A slag cooling process simulation method based on multi-source heterogeneous data

Abstract

The application provides a slag cooling process simulation method based on multi-source heterogeneous data, and belongs to the technical field of industrial process simulation and optimization. The method collects multi-source heterogeneous data generated in the boiler slag discharge process, and after preprocessing, time synchronization and space alignment processing, uses data fusion technology to extract key features and construct a fusion feature vector. Further based on the fused data, an improved simulated annealing algorithm is used to establish a slag heat conduction model, and combined with real-time monitoring data of the slag surface, internal temperature data, historical cooling data and environmental condition data, a data assimilation technology is used to dynamically update the temperature field model, and a three-dimensional model of the slag temperature field is constructed. Finally, the improved sparrow search algorithm is used to optimize the calculation of the best cooling air volume and the slag movement speed. The application can improve the slag cooling efficiency, reduce the energy consumption, and optimize the stability of the cooling process, and has significant industrial application value.

Description

Technical Field

[0001] This invention relates to the field of industrial process simulation and optimization technology, and in particular to a method for simulating slag cooling process based on multi-source heterogeneous data. Background Technology

[0002] In modern industrial production, boiler slag discharge and slag cooling processes are crucial for energy utilization and environmental protection. Traditional slag cooling methods rely primarily on empirical parameter settings and fixed cooling strategies, lacking dynamic monitoring and optimization of the cooling process. However, this experience-based approach has several shortcomings:

[0003] Low cooling efficiency: Traditional cooling methods cannot dynamically adjust cooling parameters (such as cooling air volume and slag movement speed) according to the actual temperature distribution of slag and cooling requirements, resulting in low cooling efficiency, long cooling time, and reduced production efficiency.

[0004] High energy consumption: Due to the lack of precise cooling control, traditional methods often require excessive cooling air volume to ensure the cooling effect of slag, resulting in energy waste and increased production costs.

[0005] Unstable cooling process: Traditional cooling processes cannot monitor the temperature changes inside and on the surface of slag in real time, making it difficult to deal with possible temperature anomalies or local overheating during the cooling process. This can easily lead to uneven cooling of slag and affect subsequent processing.

[0006] Lack of data-driven optimization methods: In existing technologies, the optimization of the cooling process mainly relies on human experience and fixed parameters, lacking dynamic optimization capabilities based on real-time data. With the development of industrial automation and intelligence, this experience-based approach is no longer sufficient to meet the requirements of modern industry for efficient, energy-saving, and stable production.

[0007] Insufficient Data Utilization: In modern industrial production, the boiler ash discharge process generates a large amount of multi-source heterogeneous data, including information such as temperature, air pressure, air volume, and images. However, existing technologies have failed to fully utilize this data for modeling and optimizing the cooling process, resulting in a waste of data resources.

[0008] To address the shortcomings of existing technologies, this invention proposes a slag cooling process simulation method based on multi-source heterogeneous data. By collecting, fusing, and analyzing multiple data sources and combining them with advanced algorithmic techniques, dynamic modeling and optimization of the slag cooling process are achieved. This invention can effectively improve cooling efficiency, reduce energy consumption, optimize the stability of the cooling process, and provide intelligent solutions for industrial production. Summary of the Invention

[0009] The purpose of this invention is to provide a slag cooling process simulation method based on multi-source heterogeneous data, which can improve slag cooling efficiency, reduce energy consumption, and optimize the stability of the cooling process, and has significant industrial application value.

[0010] To achieve the above objectives, the present invention provides a method for simulating slag cooling processes based on multi-source heterogeneous data, comprising the following steps:

[0011] Step S1: Collect multi-source heterogeneous data generated during boiler ash discharge;

[0012] Step S2: Preprocess the collected multi-source heterogeneous data to obtain preprocessed data;

[0013] Step S3: Perform time synchronization and spatial alignment on the preprocessed data to obtain time-synchronized and spatially aligned data;

[0014] Step S4: Using data fusion technology, the time-synchronized and spatially aligned data are fused to extract key features and construct a fused feature vector;

[0015] Step S5: Based on the fused data, a slag heat conduction model is established using an improved simulated annealing algorithm;

[0016] Step S6: Combining real-time monitoring data of slag surface, internal temperature data, historical cooling data and environmental condition data, the temperature field model is dynamically updated using data assimilation technology to construct a three-dimensional model of the slag temperature field.

[0017] Step S7: Based on the constructed three-dimensional model of the slag temperature field, the optimal cooling air volume and slag movement speed are optimized and calculated using the improved sparrow search algorithm.

[0018] Preferably, in step S1, the multi-source heterogeneous data includes: slag temperature sensor data; cooling air pressure and air volume sensor data; slag movement speed sensor data; and slag surface image data.

[0019] Preferably, in step S2, the preprocessing includes data cleaning and data normalization.

[0020] Preferably, in step S3, the time synchronization process is as follows:

[0021] T sync (t)=T sensor (t+Δt);

[0022] Among them, T sync (t) represents the data after time synchronization; T sensor This represents the original time data collected by the sensor; t represents time; Δt represents the time offset, used to correct for time differences between different sensors.

[0023] Spatial alignment is handled as follows:

[0024] X aligned (x,y,z)=x sensor (x+Δx,y+Δy,z+Δz);

[0025] Among them, X aligned (x,y,z) represents the spatially aligned data; (x,y,z) represents the spatial coordinates; x sensor Δx represents the spatial data collected by the original sensor; Δy represents the spatial offset in the x-axis direction; Δz represents the spatial offset in the y-axis direction; and Δz represents the spatial offset in the z-axis direction.

[0026] Preferably, in step S4, data fusion technology is used to fuse the time-synchronized and spatially aligned data, extract key features, and construct a fused feature vector. The specific operations are as follows:

[0027] Step S41: Extract slag temperature features, cooling air pressure and air volume features, slag movement speed features and slag surface image features from the time-synchronized and spatially aligned slag temperature data, cooling air pressure data, cooling air volume data, slag movement speed data and slag surface image data.

[0028] Step S42: Fuse the extracted features to construct a fused feature vector using a weighted concatenation method, as shown in the following formula:

[0029] F Fusion =[ω1×F T ;ω2×F P ;ω3×F Q ;ω4×F V ;ω5×F GLCM ;ω6×F Edge ];

[0030] Among them, F Fusion This represents the fused feature vector; ω1, ω2, ω3, ω4, ω5, and ω6 represent the weight coefficients of each feature, respectively. Indicates the temperature characteristics of the slag. ΔT represents the average temperature, and ΔT represents the temperature gradient. Indicates wind pressure characteristics, Indicates average wind pressure; Indicates air volume characteristics. Indicates average air volume; This indicates the velocity characteristics of slag movement. F represents the average velocity, ΔV represents the rate of change of velocity; GLCM Texture features are represented using the Gray-Level Co-occurrence Matrix (GLCM); FEdge The edge features are represented and obtained through Canny edge detection.

[0031] Step S43: Use the particle swarm optimization algorithm to optimize the weight coefficients in the fused feature vector.

[0032] Preferably, in step S43, the particle swarm optimization algorithm is used to optimize the weight coefficients in the fused feature vector. The specific operation is as follows:

[0033] Step S431: Randomly initialize the position and velocity of the particles. The position of the particles is represented as w = [ω1, ω2, ω3, ω4, ω5, ω6]. Each particle represents a set of weight coefficients, and the velocity is initialized to a random value.

[0034] Step S432: Define the fitness function;

[0035] Fitness(w) = MSE(F) Fusion );

[0036] Where Fitness(w) represents the fitness function; MSE(F) Fusion ) represents the error of the fused feature vector on the validation set;

[0037] Step S433: For each particle, calculate its fitness value and update its optimal position P. best and the global optimal position g best ;

[0038] Step S434: Update particle velocity and position;

[0039] Step S435: Ensure the weighting coefficient is within 0 to 1; if it exceeds the range, adjust it randomly.

[0040] Step S436: Repeat steps S432 to S435 until the preset number of iterations is reached;

[0041] Step S437: Recalculate the fused feature vector using the optimized weight coefficients obtained from the particle swarm optimization algorithm;

[0042]

[0043] Among them, F Final This represents the optimized fused feature vector; These represent the weight coefficients of each feature after optimization;

[0044] The optimized fused feature vector is then normalized.

[0045]

[0046] Among them, FNorm min(F) represents the normalized fused feature vector; Final ), max(F Final ) represent the minimum and maximum values ​​of the optimized fused feature vector, respectively.

[0047] Preferably, in step S5, based on the fused data, an improved simulated annealing algorithm is used to establish a slag heat conduction model, specifically as follows:

[0048] Step S51: Initialize the Grey Wolf Algorithm and run the Grey Wolf Algorithm to obtain the optimal solution x. GWO ;

[0049] Step S52: Initialize the simulated annealing algorithm, using x GWO As the initial solution;

[0050] Step S53: Combining the global search capability of the Grey Wolf algorithm and the local search capability of the Simulated Annealing algorithm, run the two algorithms alternately;

[0051] Step S54: Output the final thermal conductivity model parameters θ best And a slag heat conduction model was constructed.

[0052] Preferably, in step S54, the slag heat conduction model is as follows:

[0053]

[0054] Where T represents the temperature distribution; Represents the Laplace operator; θ best,α This represents the optimized thermal diffusivity; Q(x,t,θ) best ) represents the heat source term;

[0055]

[0056] Where ρ represents the density of the slag; c p θ represents the specific heat capacity of slag. best,λ This represents the optimized thermal conductivity.

[0057] Preferably, in step S6, the three-dimensional model of the slag temperature field is as follows:

[0058]

[0059] Where T(x,y,z,t) represents the temperature distribution; Q cooling (x,y,z,t) represents the cooling heat source term; ΔT correction (x,y,z,t) represents the temperature deviation corrected by data assimilation techniques;

[0060] Q coolingThe formula for calculating (x,y,z,t) is as follows:

[0061]

[0062] Where k represents a coefficient related to cooling efficiency; σ represents the spatial range of cooling influence; (x0, y0, z0) represents the center position of the cooling air action; Q' represents the heat of the cooling air; and P' represents the pressure of the cooling air.

[0063] ΔT correction The formula for calculating (x,y,z,t) is as follows:

[0064] ΔT correction (x,y,z,t)=ω[T surf (x,y,z,t)-T model (x,y,z,t)]+(1-ω)[T internal (x,y,z,t)-T model (x,y,z,t)];

[0065] Where ω represents the weighting factor, used to balance the influence of surface and internal temperature data; T model (x,y,z,t) represents the temperature field calculated based on the model; T surf (x,y,z,t) represents the real-time monitored temperature of the slag surface; T internal (x,y,z,t) represents the internal temperature data of the slag.

[0066] Preferably, in step S7, based on the constructed three-dimensional model of the slag temperature field, the optimal cooling air volume and slag movement velocity are optimized and calculated using an improved sparrow search algorithm. The specific operation is as follows:

[0067] Step S71: Define the optimization objective function;

[0068] f(Q air V slag )=α1×MSE temp +β1×MSE energy ;

[0069] Where α1 and β1 both represent weighting coefficients; MSE temp Mean square error (MSE) represents the temperature deviation. energy The mean square error representing cooling energy consumption;

[0070] Step S72: Initialize the sparrow population X = {X1, X2, ... XN}, where each sparrow represents a combination of cooling air volume and slag movement speed;

[0071] Step S73: Calculate the fitness value for each sparrow;

[0072] Step S74: Update the sparrow's position;

[0073]

[0074] in, This represents the position of the i-th sparrow in the (a+1)-th generation; γ represents the position of the i-th sparrow in generation a; γ represents the learning rate; L represents the position adjustment factor. Let represent the optimal solution in generation a;

[0075] Step S75: Ensure that the cooling air volume and slag movement speed are within a reasonable range.

[0076] Q air ∈[Q min Q max ];

[0077] V slag ∈[V min V max ];

[0078] Among them, Q min Q represents the minimum cooling airflow; max V represents the maximum cooling airflow. min V represents the minimum velocity of the slag movement. max This represents the maximum velocity of the slag.

[0079] Step S76: Repeat steps S72 to S75 until the preset number of iterations is reached;

[0080] Step S77: Output the optimal cooling airflow Q air,opt And the optimal slag movement speed V slag,opt .

[0081] Therefore, the present invention employs the above-mentioned method for simulating the slag cooling process based on multi-source heterogeneous data, and the beneficial technical effects are as follows:

[0082] (1) Significantly improves cooling efficiency:

[0083] This invention dynamically monitors and optimizes cooling parameters (such as cooling air volume and slag movement speed), enabling real-time adjustments based on the actual temperature distribution of the slag and cooling requirements. Compared to traditional cooling methods with fixed parameters, this significantly shortens cooling time and improves production efficiency.

[0084] (2) Reduce energy consumption and costs:

[0085] Traditional cooling methods, lacking precise control, often require excessive cooling airflow to ensure effective slag cooling, leading to energy waste. This invention optimizes cooling parameters to minimize cooling energy consumption while meeting cooling requirements, significantly reducing production costs.

[0086] (3) Enhance the stability of the cooling process:

[0087] This invention utilizes multi-source heterogeneous data and advanced data assimilation technology to monitor temperature changes inside and on the surface of slag in real time and dynamically adjust the cooling strategy. Compared with traditional methods, this invention effectively avoids problems such as localized overheating or uneven cooling, and significantly improves the stability of the cooling process.

[0088] (4) Optimization through intelligence and data-driven approaches:

[0089] This invention achieves data-driven intelligent optimization by integrating multiple data sources (such as temperature, air pressure, air volume, and images) and combining improved simulated annealing and sparrow search algorithms. Compared to traditional cooling methods that rely on human experience, this invention improves the intelligence level of the cooling process and fully leverages the value of data resources. Attached Figure Description

[0090] Figure 1 This is a flowchart of a slag cooling process simulation method based on multi-source heterogeneous data according to the present invention;

[0091] Figure 2 A flowchart for constructing a fusion feature vector model and a slag heat conduction model;

[0092] Figure 3 A flowchart for optimizing cooling parameters. Detailed Implementation

[0093] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0094] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0095] Example 1

[0096] like Figure 1 The diagram shows a flowchart of a slag cooling process simulation method based on multi-source heterogeneous data according to the present invention, which includes the following steps:

[0097] Step S1: Collect multi-source heterogeneous data generated during boiler ash discharge;

[0098] Slag temperature sensor data: A high-precision temperature sensor array monitors the temperature distribution inside and on the surface of the slag in real time. Sensors are distributed at different locations within the cooling device to obtain comprehensive temperature information. The temperature sensors have a sampling frequency of 1Hz, a temperature range of 0℃ to 1000℃, and an accuracy of ±0.5℃.

[0099] Cooling air pressure and airflow sensor data: Air pressure and airflow in the cooling system are monitored in real time using air pressure and airflow sensors. The air pressure sensor has a range of 0–100 kPa and an accuracy of ±0.1 kPa; the airflow sensor has a range of 0–1000 m³ / h. 3 / min, with an accuracy of ±1%.

[0100] Slag velocity sensor data: The velocity sensor monitors the movement speed of the slag in the cooling device. The velocity sensor has a range of 0–5 m / s and an accuracy of ±0.05 m / s.

[0101] Slag surface image data: The industrial camera resolution is set to 1920×1080 pixels, the frame rate is 10fps, and the image acquisition range covers the entire surface of the slag cooling device.

[0102] Step S2: Preprocess the collected multi-source heterogeneous data to remove noise and outliers, and normalize the data to a uniform range to obtain preprocessed data;

[0103] Step S3: Perform time synchronization and spatial alignment processing on the preprocessed data to obtain time-synchronized and spatially aligned data, so as to eliminate the time and space differences between different sensors.

[0104] Time synchronization is handled as follows:

[0105] T sync (t)=T sensor (t+Δt);

[0106] Among them, T sync (t) represents the data after time synchronization; T sensor This represents the original time data collected by the sensor; t represents time; Δt represents the time offset, used to correct for time differences between different sensors; the time offset is adjusted according to the sensor's installation location and signal transmission delay, with a range of ±0.1S.

[0107] Spatial alignment is handled as follows:

[0108] X aligned (x,y,z)=x sensor (x+Δx,y+Δy,z+Δz);

[0109] Among them, X aligned(x,y,z) represents the spatially aligned data; (x,y,z) represents the spatial coordinates; x sensor Δx represents the spatial data collected by the original sensor; Δy represents the spatial offset in the x-axis direction; Δy represents the spatial offset in the y-axis direction; and Δz represents the spatial offset in the z-axis direction. Δx, Δy, and Δz are adjusted according to the actual installation position of the sensor, with ranges of ±0.1m, ±0.1m, and ±0.05m, respectively.

[0110] like Figure 2 The diagram shows the flowchart for constructing the fusion feature vector and the slag heat conduction model.

[0111] Step S4: Using data fusion technology, the time-synchronized and spatially aligned data are fused to extract key features and construct a fused feature vector;

[0112] Step S41: Extract slag temperature features, cooling air pressure and air volume features, slag movement speed features and slag surface image features from the time-synchronized and spatially aligned slag temperature data, cooling air pressure data, cooling air volume data, slag movement speed data and slag surface image data.

[0113] Step S42: Fuse the extracted features to construct a fused feature vector using a weighted concatenation method, as shown in the following formula:

[0114] F Fusion =[ω1×F T ;ω2×F P ;ω3×F Q ;ω4×F V ;ω5×F GLCM ;ω6×F Edge ];

[0115] Among them, F Fusion This represents the fused feature vector; ω1, ω2, ω3, ω4, ω5, and ω6 represent the weight coefficients of each feature, respectively. Indicates the temperature characteristics of the slag. ΔT represents the average temperature, and ΔT represents the temperature gradient. Indicates wind pressure characteristics, Indicates average wind pressure; Indicates air volume characteristics. Indicates average air volume; This indicates the velocity characteristics of slag movement. F represents the average velocity, ΔV represents the rate of change of velocity; GLCM Texture features are represented using the Gray-Level Co-occurrence Matrix (GLCM); F Edge The edge features are represented and obtained through Canny edge detection.

[0116] Step S43: Optimize the weight coefficients in the fused feature vector using the particle swarm optimization algorithm. The specific steps are as follows:

[0117] Step S431: Randomly initialize the position and velocity of the particles. The position of the particles is represented as w = [ω1, ω2, ω3, ω4, ω5, ω6]. Each particle represents a set of weight coefficients, and the velocity is initialized to a random value.

[0118] Step S432: Define the fitness function;

[0119] Fitness(w) = MSE(F) Fusion );

[0120] Where Fitness(w) represents the fitness function; MSE(F) Fusion ) represents the error of the fused feature vector on the validation set;

[0121] Step S433: For each particle, calculate its fitness value and update its optimal position P. best and the global optimal position g best ;

[0122] Step S434: Update particle velocity and position;

[0123] Step S435: Ensure the weighting coefficient is within 0 to 1; if it exceeds the range, adjust it randomly.

[0124] Step S436: Repeat steps S432 to S435 until the preset number of iterations is reached;

[0125] Step S437: Recalculate the fused feature vector using the optimized weight coefficients obtained from the particle swarm optimization algorithm;

[0126]

[0127] Among them, F Final This represents the optimized fused feature vector; These represent the weight coefficients of each feature after optimization;

[0128] The optimized fused feature vector is then normalized.

[0129]

[0130] Among them, F Norm min(F) represents the normalized fused feature vector; Final ), max(F Final ) represent the minimum and maximum values ​​of the optimized fused feature vector, respectively.

[0131] Step S5: Based on the fused data, an improved simulated annealing algorithm is used to establish a slag heat conduction model. The specific operation is as follows:

[0132] Step S51: Initialize the Grey Wolf Algorithm and run the Grey Wolf Algorithm to obtain the optimal solution x. GWO ;

[0133] Step S52: Initialize the simulated annealing algorithm, using x GWO As the initial solution;

[0134] Step S53: Combining the global search capability of the Grey Wolf algorithm and the local search capability of the Simulated Annealing algorithm, run the two algorithms alternately;

[0135] Step S54: Output the final thermal conductivity model parameters θ best A slag heat conduction model was constructed, as follows:

[0136]

[0137] Where T represents the temperature distribution; Represents the Laplace operator; θ best,α This represents the optimized thermal diffusivity; Q(x,t,θ) best ) represents the heat source term;

[0138]

[0139] Where ρ represents the density of the slag; c p θ represents the specific heat capacity of slag. best,λ This represents the optimized thermal conductivity.

[0140] like Figure 3 The diagram shown is a flowchart for optimizing cooling parameters.

[0141] Step S6: Combining real-time monitoring data of slag surface, internal temperature data, historical cooling data and environmental condition data, the temperature field model is dynamically updated using data assimilation technology to construct a three-dimensional model of the slag temperature field.

[0142] The three-dimensional model of the slag temperature field is as follows:

[0143]

[0144] Where T(x,y,z,t) represents the temperature distribution; Q cooling (x,y,z,t) represents the cooling heat source term; ΔT correction (x,y,z,t) represents the temperature deviation corrected by data assimilation techniques;

[0145] Q cooling The formula for calculating (x,y,z,t) is as follows:

[0146]

[0147] Where k represents a coefficient related to cooling efficiency; σ represents the spatial range of cooling influence; (x0, y0, z0) represents the center position of the cooling air action; Q' represents the heat of the cooling air; and P' represents the pressure of the cooling air.

[0148] ΔT correction The formula for calculating (x,y,z,t) is as follows:

[0149] ΔT correction (x,y,z,t)=ω[T surf (x,y,z,t)-T model (x,y,z,t)]+(1-ω)[T internal (x,y,z,t)-T model (x,y,z,t)];

[0150] Where ω represents the weighting factor, used to balance the influence of surface and internal temperature data; T model (x,y,z,t) represents the temperature field calculated based on the model; T surf (x,y,z,t) represents the real-time monitored temperature of the slag surface; T internal (x,y,z,t) represents the internal temperature data of the slag.

[0151] Step S7: Based on the constructed three-dimensional model of the slag temperature field, the optimal cooling airflow and slag movement velocity are optimized and calculated using an improved sparrow search algorithm. The specific operations are as follows:

[0152] Step S71: Define the optimization objective function;

[0153] f(Q air V slag )=α1×MSE temp +β1×MSE energy ;

[0154] Where α1 and β1 both represent weighting coefficients; MSE temp Mean square error (MSE) represents the temperature deviation. energy The mean square error representing cooling energy consumption;

[0155] Step S72: Initialize the sparrow population X = {X1, X2, ... XN}, where each sparrow represents a combination of cooling air volume and slag movement speed;

[0156] Step S73: Calculate the fitness value for each sparrow;

[0157] Step S74: Update the sparrow's position;

[0158]

[0159] in, This represents the position of the i-th sparrow in the (a+1)-th generation; γ represents the position of the i-th sparrow in generation a; γ represents the learning rate; L represents the position adjustment factor. Let represent the optimal solution in generation a;

[0160] Step S75: Ensure that the cooling air volume and slag movement speed are within a reasonable range.

[0161] Q air ∈[Q min Q max ];

[0162] V slag ∈[V min V max ];

[0163] Among them, Q min Q represents the minimum cooling airflow; max V represents the maximum cooling airflow. min V represents the minimum velocity of the slag movement. max This represents the maximum velocity of the slag.

[0164] Step S76: Repeat steps S72 to S75 until the preset number of iterations is reached;

[0165] Step S77: Output the optimal cooling airflow Q air,opt And the optimal slag movement speed V slag,opt .

[0166] It is worth noting that all contents not described in detail in this invention are existing technologies and are well known to those skilled in the art.

[0167] Therefore, the above-mentioned slag cooling process simulation method based on multi-source heterogeneous data in this invention can improve slag cooling efficiency, reduce energy consumption, and optimize the stability of the cooling process, and has significant industrial application value.

[0168] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for slag cooling process simulation based on multi-source heterogeneous data, characterized in that, The method comprises the following steps: Step S1, collecting multi-source heterogeneous data generated in the process of boiler slag discharge; Step S2, preprocessing the collected multi-source heterogeneous data to obtain preprocessed data; Step S3, performing time synchronization and space alignment processing on the preprocessed data to obtain time-synchronized and space-aligned data; Step S4, using data fusion technology to fuse the time-synchronized and space-aligned data, extract key features, and construct a fusion feature vector; Step S5, based on the fused data, using an improved simulated annealing algorithm to establish a slag heat conduction model; Step S6, combining real-time monitoring data of slag surface, internal temperature data, historical cooling data, and environmental condition data, using data assimilation technology to dynamically update the temperature field model, and constructing a three-dimensional model of the slag temperature field; Step S7, based on the constructed three-dimensional model of the slag temperature field, using an improved sparrow search algorithm to optimize the calculation of the best cooling air volume and slag movement speed.

2. The method according to claim 1, wherein, In step S1, the multi-source heterogeneous data includes: slag temperature sensor data; cooling air pressure and volume sensor data; slag movement speed sensor data; and slag surface image data.

3. The method for slag cooling process simulation based on multi-source heterogeneous data according to claim 1, characterized in that, In step S2, preprocessing includes data cleaning and data normalization.

4. The method for slag cooling process simulation based on multi-source heterogeneous data according to claim 1, characterized in that, In step S3, the time synchronization processing is as follows: T sync (t) = T sensor (t + Δt); Wherein, T sync (t) represents time-synchronized data; T sensor represents the time data collected by the original sensor; t represents time; Δt represents the time offset, used to correct the time difference of different sensors; The space alignment processing is as follows: X aligned (x,y,z) = x sensor (x+Δx,y+Δy,z+Δz); wherein X aligned (x, y, z) represents the data after spatial alignment; (x, y, z) represents the spatial coordinates; x sensor represents the spatial data collected by the original sensor; Δx represents the spatial offset in the x-axis direction; Δy represents the spatial offset in the y-axis direction; and Δz represents the spatial offset in the z-axis direction.

5. The method for slag cooling process simulation based on multi-source heterogeneous data according to claim 1, characterized in that, In step S4, using data fusion technology to fuse the time-synchronized and space-aligned data, extract key features, and construct a fusion feature vector, the specific operations are as follows: Step S41, extracting slag temperature features, cooling air pressure and volume features, slag movement speed features, and slag surface image features from the time-synchronized and space-aligned slag temperature data, cooling air pressure data, cooling air volume data, slag movement speed data, and slag surface image data; Step S42, fusing the extracted features to construct a fusion feature vector, using a weighted splicing method, the formula is as follows: F Fusion = [ω1 x F T ; ω2 x F P ; ω3 x F Q ; ω4 x F V ; ω5 x F GLCM ; ω6 x F Edge ] ; wherein, F Fusion represents a fusion feature vector; ω1, ω2, ω3, ω4, ω5, ω6 respectively represent the weight coefficients of each feature; represents a slag temperature feature, represents an average temperature, and ΔT represents a temperature gradient; represents a wind pressure feature, represents an average wind pressure; represents a wind volume feature, represents an average wind volume; represents a slag movement speed feature, represents an average speed, and ΔV represents a speed change rate;F GLCM represents a texture feature, obtained by a gray level co-occurrence matrix GLCM;F Edge represents an edge feature, obtained by Canny edge detection; Step S43, using a particle swarm algorithm to optimize the weight coefficients in the fusion feature vector.

6. The method for slag cooling process simulation based on multi-source heterogeneous data according to claim 5, characterized in that, In step S43, using a particle swarm algorithm to optimize the weight coefficients in the fusion feature vector, the specific operations are as follows: Step S431, randomly initializing the position and speed of the particles, the position of the particles is represented as w=[ω1,ω2,ω3,ω4,ω5,ω6], each particle represents a set of weight coefficients, and the speed is initialized as a random value; Step S432, defining a fitness function; Fitness(w) = MSE(F Fusion ); where Fitness(w) represents the fitness function; MSE(F Fusion ) represents the error of the fusion feature vector on the validation set; Step S433, for each particle, calculate its fitness value, and update the individual optimal position P best and the global optimal position g best ; Step S434, updating the particle speed and position; Step S435, ensuring that the weight coefficients are within 0-1, and if they exceed the range, they are randomly adjusted; Step S436, repeating steps S432-S435 until a preset number of iterations is reached; Step S437, recalculating the fusion feature vector with the weight coefficients optimized by the particle swarm algorithm; wherein F Final represents the optimized fusion feature vector; respectively represent the weight coefficients of the respective features after optimization. Normalizing the optimized fusion feature vector; where F Norm represents the normalized fusion feature vector; min(F Final ), max(F Final ) represent the minimum and maximum values of the optimized fusion feature vector, respectively.

7. The method according to claim 6, wherein, In step S5, based on the fused data, using an improved simulated annealing algorithm to establish a slag heat conduction model, the specific operations are as follows: Step S51, initializing the grey wolf algorithm, running the grey wolf algorithm to obtain the optimal solution x GWO ; Step S52, initialize the simulated annealing algorithm, using x GWO as the initial solution; Step S53, combining the global search ability of the grey wolf algorithm and the local search ability of the simulated annealing algorithm, and alternately running the two algorithms; Step S54, output the final heat conduction model parameter θ best and build a slag heat conduction model.

8. The method according to claim 7, wherein, In step S54, the slag heat conduction model is as follows: where T denotes the temperature distribution; denotes the Laplacian; θ best,α denotes the optimized thermal diffusivity; Q(x, t, θ best ) denotes the heat source term; wherein p represents the density of the slag; c p represents the specific heat capacity of the slag; θ best,λ represents the optimized thermal conductivity.

9. The method according to claim 8, wherein, In step S6, the three-dimensional model of the slag temperature field is as follows: where T(x, y, z, t) represents the temperature distribution; Q cooling (x, y, z, t) represents the cooling heat source term; ΔT correction (x, y, z, t) represents the temperature bias corrected by the data assimilation technique; Q cooling The calculation formula of (x, y, z, t) is as follows: wherein k represents a coefficient related to the cooling efficiency; σ represents the spatial range of the cooling influence; (x0, y0, z0) represents the central position of the cooling air action; Q' represents the heat of the cooling air; P' represents the pressure of the cooling air; ΔT correction The calculation formula of (x, y, z, t) is as follows: ΔT correction (x,y,z,t) = ω [T surf (x,y,z,t) - T model (x,y,z,t)] + (1 - ω) [T internal (x,y,z,t) - T model (x,y,z,t)] ; where ω represents a weight factor to balance the influence of surface and internal temperature data; T model (x, y, z, t) represents the temperature field calculated based on the model; T surf (x, y, z, t) represents the real-time monitoring temperature of the slag surface; T internal (x, y, z, t) represents the internal temperature data of the slag.

10. The method for slag cooling process simulation based on multi-source heterogeneous data according to claim 9, characterized in that, In step S7, based on the constructed three-dimensional model of the slag temperature field, the improved sparrow search algorithm is used to optimize and calculate the optimal cooling air volume and the slag movement speed, and the specific operation is as follows: Step S71, define the optimization objective function; f(Q air ,V slag ) = a1 x MSE temp + b1 x MSE energy ; Wherein, α1, β1 represent weight coefficients; MSE temp MSE represents the mean square error of temperature deviation; MSE energy MSE represents the mean square error of cooling energy consumption; Step S72, initialize the sparrow population X={X1, X2,...XN}, and each sparrow represents a combination of cooling air volume and slag movement speed; Step S73, calculate the fitness value of each sparrow; Step S74, update the sparrow position; wherein, represents the position of the i-th sparrow at the a+1th generation; represents the position of the i-th sparrow at the ath generation; γ represents a learning rate; and L represents a position adjustment factor; represents the optimal solution of the ath generation; Step S75, ensure that the cooling air volume and the slag movement speed are within a reasonable range: Q air ∈[Q min ,Q max ] V slag ∈[V min ,V max ] where Q min represents the minimum value of the cooling air volume; Q max represents the maximum value of the cooling air volume; V min represents the minimum value of the slag movement speed; V max represents the maximum value of the slag movement speed; Step S76, repeat steps S72-S75 until the preset iteration number is reached; Step S77, outputting the optimal cooling air volume Q air,opt and the optimal slag movement speed V slag,opt .

Images