Multi-modal data driven fly ash dioxin low-temperature pyrolysis optimization control method and system

Abstract

The invention relates to a multi-modal data-driven fly ash dioxin low-temperature pyrolysis optimization control method and system, and the method comprises the steps: constructing a multi-modal data multi-stream multi-scale feature sharing neural network model, and simulating a fly ash dioxin low-temperature pyrolysis process; step 2, optimizing the multi-stream multi-scale feature sharing neural network model through a data driving mode; inputting current atmosphere data, fly ash feeding data and heat treatment process data into the multi-stream multi-scale feature shared neural network model to obtain an estimated value of the dioxin low-temperature pyrolysis rate; on the basis of the current estimated value of the dioxin low-temperature pyrolysis rate, the current atmosphere and the fly ash feeding condition are kept unchanged, and the fly ash dioxin low-temperature pyrolysis is optimally controlled through adjustment and control in the heat treatment process; according to the method, modeling is carried out on the fly ash dioxin pyrolysis process in a data modeling mode to adaptively adjust the pyrolysis process, so that the pyrolysis rate of dioxin is increased.

Description

Technical Field

[0001] The invention relates to the technical field of fly ash dioxin low-temperature pyrolysis optimization control, and in particular to a multi-modal data-driven fly ash dioxin low-temperature pyrolysis optimization control method and system. Background Art

[0002] With the annual increase in the amount of waste incineration, dioxins in the fly ash produced by incineration have become the main problem restricting the promotion and application of waste incineration technology. Low-temperature pyrolysis technology has gradually become one of the main technologies for reducing the dioxin content in fly ash, which can effectively promote pollution prevention and control and fly ash resource utilization. However, fly ash low-temperature pyrolysis technology is a complex system with multiple factors coupled with atmosphere, fly ash feed, and heat treatment process. Its operating conditions and pyrolysis efficiency are variable and difficult to control accurately. Modeling and analysis of the pyrolysis process is the key to improving process control and dioxin pyrolysis efficiency. In the early days, the low-temperature pyrolysis process of dioxins mostly relied on mechanism modeling and experimental analysis. On the one hand, this type of technology is difficult to accurately simulate the complex low-temperature pyrolysis process of dioxins, and its simulation accuracy is low. On the other hand, it is difficult to apply to actual working conditions. Although it is difficult to solve this problem by regulating the pyrolysis equipment system through pilot technology, there is still a low output of a single device, which makes it difficult to achieve large-scale industrialization.

[0003] Modeling the fly ash dioxin pyrolysis process is the key to improving pyrolysis efficiency. In order to get rid of the difficulties of early mechanism modeling, data-driven pyrolysis process modeling has gradually attracted attention from academia and industry. However, the existing data-driven dioxin pyrolysis control technology only considers a single or partial key factor affecting the pyrolysis efficiency, and is unable to model multiple factors such as atmosphere, fly ash feed, and thermal treatment process. There are serious deficiencies in the promotion and accuracy of pyrolysis process modeling. Summary of the invention

[0004] The object of the present invention is to provide a multi-modal data driven fly ash dioxin low-temperature pyrolysis optimization control method and system.

[0005] In order to have a basic understanding of some aspects of the disclosed embodiments, a brief summary is given below. This summary is not intended to be a general review, nor is it intended to identify key / important components or to delineate the scope of protection of these embodiments. Its only purpose is to present some concepts in a simple form as a preface to the detailed description that follows.

[0006] According to a first aspect of an embodiment of the present invention, a multimodal data-driven fly ash dioxin low-temperature pyrolysis optimization control method is provided, the method comprising:

[0007] Step 1: Construct a neural network model with multi-modal data, multi-stream and multi-scale feature sharing to simulate the low-temperature pyrolysis process of fly ash dioxin;

[0008] Step 2: Optimize the multi-stream and multi-scale feature sharing neural network model in a data-driven way;

[0009] Step 3, inputting current atmosphere data, fly ash feed data, and thermal treatment process data into a multi-stream multi-scale feature sharing neural network model to obtain an estimated value of the low-temperature thermal decomposition rate of dioxins;

[0010] Step 4: Based on the current estimated value of the dioxin low-temperature pyrolysis rate, maintain the current atmosphere and the fly ash feeding conditions unchanged, increase the dioxin low-temperature pyrolysis rate through regulation and control of the heat treatment process, optimize the control of the fly ash dioxin low-temperature pyrolysis, and form a fly ash dioxin low-temperature pyrolysis optimization system in which the boiler body and the optimization control method match each other.

[0011] In an optional embodiment, in the multimodal data multi-stream multi-scale feature sharing neural network model constructed in step 1, the input data is multimodal data, and the output data is dioxin pyrolysis rate y; the input multimodal data includes atmosphere data, fly ash feed data, and heat treatment process data; the atmosphere data characterizes two properties of atmosphere type and atmosphere concentration; the fly ash feed data characterizes two properties of fly ash dioxin concentration and fly ash feed amount; the heat treatment process data characterizes two properties of pyrolysis temperature and heat treatment time; wherein,

[0012] The expression of atmosphere data is as follows:

[0013] x1={α,β}

[0014] In the formula, α represents the type of atmosphere, and β represents the concentration of atmosphere;

[0015] The expression for fly ash feed data is as follows:

[0016] x2={φ,δ}

[0017] In the formula, φ represents the fly ash dioxin concentration, and δ represents the fly ash feed amount;

[0018] The expression of heat treatment process data is as follows:

[0019] x3={λ,η}

[0020] Where λ represents the pyrolysis temperature and η represents the heat treatment time.

[0021] In an optional embodiment, in step 2, a neural network model with multi-modal data, multi-stream and multi-scale feature sharing simulates the pyrolysis process in a data-driven manner, and optimizes the model using training samples, including:

[0022] Inputting initial multi-modal data, the initial multi-modal data includes multi-modal data including atmosphere data, fly ash feed data, and thermal treatment process data;

[0023] Provide the first convolution layer and pooling layer to extract and calculate features of multimodal data;

[0024] Providing a second convolutional layer and a pooling layer to further extract and calculate features of the processed multimodal data;

[0025] Providing a feature sharing layer to couple the multimodal data after the first convolution layer and the second convolution layer respectively to obtain the first layer feature coupling result and the second layer feature coupling result respectively;

[0026] Provide a feature concatenation layer to concatenate the first-layer feature coupling results and the second-layer feature coupling results to form concatenated features at two convolution scales;

[0027] Providing a channel attention layer to weightedly fuse the concatenated features at two convolutional scales and then re-inputting them into the feature concatenation layer, and combining them with the atmosphere data in the initially input multimodal data to form the final features for estimating the low-temperature pyrolysis rate of fly ash dioxins;

[0028] Provide a pooling layer and a softmax layer, input the final features into the pooling layer and the softmax layer, and make an estimate of the dioxin thermal decomposition rate under the current conditions.

[0029] In an optional embodiment, the coupled data of the feature sharing layer include: atmosphere concentration β, fly ash dioxin concentration φ, fly ash feed amount δ, pyrolysis temperature λ, and heat treatment time η;

[0030] Among them, the first layer of feature coupling after the first layer of convolutional layer features is:

[0031]

[0032] in, is the first layer of convolutional features, is the first layer feature coupling weight, It is the first layer feature coupling result;

[0033] After the second convolutional layer features, the second layer feature coupling is:

[0034]

[0035] in, is the second layer convolution feature, is the second layer feature coupling weight, It is the second layer feature coupling result.

[0036] In an optional embodiment, the convolution kernel size of the first convolution layer is 1×7, the convolution kernel size of the second convolution layer is 1×5, and the size of the pooling layer is 3;

[0037] Among them, the concatenation feature expression at the convolution scale of the first convolutional layer is:

[0038]

[0039] The concatenated feature expression at the convolution scale of the second convolutional layer is:

[0040]

[0041] In an optional embodiment, the channel attention layer is used to weight the concatenated features at the two convolutional scales and then re-input into the feature concatenation layer, and combined with the atmosphere data in the initially input multimodal data to form the final feature expression for estimating the low-temperature pyrolysis rate of fly ash dioxins:

[0042]

[0043] Wherein, α represents the atmosphere type, and ω1, ω2, ω3, ω4, and ω5 are weighting coefficients.

[0044] In an optional embodiment, in step 4, based on the current estimated value of the dioxin low-temperature pyrolysis rate, the current atmosphere is maintained, the fly ash feed conditions remain unchanged, the dioxin low-temperature pyrolysis rate is estimated, and an optimization function for maximizing the dioxin low-temperature pyrolysis rate y is established through the regulation and control of the thermal treatment process. The fly ash dioxin low-temperature pyrolysis heat treatment process under the condition of the maximum dioxin low-temperature pyrolysis rate is solved through the gradient descent algorithm, wherein:

[0045] {λ * ,η *}=arg max(y)

[0046] In the formula, λ * is the optimal pyrolysis temperature, η * is the optimal heat treatment time, arg max(y) is the optimal solution for the maximum low-temperature pyrolysis rate of dioxins in the process of low-temperature pyrolysis of fly ash dioxins;

[0047] The optimal pyrolysis temperature and heat treatment time are fed back to the PLC control system of the low-temperature pyrolysis boiler body to achieve optimized control of low-temperature pyrolysis of fly ash dioxins.

[0048] According to a second aspect of an embodiment of the present invention, a multimodal data-driven fly ash dioxin low-temperature pyrolysis optimization control system is provided, comprising:

[0049] A collection module, the collection module is used to collect current atmosphere data, fly ash feed data and thermal treatment process data, and transmit multimodal data to the processing module;

[0050] A processing module is used to construct a neural network model for sharing multi-modal data, multi-stream and multi-scale features, and to optimize the model through multi-module data transmitted by the acquisition module to achieve an optimized solution for estimating the low-temperature pyrolysis emission of fly ash dioxins.

[0051] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0052] Through a multimodal data-driven fly ash dioxin low-temperature pyrolysis optimization control method and system of an embodiment of the present invention, the fly ash dioxin pyrolysis process can be modeled by data modeling, and a solution that comprehensively considers multiple factors is proposed for a complex system of multi-factor coupling in the thermal treatment process, which solves the problem that mechanism modeling and existing machine learning technologies only consider a single or a small number of operating conditions; and it is possible to optimize and adjust the thermal treatment process by maximizing the pyrolysis rate objective function under known conditions such as atmosphere and fly ash feed measurement, and it is possible to adaptively adjust the pyrolysis process according to different atmospheres and fly ash feed conditions to improve the pyrolysis rate of dioxins. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] The above contents and other objects, features and advantages of the present invention will become more apparent through the following description of the embodiments of the present invention with reference to the accompanying drawings, in which:

[0054] Figure 1 It is a schematic diagram of the structure of a fly ash dioxin low temperature pyrolysis boiler;

[0055] Figure 2 is a step diagram of the method of the present invention;

[0056] Figure 3 It is a flowchart for optimizing a multi-stream, multi-scale, shared neural network model in a data-driven manner; DETAILED DESCRIPTION

[0057] The following description and accompanying drawings fully illustrate the specific embodiments of this article so that those skilled in the art can practice them. Parts and features of some embodiments may be included in or replace parts and features of other embodiments. The scope of the embodiments of this article includes the entire scope of the claims, as well as all available equivalents of the claims. Herein, the terms "first", "second", etc. are only used to distinguish one element from another, without requiring or implying any actual relationship or order between these elements. In fact, the first element can also be called the second element, and vice versa. Moreover, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that the structure, device or equipment including a series of elements includes not only those elements, but also other elements that are not explicitly listed, or also include elements inherent to such structure, device or equipment. In the absence of more restrictions, the elements defined by the sentence "including one..." do not exclude the existence of other identical elements in the structure, device or equipment including the elements. Each embodiment is described in a progressive manner herein, and each embodiment focuses on the differences from other embodiments, and the same and similar parts between the embodiments can be referred to each other.

[0058] It should be understood that, although the various steps in the flow chart are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the figure may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these sub-steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.

[0059] Each module in the device or system of the present application can be implemented in whole or in part by software, hardware, or a combination thereof. The above modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to the above modules.

[0060] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0061] Figure 1A schematic diagram of the structure of a fly ash dioxin low-temperature pyrolysis boiler is shown in the figure, which includes a feed cover, a sealing device, a cylinder part, a heating cover, internal blades, a discharge cover, a 2-speed support device, a bottom steel frame, a transmission device, and a 1-speed support device. It should be noted that the present invention provides a multi-modal data-driven fly ash dioxin low-temperature pyrolysis optimization control method and system. In the model training process, the multi-modal operation data of the designed boiler body is collected in a targeted manner, and the model convergence is achieved in a data-driven manner, so that the control model can adapt to the designed boiler body; Figure 1 That is, a boiler structure to which the method can be applied to optimize the fly ash dioxin pyrolysis process.

[0062] According to a first aspect of an embodiment of the present invention, a multimodal data driven fly ash dioxin low temperature pyrolysis optimization control method is provided. Figure 2 The steps of the method of the present invention are shown in FIG. 1 , which comprises:

[0063] Step 1, model construction: construct a neural network model with multi-modal data multi-stream multi-scale feature sharing to simulate the low-temperature pyrolysis process of fly ash dioxin. The input data is multi-modal data, and the output data is the dioxin pyrolysis rate y. The input multi-modal data includes atmosphere data, fly ash feed data, and thermal treatment process data.

[0064] Step 2: Model training: Optimize the multi-stream and multi-scale feature sharing neural network model in a data-driven way; specifically, conduct experiments on the designed pyrolysis boiler body, collect the atmosphere of the dioxin pyrolysis process, fly ash feed, operating data of the heat treatment process and pyrolysis rate as training samples, and establish a training set with a sample size of more than 6,000 cases for training;

[0065] Step 3, model response: input normalized current atmosphere data, fly ash feed data, and thermal treatment process data into the multi-stream multi-scale feature sharing neural network model to obtain an estimated value of the low-temperature thermal decomposition rate of dioxins;

[0066] Step 4, optimal feedback control: Based on the current estimated value of dioxin low-temperature pyrolysis rate, maintain the current atmosphere and the fly ash feeding conditions unchanged, improve the dioxin low-temperature pyrolysis rate through regulation and control of the heat treatment process, optimize the control of fly ash dioxin low-temperature pyrolysis, and form a fly ash dioxin low-temperature pyrolysis optimization system in which the boiler body and the optimization control method match each other.

[0067] In an optional embodiment, in the multi-modal data multi-stream multi-scale feature sharing neural network model constructed in step 1, the atmosphere data characterizes two attributes of atmosphere type and atmosphere concentration; the fly ash feed data characterizes two attributes of fly ash dioxin concentration and fly ash feed amount; the heat treatment process data characterizes two attributes of pyrolysis temperature and heat treatment time; wherein,

[0068] The expression of atmosphere data is as follows:

[0069] x1={α,β}

[0070] In the formula, α represents the type of atmosphere, and β represents the concentration of atmosphere;

[0071] The expression for fly ash feed data is as follows:

[0072] x2={φ,δ}

[0073] In the formula, φ represents the fly ash dioxin concentration, and δ represents the fly ash feed amount;

[0074] The expression of heat treatment process data is as follows:

[0075] x3={λ,η}

[0076] Where λ represents the pyrolysis temperature and η represents the heat treatment time.

[0077] In an optional embodiment, in step 2, a neural network model for sharing multi-scale features of multi-modal data multi-streams simulates the pyrolysis process in a data-driven manner. The model realizes the sharing, fusion and identification of multi-scale features of multi-modal data multi-streams. The multi-stream multi-scale features are calculated by multi-level multi-scale convolution kernels. The model is optimized using training samples, specifically including:

[0078] Inputting initial multi-modal data, the initial multi-modal data includes multi-modal data including atmosphere data, fly ash feed data, and thermal treatment process data;

[0079] Provide the first convolution layer and pooling layer to extract and calculate features of multimodal data;

[0080] Providing a second convolutional layer and a pooling layer to further extract and calculate features of the processed multimodal data;

[0081] Providing a feature sharing layer to couple the multimodal data after the first convolution layer and the second convolution layer respectively to obtain the first layer feature coupling result and the second layer feature coupling result respectively;

[0082] Provide a feature concatenation layer to concatenate the first-layer feature coupling results and the second-layer feature coupling results to form concatenated features at two convolution scales;

[0083] Providing a channel attention layer to weightedly fuse the concatenated features at two convolutional scales and then re-inputting them into the feature concatenation layer, and combining them with the atmosphere data in the initially input multimodal data to form the final features for estimating the low-temperature pyrolysis rate of fly ash dioxins;

[0084] Provide a pooling layer and a softmax layer, input the final features into the pooling layer and the softmax layer, and make an estimate of the dioxin thermal decomposition rate under the current conditions.

[0085] In an optional embodiment, the coupled data of the feature sharing layer include: atmosphere concentration β, fly ash dioxin concentration φ, fly ash feed amount δ, pyrolysis temperature λ, and heat treatment time η;

[0086] Among them, the first layer of feature coupling after the first layer of convolutional layer features is:

[0087]

[0088] in, is the first layer of convolutional features, is the first layer feature coupling weight, It is the first layer feature coupling result;

[0089] After the second convolutional layer features, the second layer feature coupling is:

[0090]

[0091] in, is the second layer convolution feature, is the second layer feature coupling weight, It is the second layer feature coupling result.

[0092] In an optional embodiment, the convolution kernel size of the first convolution layer is 1×7, the convolution kernel size of the second convolution layer is 1×5, and the size of the pooling layer is 3;

[0093] Among them, the concatenation feature expression at the convolution scale of the first convolutional layer is:

[0094]

[0095] The concatenated feature expression at the convolution scale of the second convolutional layer is:

[0096]

[0097] In an optional embodiment, the channel attention layer is used to weight the concatenated features at the two convolutional scales and then re-input into the feature concatenation layer, and combined with the atmosphere data in the initially input multimodal data to form the final feature expression for estimating the low-temperature pyrolysis rate of fly ash dioxins:

[0098]

[0099] Wherein, α represents the atmosphere type, and ω1, ω2, ω3, ω4, and ω5 are weighting coefficients.

[0100] By collecting the atmosphere of the dioxin pyrolysis process, fly ash feed, operating data of the heat treatment process and pyrolysis rate as training samples, a training set with a sample size of more than 6,000 cases was established. After a large amount of model training, the model parameters were solidified.

[0101] In an optional embodiment, in step 4, based on the current estimated value of the dioxin low-temperature pyrolysis rate, the current atmosphere is maintained, the fly ash feed conditions remain unchanged, the dioxin low-temperature pyrolysis rate is estimated, and an optimization function for maximizing the dioxin low-temperature pyrolysis rate y is established through the regulation and control of the thermal treatment process. The fly ash dioxin low-temperature pyrolysis heat treatment process under the condition of the maximum dioxin low-temperature pyrolysis rate is solved through the gradient descent algorithm, wherein:

[0102] {λ * , η *}=arg max(y)

[0103] In the formula, λ * is the optimal pyrolysis temperature, η * is the optimal heat treatment time, arg max(y) is the optimal solution for the maximum low-temperature pyrolysis rate of dioxins in the process of low-temperature pyrolysis of fly ash dioxins;

[0104] The optimal pyrolysis temperature and heat treatment time are fed back to the PLC control system of the low-temperature pyrolysis boiler body to achieve optimized control of low-temperature pyrolysis of fly ash dioxins.

[0105] The present invention is based on accurate estimation of the pyrolysis rate of dioxins, and for existing working conditions, under the conditions of accurate measurement of atmosphere and fly ash feed, by maximizing the pyrolysis rate of dioxins, and feedback optimizing the control variables of the heat treatment process, so as to optimize and adjust the heat treatment process in order to maximize the pyrolysis rate objective function. The pyrolysis process can be adaptively adjusted according to different atmospheres and fly ash feed conditions to improve the pyrolysis rate of dioxins.

[0106] In one embodiment of the present invention, a multimodal data-driven fly ash dioxin low-temperature pyrolysis optimization control system is provided, comprising:

[0107] A collection module, the collection module is used to collect current atmosphere data, fly ash feed data and thermal treatment process data, and transmit multimodal data to the processing module;

[0108] A processing module is used to construct a neural network model for sharing multi-modal data, multi-stream and multi-scale features, and to optimize the model through multi-module data transmitted by the acquisition module to achieve an optimized solution for estimating the low-temperature pyrolysis emission of fly ash dioxins.

[0109] In this embodiment, a computer device is provided, which may be a server. The computer device includes a processor, a memory and a network interface connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store static information and dynamic information data. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, the steps in the above method embodiment are implemented.

[0110] It will be appreciated by those skilled in the art that the features described in the various embodiments of the present invention may be combined and / or combined in various ways, even if such combinations and / or combinations are not explicitly described in the present invention. In particular, without departing from the spirit and teachings of the present invention, the features described in the various embodiments of the present invention may be combined and / or combined in various ways. All of these combinations and / or combinations fall within the scope of the present invention.

Claims

1. A multimodal data driven fly ash dioxin low temperature pyrolysis optimization control method, characterized in that: The method includes: Step 1: Construct a neural network model with multi-modal data, multi-stream and multi-scale feature sharing to simulate the low-temperature pyrolysis process of fly ash dioxin; Step 2: Optimize the multi-stream and multi-scale feature sharing neural network model in a data-driven way; Step 3, inputting current atmosphere data, fly ash feed data, and thermal treatment process data into a multi-stream multi-scale feature sharing neural network model to obtain an estimated value of the low-temperature thermal decomposition rate of dioxins; Step 4: Based on the current estimated value of the dioxin low-temperature pyrolysis rate, maintain the current atmosphere and the fly ash feeding conditions unchanged, increase the dioxin low-temperature pyrolysis rate through regulation and control of the heat treatment process, optimize the control of the fly ash dioxin low-temperature pyrolysis, and form a fly ash dioxin low-temperature pyrolysis optimization system in which the boiler body and the optimization control method match each other.

2. The multimodal data driven fly ash dioxin low temperature pyrolysis optimization control method according to claim 1, characterized in that: In the multimodal data multi-stream multi-scale feature sharing neural network model constructed in step 1, the input data is multimodal data, and the output data is dioxin pyrolysis rate y; the input multimodal data includes atmosphere data, fly ash feed data, and heat treatment process data; the atmosphere data characterizes two attributes: atmosphere type and atmosphere concentration; the fly ash feed data characterizes two attributes: fly ash dioxin concentration and fly ash feed amount; the heat treatment process data characterizes two attributes: pyrolysis temperature and heat treatment time; among them, The expression of atmosphere data is as follows: x1＝{α,β} In the formula, α represents the type of atmosphere, and β represents the concentration of atmosphere; The expression for fly ash feed data is as follows: x2={φ,δ} In the formula, φ represents the fly ash dioxin concentration, and δ represents the fly ash feed amount; The expression of heat treatment process data is as follows: x3={λ,η} Where λ represents the pyrolysis temperature and η represents the heat treatment time.

3. The multimodal data driven fly ash dioxin low temperature pyrolysis optimization control method according to claim 2, characterized in that: In step 2, the neural network model with multi-modal data, multi-stream and multi-scale feature sharing simulates the pyrolysis process in a data-driven manner and optimizes the model using training samples, including: Inputting initial multi-modal data, the initial multi-modal data includes multi-modal data including atmosphere data, fly ash feed data, and thermal treatment process data; Provide the first convolution layer and pooling layer to extract and calculate features of multimodal data; Providing a second convolutional layer and a pooling layer to further extract and calculate features of the processed multimodal data; Providing a feature sharing layer to couple the multimodal data after the first convolution layer and the second convolution layer respectively to obtain the first layer feature coupling result and the second layer feature coupling result respectively; Provide a feature concatenation layer to concatenate the first-layer feature coupling results and the second-layer feature coupling results to form concatenated features at two convolution scales; Providing a channel attention layer to weightedly fuse the concatenated features at two convolutional scales and then re-inputting them into the feature concatenation layer, and combining them with the atmosphere data in the initially input multimodal data to form the final features for estimating the low-temperature pyrolysis rate of fly ash dioxins; Provide a pooling layer and a softmax layer, input the final features into the pooling layer and the softmax layer, and make an estimate of the dioxin thermal decomposition rate under the current conditions.

4. The multimodal data driven fly ash dioxin low temperature pyrolysis optimization control method according to claim 3, characterized in that: The coupled data of the feature sharing layer include: atmosphere concentration β, fly ash dioxin concentration φ, fly ash feed amount δ, pyrolysis temperature λ, and heat treatment time η; Among them, the first layer of feature coupling after the first layer of convolutional layer features is: in, is the first layer of convolutional features, is the first layer feature coupling weight, It is the first layer feature coupling result; After the second convolutional layer features, the second layer feature coupling is: in, is the second layer convolution feature, is the second layer feature coupling weight, It is the second layer feature coupling result.

5. The multimodal data driven fly ash dioxin low temperature pyrolysis optimization control method according to claim 4, characterized in that: The convolution kernel size of the first convolution layer is 1×7, the convolution kernel size of the second convolution layer is 1×5, and the size of the pooling layer is 3; Among them, the concatenation feature expression at the convolution scale of the first convolutional layer is: The concatenated feature expression at the convolution scale of the second convolutional layer is:

6. A multimodal data driven fly ash dioxin low temperature pyrolysis optimization control method according to claim 5, characterized in that: The channel attention layer is used to weight the splicing features at the two convolutional scales and then re-input them into the feature splicing layer, and combined with the atmosphere data in the initial input multimodal data to form the final feature expression for the estimation of the low-temperature pyrolysis rate of fly ash dioxin: Wherein, α represents the atmosphere type, and ω1, ω2, ω3, ω4, and ω5 are weighting coefficients.

7. The multimodal data driven fly ash dioxin low temperature pyrolysis optimization control method according to claim 1, characterized in that: In step 4, based on the current estimated value of dioxin low-temperature pyrolysis rate, the current atmosphere is maintained, and the fly ash feed conditions remain unchanged to estimate the dioxin low-temperature pyrolysis rate. Through the regulation and control of the thermal treatment process, an optimization function for maximizing the dioxin low-temperature pyrolysis rate y is established. Through the gradient descent algorithm, the fly ash dioxin low-temperature pyrolysis heat treatment process under the condition of the maximum dioxin low-temperature pyrolysis rate is solved, wherein, {l} * ,or * }=arg max(y) In the formula, λ * is the optimal pyrolysis temperature, η * is the optimal heat treatment time, arg max(y) is the optimal solution for the maximum low-temperature pyrolysis rate of dioxins in the process of low-temperature pyrolysis of fly ash dioxins; The optimal pyrolysis temperature and heat treatment time are fed back to the PLC control system of the low-temperature pyrolysis boiler body to achieve optimized control of low-temperature pyrolysis of fly ash dioxins.

8. A multimodal data driven fly ash dioxin low temperature pyrolysis optimization control system, characterized in that: include: A collection module, the collection module is used to collect current atmosphere data, fly ash feed data and thermal treatment process data, and transmit multimodal data to the processing module; A processing module is used to construct a neural network model for sharing multi-modal data, multi-stream and multi-scale features, and to optimize the model through multi-module data transmitted by the acquisition module to achieve an optimized solution for estimating the low-temperature pyrolysis emission of fly ash dioxins.

Images