Heating furnace temperature optimization method based on steel temperature prediction model and improved PSO
By combining the steel temperature prediction model and the improved PSO algorithm, the heating furnace temperature setting value is optimized, and the real-time follow-up problem of heating furnace temperature regulation is solved, the heating quality and energy efficiency of the billet are improved, and intelligent temperature adjustment is achieved.
Patent Information
- Application Number
- CN202410913219.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-09
- Publication Date
- 2024-11-08
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The temperature regulation of existing heating furnaces mainly relies on manual experience, and the billet heating cannot be followed up in time, resulting in overfire or underfire of the steel billet, and the inability to measure the temperature in the furnace in real time, affecting product quality and heating furnace performance.
The heating furnace temperature optimization method based on the steel temperature prediction model and the improved PSO (particle swarm optimization) algorithm is adopted. By establishing a temperature field model and an adaptive simulated annealed particle swarm algorithm, the furnace temperature setting value is optimized to achieve intelligent temperature adjustment.
The heating quality of steel billets is improved, energy consumption and carbon emissions are reduced, heating effect is ensured, over-burning or under-burning is avoided, and the accuracy and efficiency of furnace temperature optimization is enhanced.
Smart Images

Figure CN118917172A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of combustion optimization of a steel rolling heating furnace, and in particular to a heating furnace temperature optimization method based on a steel temperature prediction model and an improved PSO. Background Art
[0002] In the industrial production process, the heating furnace is a widely used equipment used to heat the steel billet to a specific temperature to meet the needs of subsequent processing. The temperature of the steel billet has an important influence on the performance of the heating furnace and the quality of the product. Too high a temperature may cause the steel billet to overheat, deform or oxidize, while too low a temperature will affect the heating effect of the steel billet and reduce the product quality. However, due to the harsh environment in the furnace, the temperature of the steel billet cannot be directly measured, so it is impossible to judge the heating condition of the steel billet. It is too late to wait until it is out of the furnace. Therefore, combining a high-precision steel billet temperature prediction model to predict the temperature change process of the steel billet in the furnace is the key to the furnace temperature optimization process. At present, the furnace temperature setting is mostly set by manual experience in the process of regulating the furnace temperature, and it is impossible to follow up the change process in time, which may cause waste or insufficient heating, or even damage the heating furnace. Therefore, it is crucial to optimize the furnace temperature setting value of the heating furnace in combination with the steel temperature predicted by the temperature field model for the production of the heating furnace.
[0003] Xu Zhishun published in CN116184807 A that used RBF neural network to identify the model of each section of the heating furnace, and then used the PID parameters of the two-degree-of-freedom PID controller to adjust and obtain the optimal solution. This method only looks for the optimal setting value of the furnace temperature based on the heating furnace, without considering the heating condition of the steel billet, which may cause the furnace temperature to not match the heating condition of the steel billet, thereby causing the steel billet to be over-burned or under-burned. To avoid this situation, the present invention combines the steel billet prediction model to optimize the furnace temperature setting value.
[0004] Aiming at the temperature optimization demand of the heating furnace, the present invention deeply analyzes the production process flow and develops an innovative method for optimizing the temperature setting value of the heating furnace. The method combines the steel temperature prediction model and the adaptive simulated annealing particle swarm algorithm (ASAPSO) to realize the intelligent adjustment of the temperature of the heating furnace. Through this optimization, not only the heating quality of the steel billet is improved, but also the energy consumption, carbon emissions and oxidation losses are reduced. In order to accurately and quickly solve the problem of heating furnace temperature optimization, a new algorithm is used. The algorithm overcomes the problem that the traditional particle swarm algorithm is prone to fall into the local optimal solution by introducing the adaptive simulated annealing mechanism. The ASAPSO algorithm improves the accuracy and convergence speed of the calculation, effectively improves the ability to solve the furnace temperature optimization problem, and can quickly and accurately determine the optimal temperature setting of each furnace section. In terms of constructing a prediction model for the temperature distribution of the steel billet, the present invention adopts a data-driven method, establishes the relationship between the operating parameters and production conditions and the heat exchange coefficient, realizes the online identification of the heat exchange coefficient, and thus forms a steel billet temperature prediction model, which provides a solid model support for optimizing the furnace temperature setting value. Summary of the invention
[0005] The object of the present invention is to provide a heating furnace temperature optimization method based on a steel temperature prediction model and an improved PSO to solve the problems raised in the above background technology.
[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a heating furnace temperature optimization method based on a steel temperature prediction model and an improved PSO, comprising: collecting and inputting relevant information of a heating furnace and a steel billet; establishing a temperature field model, and using the temperature field model to predict the temperature of the steel billet in the furnace; constructing a heating furnace temperature optimization objective function, and constructing constraint conditions based on the temperature field model; and using an adaptive simulated annealing particle swarm algorithm to optimize and search for the optimal setting value of the furnace temperature to determine the optimal furnace temperature setting for each section.
[0007] As a preferred scheme of the heating furnace temperature optimization method based on the steel temperature prediction model and improved PSO described in the present invention, wherein: the establishment of the temperature field model and the use of the temperature field model to predict the temperature of the steel billet in the furnace include: equipping the heating furnace mouth with an infrared thermometer to obtain the steel billet outlet temperature in time, and improving the model prediction accuracy by comparing and correcting it with the predicted temperature.
[0008] As an optimal solution of the heating furnace temperature optimization method based on the steel temperature prediction model and improved PSO described in the present invention, the collection and input of relevant information of the heating furnace and the steel billet includes: size (length, width, height), density, thermal conductivity index (thermal conductivity coefficient) and specific heat capacity; the structural parameters of the heating furnace involve the length of the furnace body and the specific structural design; the working conditions and operation data in the production process are composed of the initial furnace entry temperature, the final furnace exit temperature, the rolling frequency, the heating cycle, the output, the furnace pressure, the heating strategy, the frequency of the steel billet exiting the furnace and the substances produced after combustion; in order to avoid the temperature field model changing with time due to the temperature in the heating furnace, a non-steady-state three-dimensional temperature field mathematical model is used to construct a non-steady-state heat conduction equation for the steel billet, including setting boundary conditions, and solving the steel billet temperature according to the boundary conditions.
[0009] As a preferred solution of the heating furnace temperature optimization method based on the steel temperature prediction model and the improved PSO described in the present invention, the unsteady three-dimensional temperature field mathematical model constructs the unsteady heat conduction equation of the steel billet, including:
[0010] When energy is transferred inside the billet without loss, the billet is defined as a microelement cuboid, and the heating conditions of each surface of the billet are analyzed to construct a temperature field model. The three-dimensional form of the differential equation for heat conduction of the microelement cuboid is obtained:
[0011]
[0012] Furthermore, Q is the heat source inside the object, and there is no heat generated by phase change inside the billet. The above formula can be simplified as:
[0013]
[0014] Among them, x, y, z are the length, width and height of the steel billet respectively; ρ is the density of the steel plate kg / m 3 ;c p is the specific heat capacity of the steel plate (w / kg℃); k is the thermal conductivity of the steel plate (w / m℃); T(x,y,z,t) is the three-dimensional spatial temperature distribution of the steel plate at time t; t is time;
[0015] The formula for the boundary condition is:
[0016]
[0017] Where ε is the overall heat absorption rate, T env is the ambient temperature;
[0018] The process of solving the temperature of the steel billet according to the set boundary conditions includes:
[0019] The explicit difference method is used to solve the equation. First, the space is discretized, the cuboid is divided into three-dimensional grids, and then the time is discretized. The equation is discretized using the central difference method, and finally each time step is iteratively solved according to the boundary conditions. The temperature distribution of the billet is obtained.
[0020] As a preferred solution of the heating furnace temperature optimization method based on the steel temperature prediction model and the improved PSO described in the present invention, the optimal overall heat absorption rate is obtained according to the BP neural network identification, including:
[0021] The heating furnace conditions and billet properties are used as input, and the overall heat absorption rate is used as output;
[0022] Initialize BP neural network;
[0023] The data is processed by forward propagation to output the summary heat absorption rate;
[0024] Compare the total heat absorption rate with the actual value to detect whether the error meets the standard. If it meets the standard, this process is terminated. If it does not meet the standard, the next step will be continued;
[0025] Calculate the loss;
[0026] Back propagation updates the weights of each layer of neurons;
[0027] Check whether the number of iterations reaches the limit, if it reaches the limit, output the result, if not, repeat the cycle of processing data through forward propagation, outputting the summary heat absorption rate to the back propagation to update the weights of each layer of neurons.
[0028] The actual value of the overall heat absorption rate is calculated by combining the temperature measured by the furnace mouth temperature measuring device with a three-dimensional non-steady-state heat conduction model.
[0029] As a preferred solution of the heating furnace temperature optimization method based on the steel temperature prediction model and the improved PSO described in the present invention, the furnace temperature distribution model is subjected to function integration processing in the furnace length direction as the objective function of the optimization problem, and the formula of the function integration processing is:
[0030]
[0031] Where J represents the objective function; s represents the distance from the internal position of the hot furnace to the furnace entrance; T f (s) represents the furnace temperature of the steel billet at a distance s from the furnace mouth;
[0032] Furthermore, the constraint expression of the furnace temperature setting value optimization problem is:
[0033] |T s -T * |≤ΔT s
[0034] |T s -T c |≤ΔT d
[0035] T f min_j ≤T f_j ≤T f max_j
[0036] Among them, T * ΔT is the process required temperature value (℃) reached by the surface temperature of the steel billet after heat treatment; s T is the maximum deviation temperature between the surface temperature of the billet out of the furnace and the set temperature (℃); s With T c represents the surface temperature and internal temperature of the billet out of the furnace calculated by the billet temperature prediction model (℃); ΔT d is the maximum upper limit of the temperature difference of the steel billet section out of the furnace (℃); T f_j is the furnace temperature value of each position point in different areas inside the heating furnace (℃); T f max_j and T f min_j The upper and lower limits of the furnace temperature range in furnace zone j (℃), where furnace zone j indicates that each section of the heating furnace is numbered starting from the furnace mouth, and the suffix number is the lowest and highest furnace temperature of the section. The highest and lowest furnace temperatures of each section are set as the boundary conditions for limiting the temperature of the steel billet in this section, thereby ensuring the heating effect.
[0037] As a preferred solution of the heating furnace temperature optimization method based on the steel temperature prediction model and the improved PSO described in the present invention, wherein: the optimal furnace temperature setting value is optimized and searched by using the adaptive simulated annealing particle swarm algorithm to determine the optimal furnace temperature setting of each section includes:
[0038] Initialization parameters;
[0039] Call the current temperature of the steel billet to evaluate the fitness value of the global particle and record it;
[0040] Adaptive adjustment parameters are used to adjust the inertia weight coefficient ω, individual learning factor c1 and global learning factor c2 to overcome the disadvantage that the particle swarm algorithm is prone to fall into local optimality.
[0041] Update the particle velocity and perform an iterative optimization;
[0042] Calculate the probability of accepting a new solution according to the Metropolis criterion;
[0043] Based on the Metropolis criterion, the comparison probability p i(k) and rand(), where rand() is a function that generates uniformly distributed random numbers, used to generate a random number uniformly distributed in the range [0,1]. i When (k) is greater than the number randomly generated by rand(), the new solution replaces the global optimal solution for annealing and updates the temperature, otherwise the new solution is invalid;
[0044] Determine whether the maximum number of iterations k has been reached max If it is not reached, the inertia weight coefficient ω, the individual learning factor c1 and the global learning factor c2 are adjusted by adaptive adjustment parameters to overcome the disadvantage that the particle swarm algorithm is prone to fall into the local optimum. If it is reached, the current optimal particle is output, that is, the furnace temperature setting value obtained by optimization, and the algorithm terminates.
[0045] Furthermore, the initialization parameters include: setting the particle swarm size N, initializing the inertia weight coefficient ω, setting the individual learning factor c1 and the global learning factor c2, setting the maximum number of iterations k max , randomly generate a set of furnace temperature setting values x in the solution space i (0) as the initial particle position, where i = 1, 2, ..., N, and initialize the velocity v of each particle i (0).
[0046] Furthermore, the calling of the temperature of the steel billet at this moment evaluates the fitness value of the global particle and records G best include:
[0047]
[0048] Among them, T(k) is the initial temperature of simulated annealing, μ is the set temperature reduction coefficient, G best is the optimal position of the population, T is the initial temperature, the Metropolis criterion in the simulated annealing algorithm is introduced into the iteration, the initial temperature is set according to the initial optimal value of the particle, and it decays with a certain temperature reduction coefficient μ after each iteration;
[0049] Furthermore, the inertia weight coefficient ω, the individual learning factor c1 and the global learning factor c2 are adjusted by using adaptive adjustment parameters to overcome the disadvantage that the particle swarm algorithm is prone to fall into the local optimum, including:
[0050] ω=(ω max +ω min ) / 2+tanh(-4+8×(k max -k) / k max )(ω max -ω min ) / 2
[0051] c1=c 1max -k(c1max -c 1min ) / k max
[0052] c2=c 2min -k(c 2min -c 2max ) / k max
[0053] Among them, ω max ,ω min is the maximum and minimum value of the inertia weight coefficient; k is the current iteration number; k max is the maximum number of iterations; c 1max and c 2min is the maximum and minimum value of the individual learning factor; c 2max and c 2min are the maximum and minimum values of the global learning factor.
[0054] As a preferred solution of the heating furnace temperature optimization method based on the steel temperature prediction model and the improved PSO described in the present invention, the inertia weight coefficient ω, the individual learning factor c1 and the global learning factor c2 are iteratively optimized according to the formula to update the particle speed, and the formula is:
[0055] v i (k+1)=ωv i (k)+c1r1(P best.i (k)-x i (k))+c2r2(G best -x i (k))
[0056] x i (k+1)=x i (k)+v i (k+1)
[0057] Among them, x i (k) and v i (k) represent the position and velocity of the i-th particle at the k-th iteration; r1 and r2 are random numbers; P best is the optimal position of the i-th particle;
[0058] Specifically, the temperature of the steel billet is predicted for each particle using the steel billet temperature prediction model, and then the furnace temperature setting value is found according to the predicted steel temperature. After the energy consumption is calculated using the optimal furnace temperature setting value, the new steel billet temperature is predicted according to the furnace temperature setting value, and the steel billet temperature and the furnace temperature setting value are substituted into the constraint condition formula to check whether all constraints are met. For particles that do not meet the constraints, penalties are imposed to increase their objective function values, and the optimal position of the individual particles and the optimal position of the population are updated according to the fitness of the particles after movement.
[0059] The probability p of accepting a new solution is calculated according to the Metropolis criterion i (k) include:
[0060]
[0061] Among them, E i (k) represents the internal energy of the i-th particle at the k-th iteration, that is, the fitness value of the current particle; E g represents the internal energy of the current population at the optimal point; T i Represents the current temperature. The temperature will decay linearly to a certain extent in each iteration. The optimization process is an alternating process of constantly searching for new solutions and slowly cooling down.
[0062] Compared with the prior art, the beneficial effects achieved by the present invention are: a high-precision temperature field model is established to predict the temperature of the steel billet from the time it enters the furnace to the time it leaves the furnace, eliminating the blind spot where the temperature of the steel billet cannot be known; the furnace temperature setting value is found more accurately, avoiding the loss caused by manual experience, and making up for the situation where manual experience cannot handle various changes in a timely manner. BRIEF DESCRIPTION OF THE DRAWINGS
[0063] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.
[0064] in:
[0065] Figure 1 The present invention is a flowchart of a heating furnace temperature optimization method based on a steel temperature prediction model and improved PSO according to an embodiment of the present invention.
[0066] Figure 2 The present invention is a flowchart of constructing a billet temperature prediction model according to an embodiment of the present invention.
[0067] Figure 3 The present invention is a flowchart of using the ASAPSO algorithm to optimize the furnace temperature setting value according to an embodiment of the present invention. DETAILED DESCRIPTION
[0068] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0069] See also Figure 1-Figure 3 The present invention provides a heating furnace temperature optimization method based on a steel temperature prediction model and an improved PSO, comprising:
[0070] S1: The information collected and inputted about the heating furnace and the steel billet includes: size (length, width, height), density, thermal conductivity index (thermal conductivity coefficient) and specific heat capacity; the structural parameters of the heating furnace involve the length of the furnace body and the specific structural design; the working conditions and operation data in the production process are composed of the initial furnace entry temperature, the final furnace exit temperature, the rolling frequency, the heating cycle, the output, the furnace pressure, the heating strategy, the frequency of the steel billet exiting the furnace and the substances produced after combustion.
[0071] S2: Figure 2 As shown, a temperature field model is established, and the temperature of the steel billet in the furnace is predicted using the temperature field model.
[0072] S2.1: Use the unsteady three-dimensional temperature field mathematical model to construct the unsteady heat conduction equation of the steel billet, the formula is:
[0073]
[0074] Furthermore, Q is the heat source inside the object, and there is no heat generated by phase change inside the billet, so the formula is simplified to:
[0075]
[0076] Among them, x, y, z are the length, width and height of the steel billet respectively; ρ is the density of the steel plate kg / m 3 ;c p is the specific heat capacity of the steel plate (W / kg℃); k is the thermal conductivity of the steel plate (W / m℃); T(x,y,z,t) is the three-dimensional spatial temperature distribution of the steel plate at time t; t is time.
[0077] S2.2: Set boundary conditions.
[0078] The formula for the boundary condition is:
[0079]
[0080] Where ε is the overall heat absorption rate, T env is the ambient temperature.
[0081] S2.3: Solve the billet temperature according to the set boundary conditions.
[0082] Specifically, the process includes the following:
[0083] Use the explicit difference method to solve the equation. First, discretize the space and divide the cuboid into small cubic elements. Use i, j, and l to represent the spatial nodes and n to represent the time step. Use forward differences to approximate the time derivative and use central differences to approximate the spatial derivative.
[0084] The time derivatives are:
[0085]
[0086] The spatial derivatives are:
[0087]
[0088]
[0089]
[0090] Furthermore, these differential expressions are substituted into the heat conduction equation and simplified to obtain:
[0091]
[0092] This equation needs to be repeated at every node throughout the billet until the required number of time steps is reached or the temperature stabilizes.
[0093] Furthermore, the optimal overall heat absorption rate ε is obtained according to BP neural network identification.
[0094] Specifically, the detailed training of BP neural network includes the following steps:
[0095] The heating furnace conditions and billet properties are used as input, and the overall heat absorption rate is used as output;
[0096] Initialize BP neural network;
[0097] The data is processed by forward propagation to output the summary heat absorption rate;
[0098] Compare the total heat absorption rate with the actual value to detect whether the error meets the standard. If it meets the standard, this process is terminated. If it does not meet the standard, the next step will be continued;
[0099] Calculate the loss;
[0100] Back propagation updates the weights of each layer of neurons;
[0101] Check whether the number of iterations reaches the limit, if it reaches the limit, output the result, if not, repeat the cycle of processing data through forward propagation, outputting the summary heat absorption rate to the back propagation to update the weights of each layer of neurons.
[0102] S3: Construct a heating furnace temperature optimization objective function and construct constraint conditions based on the temperature field model.
[0103] Specifically, the furnace temperature distribution model is subjected to function integration processing in the furnace length direction as the objective function of the optimization problem, and the formula is:
[0104]
[0105] Where J represents the objective function; s represents the distance from the internal position of the hot furnace to the furnace entrance; T f (s) represents the furnace temperature of the steel billet at a distance s from the furnace mouth.
[0106] Furthermore, the constraint formula for the furnace temperature setting value optimization problem is:
[0107] |T s -T * |≤ΔT s
[0108] |T s -T c |≤ΔT d
[0109] T f min_j ≤T f_j ≤T f max_j
[0110] Among them, T * ΔT is the process required temperature value (℃) reached by the surface temperature of the steel billet after heat treatment; s T is the maximum deviation temperature between the surface temperature of the billet out of the furnace and the set temperature (℃); s With T c The surface temperature and internal temperature of the billet out of the furnace calculated by the billet temperature prediction model (℃); ΔT d is the maximum upper limit of the temperature difference of the steel billet section out of the furnace (℃); T f_j is the furnace temperature value of each position point in different areas inside the heating furnace (℃); T f max_j and T f min_j The upper and lower limits of the furnace temperature range in furnace zone j (℃), where furnace zone j indicates that each section of the heating furnace is numbered starting from the furnace mouth, and the suffix number is the lowest and highest furnace temperature of the section. The highest and lowest furnace temperatures of each section are set as the boundary conditions for limiting the temperature of the steel billet in this section, thereby ensuring the heating effect.
[0111] S4: Figure 3 As shown, the adaptive simulated annealing particle swarm algorithm is used to optimize the search for the optimal furnace temperature setting value to determine the optimal furnace temperature setting for each section.
[0112] S4.1: Initialization parameters.
[0113] Specifically, set the particle swarm size N, initialize the inertia weight coefficient ω, set the individual learning factor c1 and the global learning factor c2, and set the maximum number of iterations k max , randomly generate a set of furnace temperature setting values x in the solution space i (0) as the initial particle position, where i = 1, 2, ..., N, and initialize the velocity v of each particle i (0).
[0114] S4.2: Call the current temperature of the steel billet to evaluate the fitness value of the global particle and record G best .
[0115] Specifically, the initial temperature T(k) of simulated annealing is set, the cooling coefficient μ is set, and the Metropolis criterion in the simulated annealing algorithm is introduced into the iteration. The initial temperature is set according to the initial optimal value of the particle, and decays with a certain cooling coefficient μ after each iteration.
[0116] Furthermore, the initial temperature T(k) formula is:
[0117]
[0118] Among them, G best is the optimal position of the population; T is the initial temperature.
[0119] S4.3: The inertia weight coefficient ω, the individual learning factor c1 and the global learning factor c2 are adjusted by an adaptive parameter adjustment method to overcome the shortcoming that the particle swarm algorithm is prone to fall into local optimality.
[0120] Specifically, the adjustment process is:
[0121] ω=(ω max +ω min ) / 2+tanh(-4+8×(k max -k) / k max )(ω max -ω min ) / 2
[0122] c1=c 1max -k(c 1max -c 1min ) / k max
[0123] c2=c 2min -k(c 2min -c 2max ) / k max
[0124] Among them, ω max ,ω minis the maximum and minimum value of the inertia weight coefficient; k is the current iteration number; k max is the maximum number of iterations; c 1max , c 2min are the maximum and minimum values of the individual learning factor respectively; c 2max , c 2min are the maximum and minimum values of the global learning factor respectively.
[0125] S4.4: The inertia weight coefficient ω, the individual learning factor c1 and the global learning factor c2 are iteratively optimized according to the formula to update the particle speed, the formula is:
[0126] v i (k+1)=ωv i (k)+c1r1(P best.i (k)-x i (k))+c2r2(G best -x i (k))
[0127] x i (k+1)=x i (k)+v i (k+1)
[0128] Among them, x i (k) and v i (k) represent the position and velocity of the i-th particle at the k-th iteration; r1 and r2 are random numbers; P best is the optimal position of the i-th particle;
[0129] Specifically, the temperature of the steel billet is predicted for each particle using the steel billet temperature prediction model, and then the furnace temperature setting value is found according to the predicted steel temperature. After the energy consumption is calculated using the optimal furnace temperature setting value, the new steel billet temperature is predicted according to the furnace temperature setting value, and the steel billet temperature and the furnace temperature setting value are substituted into the constraint condition formula to check whether all constraints are met. For particles that do not meet the constraints, penalties are imposed to increase their objective function values, and the optimal position of the individual particles and the optimal position of the population are updated according to the fitness of the particles after movement.
[0130] S4.5: The probability p of accepting a new solution is calculated according to the Metropolis criterion i (k).
[0131] Specifically, the probability p i The formula for (k) is:
[0132]
[0133] Among them, E i(k) represents the internal energy of the i-th particle at the k-th iteration, that is, the fitness value of the current particle; E g represents the internal energy of the current population at the optimal point; T i Represents the current temperature. The temperature will decay linearly to a certain extent in each iteration. The optimization process is an alternating process of constantly searching for new solutions and slowly cooling down.
[0134] S4.6: Based on the Metropolis criterion, compare the probability p i (k) and rand(), where rand() is a function that generates uniformly distributed random numbers, used to generate a random number uniformly distributed in the range [0,1]. i When (k) is greater than the number randomly generated by rand(), the new solution replaces the global optimal solution for annealing and updates the temperature. Otherwise, the new solution is invalid.
[0135] S4.7: Determine whether the maximum number of iterations k has been reached max If it is not reached, the inertia weight coefficient ω, the individual learning factor c1 and the global learning factor c2 are adjusted by adaptive adjustment parameters to overcome the disadvantage that the particle swarm algorithm is prone to fall into the local optimum. If it is reached, the current optimal particle is output, that is, the furnace temperature setting value obtained by optimization, and the algorithm terminates.
[0136] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.
[0137] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A heating furnace temperature optimization method based on steel temperature prediction model and improved PSO, It is characterized in that The method includes: Collect and input parameters of heating furnace and billet; Establishing a temperature field model, and using the temperature field model to predict the temperature of the steel billet in the heating furnace; Constructing a heating furnace temperature optimization objective function and constructing constraint conditions based on the temperature field model; The adaptive simulated annealing particle swarm algorithm is used to optimize the search for the optimal furnace temperature setting and determine the optimal furnace temperature setting for each section.
2. According to claim 1, a heating furnace temperature optimization method based on steel temperature prediction model and improved PSO, It is characterized in that The parameters of the steel billet include size, density, thermal conductivity index and specific heat capacity.
3. According to claim 1, a heating furnace temperature optimization method based on steel temperature prediction model and improved PSO, It is characterized in that A non-steady-state heat conduction equation of the steel billet is constructed, including setting boundary conditions, and solving the temperature of the steel billet according to the boundary conditions.
4. According to claim 3, a heating furnace temperature optimization method based on a steel temperature prediction model and an improved PSO, It is characterized in that The construction of the unsteady heat conduction equation of the steel billet includes: The steel billet is defined as a microelement cuboid, and the heating conditions of each surface of the steel billet are analyzed to obtain the three-dimensional form of the differential equation for heat conduction of the microelement cuboid: Where x, y, z are the length, width and height of the steel billet respectively; ρ is the density of the steel plate; c p is the specific heat capacity of the steel plate; k is the thermal conductivity of the steel plate; T(x, y, z, t) is the three-dimensional temperature distribution of the steel plate at time t; t is time; The boundary conditions are: Where ε is the overall heat absorption rate, T env is the ambient temperature; The process of solving the billet temperature according to the boundary conditions includes: First, the space is discretized, then the time is discretized, and finally the equation is discretized using the central difference method; according to the boundary conditions, each time step is iteratively solved to obtain the temperature distribution of the billet.
5. A heating furnace temperature optimization method based on a steel temperature prediction model and improved PSO according to any one of claims 3 to 4, It is characterized in that The process of solving the temperature of the steel billet according to the boundary conditions also includes: The optimal overall heat absorption rate ε is obtained according to BP neural network identification, including the following steps: The heating furnace conditions and billet property information are used as input, and the overall heat absorption rate is used as output; Initialize BP neural network; Outputting the optimal overall heat absorption rate ε by forward propagation processing data; Compare the optimal overall heat absorption rate ε with the actual value to detect whether the error meets the standard. If so, terminate the process; if not, calculate the loss, and back-propagate to update the weights of each layer of neurons; Check whether the limited number of iterations is reached. If reached, the result is output. If not reached, the cycle of processing data through forward propagation to output the optimal overall heat absorption rate ε and back propagating to update the weights of each layer of neurons is repeated.
6. According to claim 1, a heating furnace temperature optimization method based on steel temperature prediction model and improved PSO, It is characterized in that The construction of the heating furnace temperature optimization objective function and the construction of the constraint conditions based on the temperature field model include: The furnace temperature distribution model is processed by function integration in the furnace length direction as the objective function of the optimization problem. The formula is: Where J represents the objective function; s represents the distance from the internal position of the hot furnace to the furnace entrance; T f (s) represents the furnace temperature of the steel billet at a distance s from the furnace mouth; The constraint expression of the furnace temperature setting value optimization problem is: |T s -T * |≤ΔT s |T s -T c |≤ΔT d T fmin_j ≤T f_j ≤T fmax_j Among them, T * is the temperature value; ΔT s is the maximum deviation temperature; T s With T c is the surface temperature and internal temperature of the billet out of the furnace calculated by the billet temperature prediction model; ΔT d is the maximum upper limit of the temperature difference of the steel billet section; T f_j is the furnace temperature; T fmax_j and T fmin_j are the upper and lower limits of the range of furnace temperature in furnace zone j; The furnace zone j indicates that each section of the heating furnace is numbered starting from the furnace mouth, and the heating effect is ensured by setting the highest and lowest furnace temperatures of each section as boundary conditions that limit the temperature of the steel billet in this section.
7. According to claim 1, a heating furnace temperature optimization method based on steel temperature prediction model and improved PSO, It is characterized in that The optimization search for the optimal setting value of the furnace temperature using the adaptive simulated annealing particle swarm algorithm includes: Initialization parameters; Call the current temperature of the steel billet to evaluate the fitness value of the global particle and record the optimal position of the population G best ; For the inertia weight coefficient ω and individual learning factor c 1 and the global learning factor c 2 Adaptive adjustment parameters are used to overcome the disadvantage that particle swarm algorithm is prone to fall into local optimum. Update the particle velocity and perform an iterative optimization; Calculate the probability p of accepting a new solution according to the Metropolis criterion i (k); Based on the Metropolis criterion, the comparison probability p i (k) and rand(), where rand() is a function that generates uniformly distributed random numbers. i When (k) is greater than the number randomly generated by rand(), the new solution replaces the global optimal solution for annealing and updates the temperature, otherwise the new solution is invalid; Determine whether the maximum number of iterations k has been reached max If it is not reached, the inertia weight coefficient ω and individual learning factor c are returned. 1 and the global learning factor c 2 Adaptive adjustment parameters are used to overcome the disadvantage that the particle swarm algorithm is prone to fall into local optimality; if it reaches, the current optimal particle is output, that is, the furnace temperature setting value obtained by optimization, and the algorithm terminates.
8. According to claim 7, a heating furnace temperature optimization method based on steel temperature prediction model and improved PSO, It is characterized in that The initialization parameters include: Set the particle swarm size N, initialize the inertia weight coefficient ω, and set the individual learning factor c 1 and the global learning factor c 2 , set the maximum number of iterations k max , randomly generate a set of furnace temperature setting values x in the solution space i (0) as the initial particle position, where i is a constant, and the velocity v of each particle is initialized. i (0); The calling of the temperature of the steel billet at this moment evaluates the fitness value of the global particle and records G best include: Among them, T(k) is the initial temperature of simulated annealing, μ is the set temperature reduction coefficient, G best is the optimal position of the population, T is the initial temperature, the Metropolis criterion in the simulated annealing algorithm is introduced into the iteration, the initial temperature is set according to the initial optimal value of the particle, and it decays with a certain temperature reduction coefficient μ after each iteration; The ω, c 1 and c 2 Overcoming the shortcomings of the particle swarm algorithm that easily falls into local optimality includes: ω=(ω max +oh min ) / 2+tanh(-4+8×(k max -k) / k max )(oh max -oh min ) / 2 c 1 =c 1max -k(c 1max -c 1min ) / k max c 2 =c 2min -k(c 2min -c 2max ) / k max Among them, ω max ,ω min is the maximum and minimum value of the inertia weight coefficient; k is the current iteration number; k max is the maximum number of iterations; c 1max and c 2min is the maximum and minimum value of the individual learning factor; c 2max and c 2min are the maximum and minimum values of the global learning factor.
9. A heating furnace temperature optimization method based on a steel temperature prediction model and improved PSO according to any one of claims 7 to 8, It is characterized in that The particle speed is updated and an iterative optimization is performed, ω, c 1 and c 2 The formula is: v i (k+1)=ωv i (k)+c 1 r 1 (P best.i (k)-x i (k))+c 2 r 2 (G best -x i (k)) x i (k+1)=x i (k)+v i (k+1) Among them, x i (k) and v i (k) represent the position and velocity of the i-th particle at the k-th iteration; r 1 and r 2 is a random number; P best is the optimal position of the i-th particle.
10. The heating furnace temperature optimization method based on the steel temperature prediction model and improved PSO according to claim 7, It is characterized in that The probability p of accepting a new solution is calculated according to the Metropolis criterion i The formula for (k) is: Among them, E i (k) represents the internal energy of the i-th particle at the k-th iteration, that is, the fitness value of the current particle; E g represents the internal energy of the current population at the optimal point; T i Indicates the current temperature.
Citation Information
Patent Citations
Heating furnace temperature optimization method based on hybrid modeling and improved sand dune cat swarm algorithm
CN116127760A
E-plane metal waveguide filter optimization method based on improved particle swarm optimization
CN117272647A
Cited By
Slag cooling process simulation method based on multi-source heterogeneous data
CN120145664A
Billet whole-process temperature field forecasting method based on multi-modal data and digital twinning
CN121902429A
