-heat distribution.

[0031] More specifically, in an embodiment of the present application, the characteristic distribution difference energy coefficient between the time-series associated characteristic vector of the pull air volume and each thermal distribution row vector in the initial bond matrix of the thermal distribution is calculated to obtain a set of characteristic distribution difference energy coefficients of the pull air volume-thermal distribution, including: calculating the interaction vector between the time-series associated characteristic vector of the pull air volume and each thermal distribution row vector in the initial bond matrix of the thermal distribution to obtain a set of pull air volume-thermal distribution interaction vectors; calculating the characteristic distribution difference energy coefficient of each pull air volume-thermal distribution interaction vector in the set of pull air volume-thermal distribution interaction vectors to obtain a set of characteristic distribution difference energy coefficients of the pull air volume-thermal distribution; wherein, the calculation of the time-series associated characteristic vector of the pull air volume and the initial bond matrix of the thermal distribution is The interaction vectors between each thermal distribution row vector in the array are used to obtain a set of pull air volume-thermal distribution interaction vectors, including: respectively calculating the position difference, position point multiplication and position addition between the pull air volume time series correlation feature vector and the thermal distribution row vector to obtain the pull air volume-thermal distribution differential vector, the pull air volume-thermal distribution point product vector and the pull air volume-thermal distribution sum vector; cascading the pull air volume-thermal distribution differential vector, the pull air volume-thermal distribution point product vector and the pull air volume-thermal distribution sum vector, and then performing one-dimensional convolution encoding to obtain a pull air volume-thermal distribution multi-dimensional fusion vector; performing local window-based maximum pooling processing on the pull air volume-thermal distribution multi-dimensional fusion vector to obtain the pull air volume-thermal distribution interaction vector.

[0032] More specifically, in an embodiment of the present application, the characteristic distribution difference energy coefficient of each air pull volume-thermal distribution interaction vector in the set of the air pull volume-thermal distribution interaction vectors is calculated to obtain the set of the air pull volume-thermal distribution characteristic distribution difference energy coefficients, including: extracting the maximum value of the air pull volume-thermal distribution interaction vector respectively to obtain the air pull volume-thermal distribution interaction maximum value; calculating the mean and variance of the air pull volume-thermal distribution interaction vector respectively to obtain the air pull volume-thermal distribution interaction mean and the air pull volume-thermal distribution interaction variance; adding the air pull volume-thermal distribution interaction variance and the hyperparameter to obtain the first difference energy coefficient of the air pull volume-thermal distribution interaction; calculating The square of the difference between the maximum value of the pull air volume-thermal distribution interaction and the mean value of the pull air volume-thermal distribution interaction is obtained to obtain the pull air volume-thermal distribution interaction difference value; the pull air volume-thermal distribution interaction variance is multiplied by constant two to obtain the modulated pull air volume-thermal distribution interaction variance; the pull air volume-thermal distribution interaction difference value and the modulated pull air volume-thermal distribution interaction variance are added to the hyperparameter to obtain the second difference energy coefficient of the pull air volume-thermal distribution interaction; the first difference energy coefficient of the pull air volume-thermal distribution interaction is divided by the second difference energy coefficient of the pull air volume-thermal distribution interaction to obtain the pull air volume-thermal distribution characteristic distribution difference energy coefficient.

[0033] In an embodiment of the present application, specifically, the time series of the air volume temporal association feature vector and the semantic feature vector of the selected optimized thermal distribution map are passed through a cross-domain dynamic gating matching module based on local key matrix association to obtain the air volume-thermal distribution cross-domain dynamic semantic matching feature vector, including: passing the air volume temporal association feature vector and the time series of the semantic feature vector of the selected optimized thermal distribution map through a cross-domain dynamic gating matching module based on local key matrix association, and processing them with the following cross-domain gating matching formula to obtain the air volume-thermal distribution cross-domain dynamic semantic matching feature vector; wherein, the cross-domain gating matching formula is: .

[0034] in, represents the time series of the semantic feature vector of the heat distribution map after the selection optimization, is the time series of the semantic feature vector of the heat distribution map after the optimization selection Select the optimized semantic feature vector of the heat distribution map, is the initial bond matrix of the thermodynamic distribution, is the first bond matrix in the thermal distribution The heat distribution row vector, The time series of the semantic feature vector of the heat distribution map after the optimization is selected Select the optimized semantic feature vector of the heat distribution map, is the weight matrix, is the bias vector, is the first bond matrix in the thermal distribution The heat distribution row vector, For Perform linear transformation operations, is the time series correlation feature vector of the air pulling volume, , and They are position difference, position dot multiplication and position addition respectively. For cascade processing, is a one-dimensional convolutional coding operation, is the maximum pooling operation, It is the first in the set of airflow-heat distribution interaction vectors. The airflow-heat distribution interaction vector, To extract the maximum value of a vector, and They are The mean and variance of is a hyperparameter, It is the first in the set of the energy coefficients of the airflow-heat distribution characteristics distribution difference. The energy coefficient of the difference in the distribution of the air flow and the heat distribution characteristics, is a predetermined threshold, and To update the thermal distribution bond matrix Update the thermal distribution row vector, and are the value embedding matrix and value bias vector respectively, is the time series correlation value vector of the air pulling volume, for The transposed vector of is the length of each updated thermal distribution row vector, yes function, is the first in the sequence of the cross-domain query attention vector of the airflow-heat distribution The air volume-heat distribution cross-domain query attention vector, is the number of vectors in the sequence of the cross-domain query focus vector of the airflow-heat distribution, It is the cross-domain dynamic semantic matching feature vector of the air volume-heat distribution.

[0035] In step S137, based on the cross-domain dynamic semantic matching feature vector of the air volume-thermal distribution, a control result is obtained, and the control result is used to indicate whether the air volume of the high-temperature fan at the next time point should be increased, decreased or unchanged. Specifically, in an embodiment of the present application, based on the cross-domain dynamic semantic matching feature vector of the air volume-thermal distribution, a control result is obtained, including: passing the cross-domain dynamic semantic matching feature vector of the air volume-thermal distribution through a high-temperature fan air volume controller based on a classifier to obtain the control result. That is, the cross-domain dynamic semantic matching feature vector of the air volume-thermal distribution obtained by cross-domain query matching using the time series of the air volume temporal association feature vector and the selected optimized thermal distribution map semantic feature vector is classified and processed, so as to adaptively control the air volume of the high-temperature fan at the next time point. In this way, it is possible to quickly respond to changes in thermal distribution and air volume, so as to accurately control the size of the air volume, thereby realizing a more intelligent waste tire cracking production control.

[0036] In particular, the present application takes into account that the time series of the air pull time series association feature vector and the selected optimized thermal distribution map semantic feature vector respectively represent the energy spectrum time series distribution optimization characteristics of the bidirectional short-range and long-range time series association characteristics of the air pull and the image semantic characteristics of the thermal distribution map. When they are passed through a cross-domain dynamic gating matching module based on local key matrix association, the local key matrices of different dimensional time series distributions based on cross-modal semantics will also have cross-domain dynamic matching differences. Therefore, it is expected to improve the detail semantic aggregation expression effect of the cross-domain dynamic semantic matching feature vector of the air pull-thermal distribution based on the time series semantic representation differences.

[0037] In a preferred example, when the airflow-heat distribution cross-domain dynamic semantic matching feature vector passes through the classifier-based high-temperature fan airflow controller, the airflow-heat distribution cross-domain dynamic semantic matching feature vector is optimized, and the optimization process includes: calculating the absolute value sum of each eigenvalue of the airflow-heat distribution cross-domain dynamic semantic matching feature vector to obtain a first airflow-heat distribution cross-domain dynamic semantic matching spatial structure value, and calculating the square root of the sum of the squares of each eigenvalue of the airflow-heat distribution cross-domain dynamic semantic matching feature vector to obtain a second airflow-heat distribution cross-domain dynamic semantic matching spatial structure value: ;in, The first character representing the cross-domain dynamic semantic matching feature vector of the ventilation volume and heat distribution eigenvalues, represents the cross-domain dynamic semantic matching feature vector of the airflow-heat distribution, represents the set of real numbers, represents the length of the cross-domain dynamic semantic matching feature vector of the airflow-heat distribution, Indicates the spatial structure value of the first ventilation volume-heat distribution cross-domain dynamic semantic matching, It represents the spatial structure value of the second ventilation volume-heat distribution cross-domain dynamic semantic matching.

[0038] Each eigenvalue of the air pull volume-heat distribution cross-domain dynamic semantic matching feature vector is multiplied with the first air pull volume-heat distribution cross-domain dynamic semantic matching spatial structure value and the second air pull volume-heat distribution cross-domain dynamic semantic matching spatial structure value to obtain the first air pull volume-heat distribution cross-domain dynamic semantic matching structure reference value and the second air pull volume-heat distribution cross-domain dynamic semantic matching structure reference value corresponding to each eigenvalue.

[0039] Multiply each eigenvalue of the air pull volume-heat distribution cross-domain dynamic semantic matching feature vector by the length of the air pull volume-heat distribution cross-domain dynamic semantic matching feature vector and the square root of the length to obtain the first air pull volume-heat distribution cross-domain dynamic semantic matching scale transformation value and the second air pull volume-heat distribution cross-domain dynamic semantic matching scale transformation value corresponding to each eigenvalue.

[0040] The first air pull volume-heat distribution cross-domain dynamic semantic matching structure reference value is divided by the difference between the first air pull volume-heat distribution cross-domain dynamic semantic matching spatial structure value and the first air pull volume-heat distribution cross-domain dynamic semantic matching scale transformation value to obtain the first air pull volume-heat distribution cross-domain dynamic semantic matching transformation adjustment value.

[0041] The second air pull volume-heat distribution cross-domain dynamic semantic matching structure reference value is divided by the difference between the second air pull volume-heat distribution cross-domain dynamic semantic matching spatial structure value and the second air pull volume-heat distribution cross-domain dynamic semantic matching scale transformation value to obtain the second air pull volume-heat distribution cross-domain dynamic semantic matching transformation adjustment value.

[0042] The weighted sum of the first airflow-heat distribution cross-domain dynamic semantic matching transformation adjustment value and the second airflow-heat distribution cross-domain dynamic semantic matching transformation adjustment value is calculated to obtain each eigenvalue of the optimized airflow-heat distribution cross-domain dynamic semantic matching feature vector. Finally, the optimized airflow-heat distribution cross-domain dynamic semantic matching feature vector is passed through a high-temperature fan airflow controller based on a classifier to obtain the control result.

[0043] Here, the cross-domain dynamic semantic matching feature vector of the air flow-heat distribution is recorded as The optimization expression is: ;in, represents the second air intake volume-heat distribution cross-domain dynamic semantic matching transformation adjustment vector composed of the second air intake volume-heat distribution cross-domain dynamic semantic matching transformation adjustment values, represents the first air intake volume-heat distribution cross-domain dynamic semantic matching transformation adjustment vector composed of the first air intake volume-heat distribution cross-domain dynamic semantic matching transformation adjustment values, represents the weighted hyperparameter, It means adding by position. It means point multiplication by position. Represents the optimized cross-domain dynamic semantic matching feature vector of air flow and heat distribution.

[0044] exist ,and In the case of: ;in, Indicates the spatial structure value of the first ventilation volume-heat distribution cross-domain dynamic semantic matching, Indicates the spatial structure value of the second ventilation volume-heat distribution cross-domain dynamic semantic matching, The first character representing the cross-domain dynamic semantic matching feature vector of the ventilation volume and heat distribution eigenvalues, represents the length of the cross-domain dynamic semantic matching feature vector of the airflow-heat distribution, Indicates the second ventilation volume-heat distribution cross-domain dynamic semantic matching transformation adjustment value, Represents the first ventilation volume-heat distribution cross-domain dynamic semantic matching transformation adjustment value.

[0045] That is, with respect to the spatial structure information of the feature set of the cross-domain dynamic semantic matching feature vector of air volume-heat distribution in the high-dimensional space, the scale-based frame transformation of each eigenvalue of the cross-domain dynamic semantic matching feature vector of air volume-heat distribution is performed by taking the class norm space structured representation of the cross-domain dynamic semantic matching feature vector of air volume-heat distribution as the reference window, and the spatial structure-based frame attention weight adjustment of each eigenvalue of the cross-domain dynamic semantic matching feature vector of air volume-heat distribution is implemented to ensure the spatial transformation invariance of the cross-domain dynamic semantic matching feature vector of air volume-heat distribution under the interaction of feature space, thereby improving the perception effect of the aggregated semantics of the cross-domain dynamic semantic matching feature vector of air volume-heat distribution on the changes in detailed semantics, thereby improving the expression effect of the cross-domain dynamic semantic matching feature vector of air volume-heat distribution, and improving the accuracy of the control result obtained by passing it through the classifier-based high-temperature fan air volume controller. In this way, it is possible to quickly respond to changes in thermal distribution and air volume, and at the same time more comprehensively understand the subtle dynamic changes in parameters during the combustion process, so as to accurately control the air volume and ensure that the combustion process runs under optimal conditions, thus achieving more intelligent waste tire pyrolysis production control.

[0046] In summary, the intelligent control method for the waste tire cracking production line based on the embodiment of the present application is explained, which uses artificial intelligence-based image analysis and data processing technology to extract and optimize the semantic features of the time series data of the thermal distribution diagram in the decomposition furnace, and time-series association of the time series data of the air volume of the high-temperature fan, so as to adaptively control the air volume of the high-temperature fan at the next time point according to the cross-modal dynamic matching features between the optimized thermal distribution semantic features and the air volume time series features. In this way, it is possible to quickly respond to changes in thermal distribution and air volume, so as to accurately control the size of the air volume, thereby achieving more intelligent waste tire cracking production control.

[0047] The present application provides an intelligent control system for a waste tire cracking production line, comprising: a thermodynamic distribution map data acquisition module, used to obtain the time series of the thermodynamic distribution map in the decomposition furnace; a high-temperature fan air volume data acquisition module, used to obtain the time series of the air volume of the high-temperature fan; a thermodynamic distribution map semantic feature extraction module, used to pass the time series of the thermodynamic distribution map in the decomposition furnace through a thermodynamic distribution semantic feature extractor based on a void convolutional neural network model to obtain a time series of thermodynamic distribution map semantic feature vectors; a thermodynamic distribution map semantic feature selection and optimization module, used to pass the time series of the thermodynamic distribution map semantic feature vectors through a feature sequence optimization module based on feature significance distribution to obtain a time series of thermodynamic distribution map semantic feature vectors after selection and optimization; an air volume temporal correlation feature extraction module, used to extract the time series of the thermodynamic distribution map semantic feature vectors through a feature sequence optimization module based on feature significance distribution to obtain a time series of the thermodynamic distribution map semantic feature vectors after selection and optimization; After arranging the time series of the air volume as an air volume time series input vector according to the time dimension, the air volume time series input vector is passed through an air volume time series encoder based on a Bi-LSTM model to obtain an air volume time series associated feature vector; an air volume-thermal distribution cross-domain dynamic semantic matching feature generation module is used to pass the air volume time series associated feature vector and the time series of the selected optimized thermal distribution map semantic feature vector through a cross-domain dynamic gating matching module based on local key matrix association to obtain an air volume-thermal distribution cross-domain dynamic semantic matching feature vector; a high-temperature fan air volume adjustment result generation module is used to obtain a control result based on the air volume-thermal distribution cross-domain dynamic semantic matching feature vector, and the control result is used to indicate whether the air volume of the high-temperature fan at the next time point should increase, decrease or remain unchanged.

[0048] 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 for the waste tire pyrolysis production line have been described in detail above. Figure 1 to Figure 2 The intelligent control method for the waste tire pyrolysis production line is described in detail, and therefore, its repeated description will be omitted.

[0049] In summary, the intelligent control system for the waste tire cracking production line based on the embodiment of the present application is explained, which uses artificial intelligence-based image analysis and data processing technology to extract and optimize the semantic features of the time series data of the thermal distribution diagram in the decomposition furnace, and performs time series association on the time series data of the air volume of the high-temperature fan, so as to adaptively control the air volume of the high-temperature fan at the next time point according to the cross-modal dynamic matching features between the optimized thermal distribution semantic features and the air volume time series features. In this way, it is possible to quickly respond to changes in thermal distribution and air volume, so as to accurately control the size of the air volume, thereby achieving more intelligent waste tire cracking production control.

[0050] In other embodiments of the present application, a waste tire pyrolysis production line is also provided. It first crushes the waste tires into rubber blocks less than 30 mm to ensure that the materials can be evenly heated in the subsequent pyrolysis process and improve the pyrolysis efficiency. Subsequently, these rubber blocks are continuously conveyed by a chain plate feeder, a conveyor, a distributor and a screw feeder and fed into the pyrolysis host. The pyrolysis host is the core equipment of the entire pyrolysis process, and it operates under oxygen-free (oxygen-poor) conditions. In order to control the pyrolysis process, the heating system accurately controls the temperature of the pyrolysis host, and the material undergoes pyrolysis and cracking reactions at a set temperature (440±20℃). This temperature range is carefully selected to maximize the pyrolysis efficiency and product quality. At the set temperature, the high molecular polymers in the waste tires will decompose to produce small molecular oil, gas and solid carbon black products, which can be further processed or directly used in industrial production, such as fuel oil, chemical raw materials or supplementary materials for the rubber industry.

Claims

1. An intelligent control method for a waste tire cracking production line, comprising: crushing waste tires to obtain waste tire rubber blocks; spraying a liquid catalytic cracking oxidant on the surface of the waste tire rubber blocks, and performing a homogenization treatment to obtain a modified fuel; sending the modified fuel to a decomposition furnace, so as to perform suspended catalytic cracking and catalytic oxidation flameless combustion on the modified fuel through a bottom temperature field in the decomposition furnace to obtain a mineralizer raw material; wherein, The modified fuel is sent to a decomposition furnace, and the modified fuel is subjected to suspended catalytic cracking and catalytic oxidation flameless combustion through the bottom temperature field in the decomposition furnace to obtain a mineralizer raw material, characterized in that it includes: obtaining a time series of a thermodynamic distribution diagram in the decomposition furnace; obtaining a time series of an air volume of a high-temperature fan; passing the time series of the thermodynamic distribution diagram in the decomposition furnace through a thermodynamic distribution semantic feature extractor based on a void convolutional neural network model to obtain a time series of a thermodynamic distribution diagram semantic feature vector; passing the time series of the thermodynamic distribution diagram semantic feature vector through a feature sequence optimization module based on feature significance distribution to obtain a time series of a thermodynamic distribution diagram semantic feature vector selected and optimized. sequence; after arranging the time series of the air volume as an air volume time series input vector according to the time dimension, the air volume time series input vector is passed through an air volume time series encoder based on a Bi-LSTM model to obtain an air volume time series associated feature vector; the air volume time series associated feature vector and the time series of the selected optimized thermal distribution map semantic feature vector are passed through a cross-domain dynamic gating matching module based on local key matrix association to obtain an air volume-thermal distribution cross-domain dynamic semantic matching feature vector; based on the air volume-thermal distribution cross-domain dynamic semantic matching feature vector, a control result is obtained, and the control result is used to indicate whether the air volume of the high-temperature fan at the next time point should increase, decrease or remain unchanged.

2. The intelligent control method for the waste tire pyrolysis production line according to claim 1, characterized in that: The time series of the semantic feature vectors of the thermal distribution map is passed through a feature sequence optimization module based on feature significance distribution to obtain the time series of the semantic feature vectors of the thermal distribution map after selection and optimization, including: calculating the feature energy distribution spectrum vectors of each semantic feature vector of the thermal distribution map in the time queue of the semantic feature vectors of the thermal distribution map to obtain a sequence of semantic feature energy distribution spectrum vectors of the thermal distribution map; calculating the positional mean vector of the sequence of semantic feature energy distribution spectrum vectors of the thermal distribution map as the center vector of the semantic energy spectrum distribution of the thermal distribution map; calculating the energy distribution spectrum span factor between the center vector of the semantic energy spectrum distribution of the thermal distribution map and each semantic feature energy distribution spectrum vector of the thermal distribution map in the sequence of semantic feature energy distribution spectrum vectors of the thermal distribution map to obtain a sequence of semantic energy distribution spectrum span factors of the thermal distribution map; based on the comparison between each semantic energy distribution spectrum span factor of the thermal distribution map in the sequence of semantic energy distribution spectrum span factors of the thermal distribution map and a preset threshold, determining the time series of the semantic feature vector of the thermal distribution map after selection and optimization sequence; wherein, the energy distribution spectrum span factor between the central vector of the semantic energy spectrum distribution of the thermal distribution map and each thermal distribution map semantic feature energy distribution spectrum vector in the sequence of the semantic feature energy distribution spectrum vector of the thermal distribution map is calculated to obtain a sequence of the semantic energy distribution spectrum span factors of the thermal distribution map, including: calculating the positional difference between the central vector of the semantic energy spectrum distribution of the thermal distribution map and the semantic feature energy distribution spectrum vector of the thermal distribution map to obtain the semantic differential feature vector of the thermal distribution map; calculating the covariance matrix between the central vector of the semantic energy spectrum distribution of the thermal distribution map and the semantic feature energy distribution spectrum vector of the thermal distribution map, and calculating the inverse of the covariance matrix to obtain the semantic inverse covariance matrix of the thermal distribution map; calculating the transposed vector of the semantic differential feature vector of the thermal distribution map and the product of the semantic inverse covariance matrix of the thermal distribution map and the semantic differential feature vector of the thermal distribution map to obtain the semantic distribution value of the thermal distribution map; calculating the square root of the semantic distribution value of the thermal distribution map to obtain the semantic energy distribution spectrum span factor of the thermal distribution map.

3. The intelligent control method for the waste tire pyrolysis production line according to claim 2, characterized in that: Calculating the characteristic energy distribution spectrum vector of each thermal distribution map semantic feature vector in the time queue of the thermal distribution map semantic feature vector to obtain a sequence of thermal distribution map semantic feature energy distribution spectrum vectors, including: respectively calculating the characteristic distribution energy synergy representation vector between each thermal distribution map semantic feature vector and other thermal distribution map semantic feature vectors in the time queue of the thermal distribution map semantic feature vector to obtain a sequence of multiple thermal distribution map semantic feature distribution energy synergy representation vectors; calculating the characteristic distribution energy synergy factor vector of each sequence of thermal distribution map semantic feature distribution energy synergy representation vectors in the sequence of multiple thermal distribution map semantic feature distribution energy synergy representation vectors to obtain a sequence of thermal distribution map semantic feature energy distribution spectrum vectors.

4. The intelligent control method for the waste tire pyrolysis production line according to claim 3, characterized in that: The feature distribution energy collaborative representation vectors between each of the thermal distribution map semantic feature vectors and other thermal distribution map semantic feature vectors in the time queue of the thermal distribution map semantic feature vector are calculated respectively to obtain a sequence of multiple thermal distribution map semantic feature distribution energy collaborative representation vectors, including: extracting a predetermined thermal distribution map semantic feature vector from the time queue of the thermal distribution map semantic feature vector; performing element-by-element maximum value extraction on the predetermined thermal distribution map semantic feature vector and other thermal distribution map semantic feature vectors respectively to obtain a sequence of thermal distribution map semantic feature distribution energy collaborative representation vectors corresponding to the predetermined thermal distribution map semantic feature vector.

5. The intelligent control method for the waste tire pyrolysis production line according to claim 4, characterized in that: Calculate the feature distribution energy synergy factor vector of each sequence of the thermal distribution map semantic feature distribution energy synergy representation vectors in the sequence of the multiple thermal distribution map semantic feature distribution energy synergy representation vectors to obtain the sequence of the thermal distribution map semantic feature energy distribution spectrum vectors, including: respectively extracting the maximum value of each thermal distribution map semantic feature distribution energy synergy representation vector in the sequence of the thermal distribution map semantic feature distribution energy synergy representation vector to obtain the sequence of thermal distribution map semantic feature maximum values; respectively calculate the mean and variance of each thermal distribution map semantic feature distribution energy synergy representation vector to obtain the sequence of thermal distribution map semantic feature means and the sequence of thermal distribution map semantic feature variances; positionally add the sequence of thermal distribution map semantic feature variances and preset hyperparameters to obtain the sequence of thermal distribution map semantic feature first energy distribution coefficients; calculate the sequence of thermal distribution map semantic feature maximum values ​​and the sequence of thermal distribution map semantic feature variances; The square of the difference between the maximum value of the thermal distribution map semantic feature and the mean value of the thermal distribution map semantic feature corresponding to each group in the sequence of the mean values ​​of the thermal distribution map semantic features is calculated to obtain a sequence of thermal distribution map semantic difference values; the sequence of the thermal distribution map semantic feature variances is multiplied by constant two by position to obtain a sequence of modulated thermal distribution map semantic feature variances; the thermal distribution map semantic difference values ​​and the modulated thermal distribution map semantic feature variances corresponding to each group in the sequence of the thermal distribution map semantic difference values ​​and the sequence of the modulated thermal distribution map semantic feature variances are added to the preset hyperparameter to obtain a sequence of the second energy distribution coefficients of the thermal distribution map semantic features; the sequence of the first energy distribution coefficients of the thermal distribution map semantic features is divided by the sequence of the second energy distribution coefficients of the thermal distribution map semantic features by position to obtain the thermal distribution map semantic feature energy distribution spectrum vector composed of multiple feature distribution energy synergy factors.

6. The intelligent control method for the waste tire pyrolysis production line according to claim 5, characterized in that: The time series of the airflow volume temporal association feature vector and the selected optimized thermal distribution map semantic feature vector are passed through a cross-domain dynamic gating matching module based on local key matrix association to obtain the airflow volume-thermal distribution cross-domain dynamic semantic matching feature vector, including: linearly transforming each selected optimized thermal distribution map semantic feature vector in the time series of the selected optimized thermal distribution map semantic feature vector to obtain the thermal distribution initial key matrix, wherein each thermal distribution row vector in the thermal distribution initial key matrix is ​​each linearly transformed selected optimized thermal distribution map semantic feature vector; calculating the feature vectors between the airflow volume temporal association feature vector and each thermal distribution row vector in the thermal distribution initial key matrix. Distribute difference energy coefficients to obtain a set of pull volume-thermal distribution characteristic distribution difference energy coefficients; compare each pull volume-thermal distribution characteristic distribution difference energy coefficient in the set of pull volume-thermal distribution characteristic distribution difference energy coefficients with a predetermined threshold, and in response to the pull volume-thermal distribution characteristic distribution difference energy coefficient being less than the predetermined threshold, use the mean vector between the thermal distribution row vector corresponding to the pull volume-thermal distribution characteristic distribution difference energy coefficient and the pull volume time series associated feature vector as the updated thermal distribution row vector to obtain an updated thermal distribution key matrix; use the value embedding matrix to process the pull volume time series associated feature vector to obtain the pull volume time series associated value vector; use the pull volume The time series correlation feature vector is used as the query vector, each updated thermal distribution row vector in the updated thermal distribution key matrix is ​​used as the key vector, and the air volume time series correlation value vector is input into the cross-domain attention gate query module based on the Transformer structure to obtain a sequence of air volume-thermal distribution cross-domain query attention vectors; the positional mean vector of the sequence of air volume-thermal distribution cross-domain query attention vectors is calculated to obtain the air volume-thermal distribution cross-domain dynamic semantic matching feature vector; wherein, the air volume time series correlation feature vector is used as the query vector, each updated thermal distribution row vector in the updated thermal distribution key matrix is ​​used as the key vector, and the air volume time series correlation value vector is input into the cross-domain attention gate query module based on the Transformer structure to obtain a sequence of air volume-thermal distribution cross-domain query attention vectors; A cross-domain attention gate query module with an r structure is used to obtain a sequence of cross-domain query attention vectors of airflow-heat distribution, including: calculating the product between the query vector and the transposed vector of the updated thermal distribution row vector to obtain a query-update thermal distribution association matrix, and then dividing the query-update thermal distribution association matrix by the square root of the length of the updated thermal distribution row vector by position to obtain a query-update thermal distribution modulation matrix; inputting the query-update thermal distribution modulation matrix into a softmax function to obtain a query-update thermal distribution activation matrix; multiplying the query-update thermal distribution activation matrix by the airflow temporal association value vector to obtain the cross-domain query attention vector of the airflow-heat distribution.

7. The intelligent control method for the waste tire pyrolysis production line according to claim 6, characterized in that: Calculate the characteristic distribution difference energy coefficient between the time series associated characteristic vector of the pull air volume and each thermal distribution row vector in the initial key matrix of the thermal distribution to obtain a set of characteristic distribution difference energy coefficients of the pull air volume-thermal distribution, including: calculate the interaction vector between the time series associated characteristic vector of the pull air volume and each thermal distribution row vector in the initial key matrix of the thermal distribution to obtain a set of pull air volume-thermal distribution interaction vectors; calculate the characteristic distribution difference energy coefficient of each pull air volume-thermal distribution interaction vector in the set of pull air volume-thermal distribution interaction vectors to obtain a set of characteristic distribution difference energy coefficients of the pull air volume-thermal distribution; wherein, calculate the characteristic distribution difference energy coefficient between the time series associated characteristic vector of the pull air volume and each thermal distribution row vector in the initial key matrix of the thermal distribution Distributing the interaction vectors between the row vectors to obtain a set of airflow-heat distribution interaction vectors, including: respectively calculating the position difference, position point multiplication and position addition between the airflow time series correlation feature vector and the heat distribution row vector to obtain the airflow-heat distribution differential vector, the airflow-heat distribution point product vector and the airflow-heat distribution sum vector; cascading the airflow-heat distribution differential vector, the airflow-heat distribution point product vector and the airflow-heat distribution sum vector, and performing one-dimensional convolution encoding to obtain a multi-dimensional fusion vector of the airflow-heat distribution; performing maximum pooling processing based on a local window on the multi-dimensional fusion vector of the airflow-heat distribution to obtain the airflow-heat distribution interaction vector.

8. The intelligent control method for the waste tire pyrolysis production line according to claim 7, characterized in that: Calculate the characteristic distribution difference energy coefficient of each air pull volume-thermal distribution interaction vector in the set of the air pull volume-thermal distribution interaction vectors to obtain the set of the air pull volume-thermal distribution characteristic distribution difference energy coefficients, including: extracting the maximum value of the air pull volume-thermal distribution interaction vector respectively to obtain the air pull volume-thermal distribution interaction maximum value; respectively calculate the mean and variance of the air pull volume-thermal distribution interaction vector to obtain the air pull volume-thermal distribution interaction mean and the air pull volume-thermal distribution interaction variance; add the air pull volume-thermal distribution interaction variance and the hyperparameter to obtain the first difference energy coefficient of the air pull volume-thermal distribution interaction; calculate the air pull volume-thermal distribution interaction vector. The square of the difference between the maximum value of the pull air volume-thermal distribution interaction and the mean value of the pull air volume-thermal distribution interaction is obtained to obtain the pull air volume-thermal distribution interaction difference value; the pull air volume-thermal distribution interaction variance is multiplied by constant two to obtain the modulated pull air volume-thermal distribution interaction variance; the pull air volume-thermal distribution interaction difference value and the modulated pull air volume-thermal distribution interaction variance are added to the hyperparameter to obtain the second difference energy coefficient of the pull air volume-thermal distribution interaction; the first difference energy coefficient of the pull air volume-thermal distribution interaction is divided by the second difference energy coefficient of the pull air volume-thermal distribution interaction to obtain the pull air volume-thermal distribution characteristic distribution difference energy coefficient.

9. The intelligent control method for the waste tire pyrolysis production line according to claim 8, characterized in that: Based on the cross-domain dynamic semantic matching feature vector of air volume and thermal distribution, a control result is obtained, including: passing the cross-domain dynamic semantic matching feature vector of air volume and thermal distribution through a high-temperature fan air volume controller based on a classifier to obtain the control result.

10. An intelligent control system for a waste tire pyrolysis production line, characterized in that: include: A thermodynamic distribution diagram data acquisition module, used to obtain a time series of the thermodynamic distribution diagram in the decomposition furnace; A high-temperature fan air volume data acquisition module, used to obtain the time series of the air volume of the high-temperature fan; a thermal distribution map semantic feature extraction module, used to pass the time series of the thermal distribution map in the decomposition furnace through a thermal distribution semantic feature extractor based on a void convolutional neural network model to obtain a time series of thermal distribution map semantic feature vectors; a thermal distribution map semantic feature selection and optimization module, used to pass the time series of the thermal distribution map semantic feature vectors through a feature sequence optimization module based on feature significance distribution to obtain a time series of thermal distribution map semantic feature vectors after selection and optimization; an air volume time series association feature extraction module, used to arrange the time series of the air volume into an air volume time series input vector according to the time dimension, and then pass the air volume time series input vector through an air volume time series encoder based on a Bi-LSTM model to obtain an air volume time series association feature vector; A cross-domain dynamic semantic matching feature generation module for airflow-heat distribution is used to obtain a cross-domain dynamic semantic matching feature vector for airflow-heat distribution by passing the time series of the airflow time series association feature vector and the selected optimized heat distribution map semantic feature vector through a cross-domain dynamic gating matching module based on local key matrix association; The high-temperature fan air volume adjustment result generation module is used to obtain the control result based on the cross-domain dynamic semantic matching feature vector of the air volume-thermal distribution. The control result is used to indicate whether the air volume of the high-temperature fan at the next time point should increase, decrease or remain unchanged.

Citation Information

Patent Citations

  • A method for suspension catalytic pyrolysis and catalytic oxidation flameless combustion of waste tires

    CN107129168B

  • Waste and old tire suspension catalytic cracking and catalytic oxidation flameless combustion method

    CN107129168A

  • Temperature control system and method for roller fixation machine

    CN115951584A

  • Coal-fired power plant boiler combustion optimization method and control system

    CN117308076A

  • Goaf hidden fire source time sequence prediction method and system based on AdaBoost-LSTM

    CN118211127A

Cited By

  • Multi-dimensional power failure accurate studying and judging method

    CN119226978A

  • Children psychological counseling auxiliary system and method based on large language model

    CN119418870A