Decomposition furnace outlet temperature optimizing method, system, equipment and medium

Abstract

The invention provides a decomposing furnace outlet temperature optimizing method, system and device and a medium. The method comprises the steps that firstly, multi-source data of decomposing furnace operation are collected and preprocessed, features are extracted through an auto-encoder, and key features are screened out through a random forest to construct subsets; and then, constructing a mixed prediction model of LSTM and SVR, carrying out training by taking the feature subset as input and the outlet temperature as output, and adjusting parameters through a back propagation algorithm. And then determining an optimal combination of the coal feeding amount, the air amount and the raw material feeding amount by combining a genetic algorithm with the goal that the outlet temperature reaches a target value and the energy consumption is minimum. The optimization result is used for actual control, the temperature is monitored in real time, and if the deviation exceeds a threshold value, data are collected again for optimization. According to the method, the accuracy and the stability of outlet temperature control of the decomposing furnace are effectively improved, the energy consumption is remarkably reduced, the cross-working-condition adaptability is enhanced, the real-time performance and the efficiency of optimization are improved, and good economic benefits and quality guarantee are brought to industrial production.

Description

Technical Field

[0001] This document relates to the technical field of precalciner temperature control, and particularly to an optimization method, system, device and medium for the temperature at the outlet of the precalciner. Background Technique

[0002] In industrial production such as cement, the precise control of the temperature at the outlet of the precalciner is extremely crucial. Excessive temperature will cause over-decomposition of raw meal, equipment damage and increased energy consumption; too low temperature will result in insufficient decomposition of raw meal, affecting product quality and production efficiency.

[0003] Traditional control methods have many limitations. Operators manually adjust parameters such as coal feeding amount based on experience, which is highly subjective, difficult to ensure the accuracy and stability of temperature control, and it is difficult to make optimal decisions in a timely manner in the face of complex working conditions. Although the PID controller has the ability of automatic adjustment, in the face of the non-linear, large lag and strong coupling characteristics of the precalciner system, the control effect is not good, and it is easy to cause large fluctuations in temperature. The control method based on the mathematical model, although it predicts and controls the temperature through modeling, the operation of the precalciner is affected by various factors such as raw material characteristics and equipment status, and it is difficult to accurately model. The model parameters also need to be continuously adjusted, increasing the application difficulty.

[0004] In recent years, single machine learning algorithms such as neural networks and support vector machines have been used for temperature prediction and control, which has improved the prediction accuracy to a certain extent. However, the precalciner system is complex, and a single algorithm cannot fully mine data information and comprehensively consider the influence of various factors on temperature, and the optimization result is difficult to balance the temperature control accuracy and energy consumption optimization.

[0005] Therefore, the existing methods for controlling and optimizing the temperature at the outlet of the precalciner have deficiencies in terms of accuracy, adaptability and energy consumption. It is of great significance to develop an optimization method based on advanced machine learning algorithms. Summary of the Invention

[0006] The present invention provides an optimization method, system, device and medium for the temperature at the outlet of the precalciner, aiming to solve the above problems.

[0007] According to an embodiment of the present invention, an optimization method for the temperature at the outlet of the precalciner is provided, including:

[0008] S1. Collect multi-source data during the operation of the precalciner, clean the multi-source data, remove outliers and perform normalization processing; the multi-source data includes: the temperature at the outlet of the precalciner, the coal feeding amount, the system air volume, the raw meal feeding amount and the kiln tail waste gas composition data, and divide the processed multi-source data into a training set and a validation set according to a preset ratio;

[0009] S2. Use an autoencoder in deep learning to extract features from the processed multi-source data, calculate the importance scores of each feature through a random forest algorithm, screen out the key features, and construct a feature subset from the key features;

[0010] S3. Construct a hybrid prediction model based on a long short-term memory network and support vector regression. Use the feature subset as the input and the decomposition furnace outlet temperature as the output. Use the training set to train the hybrid prediction model, and adjust the parameters of the hybrid prediction model through the backpropagation algorithm to minimize the mean square error between the predicted temperature and the actual temperature;

[0011] S4. Based on the trained hybrid prediction model combined with a genetic algorithm, with the optimization goal of the decomposition furnace outlet temperature reaching the target value and the minimum energy consumption, determine the optimal parameter combination. The parameter combination is a combination of the coal feeding amount, the system air volume, and the raw material feeding amount; the fitness function of the genetic algorithm is constructed based on the deviation between the decomposition furnace outlet temperature and the target temperature and the energy consumption index;

[0012] S5. Apply the optimal parameter combination to the actual operation control of the decomposition furnace and monitor the decomposition furnace outlet temperature in real time.

[0013] According to an embodiment of the present invention, a system for optimizing the decomposition furnace outlet temperature is provided, including:

[0014] A data acquisition module that collects multi-source data during the operation of the decomposition furnace, cleans the multi-source data, removes outliers, and performs normalization processing; the multi-source data includes: the decomposition furnace outlet temperature, the coal feeding amount, the system air volume, the raw material feeding amount, and the kiln tail waste gas composition data, and divides the processed multi-source data into a training set and a validation set according to a preset ratio;

[0015] A feature extraction module that uses an autoencoder in deep learning to extract features from the processed multi-source data, calculates the importance scores of each feature through a random forest algorithm, screens out the key features, and constructs a feature subset from the key features;

[0016] A model construction module that constructs a hybrid prediction model based on a long short-term memory network and support vector regression. Use the feature subset as the input and the decomposition furnace outlet temperature as the output. Use the training set to train the hybrid prediction model, and adjust the parameters of the hybrid prediction model through the backpropagation algorithm to minimize the mean square error between the predicted temperature and the actual temperature;

[0017] An optimization module, based on the trained hybrid prediction model and combined with the genetic algorithm, takes the decomposition furnace outlet temperature reaching the target value and the minimum energy consumption as the optimization goal, and determines the optimal parameter combination, where the parameter combination is a combination of the coal feeding amount, the system air volume, and the raw material feeding amount; the fitness function of the genetic algorithm is constructed according to the deviation between the decomposition furnace outlet temperature and the target temperature and the energy consumption index.

[0018] A real-time control module applies the optimal parameter combination to the actual operation control of the decomposition furnace and monitors the decomposition furnace outlet temperature in real time.

[0019] According to an embodiment of the present invention, there is also provided an electronic device, including:

[0020] A processor; and,

[0021] A memory arranged to store computer-executable instructions, and when the computer-executable instructions are executed, the processor executes the steps of the method for optimizing the decomposition furnace outlet temperature as described above.

[0022] According to an embodiment of the present invention, there is also provided a storage medium for storing computer-executable instructions, and when the computer-executable instructions are executed, the steps of the method for optimizing the decomposition furnace outlet temperature as described above are implemented.

[0023] Adopting the embodiment of the present invention, taking the decomposition furnace outlet temperature reaching the target value and the minimum energy consumption as the optimization goal, constructs the fitness function of the genetic algorithm in combination with the energy consumption index during the optimization process. This optimization strategy can accurately adjust parameters such as the coal feeding amount, the system air volume, and the raw material feeding amount on the premise of ensuring production quality, avoid excessive energy consumption, realize efficient utilization of energy, and reduce the production cost of enterprises. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in one or more embodiments of this specification or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments recorded in this specification. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0025] Figure 1 It is a flowchart of a method for optimizing the decomposition furnace outlet temperature according to an embodiment of the present invention;

[0026] Figure 2 It is a schematic diagram of a system for optimizing the decomposition furnace outlet temperature according to an embodiment of the present invention. DETAILED DESCRIPTION

[0027] In order to enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the following will be combined with the drawings in one or more embodiments of this specification to clearly and completely describe the technical solutions in one or more embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this document.

[0028] Method Embodiment

[0029] According to an embodiment of the present invention, a method for optimizing the outlet temperature of a decomposition furnace is provided. Figure 1 This is a flow chart of a method for optimizing the outlet temperature of a decomposition furnace according to an embodiment of the present invention. Figure 1 As shown, a method for optimizing the outlet temperature of a decomposition furnace according to an embodiment of the present invention specifically includes:

[0030] S1. Collect multi-source data during the operation of the decomposition furnace, clean the multi-source data, remove outliers and perform normalization processing on the multi-source data; the multi-source data includes: decomposition furnace outlet temperature, coal feeding amount, system air volume, raw material feeding amount and kiln tail exhaust gas composition data, and divide the processed multi-source data into a training set and a verification set according to a preset ratio;

[0031] S1 specifically includes: continuously collecting multi-source data through sensors installed on the decomposition furnace and related equipment. Record the decomposition furnace outlet temperature, coal feed amount, system air volume, raw meal feed amount, and carbon monoxide, carbon dioxide, and oxygen content in the kiln tail exhaust gas every 5 minutes. Collect data for one month continuously as the initial data set. Perform a comprehensive check on the collected data. If it is found that the raw meal feed amount at a certain moment is negative, or the decomposition furnace outlet temperature exceeds the normal process range (such as below 700°C or above 1100°C), these data will be marked as outliers and eliminated. For a small amount of missing data, the average value of the previous and next data is used to fill in. The minimum and maximum normalization method is used to normalize the multi-source data, and the data is mapped to the [0, 1] interval.

[0032] S2. Use the autoencoder in deep learning to extract features from the processed multi-source data, calculate the importance score of each feature through the random forest algorithm, screen out key features, and construct feature subsets of the key features;

[0033] The autoencoder includes an input layer, a hidden layer and an output layer. The input layer receives processed multi-source data, the hidden layer encodes the input data and extracts potential features of the data, and the output layer decodes the encoding result of the hidden layer. The autoencoder is trained by minimizing the reconstruction error between the input data and the decoded data.

[0034] S3. Construct a hybrid prediction model based on the long short-term memory network and support vector regression. Use the feature subset as the input and the decomposition furnace outlet temperature as the output. Train the hybrid prediction model using the training set, and adjust the parameters of the hybrid prediction model through the backpropagation algorithm to minimize the mean square error between the predicted temperature and the actual temperature.

[0035] The specific construction method of the hybrid prediction model is as follows: First, use the LSTM network to learn the time series features of the data and capture the dynamic change information during the operation of the decomposition furnace. Then, use the output of the LSTM network as the input of the SVR, and the SVR further performs regression analysis on the data.

[0036] S4. Based on the trained hybrid prediction model combined with the genetic algorithm, with the optimization goal of the decomposition furnace outlet temperature reaching the target value and the minimum energy consumption, determine the optimal parameter combination. The parameter combination is the combination of the coal feeding amount, the system air volume, and the raw material feeding amount. The fitness function of the genetic algorithm is constructed according to the deviation between the decomposition furnace outlet temperature and the target temperature and the energy consumption index.

[0037] The specific operations of the genetic algorithm include:

[0038] Initialize the population: Randomly generate a certain number of individuals, and each individual represents a combination of the coal feeding amount, the system air volume, and the raw material feeding amount.

[0039] Calculate the fitness: Input the control parameters corresponding to each individual into the trained hybrid prediction model to obtain the predicted value of the decomposition furnace outlet temperature, and calculate the fitness value of each individual according to the fitness function.

[0040] Selection operation: Adopt the roulette wheel selection method to select a certain number of individuals as the parents according to the fitness values of the individuals.

[0041] Crossover operation: Perform crossover operations on the selected parent individuals to generate offspring individuals.

[0042] Mutation operation: Mutate some genes of the offspring individuals with a certain mutation probability.

[0043] Update the population: Replace the parent individuals with the offspring individuals to form a new population.

[0044] Termination condition judgment: When the preset number of iterations is met or the fitness value reaches the optimum, terminate the algorithm and output the control parameter combination corresponding to the optimal individual.

[0045] S5. Apply the optimal parameter combination to the actual operation control of the decomposition furnace and monitor the decomposition furnace outlet temperature in real time.

[0046] The method further includes: if the deviation between the outlet temperature of the precalciner and the target temperature exceeds a preset threshold, re-collect data and repeat S1 - S5.

[0047] When collecting multi-source data during the operation of the precalciner, the method further considers the influence of external environmental factors on the outlet temperature of the precalciner and performs dynamic adaptive adjustment. The specific steps are as follows:

[0048] Collect relevant data on the environment where the precalciner is located, including data on external environmental factors such as environmental temperature, environmental humidity, and atmospheric pressure, and fuse it with the multi-source data.

[0049] Use a deep belief network to mine features from the fused data containing external environmental factors, identify potential correlation features between external environmental factors and the outlet temperature of the precalciner. The deep belief network is composed of multiple restricted Boltzmann machines stacked together. Train the Boltzmann machines layer by layer through unsupervised learning to extract deep feature representations of the data.

[0050] Based on the mined environmental impact features, dynamically adjust the parameters of the hybrid prediction model and the genetic algorithm. When significant changes occur in external environmental factors, adjust the weights and biases of the long short-term memory network and support vector regression in the hybrid prediction model, as well as the parameters of the genetic algorithm.

[0051] Adopting the embodiment of the present invention has the following beneficial effects:

[0052] By comprehensively applying technologies such as autoencoders, random forests, LSTM, SVR, and genetic algorithms, it is possible to accurately predict and optimize the control of the outlet temperature of the precalciner. Fully mine the key information in multi-source data, accurately capture the dynamic changes in the operation of the precalciner, and effectively reduce temperature fluctuations. Stable and accurate control of the outlet temperature can ensure sufficient and uniform decomposition of raw meal, thus significantly improving product quality, reducing the defective rate, and enhancing the competitiveness of the enterprise in the market.

[0053] Taking the outlet temperature of the precalciner reaching the target value and minimizing energy consumption as the optimization goal, construct the fitness function of the genetic algorithm by combining energy consumption indicators during the optimization process. This optimization strategy can accurately adjust parameters such as coal feeding amount, system air volume, and raw meal feeding amount on the premise of ensuring production quality, avoid excessive energy consumption, achieve efficient utilization of energy, reduce the production cost of the enterprise, and meet the requirements of green production and sustainable development.

[0054] Considering the complex and changeable operating conditions that the decomposition furnace may face during operation, the present invention introduces a variety of innovative technologies. For example, it takes into account the impact of external environmental factors on temperature and makes dynamic adaptive adjustments, and uses transfer learning and multi-model fusion technologies to achieve optimization across different operating conditions. These measures enable the method to quickly adapt to different production conditions, raw material characteristics, and environmental changes, enhancing the adaptability and robustness of the system and reducing production instability caused by changes in operating conditions.

[0055] Combined with Internet of Things and edge computing technologies, real-time data collection, preprocessing, and analysis are achieved. Edge computing nodes can make quick decisions and adjust control parameters locally, while collaborating with the cloud server in real time, greatly shortening the data processing and decision-making time, improving the real-time performance of temperature optimization, ensuring timely response to emergencies during the operation of the decomposition furnace, and guaranteeing the continuity and stability of the production process.

[0056] During the optimization process, knowledge is accumulated for the data, model parameters, and evaluation results of each optimization, and potential relationships are mined using knowledge graph technology. This not only helps to quickly optimize in subsequent similar operating conditions but also provides a basis for the continuous optimization of models and strategies, enabling the optimization method to continuously evolve and better meet the needs of industrial production.

[0057] System embodiment

[0058] According to an embodiment of the present invention, a system for optimizing the outlet temperature of a decomposition furnace is provided. Figure 2 It is a schematic diagram of a system for optimizing the outlet temperature of a decomposition furnace according to an embodiment of the present invention. As shown in Figure 2 According to the shown, a system for optimizing the outlet temperature of a decomposition furnace according to an embodiment of the present invention specifically includes:

[0059] A data acquisition module 20, which collects multi-source data during the operation of the decomposition furnace, cleans the multi-source data, removes outliers, and normalizes it; the multi-source data includes: the outlet temperature of the decomposition furnace, the coal feeding amount, the system air volume, the raw material feeding amount, and the kiln tail waste gas component data, and divides the processed multi-source data into a training set and a validation set according to a preset ratio;

[0060] A feature extraction module 22, which uses an autoencoder in deep learning to extract features from the processed multi-source data, calculates the importance scores of each feature through a random forest algorithm, screens out key features, and constructs a feature subset with the key features;

[0061] A model construction module 24, which constructs a hybrid prediction model based on a long short-term memory network and support vector regression, uses the feature subset as the input and the outlet temperature of the decomposition furnace as the output, trains the hybrid prediction model using the training set, and adjusts the parameters of the hybrid prediction model through backpropagation algorithm to minimize the mean square error between the predicted temperature and the actual temperature.

[0062] The optimization module 26, based on the trained hybrid prediction model and combined with the genetic algorithm, takes the decomposition furnace outlet temperature reaching the target value and the minimum energy consumption as the optimization objective, and determines the optimal parameter combination, where the parameter combination is the combination of the coal feeding amount, the system air volume, and the raw material feeding amount; the fitness function of the genetic algorithm is constructed according to the deviation between the decomposition furnace outlet temperature and the target temperature and the energy consumption index;

[0063] The real-time control module 28 applies the optimal parameter combination to the actual operation control of the decomposition furnace and monitors the decomposition furnace outlet temperature in real time.

[0064] Device Embodiment 1

[0065] According to an embodiment of the present invention, there is provided an electronic device, including:

[0066] A processor; and,

[0067] A memory arranged to store computer-executable instructions, which when executed cause the processor to execute the steps of the method embodiment as described above.

[0068] Device Embodiment 2

[0069] According to an embodiment of the present invention, there is provided a storage medium for storing computer-executable instructions, which when executed implement the steps of the method embodiment as described above.

[0070] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing the outlet temperature of a decomposition furnace, characterized in that: The following steps are involved: S1. Collect multi-source data during the operation of the decomposition furnace, clean the multi-source data, remove outliers, and perform normalization processing; The multi-source data includes: decomposition furnace outlet temperature, coal feeding amount, system air volume, raw material feeding amount and kiln tail exhaust gas composition data, and the processed multi-source data is divided into a training set and a verification set according to a preset ratio; S2. Use the autoencoder in deep learning to extract features from the processed multi-source data, calculate the importance score of each feature through the random forest algorithm, screen out key features, and construct feature subsets of the key features; S3, constructing a hybrid prediction model based on long short-term memory network and support vector regression, taking the feature subset as input and the decomposition furnace outlet temperature as output, training the hybrid prediction model using the training set, and adjusting the parameters of the hybrid prediction model by back propagation algorithm to minimize the mean square error between the predicted temperature and the actual temperature; S4. Based on the trained hybrid prediction model combined with the genetic algorithm, the optimal parameter combination is determined with the goal of achieving the target value of the decomposition furnace outlet temperature and minimizing the energy consumption, wherein the parameter combination is a combination of the coal feeding amount, the system air volume and the raw material feeding amount; the fitness function of the genetic algorithm is constructed according to the deviation between the decomposition furnace outlet temperature and the target temperature and the energy consumption index; S5. Apply the optimal parameter combination to the actual operation control of the decomposition furnace and monitor the outlet temperature of the decomposition furnace in real time.

2. The method according to claim 1, characterized in that: The minimum and maximum normalization method is used to normalize multi-source data, mapping the data to the [0, 1] interval.

3. The method according to claim 1, characterized in that The autoencoder includes an input layer, a hidden layer and an output layer. The input layer receives processed multi-source data, the hidden layer encodes the input data and extracts potential features of the data, and the output layer decodes the encoding result of the hidden layer. The autoencoder is trained by minimizing the reconstruction error between the input data and the decoded data.

4. The method according to claim 1, characterized in that: The specific construction method of the hybrid prediction model is: first, use the LSTM network to learn the time series characteristics of the data to capture the dynamic change information during the operation of the decomposition furnace, and then use the output of the LSTM network as the input of the SVR, which further performs regression analysis on the data.

5. The method according to claim 1, characterized in that The specific operations of the genetic algorithm include: Initialize the population: randomly generate a certain number of individuals, each of which represents a combination of coal feeding amount, system air volume and raw meal feeding amount; Calculate fitness: Input the control parameters corresponding to each individual into the trained hybrid prediction model to obtain the predicted value of the decomposition furnace outlet temperature, and calculate the fitness value of each individual according to the fitness function; Selection operation: Roulette selection method is used to select a certain number of individuals as parents according to their fitness values; Crossover operation: Perform a crossover operation on the selected parent individuals to generate offspring individuals; Mutation operation: mutate certain genes of offspring individuals with a certain mutation probability; Update population: replace parent individuals with offspring individuals to form a new population; Termination condition judgment: When the preset number of iterations is met or the fitness value reaches the optimal value, the algorithm is terminated and the control parameter combination corresponding to the optimal individual is output.

6. The method according to claim 1, characterized in that The method further comprises: if the deviation between the decomposition furnace outlet temperature and the target temperature exceeds a preset threshold, re-collecting data and repeatedly executing S1-S5.

7. The method according to claim 1, characterized in that The method further considers the influence of external environmental factors on the outlet temperature of the decomposition furnace when collecting multi-source data during the operation of the decomposition furnace, and performs dynamic adaptive adjustment. The specific steps are as follows: Collect relevant data about the environment of the decomposition furnace, including external environmental factors such as ambient temperature, ambient humidity, and atmospheric pressure, and integrate them with multi-source data; A deep belief network is used to mine the features of the fused data containing external environmental factors, and to identify the potential correlation features between the external environmental factors and the outlet temperature of the decomposition furnace. The deep belief network is composed of multiple stacked restricted Boltzmann machines. The Boltzmann is trained layer by layer through unsupervised learning to extract the deep feature representation of the data. Based on the mined environmental impact characteristics, the parameters of the hybrid prediction model and genetic algorithm are dynamically adjusted. When the external environmental factors change significantly, the weights and biases of the long short-term memory network and support vector regression in the hybrid prediction model, as well as the parameters of the genetic algorithm, are adjusted.

8. A system for optimizing the outlet temperature of a decomposition furnace, characterized in that: The system comprises: The data acquisition module collects multi-source data during the operation of the decomposition furnace, cleans the multi-source data, removes outliers, and performs normalization processing; the multi-source data includes: decomposition furnace outlet temperature, coal feeding amount, system air volume, raw material feeding amount, and kiln tail exhaust gas composition data, and the processed multi-source data is divided into a training set and a verification set according to a preset ratio; The feature extraction module uses the autoencoder in deep learning to extract features from the processed multi-source data, calculates the importance score of each feature through the random forest algorithm, screens out key features, and constructs feature subsets of the key features; A model building module, constructing a hybrid prediction model based on long short-term memory network and support vector regression, taking the feature subset as input and the decomposition furnace outlet temperature as output, training the hybrid prediction model using the training set, and adjusting the parameters of the hybrid prediction model through a back propagation algorithm to minimize the mean square error between the predicted temperature and the actual temperature; The optimization module, based on the trained hybrid prediction model combined with the genetic algorithm, takes the decomposition furnace outlet temperature reaching the target value and the energy consumption being minimized as the optimization goal, and determines the optimal parameter combination, which is the combination of the coal feeding amount, the system air volume and the raw material feeding amount; the fitness function of the genetic algorithm is constructed according to the deviation between the decomposition furnace outlet temperature and the target temperature and the energy consumption index; The real-time control module applies the optimal parameter combination to the actual operation control of the decomposition furnace and monitors the outlet temperature of the decomposition furnace in real time.

9. An electronic device, comprising: processor; as well as, A memory arranged to store computer executable instructions, which, when executed, cause the processor to perform the steps of the method for optimizing the outlet temperature of a decomposition furnace as claimed in any one of claims 1 to 7.

10. A storage medium for storing computer executable instructions, wherein the computer executable instructions, when executed, implement the steps of the method for optimizing the decomposition furnace outlet temperature as claimed in any one of claims 1 to 7.

Images