Industrial kiln furnace temperature uniformity control system and method based on big data

Abstract

The invention discloses an industrial kiln furnace temperature uniformity control system and method based on big data, and relates to the field of temperature control. The invention divides silicon carbide heaters into regions by setting key position points and deploying temperature sensors; minimizes an objective function to obtain an optimal silicon carbide heater regional power parameter combination, and applies the optimal silicon carbide heater regional power parameter combination; constructs an initial MLP model and sets relevant parameters, trains, tests and optimizes the initial MLP model to obtain a final MLP model; collects real-time key position temperature data, calculates a real-time weighted average temperature difference, and determines whether to make adjustments based on the real-time weighted average temperature difference; uses the final MLP model to predict future key position temperature data, and determines whether to make adjustments based on the future key position temperature data; improves the accuracy and uniformity of temperature control; reduces energy consumption; and realizes temperature prediction and early adjustment.

Description

Technical Field

[0001] The present invention belongs to the field of temperature control, and in particular, relates to a system and method for controlling furnace temperature uniformity of an industrial kiln based on big data. Background Art

[0002] In the current industrial production process, especially in the ceramics, metallurgy, chemical industry and other industries, the temperature control of industrial kilns is a crucial link. The temperature uniformity and stability of industrial kilns directly affect the quality of products, production efficiency and energy consumption.

[0003] Chinese patent CN114264153B discloses a method, system and terminal for temperature monitoring and optimized operation control of aluminum melting furnace. The method uses contact and non-contact temperature measurement methods to obtain the internal temperature of aluminum liquid and the aluminum liquid temperature on the surface of the molten pool in the same smelting process, thereby establishing a data file of the two temperatures. Then, a full-cycle relationship model between the standard temperature of aluminum liquid and the surface temperature of aluminum liquid is constructed by combining production process parameters, big data analysis and neural network algorithm. The actual value of the surface temperature of aluminum liquid obtained by non-contact temperature measurement is used to calculate the real-time temperature of aluminum liquid. On this basis, the present invention proposes a dual-temperature heating and combustion optimization control strategy and system for aluminum melting furnace. The strategy automatically adjusts the fuel flow rate and the heat load of the burner according to the difference between the furnace temperature, the actual temperature of aluminum liquid and the preset temperature, thereby realizing precise control of the aluminum melting furnace. This invention effectively avoids the problems of under-burning and over-temperature of aluminum liquid, improves the quality of aluminum liquid heating, and enhances the intelligence of the aluminum melting process.

[0004] The existing temperature uniformity control system for the furnace of an industrial kiln often collects, analyzes and processes the temperature of the furnace of the industrial kiln in real time. When the temperature of the furnace of the industrial kiln is uneven, certain measures are taken to adjust it. This method lacks timeliness and cannot adjust the temperature unevenness of the furnace of the industrial kiln in advance and detect the reasons for the uneven temperature of the furnace of the industrial kiln, resulting in poor product quality and waste of resources. Summary of the invention

[0005] In response to the problems in the related technology, the present invention provides an industrial kiln furnace temperature uniformity control system and method based on big data. Through intelligent control strategies, the accuracy and uniformity of temperature control are improved, energy consumption is reduced, and temperature prediction is achieved, thereby optimizing the production process of industrial kilns.

[0006] In order to solve the above technical problems, the present invention is achieved through the following technical solutions:

[0007] S1. Set a key position point set of the industrial kiln furnace, arrange a temperature sensor at each key position point in the key position point set of the industrial kiln furnace, divide the silicon carbide heater into regions, obtain a silicon carbide heater region set, collect historical key position point temperature data and power data of each region in the silicon carbide heater region set, and obtain a combination set of historical key position point temperatures of the industrial kiln furnace and a combination set of silicon carbide heater region power parameters;

[0008] S2. According to the combination set of the temperature of the key positions of the industrial furnace and the combination set of the power parameters of the silicon carbide heater area, an objective function is set, and the objective function is minimized by using a genetic algorithm to obtain the optimal power parameter combination of the silicon carbide heater area, and the power parameter of each area in the optimal power parameter combination of the silicon carbide heater area is applied to the corresponding silicon carbide heater area;

[0009] S3, setting a first temperature collection time point set, collecting temperature data of key positions of the industrial kiln furnace at the time points in the first temperature collection time point set, obtaining a first temperature matrix, constructing an initial MLP model and setting relevant parameters, training, testing and optimizing the initial MLP model, and obtaining a final MLP model;

[0010] S4. Set the weighted average temperature difference threshold, collect real-time key position temperature data, calculate the weighted average temperature difference of the real-time key position temperature data, obtain the real-time weighted average temperature difference, and judge whether to adjust the power parameters of the silicon carbide heater area according to the real-time weighted average temperature difference; according to the real-time key position temperature data, use the final MLP model to obtain the future key position temperature data, and judge whether to adjust the power parameters of the silicon carbide heater area according to the future key position temperature data.

[0011] Preferably, the S1 comprises the following steps:

[0012] S11, set the key position point set of the industrial kiln furnace a={a1,a2,…,a i ,…,a n}, where a i represents the set i-th industrial kiln furnace key position point, and n represents the total number of industrial kiln furnace key position points;

[0013] S12, in the industrial furnace key position point set a={a1, a2,…, a i ,…,a n A temperature sensor is arranged at each key position point in the system, and the temperature sensor is used to collect temperature data of the corresponding key position point. The temperature sensor uploads the real-time collected temperature data to the database;

[0014] S13, dividing the silicon carbide heater into n regions according to the key position point set of the industrial kiln furnace, and obtaining the silicon carbide heater region set b={b1, b2, …, b i ,…,b n}, where b i represents the i-th region of the silicon carbide heater;

[0015] Allocate different heating powers according to the temperature requirements of each zone in the SiC heater zone set, and upload the real-time power of each zone to the database;

[0016] S14, set the historical time point set c={c1,c2,…,c i ,…,c m}, where c i represents the i-th time point in the historical time point set, and m represents the total number of historical time points in the historical time point set;

[0017] S15. Collect the historical time point set c={c1,c2,…,c i ,…,c m}, the key position point set of the industrial furnace is a={a1,a2,…,a i ,…,a n} and the temperature data of each key position point and the SiC heater area set b={b1,b2,…,b i ,…,b n}, and obtain the temperature combination set d={d1,d2,…,d i ,…,d m} and the SiC heater regional power parameter combination set e={e1,e2,…,e i ,…,e m}, where d i represents the temperature combination data of each key location at the historical time point i, e i represents power combination data of each area of ​​the silicon carbide heater at the historical time point i;

[0018] Through the above steps, a set of key position points of the industrial kiln furnace is set, and temperature sensors are arranged at these key position points to collect temperature data in real time; the silicon carbide heater is divided into several areas according to these key position points, and different heating powers are allocated to each area; a set of historical time points is set, and the temperature data of the key position points and the power data of the heater area at each historical time point are collected, and finally a combination set of temperature of the historical key position points of the industrial kiln furnace and a combination set of power parameters of the silicon carbide heater area are obtained, so as to realize real-time monitoring of the temperature of the industrial kiln furnace and optimal allocation of the heater power, so as to improve industrial production efficiency and product quality.

[0019] Preferably, S2 comprises the following steps:

[0020] S21. Based on the temperature combination set of the key positions in the industrial furnace, define an objective function F, where F represents the weighted average temperature difference, as follows:

[0021] ;

[0022] Where n is the number of zones in the furnace, T i is the temperature of the ith region, is the average temperature of n regions, w i The weight of the i-th region is set according to the importance and temperature sensitivity of the region;

[0023] The smaller the objective function, the more uniform the temperature in the furnace of the industrial kiln;

[0024] S22, according to the silicon carbide heater regional power parameter combination set, use genetic algorithm to find an optimal silicon carbide heater regional power parameter combination f={f1,f2,…,f i ,…,f n}, so that the objective function is minimized, where f i represents the power data of the i-th zone of the silicon carbide heater;

[0025] S23, the optimal silicon carbide heater regional power parameter combination f={f1,f2,…,f i ,…,f n}, the power parameters of each zone in are applied to the corresponding silicon carbide heater zone;

[0026] Through the above steps, an objective function is defined to measure the temperature uniformity in the furnace of the industrial kiln. The genetic algorithm is used to find the optimal combination of regional power parameters of the silicon carbide heater to achieve the optimal distribution of the temperature in the furnace, ensuring that the temperature of each area is as uniform as possible, improving the heating efficiency and product quality of the industrial kiln, and finally applying the optimal power parameters to the actual heating process to achieve precise control and optimized production.

[0027] Preferably, the S22 comprises the following steps:

[0028] S221, the silicon carbide heater regional power parameter combination set e={e1,e2,…,e i ,…,e m The number m of power parameter combinations of SiC heater regions in} is used as the size of chromosome population, and chromosome population g={g1,g2,…,g i ,…,g m}, where g i represents the i-th chromosome in the chromosome population; uses each chromosome in the chromosome population as a power parameter combination of each silicon carbide heater region; sets the maximum population iteration number to h1, the crossover rate to j, and the mutation rate to k;

[0029] S222, performing an encoding operation on each chromosome in the chromosome population to obtain an encoded chromosome population;

[0030] S223, start the iterative operation, set the current iteration number to h2, and in each round of iteration, perform selection, crossover and mutation operations on the encoded chromosome population according to the objective function F, the crossover rate j and the mutation rate k to obtain the chromosome population after the operation;

[0031] S224, decoding the operated chromosome population to obtain a decoded chromosome population;

[0032] S225, repeat S222, S223, S224, when h2 ≥ h1, stop iteration, and obtain the optimal silicon carbide heater regional power parameter combination f = {f1, f2, ..., f i ,…,f n};

[0033] Through the above steps, the framework of genetic algorithm is adopted, and the power parameter combination of the silicon carbide heater area is regarded as a chromosome population. Through operations such as encoding, iteration, selection, crossover, mutation and decoding, the power parameter combination is gradually optimized to find the optimal solution that minimizes the objective function F. It simulates the process of natural selection and improves the fitness of the power parameter combination through multiple rounds of iterations. Finally, the optimal power parameter combination is found within the set number of iterations, thereby achieving uniform distribution of temperature in the furnace of the industrial kiln and improving heating efficiency and product quality.

[0034] Preferably, S3 comprises the following steps:

[0035] S31, set the first temperature collection time point set p={p1, p2, ..., p i ,…,p m}, where p irepresents the i-th temperature collection time point in the first temperature collection time point set; the temperature of the key position points in the industrial furnace key position point set is collected at the time point in the first temperature collection time point set to obtain the first temperature matrix A, as follows,

[0036] ;

[0037] Among them, A im It represents the temperature collected at the i-th key position at the m-th temperature collection time point;

[0038] S32, constructing an initial MLP model, setting the number of nodes in the input layer of the initial MLP model to r1, the number of nodes in the hidden layer to r2, and the number of nodes in the output layer to r3, setting the initial batch size of the initial MLP model to t1, the initial learning rate to t2, and setting the training ratio and test ratio of the initial MLP model to u1 and u2 respectively;

[0039] S33, dividing the first temperature matrix A into a first temperature training matrix and a first temperature test matrix according to the training ratio and the test ratio of the initial MLP model as u1 and u2; setting a first temperature training label matrix and a first temperature test label matrix;

[0040] S34, setting a training error threshold, a maximum number of training iterations, and a current number of training iterations; inputting the first temperature training matrix and the first temperature training label matrix into the initial MLP model to train the initial MLP model, adjusting the number of nodes r1 of the input layer, the number of nodes r2 of the hidden layer, and the number of nodes r3 of the output layer of the initial MLP model according to the training results to obtain an adjusted MLP model, and continuing to train the adjusted MLP model. When the training error is ≤ the training error threshold, or the current number of training iterations is ≥ the maximum number of training iterations, the training is stopped to obtain a trained MLP model;

[0041] S35, inputting the first temperature test matrix and the first temperature test label matrix into the trained MLP model for testing, and obtaining the final MLP model after the test optimization is completed;

[0042] Through the above steps, the temperature data of key positions of the industrial kiln furnace are collected, an MLP model is constructed, and it is trained and optimized to achieve accurate prediction and control of the furnace temperature; this process includes data collection, model construction, data partitioning, model training and adjustment, as well as model testing and verification, and finally an optimized MLP model is obtained, which can improve the temperature management efficiency and accuracy of industrial kilns.

[0043] Preferably, the S35 comprises the following steps:

[0044] S351. Set the accuracy threshold v1, input the first temperature test matrix into the trained MLP model for testing to obtain a test output result, compare the test output result with the first temperature test label matrix to obtain the test accuracy v2; when v2≥v1, use the trained MLP model as the final MLP model; when v2<v1, optimize the initial batch size and initial learning rate of the trained MLP model to obtain an optimized MLP model, and use the optimized MLP model as the final MLP model.

[0045] By setting an accuracy threshold, test the trained MLP model to evaluate its performance in predicting the temperature of the industrial furnace hearth; calculate the test accuracy by comparing the model test output result with the actual historical tracking feature data test label; if the test accuracy meets or exceeds the preset threshold, confirm that the model is valid and use it as the final model; if it does not meet the threshold, optimize the initial batch size and learning rate of the model to improve the model's prediction ability.

[0046] Preferably, the optimization of the initial batch size t1 and initial learning rate t2 of the trained MLP model in S351 to obtain an optimized MLP model includes the following steps:

[0047] S3511. Construct an ant population; set the size of the ant population to J, then the ant population is represented as H={H1, H2, …, H i , …, H J}, where H i represents the i-th ant in the ant population; set the maximum number of test iterations; set the foraging search space dimension of the ant population to be two-dimensional.

[0048] S3512. Randomly generate an initial foraging position set G={(G 11 , G 12 ), (G 21 , G 22 ), …, (G i1 , G i2 ), …, (G J1 , G J2 )} according to the initial batch size and initial learning rate; G i1 and G i2 respectively represent the first-dimensional coordinate and the second-dimensional coordinate of the initial foraging position of the i-th ant in the ant population.

[0049] Set the fitness function of the ant population, and the fitness function formula is as follows:

[0050] ;

[0051] In the formula, x represents the fitness function; y represents the bias;

[0052] S3513, start the iteration operation, set the number of real-time test iterations; in each round of iteration, update the foraging position of each ant in the initial foraging position set of the ant population according to the function formula of the fitness function; and obtain the global best individual and the global best fitness in the ant population in each round of iteration;

[0053] S3514, when the real-time test iteration number is greater than or equal to the maximum test iteration number, the ant population stops the iteration operation to obtain an optimized MLP model;

[0054] The above steps simulate the foraging behavior of ants to find the optimal MLP model training parameters in the parameter space, thereby improving the prediction ability of the model.

[0055] Preferably, S4 comprises the following steps:

[0056] S41, set the weighted average temperature difference threshold, set the real-time temperature collection time point set B = {B1, B2, ..., B i ,…,B N}, where B i represents the i-th temperature collection time point in the real-time temperature collection time point set; N represents the total number of temperature collection time points in the real-time temperature collection time point set;

[0057] At a time point where the real-time temperature collection time points are concentrated, the temperature of each key position point in the industrial furnace key position point concentration is collected to obtain the real-time key position point temperature data, and the weighted average temperature difference of the real-time key position point temperature data is calculated to obtain the real-time weighted average temperature difference. When the real-time weighted average temperature difference is ≤ the weighted average temperature difference threshold, the current state is maintained; when the real-time weighted average temperature difference is greater than the weighted average temperature difference threshold, the power of the silicon carbide heater area corresponding to the key position point is adjusted to make the real-time weighted average temperature difference ≤ the weighted average temperature difference threshold, and the problem that causes the real-time weighted average temperature difference to be greater than the weighted average temperature difference threshold is detected, and after the problem is solved, the optimal silicon carbide heater area power parameter combination is restored to use;

[0058] S42, set the future temperature collection time point set C = {C1, C2, ..., C i ,…,C N}, where C i represents the i-th temperature collection time point in the future temperature collection time point set;

[0059] According to the real-time key location temperature data and the future temperature collection time point set C={C1,C2,…,C i ,…,CN}, use the final MLP model to predict the temperature of each key position point in the industrial furnace key position point concentration at the time point of the future temperature collection time point concentration, and obtain the future key position point temperature data; calculate the weighted average temperature difference of the future key position point temperature data to obtain the future weighted average temperature difference. When the future weighted average temperature difference ≤ the weighted average temperature difference threshold, keep the current state; when the future weighted average temperature difference > the weighted average temperature difference threshold, adjust the power of the silicon carbide heater area corresponding to the key position point in advance, so that the future weighted average temperature difference ≤ the weighted average temperature difference threshold, and detect the problem that causes the future weighted average temperature difference > the weighted average temperature difference threshold, and restore the use of the optimal silicon carbide heater area power parameter combination after the problem is solved;

[0060] The above steps ensure that the industrial kiln can maintain a stable temperature distribution during operation. Through real-time monitoring and predictive maintenance, the heater power can be adjusted in time to prevent temperature abnormalities, thereby improving production efficiency and product quality, reducing energy consumption and equipment failures.

[0061] An industrial kiln furnace temperature uniformity control system based on big data, used to implement the above-mentioned industrial kiln furnace temperature uniformity control method based on big data, including a data acquisition module, an optimal silicon carbide heater regional power parameter combination, a training and optimization MLP model, a real-time temperature determination module, and a future temperature data prediction and determination module;

[0062] The data acquisition module is used to collect the temperature data of the key positions of the furnace of the industrial kiln in history, the power data of the silicon carbide heater area in history; collect the temperature data of the key positions of the furnace of the industrial kiln at the time point concentrated at the first time point; collect real-time temperature data;

[0063] The optimal silicon carbide heater area power parameter combination is used to use the temperature data of key positions of the furnace of the historical industrial kiln and the power data of the historical silicon carbide heater area, and use the genetic algorithm to find the optimal silicon carbide heater area power parameter combination;

[0064] The training and optimization MLP model is used to train, test and optimize the initial MLP model according to the temperature data of the key position points of the furnace of the industrial kiln at the time point of the first time point concentration to obtain the final MLP model;

[0065] The real-time temperature determination module is used to calculate the weighted average temperature difference of the real-time temperature data, and compare it with the weighted average temperature difference threshold to determine whether to make an adjustment;

[0066] The future temperature data prediction and determination module is used to use the final MLP model to predict the temperature data of key positions of the industrial furnace hearth at future moments, calculate the weighted average temperature difference of the future temperature data, and compare it with the weighted average temperature difference threshold to determine whether to make adjustments.

[0067] The present invention has the following beneficial effects:

[0068] The accuracy and uniformity of temperature control are improved; by arranging temperature sensors at key positions of the furnace of an industrial kiln and adopting an intelligent control strategy, the present invention can monitor and adjust the furnace temperature in real time to ensure a more uniform temperature distribution, thereby improving product quality and reducing the defective rate caused by uneven temperature.

[0069] Reduced energy consumption; the present invention optimizes the regional power parameter combination of silicon carbide heaters through genetic algorithms, realizes the rational allocation and utilization of energy, effectively reduces the energy consumption of industrial kilns, improves energy utilization efficiency, and saves costs for enterprises.

[0070] The prediction and advance adjustment of temperature are realized; the present invention uses the MLP model to predict the future temperature of the furnace of the industrial kiln, and adjusts the heater power in advance according to the prediction results, thereby avoiding the occurrence of uneven temperature problems and ensuring the stability of the production process and high standards of product quality. At the same time, through real-time monitoring and problem diagnosis, the present invention can timely discover and solve the causes of uneven temperature and ensure smooth production.

[0071] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] In order to more clearly illustrate the technical solutions of the embodiments of the invention, the drawings required for describing the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, they can also obtain drawings based on these drawings without paying creative work.

[0073] Figure 1 It is a schematic diagram of a process of finding the optimal silicon carbide heater regional power parameter combination in the industrial kiln furnace temperature uniformity control system and method based on big data of the present invention;

[0074] Figure 2 It is a schematic diagram of the process flow of obtaining the final MLP model of the industrial kiln furnace temperature uniformity control system and method based on big data of the present invention;

[0075] Figure 3The present invention is a flowchart of the industrial kiln furnace temperature uniformity control system and method based on big data for determining whether to adjust the real-time and future silicon carbide heater zone power. DETAILED DESCRIPTION

[0076] The following will be combined with the drawings in the embodiments of the invention to clearly and completely describe the technical solutions in the embodiments of the invention. Obviously, the described embodiments are only part of the embodiments of the invention, not all of the embodiments. Based on the embodiments in the invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the invention.

[0077] In the description of the present invention, it is necessary to understand that the terms "opening", "upper", "lower", "top", "middle", "inside" and the like indicating orientation or positional relationship are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the components or elements referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the invention. Embodiment 1

[0078] See also Figure 1 , Figure 2 , Figure 3 The present invention discloses a method for controlling the furnace temperature uniformity of an industrial kiln based on big data, comprising the following steps:

[0079] S1. Set a key position point set of the industrial kiln furnace, arrange a temperature sensor at each key position point in the key position point set of the industrial kiln furnace, divide the silicon carbide heater into regions, obtain a silicon carbide heater region set, collect historical key position point temperature data and power data of each region in the silicon carbide heater region set, and obtain a combination set of historical key position point temperatures of the industrial kiln furnace and a combination set of silicon carbide heater region power parameters;

[0080] The S1 comprises the following steps:

[0081] S11, set the key position point set of the industrial kiln furnace a={a1,a2,…,a i ,…,a n}, where a i represents the set i-th industrial kiln furnace key position point, and n represents the total number of industrial kiln furnace key position points;

[0082] S12, in the industrial furnace key position point set a={a1, a2,…, a i ,…,a nA temperature sensor is arranged at each key position point in the system, and the temperature sensor is used to collect temperature data of the corresponding key position point. The temperature sensor uploads the real-time collected temperature data to the database;

[0083] S13, dividing the silicon carbide heater into n regions according to the key position point set of the industrial kiln furnace, and obtaining the silicon carbide heater region set b={b1, b2, …, b i ,…,b n}, where b i represents the i-th region of the silicon carbide heater;

[0084] Allocate different heating powers according to the temperature requirements of each zone in the SiC heater zone set, and upload the real-time power of each zone to the database;

[0085] S14, set the historical time point set c={c1,c2,…,c i ,…,c m}, where c i represents the i-th time point in the historical time point set, and m represents the total number of historical time points in the historical time point set;

[0086] S15. Collect the historical time point set c={c1,c2,…,c i ,…,c m}, the key position point set of the industrial furnace is a={a1,a2,…,a i ,…,a n} and the temperature data of each key position point and the SiC heater area set b={b1,b2,…,b i ,…,b n}, and obtain the temperature combination set d={d1,d2,…,d i ,…,d m} and the SiC heater regional power parameter combination set e={e1,e2,…,e i ,…,e m}, where d i represents the temperature combination data of each key location at the historical time point i, e i represents power combination data of each area of ​​the silicon carbide heater at the historical time point i;

[0087] S2. According to the combination set of the temperature of the key positions of the industrial furnace and the combination set of the power parameters of the silicon carbide heater area, an objective function is set, and the objective function is minimized by using a genetic algorithm to obtain the optimal power parameter combination of the silicon carbide heater area, and the power parameter of each area in the optimal power parameter combination of the silicon carbide heater area is applied to the corresponding silicon carbide heater area;

[0088] The S2 comprises the following steps:

[0089] S21. Based on the temperature combination set of the key positions in the industrial furnace, define an objective function F, where F represents the weighted average temperature difference, as follows:

[0090] ;

[0091] Where n is the number of zones in the furnace, T i is the temperature of the ith region, is the average temperature of n regions, w i The weight of the i-th region is set according to the importance and temperature sensitivity of the region;

[0092] S22, according to the silicon carbide heater regional power parameter combination set, use genetic algorithm to find an optimal silicon carbide heater regional power parameter combination f={f1,f2,…,f i ,…,f n}, so that the objective function is minimized, where f i represents the power data of the i-th zone of the silicon carbide heater;

[0093] The S22 comprises the following steps:

[0094] S221, the silicon carbide heater regional power parameter combination set e={e1,e2,…,e i ,…,e m The number m of power parameter combinations of SiC heater regions in} is used as the size of chromosome population, and chromosome population g={g1,g2,…,g i ,…,g m}, where g i represents the i-th chromosome in the chromosome population; uses each chromosome in the chromosome population as a power parameter combination of each silicon carbide heater region; sets the maximum population iteration number to h1, the crossover rate to j, and the mutation rate to k;

[0095] S222, performing an encoding operation on each chromosome in the chromosome population to obtain an encoded chromosome population;

[0096] S223, start the iterative operation, set the current iteration number to h2, and in each round of iteration, perform selection, crossover and mutation operations on the encoded chromosome population according to the objective function F, the crossover rate j and the mutation rate k to obtain the chromosome population after the operation;

[0097] S224, decoding the operated chromosome population to obtain a decoded chromosome population;

[0098] S225, repeat S222, S223, S224, when h2 ≥ h1, stop iteration, and obtain the optimal silicon carbide heater regional power parameter combination f = {f1, f2, ..., f i ,…,f n};

[0099] S23, the optimal silicon carbide heater regional power parameter combination f={f1,f2,…,f i ,…,f n}, the power parameters of each zone in are applied to the corresponding silicon carbide heater zone;

[0100] S3, setting a first temperature collection time point set, collecting temperature data of key positions of the industrial kiln furnace at the time points in the first temperature collection time point set, obtaining a first temperature matrix, constructing an initial MLP model and setting relevant parameters, training, testing and optimizing the initial MLP model, and obtaining a final MLP model;

[0101] The S3 comprises the following steps:

[0102] S31, set the first temperature collection time point set p={p1, p2, ..., p i ,…,p m}, where p i represents the i-th temperature collection time point in the first temperature collection time point set; the temperature of the key position points in the industrial furnace key position point set is collected at the time point in the first temperature collection time point set to obtain the first temperature matrix A, as follows,

[0103] ;

[0104] Among them, A im It represents the temperature collected at the i-th key position at the m-th temperature collection time point;

[0105] S32, constructing an initial MLP model, setting the number of nodes in the input layer of the initial MLP model to r1, the number of nodes in the hidden layer to r2, and the number of nodes in the output layer to r3, setting the initial batch size of the initial MLP model to t1, the initial learning rate to t2, and setting the training ratio and test ratio of the initial MLP model to u1 and u2 respectively;

[0106] S33. Divide the first temperature matrix A into a first temperature training matrix and a first temperature test matrix according to the training ratio u1 and the test ratio u2 of the initial MLP model; set a first temperature training label matrix and a first temperature test label matrix;

[0107] S34. Set a training error threshold, a maximum number of training iterations, and the current number of training iterations; input the first temperature training matrix and the first temperature training label matrix into the initial MLP model to train the initial MLP model, adjust the number of nodes r1 in the input layer, the number of nodes r2 in the hidden layer, and the number of nodes r3 in the output layer of the initial MLP model according to the training results to obtain an adjusted MLP model, and continue to train the adjusted MLP model. When the training error ≤ the training error threshold or the current number of training iterations ≥ the maximum number of training iterations, stop training to obtain a trained MLP model;

[0108] S35. Input the first temperature test matrix and the first temperature test label matrix into the trained MLP model for testing. After the test optimization is completed, obtain a final MLP model;

[0109] The S35 includes the following steps:

[0110] S351. Set an accuracy threshold v1, input the first temperature test matrix into the trained MLP model for testing to obtain a test output result, compare the test output result with the first temperature test label matrix to obtain a test accuracy v2; when v2 ≥ v1, use the trained MLP model as the final MLP model; when v2 < v1, optimize the initial batch size and the initial learning rate of the trained MLP model to obtain an optimized MLP model, and use the optimized MLP model as the final MLP model;

[0111] The optimization of the initial batch size t1 and the initial learning rate t2 of the trained MLP model in the S351 to obtain an optimized MLP model includes the following steps:

[0112] S3511. Construct an ant population; set the scale of the ant population to J, then the ant population is represented as H = {H1, H2, …, H i , …, H J}, H i represents the i-th ant in the ant population; set the maximum number of test iterations; set the foraging search space dimension of the ant population to be two-dimensional;

[0113] S3512, randomly generate an initial foraging location set G of the ant population according to the initial batch size and the initial learning rate = {(G 11 ,G 12 ),(G 21 ,G 22 ),…,(G i1 ,G i2 ),…,(G J1 ,G J2 )}; G i1 and G i2 Respectively represent the first dimension coordinates and the second dimension coordinates of the initial foraging position of the i-th ant in the ant population;

[0114] The fitness function of the ant population is set, and the fitness function formula is as follows:

[0115] ;

[0116] In the formula, x represents the fitness function; y represents the bias;

[0117] S3513, start the iteration operation, set the number of real-time test iterations; in each round of iteration, update the foraging position of each ant in the initial foraging position set of the ant population according to the function formula of the fitness function; and obtain the global best individual and the global best fitness in the ant population in each round of iteration;

[0118] S3514, when the real-time test iteration number is greater than or equal to the maximum test iteration number, the ant population stops the iteration operation to obtain an optimized MLP model;

[0119] S4. Set a weighted average temperature difference threshold, collect real-time key position temperature data, calculate the weighted average temperature difference of the real-time key position temperature data, obtain the real-time weighted average temperature difference, and determine whether to adjust the power parameters of the silicon carbide heater area according to the real-time weighted average temperature difference; use the final MLP model to obtain future key position temperature data according to the real-time key position temperature data, and determine whether to adjust the power parameters of the silicon carbide heater area according to the future key position temperature data;

[0120] The S4 comprises the following steps:

[0121] S41, set the weighted average temperature difference threshold, set the real-time temperature collection time point set B = {B1, B2, ..., B i ,…,B N}, where B i represents the i-th temperature collection time point in the real-time temperature collection time point set; N represents the total number of temperature collection time points in the real-time temperature collection time point set;

[0122] At a time point where the real-time temperature collection time points are concentrated, the temperature of each key position point in the industrial furnace key position point concentration is collected to obtain the real-time key position point temperature data, and the weighted average temperature difference of the real-time key position point temperature data is calculated to obtain the real-time weighted average temperature difference. When the real-time weighted average temperature difference is ≤ the weighted average temperature difference threshold, the current state is maintained; when the real-time weighted average temperature difference is greater than the weighted average temperature difference threshold, the power of the silicon carbide heater area corresponding to the key position point is adjusted to make the real-time weighted average temperature difference ≤ the weighted average temperature difference threshold, and the problem that causes the real-time weighted average temperature difference to be greater than the weighted average temperature difference threshold is detected, and after the problem is solved, the optimal silicon carbide heater area power parameter combination is restored to use;

[0123] S42, set the future temperature collection time point set C = {C1, C2, ..., C i ,…,C N}, where C i represents the i-th temperature collection time point in the future temperature collection time point set;

[0124] According to the real-time key location temperature data and the future temperature collection time point set C={C1,C2,…,C i ,…,C N}, use the final MLP model to predict the temperature of each key position point in the key position point concentration of the industrial furnace at the time point where the future temperature collection time points are concentrated, and obtain the future key position point temperature data; calculate the weighted average temperature difference of the future key position point temperature data to obtain the future weighted average temperature difference. When the future weighted average temperature difference ≤ the weighted average temperature difference threshold, maintain the current state; when the future weighted average temperature difference > the weighted average temperature difference threshold, adjust the power of the silicon carbide heater area corresponding to the key position in advance, so that the future weighted average temperature difference ≤ the weighted average temperature difference threshold, and detect the problem that causes the future weighted average temperature difference > the weighted average temperature difference threshold. After the problem is solved, restore the use of the optimal silicon carbide heater area power parameter combination. Embodiment 2

[0125] An industrial kiln furnace temperature uniformity control system based on big data, used to implement the above-mentioned industrial kiln furnace temperature uniformity control method based on big data, including a data acquisition module, an optimal silicon carbide heater regional power parameter combination, a training and optimization MLP model, a real-time temperature determination module, and a future temperature data prediction and determination module;

[0126] The data acquisition module is used to collect the temperature data of the key positions of the furnace of the industrial kiln in history, the power data of the silicon carbide heater area in history; collect the temperature data of the key positions of the furnace of the industrial kiln at the time point concentrated at the first time point; collect real-time temperature data;

[0127] The optimal silicon carbide heater area power parameter combination is used to use the temperature data of key positions of the furnace of the historical industrial kiln and the power data of the historical silicon carbide heater area, and use the genetic algorithm to find the optimal silicon carbide heater area power parameter combination;

[0128] The training and optimization MLP model is used to train, test and optimize the initial MLP model according to the temperature data of the key position points of the furnace of the industrial kiln at the time point of the first time point concentration to obtain the final MLP model;

[0129] The real-time temperature determination module is used to calculate the weighted average temperature difference of the real-time temperature data, and compare it with the weighted average temperature difference threshold to determine whether to make an adjustment;

[0130] The future temperature data prediction and determination module is used to use the final MLP model to predict the temperature data of key positions of the industrial furnace hearth at future moments, calculate the weighted average temperature difference of the future temperature data, and compare it with the weighted average temperature difference threshold to determine whether to make adjustments.

[0131] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0132] The preferred embodiments of the invention disclosed above are only used to help explain the invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the invention, so that those skilled in the art can understand and use the invention well.

Claims

1. A method for controlling furnace temperature uniformity of an industrial kiln based on big data, characterized in that: The following steps are involved: S1. Set a key position point set of the industrial kiln furnace, arrange a temperature sensor at each key position point in the key position point set of the industrial kiln furnace, divide the silicon carbide heater into regions, obtain a silicon carbide heater region set, collect historical key position point temperature data and power data of each region in the silicon carbide heater region set, and obtain a combination set of historical key position point temperatures of the industrial kiln furnace and a combination set of silicon carbide heater region power parameters; S2. According to the combination set of the temperature of the key positions of the industrial furnace and the combination set of the power parameters of the silicon carbide heater area, an objective function is set, and the objective function is minimized by using a genetic algorithm to obtain the optimal power parameter combination of the silicon carbide heater area, and the power parameter of each area in the optimal power parameter combination of the silicon carbide heater area is applied to the corresponding silicon carbide heater area; S3, setting a first temperature collection time point set, collecting temperature data of key positions of the industrial kiln furnace at the time points in the first temperature collection time point set, obtaining a first temperature matrix, constructing an initial MLP model and setting relevant parameters, training, testing and optimizing the initial MLP model, and obtaining a final MLP model; S4. Set the weighted average temperature difference threshold, collect real-time key position temperature data, calculate the weighted average temperature difference of the real-time key position temperature data, obtain the real-time weighted average temperature difference, and judge whether to adjust the power parameters of the silicon carbide heater area according to the real-time weighted average temperature difference; according to the real-time key position temperature data, use the final MLP model to obtain the future key position temperature data, and judge whether to adjust the power parameters of the silicon carbide heater area according to the future key position temperature data.

2. The method for controlling furnace temperature uniformity of an industrial kiln based on big data according to claim 1 is characterized in that: The S1 comprises the following steps: S11. Setting the key position point set of the industrial furnace; S12, arranging a temperature sensor at each key position point in the industrial furnace furnace key position point set, the temperature sensor is used to collect temperature data of the corresponding key position point; S13, dividing the silicon carbide heater into n regions according to the key position point set of the industrial kiln furnace to obtain a silicon carbide heater region set; Allocate different heating powers according to the temperature requirements of each zone in the SiC heater zone set; S14, setting a historical time point set; S15. Collect the temperature data of each key position point in the industrial furnace key position point set and the power data of each area in the silicon carbide heater area set at each time point in the historical time point set, and obtain the temperature combination set of the historical key position points of the industrial furnace and the power parameter combination set of the silicon carbide heater area.

3. The method for controlling furnace temperature uniformity of an industrial kiln based on big data according to claim 2 is characterized in that: The S2 comprises the following steps: S21. Based on the temperature combination set of the key positions of the industrial furnace, define an objective function F, where F represents the weighted average temperature difference; S22, according to the silicon carbide heater regional power parameter combination set, using a genetic algorithm to find an optimal silicon carbide heater regional power parameter combination so that the objective function is minimized; S23, applying the power parameter of each zone in the optimal silicon carbide heater zone power parameter combination to the corresponding silicon carbide heater zone.

4. The method for controlling furnace temperature uniformity of an industrial kiln based on big data according to claim 3 is characterized in that: The S22 comprises the following steps: S221, taking the number of silicon carbide heater regional power parameter combinations in the silicon carbide heater regional power parameter combination set as the size of the chromosome population, constructing a chromosome population; taking each chromosome in the chromosome population as each silicon carbide heater regional power parameter combination; setting the maximum number of population iterations to h1, the crossover rate to j, and the mutation rate to k; S222, performing an encoding operation on each chromosome in the chromosome population to obtain an encoded chromosome population; S223, start the iterative operation, set the current iteration number to h2, and in each round of iteration, perform selection, crossover and mutation operations on the encoded chromosome population according to the objective function F, the crossover rate j and the mutation rate k to obtain the chromosome population after the operation; S224, decoding the operated chromosome population to obtain a decoded chromosome population; S225. Repeat S222, S223, and S224. When h2≥h1, stop the iteration to obtain the optimal silicon carbide heater zone power parameter combination.

5. The method for controlling furnace temperature uniformity of an industrial kiln based on big data according to claim 4 is characterized in that: The S3 comprises the following steps: S31, setting a first temperature collection time point set; collecting temperature at each key position point in the industrial furnace hearth key position point set at a time point in the first temperature collection time point set, to obtain a first temperature matrix; S32, constructing an initial MLP model, setting the number of nodes in the input layer of the initial MLP model to r1, the number of nodes in the hidden layer to r2, and the number of nodes in the output layer to r3, setting the initial batch size of the initial MLP model to t1, the initial learning rate to t2, and setting the training ratio and test ratio of the initial MLP model to u1 and u2 respectively; S33, dividing the first temperature matrix into a first temperature training matrix and a first temperature test matrix according to the training ratio and the test ratio of the initial MLP model; setting a first temperature training label matrix and a first temperature test label matrix; S34, setting a training error threshold, a maximum number of training iterations, and a current number of training iterations; inputting the first temperature training matrix and the first temperature training label matrix into the initial MLP model to train the initial MLP model, adjusting the number of nodes r1 of the input layer, the number of nodes r2 of the hidden layer, and the number of nodes r3 of the output layer of the initial MLP model according to the training results to obtain an adjusted MLP model, and continuing to train the adjusted MLP model. When the training error is ≤ the training error threshold, or the current number of training iterations is ≥ the maximum number of training iterations, the training is stopped to obtain a trained MLP model; S35. Input the first temperature test matrix and the first temperature test label matrix into the trained MLP model for testing. After the test optimization is completed, the final MLP model is obtained.

6. The method for controlling furnace temperature uniformity of an industrial kiln based on big data according to claim 5 is characterized in that: The S35 comprises the following steps: S351. Set the accuracy threshold v1, input the first temperature test matrix into the trained MLP model for testing to obtain a test output result, compare the test output result with the first temperature test label matrix to obtain the test accuracy v2; when v2≥v1, use the trained MLP model as the final MLP model; when v2<v1, optimize the initial batch size and initial learning rate of the trained MLP model to obtain an optimized MLP model, and use the optimized MLP model as the final MLP model.

7. The method for controlling furnace temperature uniformity of an industrial kiln based on big data according to claim 6 is characterized in that: The optimization of the initial batch size and initial learning rate of the trained MLP model in S351 to obtain an optimized MLP model includes the following steps: S3511. Construct an ant population; set the scale of the ant population to J; set the maximum number of test iterations; set the dimension of the foraging search space of the ant population to be two-dimensional; S3512. Randomly generate an initial foraging position set of the ant population according to the initial batch size and initial learning rate; Set the fitness function of the ant population; S3513. Start the iterative operation, set the real-time test iteration number; in each round of iteration, update the foraging positions of each ant in the initial foraging position set of the ant population according to the function formula of the fitness function; and obtain the global best individual and global best fitness in the ant population in each round of iteration; S3514. When the real-time test iteration number≥the maximum number of test iterations, the ant population stops the iterative operation to obtain an optimized MLP model.

8. The method for controlling furnace temperature uniformity of an industrial kiln based on big data according to claim 7 is characterized in that: S4 includes the following steps: S41. Set the weighted average temperature difference threshold, and set the real-time temperature acquisition time point set; At the time points in the real-time temperature acquisition time point set, collect the temperature of each key position point in the key position point set of the industrial furnace hearth to obtain the real-time key position point temperature data, calculate the weighted average temperature difference of the real-time key position point temperature data to obtain the real-time weighted average temperature difference. When the real-time weighted average temperature difference≤the weighted average temperature difference threshold, maintain the current state; when the real-time weighted average temperature difference>the weighted average temperature difference threshold, adjust the power of the silicon carbide heater area corresponding to the key position point to make the real-time weighted average temperature difference≤the weighted average temperature difference threshold, and detect the problem that causes the real-time weighted average temperature difference>the weighted average temperature difference threshold. After solving the problem, restore the use of the optimal silicon carbide heater area power parameter combination; S42. Set the future temperature acquisition time point set; According to the real-time key position temperature data and the future temperature collection time point set, the temperature of each key position point in the industrial furnace key position point set at the time point in the future temperature collection time point set in the final MLP model is predicted to obtain the future key position temperature data; the weighted average temperature difference of the future key position temperature data is calculated to obtain the future weighted average temperature difference. When the future weighted average temperature difference ≤ the weighted average temperature difference threshold, the current state is maintained; when the future weighted average temperature difference > the weighted average temperature difference threshold, the power of the silicon carbide heater area corresponding to the key position point is adjusted in advance to make the future weighted average temperature difference ≤ the weighted average temperature difference threshold, and the problem that causes the future weighted average temperature difference > the weighted average temperature difference threshold is detected, and the optimal silicon carbide heater area power parameter combination is restored after the problem is solved.

9. An industrial kiln furnace temperature uniformity control system based on big data, used to implement an industrial kiln furnace temperature uniformity control method based on big data as described in any one of claims 1 to 8.

Images