Intelligent regulation and control method for solid waste collaborative utilization in calcium carbide production process

Abstract

The invention discloses an intelligent regulation and control method for solid waste collaborative utilization in a calcium carbide production process, and relates to the technical field of collaborative regulation and control, and the method comprises the following steps: collecting solid waste carbon spectral density original data of solid waste entering a batching bin in the calcium carbide production process, and carrying out clustering calculation on carbon content distribution and activation energy of different types of waste plastics to obtain a joint feature vector; establishing a solid waste carbonaceous feature matrix according to the joint feature vector; the gas recovery and residue separation process is adjusted through optimization control parameters of a digital twinborn model, and whole-process operation data is obtained by combining gas reutilization and calcium solid waste backflow; based on the whole process operation data, the change trend of the key variables is extracted, a working condition evolution rule is established, a control strategy network is trained through a reinforcement learning algorithm, and a working condition optimization instruction adaptive to the current production state is generated. Through the reinforcement learning algorithm, changes in the production environment can be adaptively coped with, the carbon conversion efficiency is improved, the energy consumption is reduced, and automatic optimization of production is achieved.

Description

Technical Field

[0001] This invention relates to the field of collaborative control technology, and in particular to an intelligent control method for the collaborative utilization of solid waste in the calcium carbide production process. Background Technology

[0002] In the calcium carbide production process, the carbon content of solid waste and the carbon conversion process are important research areas for optimizing energy consumption and production efficiency. With increasingly stringent requirements for energy costs and environmental protection, more and more research is dedicated to improving resource utilization and reducing carbon emissions in calcium carbide production. Traditional methods mainly rely on empirical formulas and simple models to estimate the carbon composition in waste materials. While these methods can provide preliminary estimates, they often fail to fully utilize multi-dimensional waste characteristic data. In recent years, the application of machine learning and data-driven models in in-furnace carbon conversion and energy consumption modeling has gradually provided new ideas for improving production efficiency and accuracy.

[0003] However, many existing methods neglect the characteristics of different types of waste and fail to accurately capture the complex relationship between solid waste carbon content distribution, activation energy, and production conditions. Existing carbon conversion prediction models often rely on single carbon content data and lack real-time optimization and control of the ratio of waste plastics to limestone and carbon conversion efficiency. Therefore, although many methods can provide useful results under specific conditions, their insufficient accuracy and real-time performance limit the optimization of carbon conversion processes and the improvement of energy efficiency in actual industrial production. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides an intelligent control method for the co-utilization of solid waste in the calcium carbide production process, which solves the problem in the prior art that there is a lack of precise control and real-time feedback between the carbon content of solid waste and energy efficiency optimization in the calcium carbide production process.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides an intelligent control method for the co-utilization of solid waste in the calcium carbide production process, comprising: Raw carbon density data of solid waste entering the batching silo during calcium carbide production were collected. Clustering calculations were performed on the carbon content distribution and activation energy of different types of waste plastics to obtain a joint feature vector. A solid waste carbon feature matrix was then established based on this joint feature vector. The ratio of waste plastics to limestone was controlled in real time according to the solid waste carbon feature matrix. The mixed raw material flow was obtained, and carbon spectrum tags and furnace temperature signals were extracted from the mixed raw material flow. A carbon balance prediction model was constructed, and the carbon spectrum tags and furnace temperature signals were input into the model to predict the carbon conversion trend within the furnace, yielding energy consumption data and carbon conversion. Status data; based on energy consumption data and carbon conversion status data, virtual prediction and real-time correction are performed on the synchronous mapping of the furnace temperature field and furnace atmosphere field through a digital twin model, generating optimized control parameters for the digital twin model; the gas recovery and residue separation processes are adjusted through the optimized control parameters of the digital twin model, and the entire process operation data is obtained by combining gas reuse and calcium solid waste recirculation; based on the entire process operation data, the changing trends of key variables are extracted and the evolution law of operating conditions is established, and the control strategy network is trained through reinforcement learning algorithm to generate operating condition optimization instructions adapted to the current production state.

[0007] As a preferred embodiment of the intelligent control method for the co-utilization of solid waste in the calcium carbide production process described in this invention, the specific steps for obtaining the joint feature vector are as follows: Noise removal, time alignment, and scale normalization were performed on the raw carbon density data of solid waste to obtain a unified cleaned dataset. Extract the carbon content, ash content, and moisture content from the unified clean dataset, and calculate the effective carbon mass fraction; Thermogravimetric analysis was performed on the unified clean dataset to obtain the thermogravimetric derivative spectrum, and the characteristic peaks in the thermogravimetric derivative spectrum were extracted to obtain the carbon content distribution. The initial activation energy is calculated based on the characteristic temperature peak related to carbon content, and the initial activation energy is corrected by regression using the isoconversion method to obtain the activation energy. The effective carbon mass fraction, carbon content distribution, and activation energy are concatenated into a joint feature vector.

[0008] As a preferred embodiment of the intelligent control method for the co-utilization of solid waste in the calcium carbide production process described in this invention, the specific steps for establishing the solid waste carbonaceous characteristic matrix are as follows: Standardization and dimensional unification are performed on the joint feature vectors to obtain a unified feature set; The unified feature set is classified according to the preset carbon conversion feature weighting rules, and the features in the unified feature set are weighted and synthesized to form a comprehensive feature value, which is then summarized into a solid waste carbon feature matrix.

[0009] As a preferred embodiment of the intelligent control method for the co-utilization of solid waste in the calcium carbide production process described in this invention, the specific steps for obtaining the mixed raw material stream are as follows: Carbonaceous characteristics data of different waste plastics and limestone were extracted from the carbonaceous characteristics matrix of solid waste; Based on the carbonaceous characteristics of different waste plastics and limestone, the feeding ratio of waste plastics and limestone is calculated to obtain the preliminary mixed raw material flow of waste plastics and limestone. Based on the initial mixed raw material stream of waste plastics and limestone, the carbonaceous characteristics of the mixed raw material stream are monitored in real time, and the feeding ratio of waste plastics and limestone is adjusted to the optimal feeding ratio to obtain the mixed raw material stream.

[0010] As a preferred embodiment of the intelligent control method for the co-utilization of solid waste in the calcium carbide production process described in this invention, the specific steps for constructing the carbon balance prediction model are as follows: The raw historical solid waste carbon density data were subjected to noise removal, normalization, and data cleaning to obtain the processed solid waste carbon dataset. Based on the processed solid waste carbon dataset, a preliminary carbon balance prediction model was established by selecting a thermodynamic model. Regression analysis was performed on the raw historical solid waste carbon spectral density data to obtain the optimized preliminary carbon balance prediction model parameters. By comparing the differences between the optimized preliminary carbon balance prediction model parameters and the original data of actual solid waste carbon spectral density, the preliminary carbon balance prediction model is corrected to obtain the final carbon balance prediction model.

[0011] As a preferred embodiment of the intelligent control method for the co-utilization of solid waste in the calcium carbide production process described in this invention, the specific steps for obtaining energy consumption data and carbon conversion status data are as follows: The carbon spectrum label of the mixed feed stream and the furnace temperature signal are input into the carbon balance prediction model to estimate the carbon conversion trend in the furnace. Based on the carbon conversion trend in the furnace, the carbon conversion state in the furnace is calculated and the evolution trend of carbon conversion is predicted to obtain carbon conversion state data. Energy consumption data is obtained by smoothing and correcting the carbon conversion state data using the Kalman filter algorithm.

[0012] As a preferred embodiment of the intelligent control method for the co-utilization of solid waste in the calcium carbide production process described in this invention, the specific steps for generating the optimized control parameters of the digital twin model are as follows: Energy consumption data and carbon conversion status data are input into a digital twin model to synchronously map the furnace temperature field and furnace atmosphere field, thereby obtaining furnace temperature data and furnace atmosphere field data. Virtual predictions were performed on furnace temperature data and furnace atmosphere field data to estimate the trends of furnace temperature change and furnace atmosphere composition change. Based on the deviation between the furnace temperature change trend and the furnace temperature data, and the deviation between the furnace atmosphere composition change trend and the furnace atmosphere field data, a correction algorithm is used to adjust the digital twin model and generate optimized control parameters for the digital twin model.

[0013] As a preferred embodiment of the intelligent control method for the co-utilization of solid waste in the calcium carbide production process described in this invention, the steps include: adjusting the gas recovery and residue separation processes through optimized control parameters using a digital twin model, and obtaining full-process operation data by combining gas reuse and calcium solid waste recirculation. The gas recovery flow rate is adjusted by optimizing the control parameters of the digital twin model, and the adjusted recovery gas flow rate data is obtained. Based on the adjusted recovered gas flow rate data, the separation efficiency and speed of the residue separation process are adjusted to obtain the adjusted residue separation data; Regression analysis and optimization were performed on the adjusted residue separation data and the adjusted recovered gas flow rate data to generate optimized gas reuse flow rate data; Based on the gas reuse flow rate data, the return flow rate and return location of the calcium solid waste are adjusted to obtain the operation data of calcium solid waste return. The adjusted recovered gas flow rate data, adjusted residue separation data, optimized gas reuse flow rate data, and calcium solid waste recirculation operation data are summarized into full-process operation data.

[0014] As a preferred embodiment of the intelligent control method for the co-utilization of solid waste in the calcium carbide production process described in this invention, the specific steps for extracting the changing trends of key variables and establishing the evolution law of operating conditions based on the full-process operation data are as follows: After cleaning and standardizing the data from the entire process, time series analysis is performed to generate the changing trends of key variables. Statistical analysis is conducted on the changing trends of key variables to identify the correlations between key variables, construct mutual influence relationships, determine the main operating conditions, and establish the laws governing the evolution of operating conditions.

[0015] As a preferred embodiment of the intelligent control method for the co-utilization of solid waste in the calcium carbide production process described in this invention, the specific steps for generating the operating condition optimization command adapted to the current production state are as follows: The control policy network is trained using reinforcement learning algorithms and then optimized through interactive learning to obtain the optimized control policy network. Based on the optimized control strategy network, the evolution of operating conditions is adjusted in real time to generate operating condition optimization instructions that are adapted to the current production status.

[0016] The beneficial effects of this invention are as follows: by extracting the changing trends of key variables from the full-process operation data, and training the control strategy network according to the reinforcement learning algorithm, the invention generates operating condition optimization instructions adapted to the current production state. This allows for dynamic adjustment of various operating parameters in the production process based on real-time monitoring data, ensuring that the production process remains in optimal condition. Through the reinforcement learning algorithm, the invention can adaptively respond to changes in the production environment, improve carbon conversion efficiency, and reduce energy consumption, thereby achieving automated optimization of production. This method not only improves resource utilization efficiency but also reduces the need for human intervention, greatly enhancing the intelligence level and production efficiency of the production process. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Fig. 1 A flowchart of an intelligent control method for the co-utilization of solid waste in the calcium carbide production process.

[0019] Fig. 2 This is a flowchart for collecting and processing carbon density data of solid waste.

[0020] Fig. 3 A flowchart for constructing the carbonaceous characteristic matrix of solid waste.

[0021] Fig. 4 A flowchart for generating operating condition optimization instructions. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Reference Figs. 1-4This is one embodiment of the present invention, which provides an intelligent control method for the co-utilization of solid waste in the calcium carbide production process, comprising the following steps: S1. Collect raw data of carbon density of solid waste entering the batching silo from the calcium carbide production process, and perform cluster calculation on the carbon content distribution and activation energy of different types of waste plastics to obtain joint feature vectors. Establish solid waste carbon feature matrix based on joint feature vectors. S1.1: Noise removal, time alignment, and scale normalization are performed on the raw carbon density data of solid waste to obtain a unified cleaned dataset; Specifically, by applying filtering algorithms, interference signals in the original solid waste carbon density data are removed to obtain noise-removed original solid waste carbon density data. The noise-removed original solid waste carbon density data is then time-aligned, and scale normalization is performed to standardize the original solid waste carbon density data from different sources to the same dimensional range, resulting in a unified cleaned dataset.

[0026] S1.2: Extract the carbon content, ash content, and moisture content from the unified clean dataset, and calculate the effective carbon mass fraction. The expression is: ; in, Indicates the effective carbon mass fraction; This indicates the carbon content in the uniformly cleaned dataset; This represents the ash content in the uniformly cleaned dataset; This indicates the moisture content in the uniformly cleaned dataset.

[0027] S1.3: Perform thermogravimetric analysis on the unified clean dataset to obtain the thermogravimetric derivative spectrum, and extract the characteristic peaks in the thermogravimetric derivative spectrum to obtain the carbon content distribution; Specifically, thermogravimetric analysis (TGA) is performed on the unified cleaning dataset. The unified cleaning dataset is heated under controlled heating rate, and the mass of the unified cleaning dataset changes with temperature, and a thermogravimetric curve is plotted. The thermogravimetric derivative spectrum is obtained by numerically differentiating the thermogravimetric curve, and the characteristic peaks in the thermogravimetric derivative spectrum are identified and analyzed to obtain the characteristic temperature peaks related to carbon content, thereby obtaining the carbon content distribution in the original solid waste carbon density data.

[0028] S1.4: Calculate the initial value of activation energy based on the characteristic temperature peak related to carbon content, and then perform regression correction on the initial value of activation energy using the isoconversion method to obtain the activation energy; Specifically, the initial activation energy is calculated based on the characteristic temperature peak related to carbon content, and the initial activation energy is corrected by regression using the carbon conversion reaction rate at different conversion rates through the isoconversion method. By comparing the carbon conversion reaction rate corresponding to different conversion rates with the preset expected activation energy value, the initial activation energy is gradually corrected to obtain a more accurate activation energy.

[0029] It should be noted that the expected activation energy value is set based on a preliminary estimate of thermodynamic models and experimental data. An exemplary value of 150 kJ / mol is based on the reaction kinetic data of solid waste materials and the results of thermogravimetric analysis experiments. The expression for calculating the initial activation energy based on the characteristic temperature peak related to carbon content is as follows: ; in, This represents the initial activation energy, expressed in J / mol. This represents the gas constant, with units of J / (mol·K); Indicates the characteristic temperature peak related to carbon content, in K; This indicates the heating rate, expressed in K / min. This represents the natural logarithm function.

[0030] S1.5: Combine the effective carbon mass fraction, carbon content distribution, and activation energy into a joint feature vector.

[0031] Specifically, the effective carbon mass fraction, carbon content distribution, and activation energy are standardized to remove unit differences, and then concatenated in the same dimension according to the order of effective carbon mass fraction, carbon content distribution, and activation energy to form a joint feature vector.

[0032] S1.6: Perform standardization and dimensional unification on the joint feature vectors to obtain a unified feature set; Specifically, each feature of the joint feature vector is standardized and converted into a standard normal distribution with a mean of zero and a variance of one, ensuring that each feature of the joint feature vector is within the same scale range. Then, through normalization, features with different units and dimensions in the joint feature vector are unified to the same scale, resulting in a unified feature set.

[0033] S1.7: Classify the unified feature set according to the preset carbon conversion feature weighting rules, perform weighted synthesis of each feature in the unified feature set to form a comprehensive feature value, and summarize it into a solid waste carbon feature matrix.

[0034] Specifically, based on the impact of each feature in the unified feature set on the carbon conversion process, the importance and weight of each feature are determined; each feature in the unified feature set is assigned a corresponding weight, and each feature is weighted and synthesized to obtain a comprehensive feature value; according to the preset carbon conversion feature weighting rules, all comprehensive feature values ​​in the unified feature set are summarized in a unified format to form a solid waste carbon feature matrix.

[0035] It should be noted that the carbon conversion feature weighting rule is set based on the understanding of the solid waste conversion process and the impact of each feature on the carbon conversion efficiency. It refers to the standard set in advance for classifying, weighting and ranking each feature when processing a unified feature set, so as to ensure that the contribution of different features matches their corresponding importance.

[0036] S2. Based on the solid waste carbonaceous characteristic matrix, the ratio of waste plastics to limestone is controlled in real time to obtain the mixed raw material flow, and carbon spectrum tags and furnace temperature signals are extracted from the mixed raw material flow. S2.1: Extract carbonaceous characteristic data of different waste plastics and limestone from the solid waste carbonaceous characteristic matrix; Specifically, based on the corresponding rows of different waste plastics and limestone in the solid waste carbonaceous characteristic matrix, the effective carbon mass fraction, carbon content distribution and activation energy are extracted and then classified according to the type identification of waste plastics and limestone to obtain carbonaceous characteristic data of different waste plastics and limestone. The screened carbonaceous characteristic data are then organized and standardized to ensure that the carbonaceous characteristic data of each type of waste plastic and limestone has consistent units and dimensions.

[0037] S2.2: Based on the carbonaceous characteristics of different waste plastics and limestone, calculate the feeding ratio of waste plastics and limestone to obtain the preliminary mixed raw material flow of waste plastics and limestone; Specifically, based on the carbonaceous characteristics of different waste plastics and limestone, the feeding ratio of waste plastics and limestone is calculated, and the feeding amount of waste plastics and limestone is adjusted according to the feeding ratio to obtain a preliminary mixed raw material flow of waste plastics and limestone.

[0038] The expression for calculating the ratio of waste plastic to limestone is: ; in, Indicates the amount of waste plastic fed in; This indicates the amount of limestone fed into the machine; Indicates the carbon content of the preset target mixed feedstock stream; Indicates the carbon content of limestone; This indicates the carbon content of waste plastics.

[0039] S2.3: Based on the initial mixed raw material stream of waste plastic and limestone, the carbonaceous characteristics of the mixed raw material stream are monitored in real time, and the feeding ratio of waste plastic and limestone is adjusted to the optimal feeding ratio to obtain the mixed raw material stream.

[0040] Specifically, based on the initial mixed raw material stream of waste plastics and limestone, the carbon content, ash content, and moisture content of the waste plastics and limestone in the mixed raw material stream are extracted by real-time monitoring of the carbonaceous characteristics of the mixed raw material stream and compared with the preset target carbon content. According to the real-time monitoring results and the preset target carbon content, the feeding ratio of waste plastics and limestone is calculated to ensure that the carbon content of the mixed raw material stream reaches the optimal feeding ratio. Then, by adjusting the feeding amount of waste plastics and limestone, the mixed raw material stream is adjusted in real time to ensure that the optimal carbon conversion efficiency and energy utilization rate are maintained during the production process.

[0041] It should be noted that the target carbon content is set based on the carbon conversion efficiency requirements of the production process, the reaction rate requirements, and the product quality standards. The target carbon content value is obtained by comparing and analyzing historical production data, laboratory analysis results, and industry standards in similar processes.

[0042] S3. Construct a carbon balance prediction model. Input the carbon spectrum label in the mixed raw material stream and the furnace temperature signal into the carbon balance prediction model to calculate the carbon conversion trend in the furnace and obtain energy consumption data and carbon conversion status data. S3.1: Remove noise, normalize, and clean the raw historical solid waste carbon density data to obtain the processed solid waste carbon dataset; Specifically, filtering algorithms are applied to remove noise from the original historical solid waste carbon density data to eliminate interference signals. The original historical solid waste carbon density data is then transformed to a uniform scale using a normalization method, so that different features have the same dimensions. Data cleaning is performed on the original historical solid waste carbon density data to remove missing values ​​and outliers, resulting in a processed solid waste carbon dataset.

[0043] S3.2: Based on the processed solid waste carbon dataset, a preliminary carbon balance prediction model is established by selecting a thermodynamic model; Specifically, by analyzing the carbon content, temperature changes, and reaction kinetic characteristics in the processed solid waste carbon dataset, a thermodynamic model matching the carbon conversion process is selected, and the various thermodynamic parameters in the thermodynamic model are initially set. The characteristics in the processed solid waste carbon dataset are input into the thermodynamic model to calculate the preliminary carbon conversion rate, adjust the thermodynamic model parameters, and establish a preliminary carbon balance prediction model.

[0044] It should be noted that the pre-training process of the thermodynamic model involves selecting a thermodynamic model that matches the carbon conversion process based on operational data related to the carbon conversion process obtained from actual experiments and historical operational data related to the carbon conversion process. The thermodynamic parameters involved in the model are then initially set and calibrated. The model is validated by analyzing the carbon content, temperature changes, and reaction kinetic characteristics in the processed solid waste carbon dataset, combined with experimental data. The parameters of the thermodynamic model are continuously optimized to ensure that it accurately reflects the carbon conversion process.

[0045] S3.3: Perform regression analysis on the raw data of historical solid waste carbon spectral density to obtain the optimized preliminary carbon balance prediction model parameters; Specifically, key features are extracted from the raw historical solid waste carbon density data and paired with historical carbon conversion data. Regression analysis is used to fit the raw historical solid waste carbon density data and the parameters of the preliminary carbon balance prediction model to minimize the prediction error and optimize the parameters of the preliminary carbon balance prediction model.

[0046] It should be noted that the key features in the original carbon density data of historical solid waste refer to the physicochemical properties that can reflect the carbon conversion performance of solid waste extracted from the carbon density data, including carbon content, ash content, pyrolysis reaction rate, reaction kinetic parameters, furnace temperature data, and furnace atmosphere field data.

[0047] S3.4: By comparing the differences between the optimized preliminary carbon balance prediction model parameters and the original data of actual solid waste carbon spectral density, the preliminary carbon balance prediction model is corrected to obtain the carbon balance prediction model.

[0048] Specifically, the differences between the optimized preliminary carbon balance prediction model parameters and the original data of actual solid waste carbon density are compared. The deviation between the prediction results of the preliminary carbon balance prediction model and the original data of actual solid waste carbon density is obtained through subtraction. The parameters of the preliminary carbon balance prediction model are adjusted according to the deviation to make the preliminary carbon balance prediction model more consistent with the carbon conversion in the actual reaction process. Through multiple correction and comparison processes, the parameters of the preliminary carbon balance prediction model are gradually optimized to obtain a more accurate carbon balance prediction model.

[0049] S3.5: Input the carbon spectrum label of the mixed feed stream and the furnace temperature signal into the carbon balance prediction model to estimate the carbon conversion trend in the furnace; Specifically, carbon spectrum analysis is performed on the mixed feed stream. The mixed feed stream is fed into a carbon spectrum analyzer to measure infrared absorption peaks, carbon-hydrogen ratio, and impurity distribution, generating carbon spectrum label data. The infrared absorption peaks, carbon-hydrogen ratio, and impurity distribution in the carbon spectrum label data are denoised and normalized. Principal component analysis (PCA) is used to extract features from the infrared absorption peaks, carbon-hydrogen ratio, and impurity distribution in the carbon spectrum label data, and carbon spectrum label data tags are added to different feature types to obtain the carbon spectrum label of the mixed feed stream. The temperature change in the furnace is monitored in real time to obtain the furnace temperature signal. After inputting the carbon spectrum label of the mixed feed stream and the furnace temperature signal into the carbon balance prediction model, the relationship between carbon content and furnace temperature is analyzed using preset association rules based on the carbon content information provided in the carbon spectrum label and the furnace temperature signal to infer the trend of carbon conversion in the furnace.

[0050] It should be noted that the association rules are set based on historical carbon conversion data, temperature changes, and thermodynamic experimental results. They are established by analyzing the regularity of carbon conversion under different conditions and are used to describe the relationship between carbon conversion trends and furnace temperature, so as to ensure that the carbon balance prediction model can accurately reflect the carbon conversion situation in the actual production process.

[0051] S3.6: Based on the carbon conversion trend in the furnace, calculate the carbon conversion state in the furnace and predict the evolution trend of carbon conversion to obtain carbon conversion state data; Specifically, based on the carbon conversion trend in the furnace, regression and statistical analysis are performed on the furnace temperature, carbon content, and carbon conversion rate. Combined with thermodynamic models and carbon balance prediction models, the carbon conversion state in the furnace is calculated. Historical carbon conversion data and preset carbon conversion rules are used to predict the evolution trend of carbon conversion. The carbon conversion state is continuously updated by real-time monitoring of changes in furnace temperature and carbon content, resulting in more accurate carbon conversion state data.

[0052] The expression for calculating the carbon conversion state in the furnace is: ; in, This indicates the carbon conversion status within the furnace, expressed as a percentage. This represents the reaction rate constant in a thermodynamic model; The proportionality constant corresponding to the carbon conversion rate is obtained by fitting the relationship between reaction rate and temperature using experimental data from thermogravimetric analysis. This indicates the furnace temperature, expressed in Kelvin (K). This indicates the carbon content in the furnace, expressed as a percentage. Represents the natural logarithm; The activation energy is determined by measuring the relationship between the carbon conversion reaction rate and temperature through a temperature-reaction rate experiment. It is obtained by fitting the carbon conversion rate data at different temperatures using the Arrhenius equation, and the unit is J / mol. The gas constant is a constant obtained by measuring the behavior of a gas at different temperatures and pressures through thermodynamic studies and verifying the idealized behavior of the gas. Its unit is J / mol·K. This indicates the carbon conversion rate, expressed in mol / s.

[0053] It should be noted that the carbon conversion law is based on historical carbon conversion data and furnace temperature change trends. It is established by analyzing the regularity of carbon conversion under different conditions and is used to describe the relationship between the carbon conversion process and the furnace temperature to ensure the accuracy and stability of carbon conversion prediction.

[0054] S3.7: The carbon conversion state data is smoothed and corrected using the Kalman filter algorithm to obtain energy consumption data.

[0055] Specifically, the recursive nature of Kalman filtering is used to dynamically estimate carbon conversion state data, removing noise and uncertainty to obtain an estimated value of carbon conversion state. The Kalman filtering algorithm is then used to weight the carbon conversion state data at each time step, and the predictions of the carbon conversion prediction model are automatically adjusted and the estimated value of carbon conversion state is updated based on the error between historical and current carbon conversion state data, resulting in smoother and more reliable carbon conversion state data. Combined with the obtained carbon conversion state data, energy consumption trends related to carbon conversion state are inferred, ensuring the accuracy and stability of energy consumption prediction. Real-time correction of the carbon conversion state data using the Kalman filtering algorithm further ensures accurate prediction of energy consumption data.

[0056] S4. Based on energy consumption data and carbon conversion status data, the synchronous mapping of the furnace temperature field and furnace atmosphere field is virtually predicted and corrected in real time through a digital twin model to generate optimized control parameters for the digital twin model. S4.1: Input energy consumption data and carbon conversion state data into the digital twin model to perform synchronous mapping of the furnace temperature field and furnace atmosphere field, and obtain furnace temperature data and furnace atmosphere field data. Specifically, the energy consumption data and carbon conversion status data are denoised and normalized to ensure that different types of energy consumption data and carbon conversion status data are within the same dimension range. The processed energy consumption data and carbon conversion status data are then input into the digital twin model. The digital twin model analyzes and integrates the processed energy consumption data and carbon conversion status data, and combines the thermodynamic principles inside the furnace to complete the synchronous mapping of furnace temperature and furnace atmosphere, calculate the changing trends of furnace temperature and furnace atmosphere composition, and obtain furnace temperature data and furnace atmosphere field data.

[0057] It should be noted that the thermodynamic principles inside the furnace refer to the energy conversion laws and material balance principles set based on the combustion, gas transport and reaction processes inside the furnace, and are used to help explain the relationship between changes in furnace temperature, changes in atmosphere composition and carbon conversion. The pre-training process of the digital twin model involves analyzing and integrating actual measured data on furnace energy consumption, carbon conversion state, furnace temperature, and changes in furnace atmosphere composition. Combined with the thermodynamic principles of the furnace, the parameters of the digital twin model are initially set and optimized. The cleaned energy consumption and carbon conversion state data are normalized to ensure that different types of energy consumption and carbon conversion state data are within the same dimensional range. Furthermore, by comparing the actual furnace temperature and atmosphere composition data with the prediction results of the digital twin model, a correction algorithm is used to adjust the parameters of the thermodynamic model, ensuring that the digital twin model accurately reflects changes in furnace temperature and atmosphere composition.

[0058] S4.2: Perform virtual prediction on the furnace temperature data and furnace atmosphere field data to estimate the trend of furnace temperature change and furnace atmosphere composition change. Specifically, by combining furnace temperature data and furnace atmosphere field data, the digital twin model uses preset thermodynamic rules and historical carbon conversion data to virtually predict changes in furnace temperature data and furnace atmosphere data. Based on the current state of the furnace and the current temperature and atmosphere data, the digital twin model simulates the relationship between the thermodynamic reactions and carbon conversion reactions in the furnace and changes in the furnace atmosphere, and calculates the trend of furnace temperature change and the trend of atmosphere composition change.

[0059] It should be noted that the thermodynamic rules are based on historical carbon conversion data and furnace thermodynamic principles. They analyze the carbon conversion patterns under different temperature, atmosphere, and reaction rate conditions, and set boundary conditions to simulate changes in furnace temperature and atmosphere composition.

[0060] S4.3: Based on the deviation between the furnace temperature change trend and the furnace temperature data, and the deviation between the furnace atmosphere composition change trend and the furnace atmosphere field data, the digital twin model is adjusted using a correction algorithm to generate optimized control parameters for the digital twin model.

[0061] Specifically, by comparing the actual furnace temperature data and furnace atmosphere field data with the predicted values ​​of the digital twin model, the deviation value is obtained. Based on the deviation value, the thermodynamic parameters of the digital twin model are adjusted using a correction algorithm, and the control parameters of the digital twin model are updated using the adjusted thermodynamic parameters to ensure that the digital twin model can more accurately reflect the actual changes in furnace temperature and atmosphere, and obtain more accurate optimized control parameters for the digital twin model.

[0062] S5. Optimize the control parameters of the digital twin model to adjust the gas recovery and residue separation process, and obtain full-process operation data by combining gas reuse and calcium solid waste reflux. S5.1: Adjust the gas recovery flow rate by optimizing the control parameters of the digital twin model to obtain the adjusted recovery gas flow rate data; Specifically, by optimizing the control parameters of the digital twin model, key control factors affecting the gas recovery flow rate are analyzed and identified. Based on the optimized control parameters of the digital twin model, combined with historical calcium carbide production operation data and the current furnace status, the gas recovery flow rate control strategy and flow rate settings are adjusted to ensure that the gas recovery flow rate reaches the preset optimal value. Based on the feedback results of the adjusted gas recovery flow rate and the prediction of the digital twin model, the gas recovery flow rate data is optimized to obtain the adjusted recovered gas flow rate data.

[0063] It should be noted that the optimal value is set based on the impact of gas recovery flow rate on production efficiency and carbon conversion under different operating conditions, combined with the relationship between gas recovery flow rate and product quality in historical operating data. An exemplary value of 5000 cubic meters per hour is determined based on historical production data, the analysis of the impact of gas recovery flow rate on carbon conversion rate and energy efficiency, and the results of laboratory simulation tests. Key control factors affecting gas recovery flow rate include furnace temperature, atmosphere composition, gas pressure, reaction rate, and furnace exhaust flow rate. Furnace temperature directly affects the reaction rate, while atmosphere composition and gas pressure affect gas flow and recovery efficiency. Changes in reaction rate can lead to fluctuations in gas production.

[0064] S5.2: Based on the adjusted recovered gas flow rate data, adjust the separation efficiency and speed of the residue separation process, and obtain the adjusted residue separation data; Specifically, the least squares method is used to perform linear regression on the adjusted recovered gas flow rate data to obtain the linear relationship between the gas recovery flow rate and the residue separation efficiency, and to determine the optimal separation efficiency and speed of the residue separation process under the current gas recovery flow rate. By adjusting the operating conditions of the residue separation process and controlling the working parameters of the residue separation equipment, the optimal separation efficiency and speed are ensured. During the adjustment process, the effect of residue separation is recorded in real time, and the operating parameters of the residue separation process are adjusted to obtain the adjusted residue separation data, so that the residue separation process meets the preset performance standards.

[0065] It should be noted that the performance standards are derived from a detailed analysis of the parameters, conditions, and historical operating results of each key link in the production process, and are used to measure the efficiency and quality of each operating process.

[0066] S5.3: Perform regression analysis and optimization on the adjusted residue separation data and the adjusted recovered gas flow rate data to generate optimized gas reuse flow rate data; Specifically, regression analysis was performed on the adjusted residue separation data and the adjusted recovered gas flow rate data to determine the mathematical relationship between them. The changing trends of the residue separation data and the recovered gas flow rate data were analyzed to identify the dynamic change patterns of the residue separation data and the recovered gas flow rate data. The control parameters of the gas recovery flow rate and the residue separation efficiency were adjusted to obtain the optimized gas reuse flow rate data.

[0067] S5.4: Based on the gas reuse flow rate data, adjust the return flow rate and return location of the calcium solid waste to obtain the operation data of calcium solid waste return; Specifically, based on the optimized gas reuse flow rate data, the return flow rate and return location of calcareous solid waste are adjusted. The return conditions of calcareous solid waste are precisely adjusted through a digital twin model to ensure that the return flow rate and return location match the actual production needs. During the adjustment process, the return flow rate setting of calcareous solid waste is dynamically optimized based on the optimized gas reuse flow rate data. The setting is reasonably made within the preset range of operating parameters for calcareous solid waste return, and the operating data of calcareous solid waste return is obtained.

[0068] It should be noted that the operating parameter range for calcium solid waste recirculation is a reasonable range set based on historical production experience, process requirements, and previous operating data. It is determined by analyzing the impact of different recirculation flow rates and recirculation locations on production efficiency and quality. This range is used to ensure that the recirculation flow rate and recirculation location are adjusted within the operating parameter range to meet actual production needs and optimize the production process.

[0069] S5.5: The adjusted recovered gas flow rate data, adjusted residue separation data, optimized gas reuse flow rate data, and calcium solid waste return operation data are summarized into full-process operation data.

[0070] Specifically, the adjusted recovered gas flow rate data, adjusted residue separation data, optimized gas reuse flow rate data, and calcium solid waste return operation data are organized, classified, and integrated according to a unified format and structure to ensure data consistency and comparability, and integrated into full-process operation data.

[0071] S6. Based on the full-process operation data, extract the changing trends of key variables and establish the working condition evolution law. Train the control strategy network through reinforcement learning algorithm to generate working condition optimization instructions that are adapted to the current production status.

[0072] S6.1: Perform time series analysis on the entire process operation data after cleaning and standardization to generate the changing trends of key variables; Specifically, outlier data and missing values ​​are removed from the entire process operation data to ensure that all data in the entire process operation data conform to a unified dimension and range, thereby guaranteeing data comparability and consistency; time series analysis is performed on each key variable of the entire process operation data to identify the correlation and fluctuation patterns between variables and generate the changing trends of key variables.

[0073] It should be noted that the key variables in the whole process operation data refer to the variables that have a significant impact on production efficiency, energy consumption, carbon conversion and process parameters in the calcium carbide production process, including furnace temperature, furnace atmosphere composition, gas recovery flow rate, residue separation efficiency, carbon content and energy consumption data.

[0074] S6.2: Conduct statistical analysis on the changing trends of key variables, identify the correlations between key variables, construct mutual influence relationships, determine the main operating conditions factors, and establish the laws governing the evolution of operating conditions.

[0075] Specifically, descriptive statistics are performed on the dataset to obtain the mean, standard deviation, maximum, and minimum values ​​of each key variable, thus revealing the distribution of the key variables. Correlation analysis is then conducted, using the Pearson correlation coefficient to quantify the linear and nonlinear relationships between the variables, determining the correlation between the key variables and obtaining the correlation matrix between them. Based on the distribution of the key variables and the correlation matrix between them, the main factors affecting the operating conditions are identified. Finally, multiple regression analysis is used to analyze the interrelationships between the key variables and establish the laws governing the evolution of the operating conditions.

[0076] S6.3: Train the control policy network using reinforcement learning algorithm, and optimize the control policy network through interactive learning to obtain the optimized control policy network; Specifically, the state space and action space of the control strategy network are updated using the Q-learning algorithm to obtain the updated control strategy network. During the production process, various data acquired in real time are interactively learned from the execution results of the updated control strategy network. The parameters of the control strategy network are adjusted based on the feedback data of the historical control strategy network and the real-time operation results to continuously optimize the decision-making process of the control strategy network. The reward mechanism is used to gradually optimize the control strategy network and generate an optimized control strategy network.

[0077] It should be noted that the various data acquired through real-time monitoring include furnace temperature, atmosphere composition, gas flow rate, and residue separation efficiency.

[0078] S6.4: Based on the optimized control strategy network, adjust the evolution of operating conditions in real time and generate operating condition optimization instructions that are adapted to the current production status.

[0079] Specifically, based on the optimized control strategy network, the real-time monitoring data of the current production status is evaluated. Combining historical production data and real-time feedback information, the deviation between the actual operating conditions and the preset operating condition target is determined. Based on the deviation between the actual operating conditions and the preset operating condition target, the trend of operating condition changes is analyzed, and operating condition optimization instructions adapted to the current production status are generated. The operating condition optimization instructions are transmitted to the control strategy network to adjust the control parameters, so that the execution of each control strategy can more accurately reflect the changes in the actual operating conditions.

[0080] It should be noted that the operating condition target is based on historical production data, experimental data, and process requirements. It is set by analyzing the performance of different production stages, combined with production efficiency, product quality standards, and energy consumption requirements, to ensure that the production process can operate at the best efficiency while achieving the optimal balance between quality and energy consumption.

[0081] In summary, this invention extracts the changing trends of key variables from the entire process operation data, trains a control strategy network using reinforcement learning algorithms, and generates operating condition optimization instructions adapted to the current production state. This allows for dynamic adjustment of various operational parameters during the production process based on real-time monitoring data, ensuring the production process remains in optimal condition. Furthermore, the reinforcement learning algorithm adaptively responds to changes in the production environment, improving carbon conversion efficiency and reducing energy consumption, thus achieving automated production optimization. This method not only improves resource utilization efficiency but also reduces the need for human intervention, significantly enhancing the intelligence level and production efficiency of the production process.

[0082] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An intelligent control method for the co-utilization of solid waste in the calcium carbide production process, characterized in that: include, Raw carbon density data of solid waste entering the batching silo from the calcium carbide production process were collected, and cluster calculations were performed on the carbon content distribution and activation energy of different types of waste plastics to obtain a joint feature vector. A solid waste carbon feature matrix was established based on the joint feature vector. The raw carbon density data of solid waste includes infrared absorption peaks, carbon-hydrogen ratio and impurity distribution information. The ratio of waste plastics to limestone is controlled in real time based on the carbonaceous characteristics matrix of solid waste to obtain the mixed raw material flow; A carbon balance prediction model is constructed. The carbon spectrum labels in the mixed feed stream and the furnace temperature signal are input into the carbon balance prediction model to calculate the carbon conversion trend in the furnace and obtain energy consumption data and carbon conversion status data. Based on energy consumption data and carbon conversion status data, the synchronous mapping of the furnace temperature field and furnace atmosphere field is virtually predicted and corrected in real time through a digital twin model, generating optimized control parameters for the digital twin model. The gas recovery and residue separation processes are adjusted by optimizing control parameters through a digital twin model, and the entire process operation data is obtained by combining gas reuse and calcium solid waste reflux. Based on the full-process operation data, the changing trends of key variables are extracted and the evolution law of working conditions is established. The control strategy network is trained through reinforcement learning algorithm to generate working condition optimization instructions that are adapted to the current production state.

2. The intelligent control method for the co-utilization of solid waste in the calcium carbide production process as described in claim 1, characterized in that: The specific steps to obtain the joint feature vector are as follows: Noise removal, time alignment, and scale normalization were performed on the raw carbon density data of solid waste to obtain a unified cleaned dataset. Extract the carbon content, ash content, and moisture content from the unified clean dataset, and calculate the effective carbon mass fraction; Thermogravimetric analysis was performed on the unified clean dataset to obtain the thermogravimetric derivative spectrum, and the characteristic peaks in the thermogravimetric derivative spectrum were extracted to obtain the carbon content distribution. The initial activation energy is calculated based on the characteristic temperature peak related to carbon content, and the initial activation energy is corrected by regression using the isoconversion method to obtain the activation energy. The effective carbon mass fraction, carbon content distribution, and activation energy are concatenated into a joint feature vector.

3. The intelligent control method for the co-utilization of solid waste in the calcium carbide production process as described in claim 1, characterized in that: The specific steps for establishing the solid waste carbonaceous characteristic matrix are as follows: Standardization and dimensional unification are performed on the joint feature vectors to obtain a unified feature set; The unified feature set is classified according to the preset carbon conversion feature weighting rules, and the features in the unified feature set are weighted and synthesized to form a comprehensive feature value, which is then summarized into a solid waste carbon feature matrix.

4. The intelligent control method for the co-utilization of solid waste in the calcium carbide production process as described in claim 1, characterized in that: The specific steps for obtaining the mixed raw material stream are as follows: Carbonaceous characteristics data of different waste plastics and limestone were extracted from the carbonaceous characteristics matrix of solid waste; Based on the carbonaceous characteristics of different waste plastics and limestone, the feeding ratio of waste plastics and limestone is calculated to obtain the preliminary mixed raw material flow of waste plastics and limestone. Based on the initial mixed raw material stream of waste plastics and limestone, the carbonaceous characteristics of the mixed raw material stream are monitored in real time, and the feeding ratio of waste plastics and limestone is adjusted to the optimal feeding ratio to obtain the mixed raw material stream.

5. The intelligent control method for the co-utilization of solid waste in the calcium carbide production process as described in claim 1, characterized in that: The specific steps for constructing the carbon balance prediction model are as follows: The raw historical solid waste carbon density data were subjected to noise removal, normalization, and data cleaning to obtain the processed solid waste carbon dataset. Based on the processed solid waste carbon dataset, a preliminary carbon balance prediction model was established by selecting a thermodynamic model. Regression analysis was performed on the raw historical solid waste carbon spectral density data to obtain the optimized preliminary carbon balance prediction model parameters. By comparing the differences between the optimized preliminary carbon balance prediction model parameters and the original data of actual solid waste carbon spectral density, the preliminary carbon balance prediction model is corrected to obtain the final carbon balance prediction model.

6. The intelligent control method for co-utilization of solid waste in the calcium carbide production process as described in claim 1, characterized in that: The specific steps for obtaining energy consumption data and carbon conversion status data are as follows: The carbon spectrum label of the mixed feed stream and the furnace temperature signal are input into the carbon balance prediction model to estimate the carbon conversion trend in the furnace. Based on the carbon conversion trend in the furnace, the carbon conversion state in the furnace is calculated and the evolution trend of carbon conversion is predicted to obtain carbon conversion state data. Energy consumption data is obtained by smoothing and correcting the carbon conversion state data using the Kalman filter algorithm.

7. The intelligent control method for the co-utilization of solid waste in the calcium carbide production process as described in claim 1, characterized in that: The specific steps for optimizing the control parameters for generating the digital twin model are as follows: Energy consumption data and carbon conversion status data are input into a digital twin model to synchronously map the furnace temperature field and furnace atmosphere field, thereby obtaining furnace temperature data and furnace atmosphere field data. Virtual predictions were performed on furnace temperature data and furnace atmosphere field data to estimate the trends of furnace temperature change and furnace atmosphere composition change. Based on the deviation between the furnace temperature change trend and the furnace temperature data, and the deviation between the furnace atmosphere composition change trend and the furnace atmosphere field data, a correction algorithm is used to adjust the digital twin model and generate optimized control parameters for the digital twin model.

8. The intelligent control method for co-utilization of solid waste in the calcium carbide production process as described in claim 1, characterized in that: The process of optimizing control parameters through a digital twin model to adjust the gas recovery and residue separation processes, and obtaining full-process operation data by combining gas reuse and calcareous solid waste recirculation, specifically involves the following steps: The gas recovery flow rate is adjusted by optimizing the control parameters of the digital twin model, and the adjusted recovery gas flow rate data is obtained. Based on the adjusted recovered gas flow rate data, the separation efficiency and speed of the residue separation process are adjusted to obtain the adjusted residue separation data; Regression analysis and optimization were performed on the adjusted residue separation data and the adjusted recovered gas flow rate data to generate optimized gas reuse flow rate data; Based on the gas reuse flow rate data, the return flow rate and return location of the calcium solid waste are adjusted to obtain the operation data of calcium solid waste return. The adjusted recovered gas flow rate data, adjusted residue separation data, optimized gas reuse flow rate data, and calcium solid waste recirculation operation data are summarized into full-process operation data.

9. The intelligent control method for the co-utilization of solid waste in the calcium carbide production process as described in claim 1, characterized in that: The specific steps for extracting the changing trends of key variables and establishing the evolution law of operating conditions based on the full-process operation data are as follows: After cleaning and standardizing the data from the entire process, time series analysis is performed to generate the changing trends of key variables. Statistical analysis is conducted on the changing trends of key variables to identify the correlations between key variables, construct mutual influence relationships, determine the main operating conditions, and establish the laws governing the evolution of operating conditions.

10. The intelligent control method for the co-utilization of solid waste in the calcium carbide production process as described in claim 9, characterized in that: The specific steps for generating the operating condition optimization instructions adapted to the current production status are as follows: The control policy network is trained using reinforcement learning algorithms and then optimized through interactive learning to obtain the optimized control policy network. Based on the optimized control strategy network, the evolution of operating conditions is adjusted in real time to generate operating condition optimization instructions that are adapted to the current production status.

Images