Steel production process energy consumption influence factor analysis and energy consumption optimization method

Abstract

The invention provides a steel production process energy consumption influence factor analysis and energy consumption optimization method, and belongs to the technical field of steel production energy conservation. The method comprises the following steps: constructing a steel production process energy consumption model; quantifying the contribution degree of input variables of the steel production process energy consumption model; quantifying the coupling strength of any two input variables; screening the input variables based on a preset threshold value, and classifying the input variables in combination with a community discovery algorithm to obtain a high-contribution-degree strong coupling variable group and a high-contribution-degree weak coupling variable group; taking the high-contribution-degree strong coupling variable group and the high-contribution-degree weak coupling variable group as input, and taking the minimum process energy consumption as a target to construct a process energy consumption optimization model; and solving the process energy consumption optimization model to determine optimization process combination parameters. According to the method, process energy consumption weak points and energy consumption influence factors are mined based on the constructed model, an energy management and control and optimization scheme is provided for field production, and energy conservation and consumption reduction of the steel production process are achieved.

Description

Technical Field

[0001] This invention relates to the field of energy-saving technology in steel production, and in particular to a method for analyzing the factors affecting energy consumption in steel production processes and optimizing energy consumption. Background Technology

[0002] The steel industry is a typical high-energy-consuming industry. Its production process generally involves multiple steps, from raw material processing, coking, sintering, ironmaking, steelmaking, rolling to finished products. This process is not only complex in its production steps, but also involves energy consumption from various energy sources. The energy efficiency level of each step has a significant impact on the overall production energy efficiency.

[0003] Existing technologies for energy consumption research in the steel industry have involved extensive studies on process mechanism models, data models, and energy conservation and consumption reduction measures. However, traditional mechanism models, when describing complex physicochemical reactions in processes, struggle to fully consider the interactions and coupling effects between various physicochemical reactions in actual production, resulting in poor model prediction accuracy. Data models, due to their data training methods, lack sufficient interpretability and can only perform trend analysis on various data, making it difficult to provide real-time and accurate suggestions for on-site production. Therefore, they often fail to achieve the expected results in actual energy-saving renovations.

[0004] Therefore, a method for analyzing the factors affecting energy consumption in steel production processes and optimizing energy consumption is needed. Summary of the Invention

[0005] In view of this, the present invention provides a method for analyzing the influencing factors and optimizing energy consumption in steel production processes. It constructs a process energy consumption model by integrating metallurgical mechanism analysis and data-driven technology through a multi-dimensional modeling approach. Based on the established energy consumption model, an energy consumption optimization model is built to identify energy consumption weaknesses and influencing factors in the processes. A comprehensive intelligent analysis method combining multiple analyses is used to conduct in-depth analysis of key factors, accurately identify energy efficiency bottlenecks, and finally formulate targeted energy-saving strategies based on the analysis results. To this end, the present invention provides the following technical solutions: An analysis of energy consumption influencing factors and a method for energy consumption optimization in steel production processes, comprising: Construct an energy consumption model for steel production processes; Quantify the contribution of input variables in the energy consumption model of steel production processes; quantify the coupling strength between any two input variables; The input variables are filtered based on a preset threshold, and the input variables are classified by combining a community detection algorithm to obtain a high-contribution strongly coupled variable group and a high-contribution weakly coupled variable group. Using high-contribution strongly coupled variable groups and high-contribution weakly coupled variable groups as inputs, a process energy consumption optimization model is constructed with the goal of minimizing process energy consumption. The energy consumption optimization model for the aforementioned process is solved to determine the optimal process combination parameters.

[0006] Furthermore, the energy consumption model for the steel production process includes:

[0007] in, Energy consumption in steel production processes This refers to the total amount of standard coal equivalent of all energy consumed in the steel production process. The amount of energy recovered during the steel production process; Output of products in steel production processes; The amount of energy recovered in the steel production process is calculated using a mechanism model, a prediction model, or a combination of a mechanism model and a prediction model. The total amount of standard coal equivalent of various energy sources consumed in the steel production process is calculated using a mechanistic model, a prediction model, or both.

[0008] Furthermore, the quantification of the coupling strength between any two input variables includes: Calculate the main effects of the parameters; The coupling strength is the sum of the main effects of the two parameters on energy consumption and the ratio of the maximum impact of the interaction between the two parameters on energy consumption.

[0009] Furthermore, the main effects of the calculated parameters include:

[0010] Let be the average energy consumption of the process at the i-th level for the k-th parameter; The total average energy consumption of all processes under all orthogonal experimental conditions; This represents the fluctuation range of the k-th parameter relative to the baseline level at the ith level. This represents the main effect of the k-th parameter.

[0011] Furthermore, the coupling strength:

[0012] in, and Let be any two parameters to be analyzed. The maximum impact of the interaction between any two parameters on energy consumption is the difference between the maximum and minimum energy consumption under all combinations of the two parameters. for The absolute value of the main effect; for The absolute value of the main effect; for The maximum horizontal fluctuation range; for The maximum horizontal fluctuation range; for and The coupling strength.

[0013] Furthermore, the contribution of the input variables in the quantitative steel production process energy consumption model includes: A single-factor model was constructed by setting variable constraints. After normalization, the individual contribution of each parameter to energy consumption was analyzed. Dimensionless processing was used to eliminate unit differences, and the energy consumption change amplitude when the parameter changes by 1% was calculated to obtain the sensitivity coefficient.

[0014] Furthermore, the step of filtering the input variables based on a preset threshold and classifying the input variables using a community detection algorithm includes: The input variables are filtered based on a preset threshold to obtain a set of high-contribution variables; A preset coupling threshold is set. If the coupling strength between any two variables in the high contribution set is greater than the coupling threshold, they are classified as high contribution strongly coupled variable groups; otherwise, they are classified as high contribution weakly coupled variable groups.

[0015] Furthermore, the step of solving the process energy consumption optimization model to determine the optimized process combination parameters includes: In the first stage, the parameters of the high-contribution, strongly coupled variable group are fixed. Then, the parameters of the high-contribution, weakly coupled variable group are optimized independently and in parallel using the particle swarm optimization algorithm until the process energy consumption meets the error requirements or the program reaches the maximum number of iterations and stops. The optimal solution of the high-contribution, weakly coupled variable group is then output. In the second stage, the optimal solution of the high-contribution, weakly coupled variable group is fixed. The high-contribution, strongly coupled variable group is then optimized collaboratively through the particle swarm optimization algorithm until the process energy consumption meets the error requirements or the program reaches the maximum number of iterations and stops. The optimal solution of the high-contribution, strongly coupled variable group is then output. The optimal solutions of the high-contribution weakly coupled variable group and the high-contribution strongly coupled variable group are used as optimization process combination parameters.

[0016] Advantages and positive effects of the present invention: 1) A high-precision process energy consumption model is constructed by using mechanism and data synergy. By quantifying the contribution of input variables and analyzing the interaction coupling strength between variables, the model can accurately diagnose the impact of key influencing parameters on energy consumption characteristics. 2) Based on community discovery and contribution thresholds, parameters are classified into high-contribution, strongly coupled and high-contribution, weakly coupled groups. A process energy consumption optimization model with an intelligent optimization strategy is constructed. After solving the model, the globally optimal process parameters and parameter combinations are output, providing accurate and precise diagnostic solutions for the field. 3) Through an adaptive feedback optimization process, optimal parameters are applied to production while continuously collecting new data. The analysis results are dynamically updated to initiate a new round of optimization, enabling the system to adapt to production changes and continuously optimize energy consumption. This also lays the foundation for dynamic model adjustments. This technology overcomes the limitations of traditional experience-based management, improving the scientific rigor and timeliness of energy consumption control through real-time model feedback optimization. It provides steel companies with energy-saving pathways for various processes, helping them achieve low-carbon production goals and promoting the industry's green transformation. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are 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] Figure 1 This is a flowchart illustrating the analysis of factors influencing energy consumption in steel production processes and the method for optimizing energy consumption in the embodiments. Figure 2 This is a schematic diagram of the multivariate iterative solution process; Figure 3 This is a flowchart illustrating the particle swarm optimization algorithm. Figure 4 To coordinate and optimize the algorithm's loop graph; Figure 5 This is a schematic diagram of the mechanism and data-driven energy consumption optimization model and intelligent analysis method of influencing factors constructed by the present invention, taking sintering as an example. Detailed Implementation

[0019] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0020] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0021] This invention provides a method for analyzing energy consumption influencing factors and optimizing energy consumption in steel production processes. Based on the construction of an energy consumption model driven by mechanism and data, an energy consumption optimization model is constructed. The energy consumption optimization model is used to perform uncorrected and corrected optimization of the process, identify the weak points in energy consumption and their influencing factors, and then comprehensively apply intelligent analysis methods combining multiple analyses to conduct in-depth analysis of key factors, accurately identify energy efficiency bottlenecks, and finally formulate targeted energy-saving strategies based on the analysis results.

[0022] like Figure 1 As shown, a method for analyzing energy consumption influencing factors and optimizing energy consumption in steel production processes includes the following steps: S1. Construct an energy consumption model for steel production processes: The steel production process includes: coking, sintering, pelletizing, blast furnace, converter and rolling.

[0023] 1. Based on the inherent laws of material transformation and energy transfer in the metallurgical process, deeply integrate the ability of neural networks to deeply mine and analyze massive production data, realize the complementary advantages of metallurgical mechanism and data-driven approach, and construct energy consumption models for each process through mechanism and data-driven collaborative approach.

[0024] Based on the overall production process in the steel industry, the characteristics and energy consumption distribution of each process are broken down into several parts, and several sub-models are constructed for each part. For calculable process energy consumption, the input consumption is calculated step by step using mechanistic methods. Meanwhile, for energy media that are significantly affected by empirical values ​​and field conditions, key parameters need to be obtained through data analysis methods, and predictions are made by constructing data models. For other energy media, simplification can be achieved by using field values ​​or empirical models. The model structure is defined according to the energy consumption situation.

[0025] 2. In order to describe some subtle changes more accurately and to better adapt to the needs of on-site production, while ensuring the accuracy of calculation, appropriate assumptions and simplifications are made to the energy consumption model of each steelmaking process. At the same time, the production boundaries of each process are set, and the material flow and energy flow are clarified.

[0026] 3. For process energy consumption analyzed using mechanistic methods, the input consumption is calculated step by step using mechanistic methods. When the energy medium requires high solution speed and accuracy and multiple iterations, the multivariate iterative method is selected as the model solution algorithm.

[0027] 4. For energy media with complex energy consumption data and missing records, a data-driven approach is used to construct an energy consumption neural network model based on the governance of on-site data. Parameters with high influence coefficients are selected as input variables for the process energy consumption model through grey relational analysis. Based on the selected influencing factors, relevant data is extracted from the on-site database using data preprocessing methods. The dataset is divided into a training set and a test set, and a neural network model is constructed based on the classification data.

[0028] 5. By synergistically coupling the mechanistic model and the data model, a process energy consumption model with mechanistic interpretability and data adaptability is constructed, and the total process energy consumption is finally calculated.

[0029] S2. Based on the established energy consumption model, conduct multi-dimensional in-depth analysis of the above input variables. That is, perform a quantitative analysis of the contribution of the input variables, introduce orthogonal experimental design to quantify the interaction coupling strength between any two variables, and classify the parameters by combining community discovery and contribution threshold, intuitively classifying the parameters into high contribution strong coupling and high contribution weak coupling groups.

[0030] 2. Identify key influencing factors based on the energy consumption model of steel production processes.

[0031] 3. Perform contribution quantification analysis on the input variables.

[0032] First, set reasonable constraints on the influencing factors, keep other parameters unchanged, change the value of one parameter within the constraints, and substitute it into the energy consumption model to obtain a data table within the constraints.

[0033] By normalizing the data, the impact of each factor on energy consumption and energy medium can be visually displayed. This allows for a single-factor analysis of energy consumption, providing a clear view of the influence of each factor on energy consumption and energy medium.

[0034] 3. Utilize normalization analysis to evaluate the individual contribution of a single factor and the combined effect of multiple factors under multi-factor influence. Introduce dimensionless processing methods to eliminate unit differences and deeply analyze the specific impact of different parameters on process energy consumption. To analyze the changing trend of absolute energy consumption in detail, calculate the specific energy consumption values ​​corresponding to different changes in each parameter. Perform normalized sensitivity analysis on the calculation results, and under stable raw materials, operating conditions, and working conditions, rank the parameters according to their sensitivity to process energy consumption to determine the priority control parameters and control ranges in production.

[0035] 4. To further evaluate the coupling effect of various factors under multiple influences, an orthogonal experimental design method is introduced. An orthogonal experimental table is selected based on the influencing factors, and baseline levels of parameters are set. Through calculations of parameter combinations, the fluctuation of process energy consumption with each parameter combination is analyzed. Furthermore, the interaction intensity calculation method is used to quantify the coupling effect between factors. Through orthogonal experimental design, the main effects and coupling effects of each factor on energy consumption are comprehensively examined within a limited number of calculations, reflecting the strength of the combined effect of each factor on energy consumption.

[0036] 5. Utilize threshold setting and community discovery to intelligently classify parameters. First, set the contribution threshold, which is automatically set using the quantile method based on the distribution of the contribution list.

[0037] S3. Using key influencing factors as input changes, construct a process energy consumption optimization model with the goal of minimizing process energy consumption.

[0038] For variables with high contribution and weak coupling, the model employs an independent parallel optimization strategy. For groups of variables with high contribution and strong coupling, the model employs a collaborative optimization strategy. By solving this optimization model with built-in step-by-step collaborative logic, the model ultimately outputs the globally optimal combination of process parameters.

[0039] S4. Solve the process energy consumption optimization model to determine the optimal process combination parameters; apply the optimized process combination parameters to actual production and continuously collect new production data. Use the new data to dynamically update the contribution and coupling strength analysis results, and then initiate a new round of optimization. This step enables the entire system to adapt to changes in raw materials, equipment, and operating conditions, achieving continuous improvement in energy consumption performance and dynamic adjustment of the model.

[0040] Example Based on this method, we analyze the factors affecting energy consumption in the sintering process and optimize energy consumption, including: S1. Constructing an energy consumption model for the sintering process: 1. Sintering process mechanism model: By employing methods such as material balance and heat balance, we gain a deeper understanding of the coupling mechanism of thermal, mass, and chemical reactions during the sintering process. Material balance analysis ensures a reasonable ratio of raw material input to product output, providing a data foundation for energy consumption calculation; heat balance considers the relationship between energy input and output, assesses heat transfer and utilization efficiency, and thus optimizes the combustion and heating process. The calculation process of the solid fuel consumption model mainly includes four steps: (1) initial condition setting; (2) material balance calculation; (3) heat balance calculation; (4) ... Figure 2 As shown, the solid fuel consumption is calculated iteratively. First, the field data is normalized, and initial input values ​​for the model are set (such as solid fuel ratio, return ore ratio, and heat loss ratio). Then, based on the physicochemical reaction characteristics of the sintering process, the masses of iron-containing raw materials, dolomite, and limestone are calculated through Fe, MgO, and basicity balance calculations, while simultaneously calculating the iron grade, basicity, and MgO content in the sinter. Based on the solid material balance, the sintering waste gas volume and its composition are calculated, and further heat balance calculations are performed to obtain the heat loss rate under the current input conditions. The calculation results are compared with the initially set heat loss rate. If the error is within the allowable range, the solid fuel consumption is output; if it does not meet the requirements, the solid fuel input is corrected through a feedback adjustment algorithm set within the model, and the process continues iterating until the model accuracy requirements are met.

[0041] The main purpose of calculating the Fe balance is to determine the amount of iron-containing raw materials used in the sintering process. The Fe balance relationship in the sintering process is as follows:

[0042] in, The amount of iron-containing raw materials used, in kg; The iron content in iron-containing raw materials, % The amount of flux used. The iron content in the flux, % For the amount of solid fuel used, The iron content in solid fuels, % For the production of sintered ore, The iron content in sintered ore.

[0043] After determining the amount of iron-containing raw materials used, further calculation of the MgO balance can yield the amount of dolomite flux used during sintering. The MgO balance relationship in the sintering process is as follows:

[0044] in, The MgO content in iron-containing raw materials, % The MgO content in the flux, % The MgO content in solid fuel, % The MgO content in the sinter is %.

[0045] After determining the amounts of iron-containing raw materials and dolomite used, the amount of limestone flux used during sintering can be calculated by performing basicity balance calculations. The basicity balance relationship in the sintering process is as follows:

[0046] in, , , The CaO content in iron-containing raw materials, flux, and solid fuel, respectively, is % (%). , , These are iron-containing raw materials, fluxes, and solid fuels. content,%; The CaO content in sinter is %; In sintered ore content,%.

[0047] Gas balance mainly targets CO and other gases in exhaust gas. , , , , The equilibrium is calculated, and the equilibrium relationships are as follows: CO balance relationship:

[0048] in, The carbon content in iron-containing raw materials, % The carbon content in the flux, % The C content in solid fuels, % The percentage of unreacted carbon, % To produce sintering exhaust gas, , The CO content in the sintering exhaust gas is expressed as %.

[0049] Balance relationship:

[0050] in, For gas consumption, ; , , CO in coal gas, , content,%; The flux loss rate, % In sintering waste gas content,%.

[0051] Balance relationship:

[0052] in, For coal gas content,%; Water quantity for sintering, kg; In sintering waste gas content,%.

[0053] Balance relationship:

[0054] in, For coal gas content,%; For sintering air volume, ; For sintering air content,%; This refers to the amount of air leakage. ; For leaky air content,%; In sintering waste gas content,%.

[0055] Balance relationship:

[0056] in, The excess sintering exhaust coefficient is -; For sintering air content,%; For leaky air content,%; In sintering waste gas content,%; , , They are respectively in coal gas , , content,%.

[0057] Balance relationship:

[0058] After setting the initial value, among which, For iron-containing raw materials content,%; In the flux content,%; , , respectively in fuel Volatile matter, sulfur content, % The sintering desulfurization rate is %; In sintering waste gas content,%.

[0059] Heat balance: The heat received during the sintering process includes two parts: physical heat from the raw materials and chemical heat from the sintering process. Physical heat refers to the heat introduced by the raw materials, gas, air, etc., while chemical heat refers to the heat released during the combustion of fixed carbon and gas, as well as the heat of reaction of sulfides, oxides, etc. The specific heat balance relationship is as follows:

[0060] Among them, heat revenue items include , , , , , , , These are the physical heat of the mixture, the physical heat of the base material, the physical heat of the gas, the physical heat of the air, the heat of combustion of fixed carbon, the chemical heat of ignition gas, the heat of sulfide reaction, and the heat of slag formation, in kJ / t; the heat expenditure items include... , , , , , These are the heat of chemical reaction, heat of carbonate decomposition, heat of water evaporation, heat of sintering cake, heat of sintering waste gas, and heat loss, respectively, in kJ / t.

[0061] After the above steps, the accounting for other solid materials in the production of finished sintered ore has been completed. After determining the gas consumption and performing gas balance calculations, a heat balance analysis of the sintering process can be further conducted to determine the current heat loss rate. The calculation formula is as follows:

[0062] in, The current heat loss rate is %; , Heat income and heat expenditure are respectively expressed in kJ / t.

[0063] Based on the calculated heat loss rate, the solid fuel consumption can be adjusted and corrected. Typically, heat dissipation and radiative heat exchange account for 5% to 10% of the total heat input, which can be used to determine if the fuel ratio is reasonable: if the heat loss rate exceeds the upper limit, it indicates excessive fuel addition, and the fuel ratio should be appropriately reduced; if it is below the lower limit, it indicates insufficient fuel, and the fuel input should be increased. When the calculated heat loss rate meets the error requirement, the solid fuel consumption required for producing the finished sinter can be output.

[0064] For process energy consumption analyzed using mechanistic methods, the input consumption is calculated step by step using mechanistic methods. In this embodiment, the mass of solid fuel consumed, specifically the mass of coke powder and the mass of anthracite, is calculated using a multivariate iterative algorithm.

[0065] 2. A data-driven approach was used to construct a neural network model for energy consumption. Parameters with high influence coefficients were selected as input variables for the process energy consumption model through grey relational analysis.

[0066] In the sintering process, grey relational analysis algorithms are commonly used to evaluate the relationship between different production parameters (such as furnace temperature, ventilation volume, fuel consumption, and material composition) and energy consumption. This method calculates the grey relational degree between each factor and energy consumption by comparing actual observation data with theoretical models, thereby revealing which factors have a greater impact on energy consumption and which have a relatively smaller impact. This allows for the effective identification of key influencing factors even when data is insufficient or the system is complex, providing a theoretical basis for optimizing energy conservation and consumption reduction in the sintering process.

[0067] 1) Normalize input variables to eliminate differences between different units of measurement, allowing variables to be compared on the same scale. Range standardization:

[0068] in, It is the j-th value in the original data sequence; It is the minimum value of the i-th sequence; It is the maximum value of the i-th sequence; It is the normalized data.

[0069] 2) Construct the difference sequence: The difference sequence represents the degree of difference between each data point and the reference sequence. For each pair of data points... and (Reference sequence), its differential sequence The calculation formula is:

[0070] in, It is a normalized comparison sequence; It is the normalized reference sequence (usually the target sequence).

[0071] The maxima and minima of the differential sequences are used to calculate the numerator and denominator of the correlation coefficient, as shown in the following formula:

[0072]

[0073] Calculate the correlation coefficient The correlation coefficient reflects the strength of the relationship between each pair of data points, and the formula for calculation is:

[0074] In the formula, The resolution coefficient is, under normal circumstances, The value is 0.5.

[0075] Calculate the correlation Association strength is the overall strength of the relationship for each reference sequence, and is usually obtained by averaging all association coefficients, as shown in the following formula:

[0076] Based on the magnitude of the correlation coefficient, the dependence between factors can be divided into three intervals: high dependence ( ∈(0.5, 0.8) indicates a strong relationship between variables, which requires close attention; moderate dependency ( ∈(0.5, 0.8) indicates a certain correlation, but the influence is weak and can be used as a reference factor for appropriate optimization; low dependence ( If ∈(0, 0.5), it indicates that there is almost no significant relationship between the variables, and the impact on the target variable is small, which can be ignored in the optimization process.

[0077] Based on the characteristics of the data, a suitable neural network model is selected for energy consumption prediction. After the model is trained, it is validated using a test set, with the mean absolute percentage error (MAE) used as the primary metric.

[0078] By synergistically coupling the mechanistic model and the data model, a process energy consumption model with mechanistic interpretability and data adaptability is constructed, and the total process energy consumption is finally calculated.

[0079] .

[0080] in, Energy consumption of the sintering process during the statistical period, kgce / t; , , , , , , , The figures are respectively the consumption and waste heat recovery of coke, coal, coke oven gas, blast furnace gas, electricity, industrial water, and nitrogen, in kgce; The value is the yield of sintered ore, expressed in tons (t).

[0081] S2. Quantify the contribution of input variables in the energy consumption model of steel production processes; quantify the coupling strength between any two input variables: The input variables for the mechanistic model of the sintering process include: The input variables for the solid combustion model, gas model, power consumption model, and waste heat recovery model should be determined, and the operable operating parameters should be determined in combination with the actual site conditions.

[0082] Input variables include: sinter basicity, sinter MgO percentage, sinter FeO percentage, sinter temperature, return ore mix ratio, bottom material quality, quicklime quality, water ratio, air leakage rate, bellows negative pressure, ignition air-fuel ratio, insulation air-fuel ratio, flue gas temperature, sinter bed thickness, bulk density, and exhaust coefficient.

[0083] Contribution of input variables in the energy consumption model for steel production processes: 1) Set reasonable constraints on the influencing factors of the sintering process, as shown in Table 1: Table 1

[0084] 2) Keeping other influencing parameters constant, substitute the above variables into the energy consumption model to calculate the process energy consumption under different values, using the parameter index corresponding to the actual minimum energy consumption value as the benchmark. Calculate the sensitivity coefficient using linear fitting and the least squares method, i.e., the impact of a 1% change in the parameter on energy consumption. Details are as follows:

[0085] in, This represents the impact of a 1% change in parameters on energy consumption. For actual parameters, These are standard parameters.

[0086] Assuming a fitted curve:

[0087] Where b is the slope, i.e., the sensitivity coefficient; and a is the baseline energy consumption value.

[0088] Calculate b using the least squares method:

[0089] That is, calculate the change in process energy consumption when the dimensionless parameter changes by one unit, and denot it as .

[0090] Calculate the percentage of change, i.e., the contribution:

[0091] This represents the sensitivity coefficient of each parameter to process energy consumption.

[0092] Based on the influencing factors, an orthogonal experimental table is selected, and the baseline level of the parameters is set. Through the calculation of parameter combinations, the fluctuation of process energy consumption with each parameter combination is analyzed, and the coupling effect between various factors is quantified by using the interaction intensity calculation method.

[0093] 3) Calculate the main effect of a parameter. The main effect refers to the average impact of different levels of a single parameter on energy consumption. The calculation formula is: Main effect of a parameter = Level fluctuation amplitude / Average energy consumption of all parameters at a certain level. Total average energy consumption of all experiments.

[0094]

[0095] in, Let k be the average energy consumption of the process at the i-th level for the k-th parameter. The total average energy consumption of the process under all orthogonal experimental conditions is given. This represents the fluctuation range of the k-th parameter relative to the baseline level at the ith level. This represents the main effect of the k-th parameter, i.e., the average change in process energy consumption when the parameter changes by 1%.

[0096] Coupling strength:

[0097] in, and Let be any two parameters to be analyzed. The maximum impact of the interaction between any two parameters on energy consumption is the difference between the maximum and minimum energy consumption under all combinations of the two parameters. for The absolute value of the main effect; for The absolute value of the main effect; for The maximum horizontal fluctuation range; for The maximum horizontal fluctuation range; for and The coupling strength.

[0098] S4. Based on a preset threshold, the input variables are filtered, and the input variables are classified by combining the community detection algorithm to obtain high-contribution strongly coupled variable groups and high-contribution weakly coupled variable groups. Variables ranking in the top 30% of contribution are defined as high-contribution variables. A candidate set H of high-contribution variables is then selected.

[0099] In this embodiment, the high contribution candidate set includes: sinter basicity, water ratio, and air leakage rate.

[0100] A preset coupling threshold is set. If the coupling strength between any two variables in the high contribution set is greater than the coupling threshold, they are classified as high contribution strongly coupled variable groups; otherwise, they are classified as high contribution weakly coupled variable groups.

[0101] In this embodiment, a high-contribution strongly coupled variable group and a high-contribution weakly coupled variable group are obtained based on the community detection algorithm: The community detection method takes five parameters as examples: water ratio, quicklime quality, return ore ratio, air leakage rate, and sinter basicity. First, a threshold is set and a high-contribution candidate set H is selected. For example, a reasonable contribution threshold Tc=0.5 is set. Based on the sensitivity analysis and orthogonal experiments above, key factors significantly higher than other parameters in the graph are selected. Variables with a contribution ≥0.5 are selected: H=[A, B, C] (i.e., sinter basicity, water ratio, and air leakage rate). The parts corresponding to A, B, and C are extracted from the global matrix. The network is constructed and thresholds are set. The nodes are A, B, and C. An edge weight threshold of 0.3 is set. Only when the coupling strength between two variables is greater than 0.3 is it considered that they have a cooperative relationship that needs attention.

[0102] Result: In the submatrix, only BC (0.412) satisfies the condition. The coupling strengths of A and B, and A and C are both less than 0.3. Therefore, the constructed network contains only one edge: BC (weight 0.412).

[0103] Then the community detection algorithm is executed: {B:0,C:0} (B - water ratio and C - air leakage rate are assigned to the same community 0), {A:1} (A - sinter basicity is assigned to a separate community 1). Applying the automatic classification rules, community 0 [B,C] contains 2 nodes that are high contribution - strongly coupled. Community 1 [A] has only 1 node that is a high contribution - weakly coupled parameter. Final classification results: High contribution - strongly coupled group: [[B - water ratio, C - air leakage rate]], High contribution - weakly coupled parameter: [A - sinter basicity].

[0104] Finally, summarizing the core coupling relationships, for strongly coupled combinations such as the return ore ratio and air leakage rate, the increased return ore will change the permeability of the material layer, thus affecting the air leakage distribution; both need to be adjusted simultaneously. For weakly coupled combinations such as the water ratio and sinter basicity, the water ratio affects the material moisture content, and the basicity affects the mineral composition; the two have a low correlation in their mechanisms of action and can be optimized separately. Orthogonal experiments can avoid energy consumption rebound caused by adjusting a single parameter, ensuring the scientific validity and feasibility of the energy-saving scheme.

[0105] The sintering process was analyzed using normalization and orthogonal experimental design, revealing the impact of five sintering parameters—water ratio, quicklime quality, external ore mix ratio, air leakage rate, and sinter basicity—on energy consumption from both single-factor and multi-factor perspectives. Single-factor analysis showed that a 1% increase in water ratio increased sintering energy consumption by 1.44 kgce / t; a 1 kg increase in quicklime quality decreased energy consumption by 0.15 kgce / t; a 1% increase in external ore mix ratio still increased energy consumption by 0.31 kgce / t; a 1% increase in air leakage rate increased energy consumption by 0.18 kgce / t; and a 0.05 increase in basicity increased energy consumption by 0.44 kgce / t. The relationship between each influencing factor and process energy consumption was quantified.

[0106] Multifactor normalization analysis shows that, under stable raw materials, operating conditions, and working conditions, the sensitivity of each parameter to process energy consumption, from highest to lowest, is as follows: sinter basicity, water supply, air leakage rate, quicklime quality, and return ore proportion. Enterprises should pay close attention to these influencing factors.

[0107] Multi-factor orthogonal experimental analysis showed that the coupling strength between the external ratio of return ore and the air leakage rate was the strongest (0.493), the coupling strength between the water ratio and the quicklime mass was 0.407, the coupling strength between the water ratio and the external ratio of return ore was 0.425, the coupling strength between the water ratio and the air leakage rate was 0.412, the coupling strength between the quicklime mass and the external ratio of return ore was 0.488, the coupling strength between the quicklime mass and the air leakage rate was 0.475, and the coupling strength between the water ratio and the basicity of sinter was the weakest (0.163). The air leakage rate should also be considered when changing the external ratio of return ore.

[0108] 4. Based on contribution and coupling, a process energy consumption optimization model with an intelligent optimization strategy is constructed. This model's solution algorithm is deeply customized. For variables with high contribution and weak coupling, the model employs an independent parallel optimization strategy. For variable groups with high contribution and strong coupling, the model adopts a collaborative optimization strategy. Solving through this optimization model with built-in step-by-step collaborative logic, the globally optimal combination of process parameters is ultimately output.

[0109] Using the classification results of "contribution-coupling" as input to the optimization strategy, the set of variables with high contribution and weak coupling is clearly defined. and sets of variables with high contribution .

[0110] First, set a baseline value: Establish an initial baseline value for all optimization variables, which can be a historical average value. Then, configure a global optimization algorithm, Particle Swarm Optimization (PSO), as the underlying solver. The solution process is as follows: Figure 3 As shown.

[0111] The first stage uses a particle swarm optimization algorithm to iterate until the process energy consumption meets the error requirements or the program reaches the maximum number of iterations and stops. In the second stage, the parameters of the high-contribution, strongly coupled variable group are fixed, while the parameters of the high-contribution, weakly coupled variable group are optimized independently and in parallel until the process energy consumption meets the error requirements or the program reaches the maximum number of iterations and stops, and the optimal solution of the high-contribution, weakly coupled variable group is output. The third stage involves fixing the optimal solution of the high-contribution, strongly coupled variable group, and then co-optimizing the high-contribution, strongly coupled variable group until the process energy consumption meets the error requirements or the program reaches the maximum number of iterations and stops, outputting the optimal solution of the high-contribution, strongly coupled variable group.

[0112] 1. Fix the parameters of the high-contribution, strongly coupled variable group in the segment so that it does not participate in the optimization in this stage.

[0113] Construct sub-optimization problems:

[0114] Among them, only To optimize the variables, an optimization algorithm is run to search for the optimal solution for the leverage variables. Because these variables are weakly coupled, the algorithm can efficiently optimize within their respective dimensions. Once the algorithm reaches convergence (e.g., the number of iterations or accuracy requirements), the set of optimal solutions for the leverage variables found at that point is recorded.

[0115] High-contribution, strongly coupled variable groups work together for optimization.

[0116] The optimal solution is obtained by fixing the set of loosely coupled variables with high contributions. Releasing all loosely coupled variables with high contributions transforms the optimization problem into:

[0117] in, To optimize the variables, the optimization algorithm is run again. The algorithm's task is to find the optimal synergy within each core variable group. For example, for group... The algorithm will learn and To determine the interrelationships between them and find the combination that minimizes energy consumption. Once the algorithm converges again, record the optimal solution set for the core variable group. .

[0118] After optimizing each of the two sets of variables individually, the optimal combination of parameters will be obtained. The system incorporates a built-in iterative feedback coordination loop. This loop sets a maximum number of coordination attempts and initializes the current optimal solution and optimal energy consumption. In each loop, the system executes two core optimization phases sequentially: First, it fixes the current optimal solution for strongly coupled variables and performs independent parallel optimization on weakly coupled variables with high contributions, seeking their local optimal solutions under the current system state. Then, it fixes the optimization results of the weakly coupled variables obtained in this round and releases all strongly coupled variables for collaborative optimization to explore the best cooperation relationship between variable groups. After each loop, the system evaluates the performance of the new solution. If the energy consumption index is better than the historical best value, it updates the global optimal solution and decides whether to start the next round of coordination; if the quality of the solution does not improve, the loop terminates early. This mechanism, through multiple fine-tuning and feedback, ensures that the independent optimality of weakly coupled variables and the collaborative optimality of strongly coupled variable groups are mutually adaptive and mutually reinforcing, ultimately approaching the true global optimal solution, rather than a simple superposition of local optimal solutions. The final global optimal solution is shown in Table 2. The loop diagram is as follows. Figure 4 As shown.

[0119] Table 2 Optimal Parameter Combinations

[0120] To further enhance the system's adaptability and robustness, the optimized parameter combination is applied to actual production, and new production data is continuously collected. The contribution and coupling strength analysis results are dynamically updated using this new data, initiating a new round of optimization. This process enables the entire system to adapt to changes in raw materials, equipment, and operating conditions, achieving continuous improvement in energy efficiency and dynamic adjustment of the model.

[0121] Application of optimization results and acquisition of new data. The optimal combination of process parameters output by the optimization model is applied to the on-site production control system to guide actual production operations. During the system's execution of this set of parameters, the following data for a complete production cycle (e.g., a shift, a day) are collected synchronously: the actual executed process parameter values ​​as input parameters and the actual process energy consumption within that cycle as output results. There are also potential fluctuations in boundary conditions such as raw material composition and ambient temperature during this period.

[0122] Data validity validation and preprocessing Operating condition consistency check: This assesses whether the production status is stable and continuous during the new data collection period, eliminating ineffective data caused by abnormal operating conditions such as downtime or malfunctions. The collected high-frequency data is aggregated and aligned to form a valid new sample dataset that matches the optimization cycle. Calculate the actual effect of the optimized solution. To verify whether the energy-saving effect has met expectations.

[0123] Dynamic updates of models and knowledge The system does not completely retrain the model with new data each time; instead, it employs an efficient incremental update strategy. This involves updating the model with a valid new sample dataset. Incorporate it into a historical database, or use it to create a rolling dataset within a time window (such as the most recent three months). Based on this updated dataset, rerun contribution analyses (such as linear regression) and coupling strength analyses (such as orthogonal experiments). Output the updated "contribution-coupling" parameter set.

[0124] The system compares the updated "contribution-coupling" parameter set with the previous version. If the contribution ranking of key factors or the core coupling relationship changes significantly (e.g., the contribution of a factor changes by more than 10%, or a new strong coupling relationship emerges), the system automatically triggers an alert. The system inputs the new classification parameter combination into the optimization model building module, automatically initiating a new round of "step-by-step collaborative optimization" process. A set of optimal parameter combinations adapted to the new production conditions is generated, starting a new round of "application-collection-update" cycle.

[0125] This method constructs energy consumption optimization models for typical steel production processes and intelligently analyzes the factors affecting process energy consumption. Deeply rooted in cutting-edge mechanism- and data-driven approaches, it analyzes the intrinsic mechanisms of energy conversion and consumption in steel production processes by deeply mining metallurgical mechanisms and on-site production data, constructing process energy consumption models including coking, sintering, ironmaking, steelmaking, and rolling. Based on these models, optimization models are simultaneously constructed using process energy consumption as a single objective function. Building upon these models, the method explores the energy consumption characteristics of each process and their influencing factors, deeply analyzing key factors affecting process energy consumption and revealing the different ways and effects of on-site parameters on process energy consumption. Based on the analysis results and referencing on-site production, it proposes energy-saving and consumption-reducing pathways and optimization schemes for each process, making a positive contribution to promoting energy conservation and consumption reduction in steel production processes.

[0126] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for analyzing factors affecting energy consumption and optimizing energy consumption in steel production processes, characterized in that, include: Construct an energy consumption model for steel production processes; The contribution of input variables in the energy consumption model of steel production processes is quantified. Quantify the coupling strength between any two input variables; The input variables are filtered based on a preset threshold, and the input variables are classified by combining a community detection algorithm to obtain a high-contribution strongly coupled variable group and a high-contribution weakly coupled variable group. Using high-contribution strongly coupled variable groups and high-contribution weakly coupled variable groups as inputs, a process energy consumption optimization model is constructed with the goal of minimizing process energy consumption. The energy consumption optimization model for the aforementioned process is solved to determine the optimal process combination parameters.

2. The method according to claim 1, characterized in that, The energy consumption model for the steel production process includes: in, Energy consumption in steel production processes This refers to the total amount of standard coal equivalent of all energy consumed in the steel production process. The amount of energy recovered during the steel production process; Output of products in steel production processes; The amount of energy recovered in the steel production process is calculated using a mechanism model, a prediction model, or a combination of a mechanism model and a prediction model. The total amount of standard coal equivalent of various energy sources consumed in the steel production process is calculated using a mechanistic model, a prediction model, or both.

3. The method according to claim 1, characterized in that, The quantization of the coupling strength between any two input variables includes: Calculate the main effects of the parameters; The coupling strength is the sum of the main effects of the two parameters on energy consumption and the ratio of the maximum impact of the interaction between the two parameters on energy consumption.

4. The method according to claim 1, characterized in that, The main effects of the calculated parameters include: Let be the average energy consumption of the process at the i-th level for the k-th parameter; The total average energy consumption of all processes under all orthogonal experimental conditions; This represents the fluctuation range of the k-th parameter relative to the baseline level at the ith level. This represents the main effect of the k-th parameter.

5. The method according to claim 1, characterized in that, The coupling strength: in, and Let be any two parameters to be analyzed. The maximum impact of the interaction between any two parameters on energy consumption is the difference between the maximum and minimum energy consumption under all combinations of the two parameters. for The absolute value of the main effect; for The absolute value of the main effect; for The maximum horizontal fluctuation range; for The maximum horizontal fluctuation range; for and The coupling strength.

6. The method according to claim 1, characterized in that, The contribution of the input variables in the quantitative steel production process energy consumption model includes: A single-factor model was constructed by setting variable constraints. After normalization, the individual contribution of each parameter to energy consumption was analyzed. Dimensionless processing was used to eliminate unit differences, and the energy consumption change amplitude when the parameter changes by 1% was calculated to obtain the sensitivity coefficient.

7. The method according to claim 1, characterized in that, The process of filtering the input variables based on a preset threshold and classifying the input variables using a community detection algorithm includes: The input variables are filtered based on a preset threshold to obtain a set of high-contribution variables; A preset coupling threshold is set. If the coupling strength between any two variables in the high contribution set is greater than the coupling threshold, they are classified as high contribution strongly coupled variable groups; otherwise, they are classified as high contribution weakly coupled variable groups.

8. The method according to claim 1, characterized in that, The step of solving the process energy consumption optimization model to determine the optimized process combination parameters includes: In the first stage, the parameters of the high-contribution, strongly coupled variable group are fixed. Then, the parameters of the high-contribution, weakly coupled variable group are optimized independently and in parallel using the particle swarm optimization algorithm until the process energy consumption meets the error requirements or the program reaches the maximum number of iterations and stops. The optimal solution of the high-contribution, weakly coupled variable group is then output. The second stage fixes the optimal solution of the high-contribution, weakly coupled variable group. Through the particle swarm optimization algorithm, the high-contribution, strongly coupled variable group is optimized collaboratively until the process energy consumption meets the error requirements or the program reaches the maximum number of iterations and stops, outputting the optimal solution of the high-contribution, strongly coupled variable group. The optimal solutions of the high-contribution weakly coupled variable group and the high-contribution strongly coupled variable group are used as optimization process combination parameters.

Images