Method and equipment for optimizing uniform temperature rise of frozen coal train based on Morris-CNN (Convolutional Neural Network) fusion

Abstract

The invention relates to a frozen coal train uniform temperature rise optimization method based on Morris-CNN fusion. The method comprises the steps that model structure parameters and electromagnetic parameters are determined; a Morris method is adopted, and parameters are selected; obtaining a sample space; obtaining a plurality of grid units; obtaining the temperature value of the temperature measuring point; calculating temperature rise efficiency and uniformity indexes; forming a sample data set to train a convolutional neural network; searching values of five parameters corresponding to the optimal temperature rise efficiency and the optimal uniformity; and the final optimal temperature rise efficiency and the optimal uniformity are obtained. According to the method, a structure optimization design based on the combination of Morris global sensitivity analysis and the convolutional neural network is adopted, the method combines a mathematical method and a statistical method, modeling and analysis are carried out on the influence of multiple variables, and the efficiency and the effect are improved; according to the method, the temperature rise uniformity and the temperature rise efficiency are optimized by applying a multi-objective optimization method, the gradient descent characteristic of the convolutional neural network is utilized, the uniformity and the temperature rise efficiency are balanced, and the optimal parameters are obtained to guide the quick and uniform temperature rise design of the frozen coal train.

Description

Technical Field

[0001] The present invention relates to the technical field of optimizing temperature rise efficiency and uniformity of electromagnetic heating frozen coal melting ice, and in particular to a method and device for optimizing uniform temperature rise of a frozen coal train based on Morris-CNN fusion. Background Art

[0002] Winter coal transportation efficiency is significantly hampered by the problem of frozen coal. In low temperatures, residual moisture in the coal freezes and adheres to the railcar surfaces, resulting in inefficient unloading. Currently, manual cleaning is the only method used, which is labor-intensive, dangerous, and time-consuming. Furthermore, residual frozen coal can easily lead to overloading of trains, increasing safety risks such as derailments.

[0003] The main treatment methods currently available include manual cleaning, mechanical cutting, vibration, and electroosmosis. Manual cleaning primarily involves shoveling, digging, and hammering. While the advantages are simple implementation and the lack of equipment, the disadvantages are high labor costs, a long cleaning cycle, and low efficiency. Mechanical cutting utilizes metal brushes, hobs, or specialized cutting equipment such as shovels and pliers to forcibly remove the frozen coal. Although this method improves efficiency, it causes significant damage to the carriage. Vibration methods, including electromagnetic vibration and electric vibration, cause the mine car to resonate during reversal, thereby dislodging the frozen coal. However, the frozen coal cannot be completely shaken off, and the vibration motor has a short lifespan. Electroosmosis uses the frozen coal binder in the mine car as a medium, the carriage as a cathode, and a specially designed anode in contact with the frozen coal. This method requires high moisture content in the frozen coal, moves slowly, and is ineffective. Summary of the Invention

[0004] In order to solve the problems of low efficiency of frozen coal removal in existing trains and the conventional electromagnetic heating method, which makes it impossible to achieve uniform temperature rise in the frozen coal carriage due to low heat in the middle and low temperature at both ends, the primary purpose of the present invention is to provide a uniform temperature rise optimization method for frozen coal trains based on Morris-CNN fusion, which optimizes the temperature rise uniformity and temperature rise efficiency through a multi-objective optimization method, balances the uniformity and temperature rise efficiency, and obtains the optimal parameters to guide the design of rapid and uniform temperature rise of frozen coal trains.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a method for optimizing the uniform temperature rise of a frozen coal train based on Morris-CNN fusion, the method comprising the following steps in sequence:

[0006] (1) Determine the model structure parameters and electromagnetic parameters, and establish a frozen coal train electromagnetic heating simulation model based on the train carrying frozen coal and the induction heating coil; the model structure parameters include the left and right coupling distance, the lower coupling distance, the upper coupling distance, the longitudinal length of the coil around the car, the number of coil partitions, and the distribution of the number of turns within the coil partition; the electromagnetic parameters include ampere-turns, electromagnetic induction heating current, and frequency;

[0007] (2) Using the Morris method, select parameters from the model structural parameters and electromagnetic parameters, and select the top five parameters that have the greatest impact on temperature rise efficiency and uniformity;

[0008] (3) Based on the first five parameters screened out, the maximum and minimum distance Latin hypercube sampling method is used to obtain the sample space;

[0009] (4) A differentiated meshing strategy is used for the electromagnetic heating simulation model of the frozen coal train. The mesh of small components is adaptively meshed, the boundary layer is optimized, the cylindrical structure is swept, and the mesh is divided to obtain multiple mesh units.

[0010] (5) Select temperature measurement points at equal intervals in the grid cells. The selected temperature measurement points evenly cover the interface between the carriage and the frozen coal, and obtain the temperature values ​​of the temperature measurement points.

[0011] (6) Calculate the temperature rise efficiency and uniformity index of the train carriage carrying frozen coal based on the temperature values ​​at each temperature measurement point;

[0012] (7) The sample space is mapped one-to-one to the temperature rise efficiency and uniformity index of the train carriages carrying frozen coal to form a sample data set. The sample data set is used to train the convolutional neural network to obtain the trained convolutional neural network, and the relationship between the input parameters and the output values ​​is constructed;

[0013] (8) Using the gradient descent properties of the convolutional neural network itself, find the values ​​of the five parameters corresponding to the optimal temperature rise efficiency and optimal uniformity of the trained convolutional neural network;

[0014] (9) The values ​​of the five parameters corresponding to the optimal temperature rise efficiency and optimal uniformity of the trained convolutional neural network are input into the electromagnetic heating simulation model of the frozen coal train and recalculated to obtain the final optimal temperature rise efficiency and optimal uniformity.

[0015] In step (1), the electromagnetic induction heating current and frequency are selected according to the skin depth formula:

[0016] ;

[0017] Where: δ is the penetration depth, μ is the relative magnetic permeability, f is the frequency, and ρ is the resistivity;

[0018] Based on the thickness of the frozen coal train carriage that requires electromagnetic induction heating, the electromagnetic induction heating current range is estimated according to the heating power requirement, and the ampere-turns are estimated based on the required magnetic field strength. According to railway safety regulations, the left and right coupling distances, lower coupling distances, and upper coupling distances between the coil and the carriage are selected. Based on the geometric parameters of the frozen coal train carriage, the longitudinal length of the coil around the carriage, the number of coil partitions, and the distribution of the number of turns within the coil partitions are determined, and finally a frozen coal train electromagnetic heating simulation model is established.

[0019] Step (2) specifically refers to: using the Morris method, selecting parameters from the model structural parameters and electromagnetic parameters, and screening key variables by calculating the basic influence of the parameters and comparing them. The basic influence EE is expressed as:

[0020] ;

[0021] Where, Indicates the The basic influence of the parameters; Represents the model structural parameters and electromagnetic parameters; Indicates the small perturbation of the model structure parameters and electromagnetic parameters; Indicates the temperature value at which the model structural parameters and electromagnetic parameters do not undergo small perturbation changes; X is the parameter that does not undergo perturbation; when the basic influence The larger the value, the greater the impact on the heating effect.

[0022] In step (3), the filtered parameters are evenly divided into N1 equal parts, and a parameter is randomly selected from each part. A random permutation is independently generated for each part to form an N1×D sample space, where D is the required number of samples. The minimum coupling distance is defined as the minimum value of the spacing between all pairs of sample points:

[0023] ;

[0024] Where: is the coordinate of the i-th sample point; is the coordinate of the jth sample point;

[0025] By repeatedly selecting parameters randomly, a new sample space is constructed, and finally Maximize and obtain the Latin hypercube sample space with the maximum and minimum distance.

[0026] In step (6), the temperature rise efficiency index The calculation formula is as follows:

[0027] ;

[0028] Where, is the temperature of the grid point of the grid unit after heating, is the temperature of the grid point before heating, is the average temperature, is the heating grid unit area;

[0029] Uniformity index for:

[0030] ;

[0031] Where N is the number of surfaces in the cabin heating area in the electromagnetic heating simulation model where the grid unit is located, is the area of ​​the i-th heating grid unit.

[0032] In step (7), when training the convolutional neural network, a fixed learning rate scheduling mechanism is used to set the initial learning rate to 0.001. Relying on the adaptive characteristics of the Adam optimizer, the gradient update direction is dynamically adjusted within 50 training rounds. A batch dynamic allocation strategy is used. Based on the total number of samples and memory limitations, the number of iterations per round is set to 6 times, and a total of 300 parameter updates are completed. Finally, a trained convolutional neural network with a relationship between input parameters and output parameters is obtained.

[0033] Another object of the present invention is to provide an electronic device, comprising:

[0034] processor; and

[0035] A memory, wherein computer program instructions are stored in the memory, and when the computer program instructions are executed by the processor, the processor executes the above-mentioned method for optimizing the uniform temperature rise of a frozen coal train based on Morris-CNN fusion.

[0036] The present invention also provides a computer-readable storage medium having computer program instructions stored thereon. When the computer program instructions are executed by a processor, the processor executes the above-mentioned method for optimizing the uniform temperature rise of a frozen coal train based on Morris-CNN fusion.

[0037] It can be seen from the above technical solution that the beneficial effects of the present invention are: First, the present invention is based on an efficient global sensitivity analysis method for trajectory sampling, which realizes key variable screening by calculating the mean and standard deviation of the basic effects of the parameters, effectively screening the key parameters of electromagnetic heating and reducing the dimension of the convolutional neural network; Second, the present invention adopts a structural optimization design based on the combination of Morris global sensitivity analysis and convolutional neural network. This method combines mathematical methods and statistical methods to model and analyze the influence of multiple variables, thereby improving efficiency and effect; Third, the present invention uses a multi-objective optimization method to optimize the temperature rise uniformity and temperature rise efficiency, and uses the gradient descent characteristics of the convolutional neural network to balance the uniformity and temperature rise efficiency, and obtain the optimal parameters to guide the rapid and uniform temperature rise design of the frozen coal train. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 is a flow chart of the present invention;

[0039] Figure 2 Schematic diagram of Morris sensitivity analysis in Example 1 of the present invention;

[0040] Figure 3This is a schematic diagram of the coil distribution structure division in the first embodiment of the present invention;

[0041] Figure 4 This is a schematic diagram of the arrangement of temperature measurement points in the carriage side wall model in the first embodiment of the present invention;

[0042] Figure 5 This is a diagram illustrating how the loss value of a convolutional neural network changes with iteration in the first embodiment of the present invention;

[0043] Figure 6 This is a comparison chart of temperature uniformity before and after optimization in Example 1 of the present invention. DETAILED DESCRIPTION

[0044] like Figure 1 As shown in FIG, a method for optimizing uniform temperature rise of a frozen coal train based on Morris-CNN fusion is provided, which includes the following steps in sequence:

[0045] (1) Determine the model structure parameters and electromagnetic parameters, and establish a frozen coal train electromagnetic heating simulation model based on the train carrying frozen coal and the induction heating coil; the model structure parameters include the left and right coupling distance, the lower coupling distance, the upper coupling distance, the longitudinal length of the coil around the car, the number of coil partitions, and the distribution of the number of turns within the coil partition; the electromagnetic parameters include ampere-turns, electromagnetic induction heating current, and frequency;

[0046] (2) Using the Morris method, select parameters from the model structural parameters and electromagnetic parameters, and select the top five parameters that have the greatest impact on temperature rise efficiency and uniformity;

[0047] (3) Based on the first five parameters screened out, the maximum and minimum distance Latin hypercube sampling method is used to obtain the sample space;

[0048] (4) A differentiated meshing strategy is used for the electromagnetic heating simulation model of the frozen coal train. The mesh of small components is adaptively meshed, the boundary layer is optimized, the cylindrical structure is swept, and the mesh is divided to obtain multiple mesh units.

[0049] (5) Select temperature measurement points at equal intervals in the grid cells. The selected temperature measurement points evenly cover the interface between the carriage and the frozen coal, and obtain the temperature values ​​of the temperature measurement points.

[0050] (6) Calculate the temperature rise efficiency and uniformity index of the train carriage carrying frozen coal based on the temperature values ​​at each temperature measurement point;

[0051] (7) The sample space is mapped one-to-one to the temperature rise efficiency and uniformity index of the train carriages carrying frozen coal to form a sample data set. The sample data set is used to train the convolutional neural network to obtain the trained convolutional neural network, and the relationship between the input parameters and the output values ​​is constructed;

[0052] (8) Using the gradient descent properties of the convolutional neural network itself, find the values ​​of the five parameters corresponding to the optimal temperature rise efficiency and optimal uniformity of the trained convolutional neural network;

[0053] (9) The values ​​of the five parameters corresponding to the optimal temperature rise efficiency and optimal uniformity of the trained convolutional neural network are input into the electromagnetic heating simulation model of the frozen coal train and recalculated to obtain the final optimal temperature rise efficiency and optimal uniformity.

[0054] In step (1), the electromagnetic induction heating current and frequency are selected according to the skin depth formula:

[0055] ;

[0056] Where: δ is the penetration depth, μ is the relative magnetic permeability, f is the frequency, and ρ is the resistivity;

[0057] Based on the thickness of the frozen coal train carriage that requires electromagnetic induction heating, the electromagnetic induction heating current range is estimated according to the heating power requirement, and the ampere-turns are estimated based on the required magnetic field strength. According to railway safety regulations, the left and right coupling distances, lower coupling distances, and upper coupling distances between the coil and the carriage are selected. Based on the geometric parameters of the frozen coal train carriage, the longitudinal length of the coil around the carriage, the number of coil partitions, and the distribution of the number of turns within the coil partitions are determined, and finally a frozen coal train electromagnetic heating simulation model is established.

[0058] Step (2) specifically refers to: using the Morris method, selecting parameters from the model structural parameters and electromagnetic parameters, and screening key variables by calculating the basic influence of the parameters and comparing them. The basic influence EE is expressed as:

[0059] ;

[0060] Where, Indicates the The basic influence of the parameters; Represents the model structural parameters and electromagnetic parameters; Indicates the small perturbation of the model structure parameters and electromagnetic parameters; Indicates the temperature value at which the model structural parameters and electromagnetic parameters do not undergo small perturbation changes; X is the parameter that does not undergo perturbation; when the basic influence The larger the value, the greater the impact on the heating effect.

[0061] In step (3), the filtered parameters are evenly divided into N1 equal parts, and a parameter is randomly selected from each part. A random permutation is independently generated for each part to form an N1×D sample space, where D is the required number of samples. The minimum coupling distance is defined as the minimum value of the spacing between all pairs of sample points:

[0062] ;

[0063] Where: is the coordinate of the i-th sample point; is the coordinate of the jth sample point;

[0064] By repeatedly selecting parameters randomly, a new sample space is constructed, and finally Maximize and obtain the Latin hypercube sample space with the maximum and minimum distance.

[0065] In step (6), the temperature rise efficiency index The calculation formula is as follows:

[0066] ;

[0067] Where, is the temperature of the grid point of the grid unit after heating, is the temperature of the grid point before heating, is the average temperature, is the heating grid unit area;

[0068] Uniformity index for:

[0069] ;

[0070] Where N is the number of surfaces in the cabin heating area in the electromagnetic heating simulation model where the grid unit is located, is the area of ​​the i-th heating grid unit.

[0071] In step (7), when training the convolutional neural network, a fixed learning rate scheduling mechanism is used to set the initial learning rate to 0.001. Relying on the adaptive characteristics of the Adam optimizer, the gradient update direction is dynamically adjusted within 50 training rounds to avoid the computational burden of manual learning rate decay. A batch dynamic allocation strategy is adopted. Based on the total number of samples and memory limitations, the number of iterations per round is set to 6, and a total of 300 parameter updates are completed. The small batch gradient descent is used to balance computational efficiency and gradient stability. The spatial dimension compression of the pooling layer and the shallow network architecture are used to limit the model complexity, effectively suppressing the risk of overfitting, and finally obtaining a trained convolutional neural network that constructs the relationship between input and output parameters.

[0072] Example 1

[0073] Based on the real freight train and electromagnetic induction coil, the simulation model is determined and constructed according to its real shape and physical parameters.

[0074] The parameters affecting train coal freezing and their probability distribution are shown in Table 1:

[0075] Table 1

[0076]

[0077] like Figure 2 As shown, the number of coil partitions, the degree of dispersion in the normal distribution of coil turns, the selection of key parameters of V distribution and U distribution, ampere-turns, and the longitudinal length of the coil are factors with greater influence. When other factors such as ampere-turns remain unchanged, the change in current has little effect.

[0078] Figure 3 This is a diagram illustrating the coil zoning and coil laying methods within the coil zones of a frozen coal train. The overall trend in coil laying is sparse in the middle and dense at both ends.

[0079] The entire model is coarsely meshed based on the basic precision grid of the electromagnetic field. On this basis, adaptive meshing is performed on complex precision components such as the bogie. The coarse mesh of the entire model is refined with four times the grid density to increase the calculation accuracy of the electromagnetic heating simulation model. The rail area is subjected to sweep optimization meshing. The cylindrical area is subjected to swept boundary layer optimization to make the electromagnetic field calculation value accurate. The carriage part is meshed with a boundary layer according to the electromagnetic induction heating penetration depth.

[0080] The side walls and bottom of the carriage are divided into 75 uniform temperature measurement areas, such as Figure 4 As shown in the figure, the integral value of the temperature surface of each area is calculated as the average temperature of the area, and the temperature of each partition is compared with the surface integral temperature.

[0081] After calculating the temperature deviation using the temperature difference between the two side walls and the bottom of the carriage as follows, the average of the temperature deviations on the three surfaces is selected as the impact on temperature uniformity in this situation. The larger the deviation value, the greater the impact.

[0082] The input data dimension is 1 in height and 4 in width, including the length of the coil wrapped around the car, the U-shaped distribution of the coil turns, the V-shaped distribution of the coil turns, and the normal distribution of the coil turns. Construct 32 convolution kernels of size 1×2, which do not move in the height direction and cover 2 adjacent feature bands in the width direction to capture local lateral correlation. Then use ReLU to set negative values ​​to zero, retain positive features, and enhance nonlinear expression capabilities. Construct 64 neurons and link all outputs of the previous layer to the magnetic layer to achieve feature fusion and high-order abstraction. Finally, go to the output layer of two neurons, corresponding to two continuous output values, and its loss value is as follows Figure 5 shown.

[0083] Based on the uniformity of temperature rise and the optimal efficiency, the optimal distribution of the model coils and the optimal length of the coils around the car are found.

[0084] The optimal coil distribution and coil winding length are introduced into the electromagnetic heating simulation model of the frozen coal train, such as Figure 6As shown, a comparative analysis of their efficiency and uniformity is conducted.

[0085] In summary, the present invention adopts an efficient global sensitivity analysis method based on trajectory sampling, which realizes key variable screening by calculating the mean and standard deviation of the basic effects of parameters, effectively screens the key parameters of electromagnetic heating and reduces the dimension of the convolutional neural network; the present invention adopts a structural optimization design based on the combination of Morris global sensitivity analysis and convolutional neural network. This method combines mathematical methods and statistical methods to model and analyze the influence of multiple variables, thereby improving efficiency and effect; the present invention uses a multi-objective optimization method to optimize the temperature rise uniformity and temperature rise efficiency, utilizes the gradient descent characteristics of the convolutional neural network, balances uniformity and temperature rise efficiency, and obtains the optimal parameters to guide the rapid and uniform temperature rise design of the frozen coal train.

[0086] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions merely illustrate the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for optimizing uniform temperature rise of frozen coal trains based on Morris-CNN fusion, characterized by: The method comprises the following steps in sequence: (1) Determine the model structure parameters and electromagnetic parameters, and establish a frozen coal train electromagnetic heating simulation model based on the train carrying frozen coal and the induction heating coil; the model structure parameters include the left and right coupling distance, the lower coupling distance, the upper coupling distance, the longitudinal length of the coil around the car, the number of coil partitions, and the distribution of the number of turns within the coil partition; the electromagnetic parameters include ampere-turns, electromagnetic induction heating current, and frequency; (2) Using the Morris method, select parameters from the model structural parameters and electromagnetic parameters, and select the top five parameters that have the greatest impact on temperature rise efficiency and uniformity; (3) Based on the first five parameters screened out, the maximum and minimum distance Latin hypercube sampling method is used to obtain the sample space; (4) A differentiated meshing strategy is used for the electromagnetic heating simulation model of the frozen coal train. The mesh of small components is adaptively meshed, the boundary layer is optimized, the cylindrical structure is swept, and the mesh is divided to obtain multiple mesh units. (5) Select temperature measurement points at equal intervals in the grid cells. The selected temperature measurement points evenly cover the interface between the carriage and the frozen coal, and obtain the temperature values ​​of the temperature measurement points. (6) Calculate the temperature rise efficiency and uniformity index of the train carriage carrying frozen coal based on the temperature values ​​at each temperature measurement point; (7) The sample space is mapped one-to-one to the temperature rise efficiency and uniformity index of the train carriages carrying frozen coal to form a sample data set. The sample data set is used to train the convolutional neural network to obtain the trained convolutional neural network, and the relationship between the input parameters and the output values ​​is constructed; (8) Using the gradient descent properties of the convolutional neural network itself, find the values ​​of the five parameters corresponding to the optimal temperature rise efficiency and optimal uniformity of the trained convolutional neural network; (9) The values ​​of the five parameters corresponding to the optimal temperature rise efficiency and optimal uniformity of the trained convolutional neural network are input into the electromagnetic heating simulation model of the frozen coal train and recalculated to obtain the final optimal temperature rise efficiency and optimal uniformity.

2. The method for optimizing uniform temperature rise of frozen coal trains based on Morris-CNN fusion according to claim 1 is characterized in that: In step (1), the electromagnetic induction heating current and frequency are selected according to the skin depth formula: ； Where: δ is the penetration depth, μ is the relative magnetic permeability, f is the frequency, and ρ is the resistivity; Based on the thickness of the frozen coal train carriage that requires electromagnetic induction heating, the electromagnetic induction heating current range is estimated according to the heating power requirement, and the ampere-turns are estimated based on the required magnetic field strength. According to railway safety regulations, the left and right coupling distances, lower coupling distances, and upper coupling distances between the coil and the carriage are selected. Based on the geometric parameters of the frozen coal train carriage, the longitudinal length of the coil around the carriage, the number of coil partitions, and the distribution of the number of turns within the coil partitions are determined, and finally a frozen coal train electromagnetic heating simulation model is established.

3. The method for optimizing uniform temperature rise of frozen coal trains based on Morris-CNN fusion according to claim 1 is characterized in that: Step (2) specifically refers to: using the Morris method, selecting parameters from the model structural parameters and electromagnetic parameters, and screening key variables by calculating the basic influence of the parameters and comparing them. The basic influence EE is expressed as: ； Where, Indicates the The basic influence of the parameters; Represents the model structural parameters and electromagnetic parameters; Indicates the small perturbation of the model structure parameters and electromagnetic parameters; Indicates the temperature value at which the model structural parameters and electromagnetic parameters do not undergo small perturbation changes; X is the parameter that does not undergo perturbation; when the basic influence The larger the value, the greater the impact on the heating effect.

4. The method for optimizing uniform temperature rise of frozen coal trains based on Morris-CNN fusion according to claim 1 is characterized in that: In step (3), the filtered parameters are evenly divided into N1 equal parts, and a parameter is randomly selected from each part. A random permutation is independently generated for each part to form an N1×D sample space, where D is the required number of samples. The minimum coupling distance is defined as the minimum value of the spacing between all pairs of sample points: ； Where: is the coordinate of the i-th sample point; is the coordinate of the jth sample point; By repeatedly selecting parameters randomly, a new sample space is constructed, and finally Maximize and obtain the Latin hypercube sample space with the maximum and minimum distance.

5. The method for optimizing uniform temperature rise of frozen coal trains based on Morris-CNN fusion according to claim 1 is characterized in that: In step (6), the temperature rise efficiency index The calculation formula is as follows: ； Where, is the temperature of the grid point of the grid unit after heating, is the temperature of the grid point before heating, is the average temperature, is the heating grid unit area; Uniformity index for: ； Where N is the number of surfaces in the cabin heating area in the electromagnetic heating simulation model where the grid unit is located, is the area of ​​the i-th heating grid unit.

6. The method for optimizing uniform temperature rise of frozen coal trains based on Morris-CNN fusion according to claim 1 is characterized by: In step (7), when training the convolutional neural network, a fixed learning rate scheduling mechanism is used to set the initial learning rate to 0.

001. Relying on the adaptive characteristics of the Adam optimizer, the gradient update direction is dynamically adjusted within 50 training rounds. A batch dynamic allocation strategy is used. Based on the total number of samples and memory limitations, the number of iterations per round is set to 6 times, and a total of 300 parameter updates are completed. Finally, a trained convolutional neural network with a relationship between input parameters and output parameters is obtained.

7. An electronic device comprising: processor; as well as A memory, wherein computer program instructions are stored in the memory, and when the computer program instructions are executed by the processor, the processor executes the uniform temperature rise optimization method for a frozen coal train based on Morris-CNN fusion according to any one of claims 1 to 6.

8. A computer-readable storage medium having computer program instructions stored thereon, wherein when the computer program instructions are executed by a processor, the processor executes the method for optimizing uniform temperature rise of a frozen coal train based on Morris-CNN fusion according to any one of claims 1 to 6.

Images