Coal blending combustion expert decision-making system and method based on coal yard intelligent centralized control technology

Abstract

The embodiment of the invention discloses a coal blending combustion expert decision making system and method based on a coal yard intelligent centralized control technology, relates to the field of intelligent decision making, and adopts an AI-based data analysis and coding technology to carry out time sequence space aggregation and operation state feature extraction on a time sequence data set of boiler operation parameters. And meanwhile, semantic embedding combination is performed on the quality parameters of the first coal type and the second coal type, so that decision suggestions are automatically obtained according to continuous prompt cross-modal combined representation between the quality parameter combined characteristics of the first coal type and the second coal type and the boiler operation state multi-modal time sequence combined characteristics. Therefore, the operation state of the boiler can be monitored in real time, the optimal coal type proportion can be rapidly adjusted according to the current working condition, the high efficiency and stability of the combustion process are ensured, the subjective influence of manual judgment is reduced, the scientificity and accuracy of decision making are enhanced, and intelligent management of coal blending combustion is achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent decision-making, and more specifically, relates to an expert decision-making system and method for coal blending and burning based on intelligent centralized control technology of a coal yard. Background Art

[0002] In modern thermal power plants, coal is the main source of fuel. In order to improve combustion efficiency, reduce operating costs and reduce environmental pollution, power plants usually need to mix multiple types of coal, that is, blended coal combustion. Through scientific and reasonable blended coal combustion, the combustion process can be optimized, thermal efficiency can be improved, harmful gas emissions can be reduced, equipment service life can be extended, and stable operation of the boiler under different working conditions can be ensured. In addition, blended coal combustion can also flexibly respond to coal quality fluctuations and equipment failures, improving the reliability and economy of the overall operation.

[0003] However, traditional coal blending and combustion methods mainly rely on empirical formulas and manual judgment. Specifically, empirical formulas are usually established based on historical data and fixed rules. These formulas may perform well under specific conditions, but their accuracy and adaptability are often insufficient when faced with complex and changeable operating conditions. For example, when the coal quality changes or the boiler operating status fluctuates, the empirical formula may not be adjusted in time, resulting in decreased combustion efficiency or increased emissions. Secondly, although manual judgment can combine the operator's experience and intuition, there is obvious subjectivity and inconsistency in actual operation. Different operators may have different judgment standards and operating habits, which leads to instability and non-repeatability of coal blending and combustion schemes.

[0004] Therefore, an expert decision-making scheme for coal blending and combustion based on intelligent centralized control technology of coal yards is desired. Summary of the invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art, and discloses a coal blending and burning expert decision system and method based on coal yard intelligent centralized control technology.

[0006] In a first aspect, an embodiment of the present invention provides a coal blending and burning expert decision system based on coal yard intelligent centralized control technology, comprising:

[0007] An environmental data acquisition module, used to acquire quality parameters of the first type of coal and quality parameters of the second type of coal;

[0008] Environmental data processing and encoding module, used to obtain time series data of boiler operating parameters;

[0009] A boiler operation state encoding module is used to perform data time series spatial aggregation and operation state feature extraction on the time series data set of the boiler operation parameters to obtain a multi-modal time series joint semantic encoding feature map of the boiler operation state;

[0010] A first and second quality parameter embedding and combining module, used for embedding and combining the quality parameter of the first type of coal and the quality parameter of the second type of coal to obtain a first type of coal-second type of coal quality parameter combined embedding coding vector;

[0011] A boiler quality state joint coding module, used for performing a cross-modal combination of the first coal type-second coal type quality parameter joint embedding coding vector and the boiler operation state multimodal temporal joint semantic coding feature map with continuous prompt guidance to obtain a coal type quality-boiler state joint coding feature map;

[0012] A decision module is used to obtain a decision suggestion based on the coal quality-boiler status joint coding feature map, wherein the decision suggestion is a first ratio of the first coal type and a second ratio of the second coal type, and the sum of the first ratio and the second ratio is 1.

[0013] In some possible embodiments, the boiler operating parameters include temperature values, load values, and pressure values.

[0014] In some possible embodiments, the boiler operation status encoding module includes:

[0015] A boiler operating parameter time series data sorting unit, used for performing data time series spatial aggregation on the time series data set of the boiler operating parameters according to the time dimension and the parameter sample dimension to obtain a multimodal time series aggregation matrix of the boiler operating parameters;

[0016] The boiler operation state feature extraction unit is used to input the boiler operation parameter multimodal time series aggregation matrix into the boiler operation state feature extractor based on the deep separable convolutional neural network model to obtain the boiler operation state multimodal time series joint semantic coding feature map.

[0017] In some possible embodiments, the first and second quality parameters are embedded in a joint module, including:

[0018] A first and second coal quality parameter embedding coding unit, configured to embed the quality parameter of the first coal type and the quality parameter of the second coal type using a quality parameter embedding coding matrix to obtain a first coal quality parameter embedding coding vector and a second coal quality parameter embedding coding vector;

[0019] The first and second coal type quality parameter embedded coding vector cascade unit is used to cascade the first coal type quality parameter embedded coding vector and the second coal type quality parameter embedded coding vector to obtain the first coal type-second coal type quality parameter joint embedded coding vector.

[0020] In some possible embodiments, the boiler quality status joint encoding module includes:

[0021] A boiler operation state local feature decomposition unit, used for performing local feature decomposition along the channel dimension on the boiler operation state multimodal temporal joint semantic coding feature map to obtain a set of boiler operation state multimodal temporal joint semantic local feature matrices;

[0022] A coal quality-boiler status local query unit, used to use the first coal type-second coal type quality parameter joint embedding coding vector as a query vector and each boiler operating status multimodal temporal joint semantic local feature matrix in the set of boiler operating status multimodal temporal joint semantic local feature matrices as a key matrix, and input the query vector and the key matrix into a prompt learning network based on a converter structure to obtain a set of coal quality-boiler status local query prompt semantic coding vectors;

[0023] The boiler operating status multimodal joint unit is used to perform cross-modal mask feature aggregation on the set of the boiler operating status multimodal time series joint semantic local feature matrix based on the first coal type-second coal type quality parameter joint embedding coding vector and the set of coal type quality-boiler status local query prompt semantic coding vectors.

[0024] In some possible embodiments, the coal quality-boiler status local query unit is used to:

[0025] Calculate the similarity scores between the first coal type-second coal type quality parameter joint embedding coding vector and each row vector in the boiler operation state multimodal temporal joint semantic local feature matrix to obtain a set of coal type quality-boiler state similarity scores;

[0026] Standardizing each coal type quality-boiler state similarity score in the set of coal type quality-boiler state similarity scores to obtain a set of coal type quality-boiler state weight coefficients;

[0027] Based on the set of coal quality-boiler status weight coefficients, the weighted sum of each row vector in the boiler operating status multimodal time series joint semantic local feature matrix is ​​calculated to obtain the coal quality-boiler status local query prompt semantic encoding vector.

[0028] In some possible embodiments, the boiler operation status multi-modal combined unit is used to:

[0029] Input each coal type quality-boiler status local query prompt semantic coding vector in the set of the first coal type-second coal type quality parameter joint embedding coding vector and the coal type quality-boiler status local query prompt semantic coding vector into a cross-modal mask weaving network based on prompt information to obtain a set of quality parameter-boiler status cross-modal local feature mask weight matrices based on prompt information;

[0030] Calculate the set of the quality parameter based on prompt information-boiler state cross-modal local feature mask weight matrix and the set of the boiler operation state multimodal temporal joint semantic local feature matrix, and obtain the set of coal type quality-boiler state cross-modal local granularity significant interaction matrices by multiplying the position points between each corresponding quality parameter based on prompt information-boiler state cross-modal local feature mask weight matrix and the boiler operation state multimodal temporal joint semantic local feature matrix;

[0031] The set of the coal quality-boiler status cross-modal local granularity significant interaction matrices is feature aggregated to obtain the coal quality-boiler status joint coding feature map.

[0032] In some possible embodiments, the decision module is used to: input the coal quality-boiler status joint coding feature map into the classifier-based coal blending and combustion expert decision module to obtain the decision recommendation, and the decision recommendation is a first ratio of the first coal type and a second ratio of the second coal type, and the sum of the first ratio and the second ratio is 1.

[0033] In a second aspect, an embodiment of the present invention provides an expert decision method for coal blending and burning based on intelligent centralized control technology of a coal yard, comprising:

[0034] Obtaining quality parameters of a first type of coal and quality parameters of a second type of coal;

[0035] Obtain time series data of boiler operating parameters;

[0036] Performing data time series spatial aggregation and operation state feature extraction on the time series data set of the boiler operation parameters to obtain a multimodal time series joint semantic coding feature map of the boiler operation state;

[0037] Embedding and combining the quality parameter of the first type of coal and the quality parameter of the second type of coal to obtain a first type of coal-second type of coal quality parameter joint embedded coding vector;

[0038] Performing a cross-modal combination of the first coal type-second coal type quality parameter joint embedding coding vector and the boiler operation status multimodal temporal joint semantic coding feature map with continuous prompt guidance to obtain a coal type quality-boiler status joint coding feature map;

[0039] Based on the coal quality-boiler status joint coding feature diagram, a decision recommendation is obtained, and the decision recommendation is a first ratio of the first coal type and a second ratio of the second coal type, and the sum of the first ratio and the second ratio is 1.

[0040] In some possible embodiments, performing data temporal spatial aggregation and operating state feature extraction on the time series data set of the boiler operating parameters to obtain a multimodal temporal joint semantic coding feature map of the boiler operating state includes:

[0041] Performing data time series spatial aggregation on the time series data set of the boiler operating parameters according to the time dimension and the parameter sample dimension to obtain a multimodal time series aggregation matrix of the boiler operating parameters;

[0042] The boiler operating parameter multimodal time series aggregation matrix is ​​input into a boiler operating state feature extractor based on a deep separable convolutional neural network model to obtain a multimodal time series joint semantic encoding feature map of the boiler operating state.

[0043] Compared with the prior art, the coal blending and burning expert decision system and method based on the coal yard intelligent centralized control technology provided by the embodiment of the present invention adopts AI-based data analysis and coding technology to perform time-series spatial aggregation and operation status feature extraction on the time series data set of boiler operation parameters, and at the same time, semantic embedding and union of the quality parameters of the first type of coal and the second type of coal are respectively performed, so as to automatically obtain decision suggestions based on the continuous prompt cross-modal joint representation between the joint features of the first type of coal and the second type of coal quality parameters and the multimodal time series joint features of the boiler operation status. In this way, the operating status of the boiler can be monitored in real time, and the optimal coal ratio can be quickly adjusted according to the current operating conditions to ensure the efficiency and stability of the combustion process, reduce the subjective influence of human judgment, thereby enhancing the scientificity and accuracy of the decision, and realizing the intelligent management of coal blending and burning. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The above and other features, advantages and aspects of the embodiments of the present invention will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the accompanying drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and that components and elements are not necessarily drawn to scale.

[0045] Figure 1 : is a system block diagram of a coal blending and combustion expert decision system based on coal yard intelligent centralized control technology according to an embodiment of the present invention;

[0046] Figure 2 Schematic diagram of data flow of the coal blending and combustion expert decision-making system based on the coal yard intelligent centralized control technology according to an embodiment of the present invention;

[0047] Figure 3 A block diagram of a boiler operation status encoding module in a coal blending and combustion expert decision system based on a coal yard intelligent centralized control technology according to an embodiment of the present invention;

[0048] Figure 4A block diagram of a boiler quality status joint encoding module in a coal blending and combustion expert decision system based on a coal yard intelligent centralized control technology according to an embodiment of the present invention;

[0049] Figure 5 The present invention is a flowchart of an expert decision-making method for coal blending and combustion based on intelligent centralized control technology of a coal yard according to an embodiment of the present invention. DETAILED DESCRIPTION

[0050] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. Based on the described embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work are within the scope of protection of the present invention.

[0051] Unless otherwise specified, the technical terms or scientific terms used in the embodiments of the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. "Including" or "comprising" used in the embodiments of the present invention neither limit the shapes, numbers, steps, actions, operations, components, originals and / or their groups mentioned, nor exclude the appearance or addition of one or more other different shapes, numbers, steps, actions, operations, components, originals and / or their groups, or the addition of these. In addition, the terms "first" and "second" are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number and order of the indicated technical features. Thus, the features defined as "first" and "second" can explicitly or implicitly include one or more of the features. In the description of the embodiments of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined.

[0052] Unless otherwise specifically stated, the relative arrangement of the components and steps, numerical expressions and numerical values ​​set forth in these embodiments do not limit the scope of the present invention. At the same time, it should be understood that for ease of description, the sizes of the various parts shown in the drawings are not drawn according to the actual proportional relationship, and the techniques, methods and devices known to ordinary technicians in the relevant fields may not be discussed in detail, but where appropriate, the techniques, methods and devices shown should be considered as part of the authorized specification. In all examples shown and discussed here, any specific other examples may have different values. It should be noted that similar symbols and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further discussed in subsequent drawings.

[0053] In the description of the embodiments of the present invention, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In the embodiments of the present invention, the schematic representations of the above terms do not necessarily refer to the same embodiments or examples. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in the embodiments of the present invention and the features of the different embodiments or examples, without contradiction.

[0054] In modern thermal power plants, coal is the main fuel. The technology of coal blending and combustion can be used to improve combustion efficiency, reduce costs, reduce pollution, and ensure the stable operation of boilers under different operating conditions. Specifically, this technology improves thermal efficiency, reduces harmful gas emissions, and extends equipment life by optimizing the combustion process. However, traditional coal blending and combustion methods rely on empirical formulas and manual judgment. These methods often lack sufficient accuracy and adaptability when faced with changes in coal quality and fluctuations in boiler operating conditions, resulting in reduced combustion efficiency or increased emissions. In addition, the subjectivity and inconsistency of manual judgment also affect the stability and repeatability of coal blending and combustion schemes.

[0055] In view of the above technical problems, the technical concept of the present invention is to obtain the quality parameters of the first type of coal and the quality parameters of the second type of coal, and obtain the time series data set of boiler operating parameters (temperature value, load value and pressure value), and use AI-based data analysis and coding technology to perform time-series spatial aggregation and operation status feature extraction on the time series data set of boiler operating parameters, and at the same time, semantic embedding and union are performed on the quality parameters of the first type of coal and the second type of coal respectively, so as to automatically obtain decision suggestions based on the continuous prompt cross-modal joint representation between the joint features of the first type of coal-second type of coal quality parameters and the multimodal time series joint features of the boiler operating status. In this way, the operating status of the boiler can be monitored in real time, and the optimal coal ratio can be quickly adjusted according to the current operating conditions to ensure the efficiency and stability of the combustion process, reduce the subjective influence of human judgment, thereby enhancing the scientificity and accuracy of the decision, and realizing the intelligent management of coal blending and combustion.

[0056] Figure 1 4 is a system block diagram of a coal blending and combustion expert decision-making system based on coal yard intelligent centralized control technology according to an embodiment of the present invention. Figure 2 This is a data flow diagram of the coal blending and burning expert decision system based on the coal yard intelligent centralized control technology according to an embodiment of the present invention. Figure 1 and Figure 2As shown, in the coal blending expert decision system 100 based on the intelligent centralized control technology of the coal yard, it includes: an environmental data acquisition module 110, which is used to obtain the quality parameters of the first coal type and the quality parameters of the second coal type; an environmental data processing and encoding module 120, which is used to obtain the time series data of the boiler operation parameters; a boiler operation state encoding module 130, which is used to perform data time series space aggregation and operation state feature extraction on the time series data set of the boiler operation parameters to obtain a multi-modal time series joint semantic coding feature map of the boiler operation state; a first and second quality parameter embedding joint module 140, which is used to perform data time series space aggregation and operation state feature extraction on the time series data set of the boiler operation parameters to obtain a multi-modal time series joint semantic coding feature map of the boiler operation state; and a first and second quality parameter embedding joint module 140, which is used to perform data time series space aggregation and operation state feature extraction on the time series data set of the boiler operation parameters to obtain a multi-modal time series joint semantic coding feature map of the boiler operation state. The first coal type and the second coal type quality parameter are jointly embedded and combined to obtain a first coal type-second coal type quality parameter joint embedded coding vector; the boiler quality status joint coding module 150 is used to perform a cross-modal combination of the first coal type-second coal type quality parameter joint embedded coding vector and the boiler operation status multimodal temporal joint semantic coding feature map with continuous prompt guidance to obtain a coal type quality-boiler status joint coding feature map; the decision module 160 is used to obtain a decision suggestion based on the coal type quality-boiler status joint coding feature map, and the decision suggestion is a first proportion of the first coal type and a second proportion of the second coal type, and the sum of the first proportion and the second proportion is 1.

[0057] In an embodiment of the present invention, the environmental data acquisition module 110 and the environmental data processing and encoding module 120 are respectively used to obtain the quality parameters of the first coal type and the quality parameters of the second coal type and the time series data for obtaining the boiler operation parameters. It should be understood that the quality parameters of the coal type specifically refer to various indicators used to describe and evaluate the quality and combustion characteristics of coal, including but not limited to calorific value, ash content, moisture, volatile matter, sulfur content, etc. The difference in quality parameters of different coal types will directly affect their combustion characteristics and thermal efficiency. By obtaining the quality parameters of the two coal types, their combustion characteristics and emission characteristics can be understood, and then the ratio can be optimized to achieve higher combustion efficiency and lower emissions. The time series data of the boiler operation parameters specifically refers to the data set of the operating status and load conditions of the boiler at different time points, specifically including parameter indicators such as temperature, load and pressure. Boiler temperature is an important parameter that affects the combustion efficiency and safe operation of the boiler; boiler load affects the operating status and combustion characteristics of the boiler; changes in boiler pressure will affect the combustion process and the safety of the boiler. According to the changes in these parameters, the ratio of coal types can be adjusted in time to meet the needs of boiler operation. For example, when the boiler load increases, it may be necessary to increase the proportion of high calorific value coal to improve combustion efficiency. In general, by combining the quality parameters of the coal type with the boiler operating parameters, the model can identify the optimal coal ratio under specific operating conditions. This comprehensive analysis ensures that the final decision recommendation is not only based on the properties of the coal type, but also takes into account the actual operating status of the boiler, which is conducive to achieving an efficient and stable combustion process.

[0058] In the present invention, obtaining accurate quality parameters of the first coal type and the second coal type, as well as a time series data set of boiler operating parameters, is the basis for formulating a scientific and reasonable coal blending and combustion plan. The following will elaborate on the data acquisition and implementation of these two aspects:

[0059] The first step is to obtain the quality parameter data of coal. When the coal enters the site, representative coal samples are collected from the first and second coal types of different batches and sources. The collection process must follow relevant standards and specifications to ensure that the coal samples can accurately reflect the overall characteristics of the batch of coal. Note that the collected coal samples should be properly preserved to prevent them from being affected by weathering, oxidation and other factors that change their quality. After the coal samples are collected, they will enter the laboratory analysis stage. The collected coal samples are sent to a professional laboratory, where they are comprehensively and carefully analyzed. The analysis items include key quality parameters such as calorific value, ash content, moisture, volatile matter, and sulfur content. The determination of each parameter depends on precise instruments and strict analysis methods. For example, when determining the calorific value, an oxygen bomb calorimeter is used to burn the coal sample in a closed environment filled with oxygen, and the calorific value is determined by measuring the heat released during the combustion process. The ash content is determined by the high-temperature burning method. The coal sample is burned in a high-temperature furnace to constant weight, and the mass difference before and after burning is the ash content. The moisture content can be determined by drying method. The coal sample is dried to constant weight at a specific temperature. The weight lost is the moisture content. The volatile matter is determined by distillation method. The coal sample is heated under air-tight conditions to make the volatile matter in the coal sample escape. The volatile matter content is determined by measuring the amount of volatile matter. The sulfur content is determined by chemical analysis methods, such as the Eschka method or coulometric titration method. The sulfur in the coal sample is converted into a measurable substance through chemical reactions to determine the sulfur content. During the entire analysis process, the experimenters strictly follow the relevant operating procedures and standards to ensure that each measurement result is accurate and reliable. After the analysis is completed, the measurement results of each quality parameter are recorded in detail and organized into a standardized data format to form the quality parameter data sets of the first coal type and the second coal type. These data sets will serve as important basic data for subsequent coal blending decisions. In order to facilitate the management and application of coal quality parameter data, data entry and management are required. The coal quality parameter data obtained by laboratory analysis is accurately entered into the production management system of the power plant or the special coal blending management database. During the data entry process, strict data verification should be carried out to avoid data entry errors by comparing with the original experimental records and setting data rationality range checks. For example, for calorific value data, it is necessary to check whether it is within a reasonable numerical range. If there is data that is obviously deviated from the normal range, it should be verified and corrected in time. After the data entry is completed, the data is classified and managed, and stored according to multiple dimensions such as coal type, batch, sampling time, etc. For example, different folders or data tables are established according to the coal type classification, and each coal type is further subdivided according to batches, and information such as sampling time is recorded at the same time, which can facilitate subsequent query, call and analysis. In order to ensure the security and integrity of the data, the data must be backed up regularly. Off-site backup and redundant storage can be used to prevent data loss due to hardware failure, human error or other unexpected situations.

[0060] Then obtain the time series data of boiler operation parameters. First of all, the sensor installation and calibration work should be done well. High-precision temperature sensors, pressure sensors and load sensors are installed at key parts of the boiler. The installation location of the temperature sensor is selected in the area that can accurately reflect the combustion temperature in the boiler, such as the furnace outlet, superheater outlet and other parts, and it is necessary to ensure that the sensor can adapt to the high temperature environment in the boiler, and its high temperature resistance and measurement accuracy must meet the requirements. The pressure sensor is installed at key pressure monitoring points such as steam pipes and water supply pipes, and can accurately measure the changes in different pressure ranges. There must be appropriate sensors covering the low pressure to high pressure areas. The load sensor is connected to the power system of the boiler to accurately measure the real-time load of the boiler. After these sensors are installed, they need to be calibrated regularly to ensure that their measurement accuracy always meets the requirements. The calibration process uses standard measuring instruments for comparison. For example, for temperature sensors, standard thermometers are used to calibrate at different temperature points, and the measured values ​​of the sensor are compared with the measured values ​​of the standard thermometer, the error is calculated and the output value of the sensor is corrected. For pressure sensors, standard pressure sources are used for calibration. By applying a known standard pressure to the sensor, the output of the sensor is adjusted to be consistent with the standard pressure value. For load sensors, the accuracy of load measurement is ensured by comparing and calibrating them with the power plant's electricity metering device or steam flow metering device.

[0061] After the sensor is installed and calibrated, the data acquisition system should be configured. A data acquisition system connected to the sensor is established, which can collect the signal output by the sensor in real time and convert it into a digital signal for storage and processing. The data acquisition system should have sufficient sampling frequency to accurately capture the dynamic changes of the boiler operating parameters. For temperature and pressure parameters, since they change relatively quickly, the sampling frequency can be set to several to dozens of times per second, such as 5 or 10 times per second, so that the fluctuation of the parameters can be recorded in detail. For load parameters, the speed of change is relatively slow, but it is also necessary to adjust the sampling frequency appropriately according to the operating characteristics of the boiler and the load adjustment frequency. Generally, it can be set to 1 time per second or 1 time per few seconds. At the same time, set the parameters of the data acquisition system, including data storage format, storage path, data transmission method, etc. The data storage format should be selected in a format suitable for subsequent data analysis and processing, such as the common CSV format or database-specific format. The storage path should be clear and easy to manage to ensure that the data can be stored in an orderly manner. The data transmission method must ensure that the data can be transmitted to the subsequent data analysis and processing links in a timely and stable manner. Wired transmission (such as Ethernet) or wireless transmission (such as Wi-Fi, 4G / 5G, etc.) can be used, and the selection should be based on the actual network environment and needs of the power plant.

[0062] Finally, there is the data collection and storage link. The data collection system continuously collects temperature, load and pressure data during the operation of the boiler to form a time series data set. During the collection process, the data should be monitored in real time, and abnormal data points should be discovered in time by setting data thresholds, change rate limits, etc. For example, for temperature values, a normal operating range is set. When the temperature exceeds the range or the temperature change rate exceeds a certain value, the system should issue an alarm so that the staff can promptly troubleshoot the fault, which may be caused by abnormal combustion, sensor failure or other operating problems. For pressure values, a reasonable pressure range and change threshold are also set. When the pressure suddenly changes or continuously deviates from the normal range, it should be checked in time. The monitoring of load values ​​should not be ignored. If the load fluctuates abnormally, it may affect the stable operation of the boiler. The collected time series data is stored in the database in chronological order, and the storage structure should facilitate data query and analysis. Relational databases or time series databases can be used to store data. Relational databases are suitable for complex associative queries and transaction processing of data, while time series databases are better at processing data stored in chronological order and can quickly perform queries and data analysis based on time ranges. At the same time, appropriate indexes are established, such as indexes based on time dimensions such as date and hour, to facilitate quick query of boiler operating parameter data within a specific time period and improve data retrieval efficiency.

[0063] In the embodiment of the present invention, the boiler operation state encoding module 130 is used to perform data time series spatial aggregation and operation state feature extraction on the time series data set of the boiler operation parameters to obtain a boiler operation state multimodal time series joint semantic encoding feature map. Specifically, Figure 3 FIG. 1 is a block diagram of a boiler operation status encoding module in a coal blending expert decision-making system based on coal yard intelligent centralized control technology according to an embodiment of the present invention. Figure 3 As shown, the boiler operation status encoding module 130 includes: a boiler operation parameter time series data sorting unit 131, which is used to perform data time series spatial aggregation on the time series data set of the boiler operation parameters according to the time dimension and the parameter sample dimension to obtain a boiler operation parameter multimodal time series aggregation matrix; a boiler operation status feature extraction unit 132, which is used to input the boiler operation parameter multimodal time series aggregation matrix into a boiler operation status feature extractor based on a deep separable convolutional neural network model to obtain the boiler operation status multimodal time series joint semantic coding feature map.

[0064] In an embodiment of the present invention, the boiler operating parameter time series data sorting unit 131 is used to perform data time series spatial aggregation on the time series data set of the boiler operating parameters according to the time dimension and the parameter sample dimension to obtain a multimodal time series aggregation matrix of the boiler operating parameters. Accordingly, considering that during the operation of the boiler, the various parameters (such as temperature, pressure, and load) in the boiler operating parameters present complex patterns with the changes in time and space, and these parameters each constitute a different mode, that is, these parameters have their own time series characteristic patterns and change information, and there is a mutual time series correlation between these parameters in the time dimension. Based on this, in order to effectively capture these time series change patterns and correlation relationships, in the technical solution of the present invention, the time series data set of the boiler operating parameters is aggregated in the time dimension and the parameter sample dimension to obtain a multimodal time series aggregation matrix of the boiler operating parameters.

[0065] In an embodiment of the present invention, the boiler operation state feature extraction unit 132 is used to input the boiler operation parameter multimodal time series aggregation matrix into the boiler operation state feature extractor based on the deep separable convolutional neural network model to obtain the boiler operation state multimodal time series joint semantic coding feature map. It should be understood that the boiler operation parameter multimodal time series aggregation matrix contains various types of time series data (such as temperature, pressure, load), which implies the complex correlation relationship and pattern structure of each parameter in the time series, and the deep separable convolutional neural network model can not only significantly reduce the amount of calculation and the number of parameters through the encoding method of combining deep convolution and point-by-point convolution, but also can capture the local relationship between boiler operation parameters more finely through this independent convolution operation model, and then capture richer local features. Therefore, in the technical solution of the present invention, the boiler operation parameter multimodal time series aggregation matrix is ​​input into the boiler operation state feature extractor based on the deep separable convolutional neural network model to extract and capture the deeper and more complex time series joint features between multiple parameters in the operation parameters, and obtain the boiler operation state multimodal time series joint semantic coding feature map.

[0066] In an embodiment of the present invention, the first and second quality parameter embedding joint module 140 is used to embed and combine the quality parameters of the first type of coal and the quality parameters of the second type of coal to obtain a first type of coal-second type of coal quality parameter joint embedded coding vector. Specifically, in an embodiment of the present invention, the first and second quality parameter embedding joint module includes: a first and second type of coal quality parameter embedding coding unit, which is used to embed and code the quality parameters of the first type of coal and the quality parameters of the second type of coal using a quality parameter embedding coding matrix to obtain a first type of coal quality parameter embedded coding vector and a second type of coal quality parameter embedded coding vector; a first and second type of coal quality parameter embedded coding vector cascade unit, which is used to cascade the first type of coal quality parameter embedded coding vector and the second type of coal quality parameter embedded coding vector to obtain the first type of coal-second type of coal quality parameter joint embedded coding vector.

[0067] It should be understood that the quality parameters of the first type of coal and the quality parameters of the second type of coal respectively express the parameter information of the two qualities of coal, such as ash, volatile matter, fixed carbon, sulfur content, etc. Therefore, in order to convert it into a unified vector form, facilitate computer processing and analysis, and more clearly understand the semantics of each quality parameter, in the technical solution of the present invention, the quality parameters of the first type of coal and the quality parameters of the second type of coal are embedded and encoded using a quality parameter embedding coding matrix to map parameter values ​​of different ranges and units to a standardized space while retaining the key information of the coal quality parameters, thereby obtaining the first type of coal quality parameter embedded coding vector and the second type of coal quality parameter embedded coding vector.

[0068] In particular, in an embodiment of the present invention, a quality parameter embedding coding matrix is ​​used to embed the quality parameters of the first type of coal and the quality parameters of the second type of coal to obtain a first type of coal quality parameter embedding coding vector and a second type of coal quality parameter embedding coding vector. One possible implementation method can be: first, a quality parameter embedding coding matrix should be prepared in advance, and the embedding coding matrix is ​​a model obtained by pre-training a large amount of coal quality parameter information. Each row in the matrix represents an embedding vector of a quality parameter, and each column represents a dimension of the embedding vector. Assume that the scale of this embedding coding matrix is ​​V×d, where V represents the number of types of quality parameters and d represents the dimension of the embedding vector. Then, the quality parameters of the first type of coal and the second type of coal are obtained, and these parameters generally cover indicators such as ash content, volatile matter, fixed carbon, and sulfur content. Assume that the quality parameter of the first type of coal is p1=[P11,P12,...,p1v], and the quality parameter of the second type of coal is P2=[P21,P22,...,P2v], where Pij represents the value of the jth quality parameter of the i-th type of coal. Subsequently, these quality parameters are transformed with the help of the quality parameter embedding coding matrix to obtain the final coal quality parameter embedding coding vector. The specific operation steps are as follows: a. For each quality parameter of the first type of coal, the corresponding embedding vector is retrieved from the embedding coding matrix. For example, the ash value of the first type of coal corresponds to the first row in the embedding coding matrix, the volatile value corresponds to the second row, and so on. For each quality parameter of the second type of coal, the corresponding embedding vector is also searched in the embedding coding matrix in this way. b. The value of each quality parameter is weighted and summed with its corresponding embedding vector to generate the final embedding coding vector. Specifically, for the first type of coal, the value of each quality parameter is multiplied by its corresponding embedding vector, and all the product results are accumulated to obtain the first type of coal quality parameter embedding coding vector. Similarly, for the second type of coal, the value of each quality parameter is multiplied by the corresponding embedding vector and then accumulated to obtain the second type of coal quality parameter embedding coding vector.

[0069] Then, in order to be able to fuse the first type of coal quality parameter embedded coding vector and the second type of coal quality parameter embedded coding vector to obtain a comprehensive representation used to represent all quality parameter information of the two types of coal and provide data support for subsequent quality decisions, in the technical solution of the present invention, the first type of coal quality parameter embedded coding vector and the second type of coal quality parameter embedded coding vector are cascaded to obtain a first type of coal-second type of coal quality parameter joint embedded coding vector. In this way, the quality parameters of the two types of coal can be comprehensively considered to provide richer feature representation for subsequent processing and analysis.

[0070] In the embodiment of the present invention, the boiler quality state joint coding module 150 is used to perform a cross-modal combination of the first coal type-second coal type quality parameter joint embedding coding vector and the boiler operation state multimodal temporal joint semantic coding feature map with continuous prompt guidance to obtain a coal type quality-boiler state joint coding feature map. Specifically, Figure 4 FIG. 1 is a block diagram of a boiler quality status joint coding module in a coal blending expert decision-making system based on coal yard intelligent centralized control technology according to an embodiment of the present invention. Figure 4 As shown, the boiler quality state joint encoding module 150 includes: a boiler operating state local feature decomposition unit 151, which is used to perform local feature decomposition along the channel dimension on the boiler operating state multimodal temporal joint semantic coding feature map to obtain a set of boiler operating state multimodal temporal joint semantic local feature matrices; a coal type quality-boiler state local query unit 152, which is used to use the first coal type-second coal type quality parameter joint embedding coding vector as a query vector and each boiler operating state multimodal temporal joint semantic local feature matrix in the set of boiler operating state multimodal temporal joint semantic local feature matrices as a key matrix, and input the query vector and the key matrix into a prompt learning network based on a converter structure to obtain a set of coal type quality-boiler state local query prompt semantic coding vectors; a boiler operating state multimodal joint unit 153, which is used to perform cross-modal mask feature aggregation on the set of boiler operating state multimodal temporal joint semantic local feature matrices based on the first coal type-second coal type quality parameter joint embedding coding vector and the set of coal type quality-boiler state local query prompt semantic coding vectors.

[0071] It should be understood that the first type of coal-second type of coal quality parameter joint embedding coding vector and the boiler operating state multimodal temporal joint semantic coding feature map respectively represent the characteristics of the two types of coal quality parameters in terms of physical and chemical properties and the temporal information of the multi-faceted data of the boiler operating state changing over time. Therefore, in order to effectively integrate the characteristics of the two modes, so as to better capture the complex relationship between the coal quality parameters and the boiler operating state, and generate a representative representation of the multimodal information of both the coal quality parameters and the boiler operating state, so as to provide a more comprehensive perspective to provide decision support for subsequent blending, in the technical solution of the present invention, the first type of coal-second type of coal quality parameter joint embedding coding vector and the boiler operating state multimodal temporal joint semantic coding feature map are cross-modally combined with continuous prompt guidance to obtain the coal quality-boiler state joint coding feature map.

[0072] In detail, it is first necessary to perform local feature decomposition along the channel dimension on the multimodal temporal joint semantic coding feature map of the boiler operating state to obtain a set of local feature matrices of the multimodal temporal joint semantic coding of the boiler operating state. Through local feature decomposition, the characteristics of each channel (such as temperature, load, pressure) in the multimodal temporal joint semantic coding feature map of the boiler operating state can be better analyzed, thereby providing data support for subsequent fine-grained interactions. The above process can be expressed as:

[0073] Decompos(F2)={M1,M2,...,M i ,...,M n}

[0074] Among them, F2 is the characteristic diagram of the melt state, M1, M2, M i and M n are respectively the 1st, 2nd, i-th and n-th local feature matrices of the molten liquid state in the set of the local feature matrices of the molten liquid state, and Decompose(·) represents the feature decomposition operation on the input data.

[0075] More specifically, in an embodiment of the present invention, the coal type quality-boiler status local query unit is used to: calculate the similarity score between the first coal type-second coal type quality parameter joint embedded coding vector and each row vector in the boiler operating status multimodal temporal joint semantic local feature matrix to obtain a set of coal type quality-boiler status similarity scores; standardize each coal type quality-boiler status similarity score in the set of coal type quality-boiler status similarity scores to obtain a set of coal type quality-boiler status weight coefficients; based on the set of coal type quality-boiler status weight coefficients, calculate the weighted sum of each row vector in the boiler operating status multimodal temporal joint semantic local feature matrix to obtain the coal type quality-boiler status local query prompt semantic coding vector. The above process can be expressed as:

[0076] M i = {v i1 ,v i2 ,...,v ij ,...,v um}

[0077]

[0078] Among them, M i is the i-th local characteristic matrix of the melt state in the set of local characteristic matrices of the melt state, v i1 、v i2 、v ij and v im M iThe first, second, jth and mth row vectors in , Transformer(·,·) is the Transformer encoding, v1 is the melting temperature time series correlation feature vector, v ij T Yes ij The transposed vector of is the matrix multiplication, ‖·‖ is the bi-norm of the vector, s j For v1 and v ij The melt state-temperature similarity score between them, exp(·) represents the exponential function value with the natural constant e as the base, v ik It is M i The kth row vector in s k For v1 and v ik The melt state-temperature similarity score between j v ij The corresponding melt state-temperature weight coefficient, m is M i The number of row vectors in v ti It is M i The corresponding melt state-temperature local query prompt semantic encoding vector, ⊙ represents the point product by position.

[0079] That is, the first coal type-second coal type quality parameter joint embedding coding vector is used as the query vector and each boiler operating state multimodal temporal joint semantic local feature matrix in the set of the boiler operating state multimodal temporal joint semantic local feature matrix is ​​used as the key matrix, which is input into the prompt learning network based on the converter structure to utilize the self-attention mechanism of the converter structure to capture the feature information of different quality parameters of different coal types and the correlation between them and the boiler operating parameters, so as to have a more comprehensive understanding of the relationship between coal type quality parameters and boiler operating status, which in turn helps to optimize the coal blending strategy so that the proportion of coal types is more in line with the actual operating needs of the boiler.

[0080] More specifically, in an embodiment of the present invention, the boiler operating state multimodal joint unit is used to: input each coal type quality-boiler state local query prompt semantic coding vector in the set of the first coal type-second coal type quality parameter joint embedding coding vector and the coal type quality-boiler state local query prompt semantic coding vector into a cross-modal mask weaving network based on prompt information to obtain a set of quality parameter-boiler state cross-modal local feature mask weight matrices based on prompt information; calculate the point-by-point multiplication between each corresponding set of quality parameter-boiler state cross-modal local feature mask weight matrices based on prompt information and the set of the boiler operating state multimodal temporal joint semantic local feature matrices to obtain a set of coal type quality-boiler state cross-modal local granularity significant interaction matrices; feature aggregate the set of the coal type quality-boiler state cross-modal local granularity significant interaction matrices to obtain the coal type quality-boiler state joint coding feature map. The above process can be expressed as:

[0081]

[0082] F 1-2 =Concat{M1⊙S t1 ,M2⊙S t2 ,...,M i ⊙S ti ...,M n ⊙S tn}

[0083] Among them, v1 is the melting temperature time series correlation feature vector, v ti T v ti The transposed vector of ti T The scale of the melt state-temperature cue association matrix obtained by multiplying it with v1 is the length of the matrix multiplied by the width of the matrix. softmax(·) is the softmax function, S ti It is M i The corresponding melt state-temperature cross-modal local feature mask weight matrix based on the prompt information, M1, M2, M i and M n are the first, second, i-th and n-th local characteristic matrices of the melt state in the set of local characteristic matrices of the melt state, S t1 , S t2 , S ti and S tn They are M1, M2, M i and M nThe corresponding melt state-temperature cross-modal local feature mask weight matrix based on the prompt information, ⊙ is the position point multiplication, Concat{·,·...,·} is the feature concatenation along the channel dimension, F 1-2 is the melting state-temperature cross-modal guided joint encoding feature map.

[0084] Here, the generated first coal type-second coal type quality parameter joint embedding coding vector and the set of coal type quality-boiler status local query prompt semantic coding vectors are input into the cross-modal mask weaving network based on the prompt information to obtain a set of quality parameter-boiler status cross-modal local feature mask weight matrices. That is, the cross-modal mask weaving network can reduce the influence of coal type quality information noise that is not related to the boiler operating status by creating a mask that emphasizes the most relevant part of the modal interaction. And by generating a set of cross-modal local feature mask weight matrices, it is possible to highlight the important interaction parts between modalities, that is, those coal type quality features that are most relevant to the boiler operating status, thereby ensuring that the final output features can focus on the most representative areas.

[0085] Subsequently, the set of mask weight matrices of the quality parameter-boiler state cross-modal local feature based on the prompt information is multiplied by the corresponding matrix in the set of multimodal temporal joint semantic local feature matrices of the boiler operation state by position, so as to obtain the set of coal quality-boiler state cross-modal local granularity significant interaction matrices. Here, the degree of fusion of coal quality parameter features and boiler operation state features can be further strengthened, so that the final interaction matrix can accurately reflect the correlation and importance between different features. Finally, by performing feature aggregation on the set of coal quality-boiler state cross-modal local granularity significant interaction matrices, the coal quality-boiler state joint coding feature map is obtained to more comprehensively and accurately reflect the various relationships between coal quality parameters and boiler operation states in the process of coal blending and combustion.

[0086] In an embodiment of the present invention, the decision module 160 is used to obtain a decision suggestion based on the coal quality-boiler status joint coding feature map, and the decision suggestion is the first ratio of the first coal type and the second ratio of the second coal type, and the sum of the first ratio and the second ratio is 1. Specifically, in an embodiment of the present invention, the decision module is used to: input the coal quality-boiler status joint coding feature map into the classifier-based coal blending expert decision module to obtain the decision suggestion, and the decision suggestion is the first ratio of the first coal type and the second ratio of the second coal type, and the sum of the first ratio and the second ratio is 1. It should be understood that the coal quality-boiler status joint coding feature map integrates the coal quality parameter characteristics and the multi-modal temporal joint semantic coding feature information of the boiler operation status, and can fully reflect the coal quality and the real-time status of the boiler operation. Considering that the classifier is a machine learning model, it can learn the mapping relationship between input data and output data through training with a large amount of labeled training data. When new input data is input, the learned mapping mode can be used to analyze and judge the information contained in it, and finally output the classification result. By using the coal quality-boiler status joint coding feature map as the input of the classifier, a decision suggestion for the allocation ratio of the first coal type and the second coal type can be obtained, thereby making the coal blending decision more accurate and reliable. In particular, in an embodiment of the present invention, the coal quality-boiler status joint coding feature map is input into the coal blending expert decision module based on the classifier to obtain a decision suggestion for representing the first ratio of the first coal type and the second ratio of the second coal type. One possible implementation method can be: each coal quality-boiler status joint coding feature matrix in the coal quality-boiler status joint coding feature map is expanded into a one-dimensional feature vector according to a row vector or a column vector, and then cascaded to obtain a coal type ratio classification feature vector; the coal type ratio classification feature vector is fully connected encoded using the fully connected layer of the classifier to obtain a coal type ratio classification fully connected coding feature vector; the coal type ratio classification fully connected coding feature vector is input into the Softmax classification function of the classifier to obtain the probability value of the coal quality-boiler status joint coding feature map belonging to each classification label, and each classification label represents a different allocation ratio of the first coal type and the second coal type.

[0087] In summary, the coal blending and burning expert decision system 100 based on the coal yard intelligent centralized control technology according to the embodiment of the present invention is explained, which adopts AI-based data analysis and coding technology to perform time-series spatial aggregation and operation status feature extraction on the time series data set of boiler operating parameters, and at the same time, semantic embedding and union are performed on the quality parameters of the first type of coal and the second type of coal respectively, so as to automatically obtain decision suggestions based on the continuous prompt cross-modal joint representation between the joint features of the first type of coal and the second type of coal quality parameters and the multimodal time series joint features of the boiler operating status. In this way, the operating status of the boiler can be monitored in real time, and the optimal coal ratio can be quickly adjusted according to the current operating conditions to ensure the efficiency and stability of the combustion process, reduce the subjective influence of human judgment, thereby enhancing the scientificity and accuracy of the decision, and realizing the intelligent management of coal blending and burning.

[0088] Figure 5 Flow chart of the expert decision-making method for coal blending and burning based on the intelligent centralized control technology of coal yard according to an embodiment of the present invention. Figure 5 As shown, in the expert decision method for coal blending and combustion based on the intelligent centralized control technology of the coal yard, the method includes: S110, obtaining the quality parameters of the first type of coal and the quality parameters of the second type of coal; S120, obtaining the time series data of the boiler operating parameters; S130, performing data time series space aggregation and operation state feature extraction on the time series data set of the boiler operating parameters to obtain a boiler operating state multimodal time series joint semantic coding feature map; S140, embedding and combining the quality parameters of the first type of coal and the quality parameters of the second type of coal to obtain a first type of coal-second type of coal quality parameter joint embedded coding vector; S150, performing continuous prompt guidance cross-modal combination on the first type of coal-second type of coal quality parameter joint embedded coding vector and the boiler operating state multimodal time series joint semantic coding feature map to obtain a coal type quality-boiler state joint coding feature map; S160, based on the coal type quality-boiler state joint coding feature map, obtaining a decision suggestion, the decision suggestion is a first proportion of the first type of coal and a second proportion of the second type of coal, and the sum of the first proportion and the second proportion is 1.

[0089] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned coal blending and burning expert decision-making method based on the intelligent centralized control technology of the coal yard have been referred to above. Figures 1 to 4 It has been introduced in detail in the description of the coal blending and combustion expert decision system based on the intelligent centralized control technology of the coal yard, and therefore, its repeated description will be omitted.

[0090] In summary, the expert decision-making method for coal blending and burning based on the coal yard intelligent centralized control technology according to the embodiment of the present invention is explained, which uses AI-based data analysis and coding technology to perform time-series spatial aggregation and operation status feature extraction on the time series data set of boiler operating parameters, and at the same time, semantic embedding and union are performed on the quality parameters of the first type of coal and the second type of coal respectively, so as to automatically obtain decision suggestions based on the continuous prompt cross-modal joint representation between the joint features of the first type of coal and the second type of coal quality parameters and the multimodal time series joint features of the boiler operating status. In this way, the operating status of the boiler can be monitored in real time, and the optimal coal ratio can be quickly adjusted according to the current operating conditions to ensure the efficiency and stability of the combustion process, reduce the subjective influence of human judgment, thereby enhancing the scientificity and accuracy of the decision, and realizing the intelligent management of coal blending and burning.

[0091] It is to be understood that the above embodiments are merely exemplary embodiments used to illustrate the principles of the present invention, but the present invention is not limited thereto. For those of ordinary skill in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also considered to be within the scope of protection of the present invention.

Claims

1. A coal blending and combustion expert decision system based on coal yard intelligent centralized control technology, characterized in that: include: An environmental data acquisition module, used to acquire quality parameters of the first type of coal and quality parameters of the second type of coal; Environmental data processing and encoding module, used to obtain time series data of boiler operating parameters; A boiler operation state encoding module, used for performing data time series spatial aggregation and operation state feature extraction on the time series data set of the boiler operation parameters to obtain a multi-modal time series joint semantic encoding feature map of the boiler operation state; A first and second quality parameter embedding and combining module, used for embedding and combining the quality parameter of the first type of coal and the quality parameter of the second type of coal to obtain a first type of coal-second type of coal quality parameter combined embedding coding vector; A boiler quality state joint coding module, used for performing a cross-modal combination of the first coal type-second coal type quality parameter joint embedding coding vector and the boiler operation state multimodal temporal joint semantic coding feature map with continuous prompt guidance to obtain a coal type quality-boiler state joint coding feature map; A decision module is used to obtain a decision suggestion based on the coal quality-boiler status joint coding feature map, wherein the decision suggestion is a first ratio of the first coal type and a second ratio of the second coal type, and the sum of the first ratio and the second ratio is 1.

2. The coal blending and burning expert decision system based on coal yard intelligent centralized control technology according to claim 1 is characterized in that: The boiler operating parameters include temperature value, load value and pressure value.

3. The coal blending and burning expert decision system based on coal yard intelligent centralized control technology according to claim 2 is characterized in that: The boiler operation status encoding module comprises: A boiler operating parameter time series data collating unit, used for performing data time series spatial aggregation on the time series data set of the boiler operating parameters according to the time dimension and the parameter sample dimension to obtain a multimodal time series aggregation matrix of the boiler operating parameters; The boiler operation state feature extraction unit is used to input the boiler operation parameter multimodal time series aggregation matrix into the boiler operation state feature extractor based on the deep separable convolutional neural network model to obtain the boiler operation state multimodal time series joint semantic coding feature map.

4. The coal blending and burning expert decision system based on coal yard intelligent centralized control technology according to claim 3 is characterized in that: The first and second quality parameters are embedded in a joint module, including: A first and second coal type quality parameter embedding coding unit, configured to embed the quality parameter of the first coal type and the quality parameter of the second coal type using a quality parameter embedding coding matrix to obtain a first coal type quality parameter embedding coding vector and a second coal type quality parameter embedding coding vector; The first and second coal type quality parameter embedded coding vector cascade unit is used to cascade the first coal type quality parameter embedded coding vector and the second coal type quality parameter embedded coding vector to obtain the first coal type-second coal type quality parameter joint embedded coding vector.

5. The coal blending and burning expert decision system based on coal yard intelligent centralized control technology according to claim 4 is characterized in that: The boiler quality status joint coding module includes: A boiler operation state local feature decomposition unit, used for performing local feature decomposition along the channel dimension on the boiler operation state multimodal temporal joint semantic coding feature map to obtain a set of boiler operation state multimodal temporal joint semantic local feature matrices; A coal quality-boiler status local query unit, used to use the first coal type-second coal type quality parameter joint embedding coding vector as a query vector and each boiler operating status multimodal temporal joint semantic local feature matrix in the set of boiler operating status multimodal temporal joint semantic local feature matrices as a key matrix, and input the query vector and the key matrix into a prompt learning network based on a converter structure to obtain a set of coal quality-boiler status local query prompt semantic coding vectors; The boiler operating status multimodal joint unit is used to perform cross-modal mask feature aggregation on the set of the boiler operating status multimodal time series joint semantic local feature matrix based on the first coal type-second coal type quality parameter joint embedding coding vector and the set of coal type quality-boiler status local query prompt semantic coding vectors.

6. The coal blending and burning expert decision system based on coal yard intelligent centralized control technology according to claim 5 is characterized in that: The coal quality-boiler status local query unit is used to: Calculate the similarity scores between the first coal type-second coal type quality parameter joint embedding coding vector and each row vector in the boiler operation state multimodal temporal joint semantic local feature matrix to obtain a set of coal type quality-boiler state similarity scores; Standardizing each coal type quality-boiler state similarity score in the set of coal type quality-boiler state similarity scores to obtain a set of coal type quality-boiler state weight coefficients; Based on the set of coal quality-boiler status weight coefficients, the weighted sum of each row vector in the boiler operating status multimodal time series joint semantic local feature matrix is ​​calculated to obtain the coal quality-boiler status local query prompt semantic encoding vector.

7. The coal blending and burning expert decision system based on coal yard intelligent centralized control technology according to claim 6 is characterized in that: The boiler operation status multi-modal joint unit is used for: Input each coal type quality-boiler status local query prompt semantic coding vector in the set of the first coal type-second coal type quality parameter joint embedding coding vector and the coal type quality-boiler status local query prompt semantic coding vector into a cross-modal mask weaving network based on prompt information to obtain a set of quality parameter-boiler status cross-modal local feature mask weight matrices based on prompt information; Calculate the set of the quality parameter based on prompt information-boiler state cross-modal local feature mask weight matrix and the set of the boiler operation state multimodal temporal joint semantic local feature matrix, and obtain the set of coal type quality-boiler state cross-modal local granularity significant interaction matrices by multiplying the position points between each corresponding quality parameter based on prompt information-boiler state cross-modal local feature mask weight matrix and the boiler operation state multimodal temporal joint semantic local feature matrix; The set of the coal quality-boiler status cross-modal local granularity significant interaction matrices is feature aggregated to obtain the coal quality-boiler status joint coding feature map.

8. The coal blending and burning expert decision system based on coal yard intelligent centralized control technology according to claim 7 is characterized in that: The decision module is used to: input the coal quality-boiler status joint coding feature map into the classifier-based coal blending and combustion expert decision module to obtain the decision recommendation, and the decision recommendation is a first ratio of the first coal type and a second ratio of the second coal type, and the sum of the first ratio and the second ratio is 1.

9. An expert decision-making method for coal blending and combustion based on intelligent centralized control technology of coal yard, characterized in that: include: Obtaining quality parameters of a first type of coal and quality parameters of a second type of coal; Obtain time series data of boiler operating parameters; Performing data time series spatial aggregation and operation state feature extraction on the time series data set of the boiler operation parameters to obtain a multimodal time series joint semantic coding feature map of the boiler operation state; Embedding and combining the quality parameter of the first type of coal and the quality parameter of the second type of coal to obtain a first type of coal-second type of coal quality parameter joint embedded coding vector; Performing a cross-modal combination of the first coal type-second coal type quality parameter joint embedding coding vector and the boiler operation status multimodal temporal joint semantic coding feature map with continuous prompt guidance to obtain a coal type quality-boiler status joint coding feature map; Based on the coal quality-boiler status joint coding feature diagram, a decision recommendation is obtained, and the decision recommendation is a first ratio of the first coal type and a second ratio of the second coal type, and the sum of the first ratio and the second ratio is 1.

10. The expert decision-making method for coal blending and combustion based on coal yard intelligent centralized control technology according to claim 9 is characterized in that: Performing data temporal spatial aggregation and operating state feature extraction on the time series data set of the boiler operating parameters to obtain a multimodal temporal joint semantic coding feature map of the boiler operating state, including: Performing data time series spatial aggregation on the time series data set of the boiler operating parameters according to the time dimension and the parameter sample dimension to obtain a multimodal time series aggregation matrix of the boiler operating parameters; The boiler operating parameter multimodal time series aggregation matrix is ​​input into a boiler operating state feature extractor based on a deep separable convolutional neural network model to obtain a multimodal time series joint semantic encoding feature map of the boiler operating state.

Images