A non-oriented silicon steel production path design method

Abstract

The application discloses a non-oriented silicon steel production path design method, and the method comprises the following steps: establishing a non-oriented silicon steel performance prediction model; when a certain production process is completed, designing path combination of a subsequent process; extracting process actual of the completed process and process setting data of different production lines of an uncompleted process; substituting the data into the magnetic performance prediction model to predict magnetic performance and mechanical performance of each path non-oriented silicon steel; comparing multiple performance prediction values of multiple paths; and recommending a path with optimal performance, highest yield, optimal cost or highest cost performance according to a designed path selection rule. The application realizes design of the non-oriented silicon steel production path, and can effectively improve the qualified rate and performance stability of the non-oriented silicon steel product, save cost and reduce energy consumption.

Description

Technical Field

[0001] This invention relates to the field of silicon steel production technology, and in particular to a method for designing a production path for non-oriented silicon steel. Background Technology

[0002] Non-oriented silicon steel, also known as non-oriented electrical steel, is a type of iron-silicon alloy with high magnetic permeability and low iron loss. It is a key soft magnetic material and is mainly used to manufacture the cores of electrical equipment such as transformers, motors, and generators. Its high performance and versatility make it one of the indispensable core materials in modern electrical equipment. With the continuous improvement of energy efficiency requirements, non-oriented silicon steel is being used more and more widely in the fields of new energy, energy-saving technology, and smart grids.

[0003] With the rapid development of computer technology, big data has been applied in various industries. The steel industry can effectively improve enterprise productivity and efficiency, optimize business decision-making processes, and make enterprises more competitive, creative, vibrant, and intelligent in the steel market by using the large amount of data stored in various aspects such as production management, process optimization, and process improvement.

[0004] The production process of non-oriented silicon steel includes continuous casting, hot rolling, normalizing, cold rolling, and annealing. Precise control of these processes ensures material consistency and high performance. Each process involves multiple production lines or units, resulting in a complex combination of process paths. Furthermore, different production lines or units have varying equipment and environmental conditions, leading to differences in process parameters. Therefore, the selection of process paths not only affects production efficiency and cost but also directly impacts the quality and performance of the final product. Thus, each process must be carefully planned and optimized before production to ensure the reliability and high efficiency of the final non-oriented silicon steel product.

[0005] Currently, the selection of production process routes for non-oriented silicon steel is mostly based on long-term production experience and actual production conditions. Although this can ensure normal and stable operation in most cases, it fails to fully consider various factors such as performance, output, and cost, and also fails to comprehensively evaluate multiple properties. It can only choose a relatively "error-free" but not necessarily optimal production route.

[0006] Big data modeling offers the possibility of selecting the optimal production path. By building performance prediction models using historical data, and designing production path combinations for subsequent processes before important production steps, the model predicts various performance characteristics for each path. Based on the prediction results, it recommends the path with the best performance, highest output, lowest cost, and highest cost-effectiveness. To a certain extent, this can improve product performance and stability, while also saving costs and increasing efficiency.

[0007] In the research field related to performance prediction of non-oriented silicon steel, current studies mainly focus on the influence of alloying elements and annealing processes on performance, and on temperature adjustment through performance prediction. However, research on production path selection and recommendation is scarce, with only one method using machine learning or reinforcement learning to build a path selection model using paths as training data. For example, invention patent CN103823974A discloses a principal component regression analysis method involving factors affecting the performance of non-oriented silicon steel. This method can comprehensively study the influence of inclusions, texture, and grain size on the performance of non-oriented silicon steel, identifying factors that significantly affect performance. Invention patent CN114139350A discloses a method for predicting iron loss in the continuous annealing process of non-oriented silicon steel. By inputting composition and process parameters into the regression equation of the established prediction model, the predicted iron loss value is obtained. Adjustments are made for cases that do not meet the target range, and this model can improve product quality. For example, invention patent CN118446079A discloses an additive manufacturing path process collaborative optimization method based on deep reinforcement learning. It trains a path planning reinforcement learning model and a process optimization reinforcement learning model separately, and then couples the two. By inputting several better actions of the path planning model into the process optimization model in real time, the optimal path and process combination is obtained. The optimal path and process combination is then input into the path planning model for deposition state update. Through continuous iteration, the optimal process path combination is obtained.

[0008] Overall, there is a considerable amount of research on the factors influencing the performance of non-oriented silicon steel, especially its microstructure and composition; however, research on data models is relatively limited, focusing primarily on data modeling methods; and research on industrial production path recommendation is almost nonexistent. Furthermore, there is a lack of research combining performance prediction models with production path design recommendations in specific industrial applications. Therefore, it is necessary to improve this structure to overcome these shortcomings. Summary of the Invention

[0009] The purpose of this invention is to provide a method for designing a production path for non-oriented silicon steel, thereby optimizing the production path, improving the pass rate and performance stability of non-oriented silicon steel products, and reducing production costs.

[0010] The above-mentioned technical objective of this invention has been achieved by the following technical solutions:

[0011] A method for designing a production path for non-oriented silicon steel, specifically including the following steps:

[0012] Step 1: Collect historical production data of non-oriented silicon steel, process the acquired historical production data, and establish a performance prediction model for non-oriented silicon steel based on the processed data;

[0013] Step 2: When a certain production process is completed, design the path combination for subsequent processes;

[0014] Step 3: Based on the production path combination designed in Step 2, extract the process performance of completed processes and the process setting data of different production lines for uncompleted processes;

[0015] Step 4: Input the process parameters extracted in Step 3 into the non-oriented silicon steel performance prediction model in Step 1.

[0016] Step 5: Predict the magnetic and mechanical properties of non-oriented silicon steel under various path combinations using the non-oriented silicon steel performance prediction model, compare the multiple performance prediction values ​​of multiple paths, and output the comparison results.

[0017] Step 6: Based on the comparison results of different paths obtained in Step 5, make a judgment according to the preset conditions and select the optimal path for production.

[0018] A further provision of the present invention is that step 1 specifically includes the following steps:

[0019] S1: Collect historical production data of non-oriented silicon steel, analyze the data, and preprocess the data. After the data preprocessing is completed, establish a performance prediction model for non-oriented silicon steel for different production line combinations.

[0020] The data preprocessing includes data deduplication, noise reduction, and handling of missing values. After preprocessing, correlation analysis and feature selection are performed. A performance prediction model for non-oriented silicon steel is established using machine learning methods. The performance predicted by the silicon steel model includes the magnetic and mechanical properties of non-oriented silicon steel.

[0021] A further provision of the present invention is that step 2 specifically includes the following steps:

[0022] S2: Before the production plan for hot rolling / normalizing annealing / cold rolling / SACL annealing is issued, design different combinations of production paths for subsequent processes.

[0023] Specifically, based on the production requirements and process characteristics of processes such as hot rolling, normalizing annealing, cold rolling, and SACL annealing, different production path combinations for subsequent processes are designed, a technical feasibility analysis is conducted for each production path combination, and the optimal production path combination is selected based on the evaluation results.

[0024] A further provision of the present invention is that step 3 specifically includes the following steps:

[0025] S3: Based on the production path combination designed in S2, retrieve the steelmaking composition data, steelmaking process parameters, hot rolling process parameters, normalizing process parameters, cold rolling process parameters, SACL process parameters, contract performance requirements, internal control requirements, specification parameters, historical output, and average downstream processing cost under different production line combinations.

[0026] A further provision of the present invention is that step 4 specifically includes the following steps:

[0027] S4: Substitute the parameters retrieved in step S3 into the non-oriented silicon steel performance prediction model, call the non-oriented silicon steel performance prediction model, predict the performance of the planned non-oriented silicon steel on different production line combinations, analyze the performance prediction results output by the model, and determine the optimal production line combination.

[0028] A further provision of the present invention is that step 5 specifically includes the following steps:

[0029] S5: Based on the contract performance requirements and internal control requirements, compare and select multiple paths according to the path recommendation rules based on the forecast results of different production lines.

[0030] A further provision of the present invention is as follows: In the path recommendation rule of step S5, under normal circumstances, the priority of the magnetic properties of non-oriented silicon steel is considered to be higher than that of mechanical properties. Among the magnetic properties of silicon steel, iron loss has the highest priority. If the focus is on the three properties of iron loss, magnetic induction, and yield strength, then the weight of iron loss is set to 4, the weight of magnetic induction is set to 3, and the weight of yield strength is set to 2. When the predicted performance is qualified, the corresponding performance weight value is taken; otherwise, it is 0. The sum of the three properties is denoted as N, and N is used to represent the performance qualification of the production line. If different performances are focused on, the performance weights can be designed according to the number and importance of the performances.

[0031] A further provision of the present invention is that the path recommendation rule in step S5 specifically includes the following steps:

[0032] When recommending paths, multiple types of path recommendations are performed, including optimal performance recommendation, most frequently used recommendation, lowest cost recommendation, and highest cost-effectiveness recommendation. The cost-effectiveness Ce calculation formula is shown below:

[0033] Ce = f(P_xn, yield, cost);

[0034] Where P_xn represents the predicted value of the key performance indicator, yield represents the historical output of the production line, and cost represents the average processing cost of the subsequent process.

[0035] A further provision of the present invention is that the path selection in step S5 specifically includes the following steps:

[0036] S51: Calculate the pass rate N of the three types of performance for each production line, and judge the pass rate N;

[0037] S52: If there is a path with N=9, that is, a path whose performance in all three prediction methods is satisfactory, then recommend several types of paths normally according to the recommendation requirements.

[0038] S53: If N≠9, meaning there are one or more performance characteristics that are predicted as unqualified on all paths, then the path with the highest qualification N value should be recommended.

[0039] S54: Regarding the path recommendation in step S53, if there are multiple calculated paths, the path with the best performance prediction is selected for recommendation; that is, the path with the highest iron loss pass rate, the highest magnetic induction pass rate, and the highest yield pass rate are selected in sequence for recommendation; ensuring that the recommended path is closest to the performance requirements.

[0040] S55: After performing the path recommendation step in S54, if the recommended path is a single path, then output that path as the optimal path; if there are still multiple recommended paths, then compare them in the order of importance of performance, yield, and cost.

[0041] When recommending the best-performing path, if multiple paths have the same and optimal critical performance values, the path with the highest output will be recommended as the optimal path.

[0042] If multiple paths have the same and lowest cost when recommending the lowest-cost path, then the path with the best critical performance is selected as the optimal path recommendation.

[0043] A further provision of this invention is that the pass rate of path selection is determined by calculating the probability that the predicted performance value meets the upper and lower limits of the performance requirements, and selecting a probability of ±2σ for the determination. The pass rate calculation formula is as follows:

[0044] CDF_Y=CDF('NORMAL',(Y_MAX-P_Y) / Y_STD)-CDF('NORMAL',(Y_MIN-P_Y) / Y_STD);

[0045] Where Y represents the performance index; Y_MAX represents the maximum performance value, Y_MIN represents the maximum performance value, P_Y represents the predicted performance value, and Y_STD represents the standard deviation of the predicted performance value.

[0046] In summary, the present invention has the following beneficial effects:

[0047] This invention provides a method for designing production paths for non-oriented silicon steel, which further optimizes performance and cost while meeting contractual requirements. Based on a performance prediction model for non-oriented silicon steel, this invention calculates the predicted performance values ​​for each production path, enabling the recommendation of non-oriented silicon steel production paths. This can improve the pass rate of non-oriented silicon steel products and reduce production costs. This control method covers a wide range of production lines and steel grades and can be widely applied in the field of performance control for non-oriented silicon steel. Attached Figure Description

[0048] Figure 1 This is a flowchart illustrating the non-oriented silicon steel production path design method of the present invention.

[0049] Figure 2 This is a flowchart illustrating an embodiment of the non-oriented silicon steel production path design method of the present invention. Detailed Implementation

[0050] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below with reference to the figures and specific embodiments.

[0051] like Figure 1 As shown, the present invention proposes a method for designing a production path for non-oriented silicon steel, which specifically includes the following steps:

[0052] Step 1: Collect historical production data of silicon steel, process the acquired historical production data, and establish a performance prediction model for non-oriented silicon steel based on the processed data;

[0053] Step 2: When a certain production process is completed, design the path combination for subsequent processes;

[0054] Step 3: Based on the production path combination designed in Step 2, extract the process performance of completed processes and the process setting data of different production lines for uncompleted processes;

[0055] Step 4: Input the process parameters extracted in Step 3 into the non-oriented silicon steel performance prediction model in Step 1.

[0056] Step 5: Predict the magnetic and mechanical properties of non-oriented silicon steel for each path combination using the non-oriented silicon steel performance prediction model, compare the multiple performance prediction values ​​of multiple paths, and output the comparison results.

[0057] Step 6: Based on the comparison results of different paths obtained in Step 5, make a judgment according to the preset conditions and select the optimal path for production.

[0058] Step 1 specifically includes the following steps:

[0059] S1: Collect historical production data of non-oriented silicon steel, analyze the data, and preprocess the data. After the data preprocessing is completed, establish a performance prediction model for non-oriented silicon steel for different production line combinations.

[0060] Data preprocessing includes data deduplication, noise reduction, and handling of missing values. After preprocessing, correlation analysis and feature selection are performed. A machine learning method is used to establish a performance prediction model for non-oriented silicon steel. Preferably, data preprocessing includes the following steps:

[0061] S11: Delete invalid and duplicate data with missing values ​​from the dataset; remove fluctuations and noise in the data through data smoothing to make the data smoother and more continuous; based on observations of the entire dataset, delete or impute missing data (e.g., using the temporal correlation of time series data to fill missing values ​​with observations from the previous time point). Preferably, the correlation analysis includes the following steps:

[0062] S12: Use correlation coefficients to evaluate the correlation between various features, and based on the correlation analysis results, select features that are highly correlated with the target performance. Preferably, the correlation analysis includes the following steps:

[0063] S13: Based on the results of correlation analysis, directly select features related to the target performance, use machine learning algorithms to evaluate the performance of the features, and select the optimal feature combination. Preferably, establishing a performance prediction model for non-oriented silicon steel using machine learning algorithms includes the following steps:

[0064] S14: Divide the preprocessed data into training and test sets. Use the training set data to train the selected machine learning algorithm to obtain a prediction model. Use the test set data to evaluate the trained model, calculate the prediction error to evaluate the model's performance, optimize the model based on the evaluation results, and deploy the optimized model to the production environment for real-time prediction of the performance of non-oriented silicon steel.

[0065] Specifically, the performance predicted by the non-oriented silicon steel model includes the magnetic properties (e.g., iron loss P15 / 50, P10 / 400, magnetic induction B50, etc.) and mechanical properties (e.g., yield strength, tensile strength, elongation, etc.). The specific prediction indicators are determined by the contract, and the performance prediction parameters of the non-oriented silicon steel performance prediction model can be adjusted according to the indicators specified in the contract.

[0066] Step 2 specifically includes the following steps:

[0067] S2: Before the production plan for hot rolling / normalizing annealing / cold rolling / SACL annealing is issued, design different combinations of production paths for subsequent processes.

[0068] Specifically, based on the production requirements and process characteristics of processes such as hot rolling, normalizing annealing, cold rolling, and SACL annealing, different production path combinations for subsequent processes are designed; a technical feasibility analysis is conducted for each production path combination, including process stability and equipment compatibility; and the optimal production path combination is selected based on the evaluation results.

[0069] Specifically, before the hot rolling production plan is issued, different production path combinations for subsequent processes are designed: these include hot rolling + normalizing annealing + cold rolling + SACL annealing.

[0070] Before the normalizing and annealing production plan was issued, different production path combinations for subsequent processes were designed, including: normalizing and annealing + cold rolling + SACL annealing.

[0071] Before the cold rolling production plan was issued, different production path combinations for subsequent processes were designed, including: cold rolling + SACL annealing;

[0072] Before the SACL annealing production plan is issued, different production path combinations for subsequent processes are designed, including SACL annealing.

[0073] In this embodiment, each process includes multiple mills. For example, hot rolling includes 2 mills, normalizing includes 3 mills, cold rolling includes 3 mills, and SACL includes 3 mills. Before the hot rolling plan is issued, the subsequent process combination is: hot rolling + normalizing annealing + cold rolling + SACL annealing. Theoretically, 54 subsequent process production paths can be designed.

[0074] Step 3 specifically includes the following steps:

[0075] S3: Based on the production path combination designed in S2, retrieve the steelmaking composition data, steelmaking process parameters, hot rolling process parameters, normalizing process parameters, cold rolling process parameters, SACL process parameters, contract performance requirements, internal control requirements, specification parameters, historical output, and average downstream processing cost under different production line combinations.

[0076] Step S3 retrieves parameters and requirements for different production line combinations based on the production path combination, which can optimize the production process, reduce waiting time and resource waste in production, thereby improving production efficiency.

[0077] Specifically, steelmaking composition data includes elements such as Si, Al, Mn, P, Ti, S, C, N, Nb, and V; steelmaking process parameters include ladle slag thickness and final free oxygen content after decarburization; hot rolling process parameters include furnace inlet temperature, furnace time, furnace outlet temperature, roughing rolling temperature, finishing rolling temperature, and coiling temperature; normalizing process parameters include soaking zone furnace temperature and heating zone furnace temperature; cold rolling process parameters include cold rolling thickness and cold rolling reduction rate; and SACL process parameters include annealing rate, annealing soaking zone furnace temperature, annealing heating zone furnace temperature, and annealing furnace tension. Force, etc.; Specification parameters include strip thickness, width, etc.; Contract performance requirements include minimum magnetic induction, maximum iron loss, and minimum yield strength requirements for product delivery; and / or minimum and maximum iron loss requirements for product delivery; Internal control requirements include performance internal control requirements for actual production, i.e., upper and lower limits and target values ​​of performance indicators; Historical output refers to the production volume of this production line combination in the past year, and the specific statistical time span can be flexibly adjusted as needed; Average downstream processing cost includes the sum of the average processing costs of each process production line.

[0078] Among them, the contract performance requirements are the release requirements set by the user and determined according to different user needs; the internal control requirements are the performance control requirements formed based on long-term actual production, and the performance control requirements have a higher priority than the contract performance requirements; historical output is the production volume of a specific grade and steel tapping mark within a certain period of time under a specific production path; the average downstream processing cost is the sum of the average processing costs of the downstream production lines under a specific production path, which is affected by various factors such as equipment and temperature, and is a variable value.

[0079] Before the hot rolling plan is issued, the steelmaking composition data and steelmaking process parameters retrieved are the actual production values ​​that have already gone through the process;

[0080] The hot rolling process parameters, normalizing process parameters, cold rolling process parameters, and SACL process parameters are the process settings for the production path designed in step S2;

[0081] Before the normalization annealing plan was issued, the steelmaking composition data, steelmaking process parameters, and hot rolling process parameters retrieved were the actual production values ​​that had already gone through the process.

[0082] The normalizing process parameters, cold rolling process parameters, and SACL process parameters are the process settings for the production path designed in step S2.

[0083] Before the cold rolling plan was issued, the steelmaking composition data, steelmaking process parameters, hot rolling process parameters, and normalizing process parameters retrieved were the actual production values ​​that had already gone through the process.

[0084] The cold rolling process parameters and SACL process parameters are the process settings for the production path designed in step S2.

[0085] Before the SACL annealing plan was issued, the steelmaking composition data, steelmaking process parameters, hot rolling process parameters, normalizing process parameters, and cold rolling process parameters retrieved were the actual production values ​​that had already gone through the process.

[0086] The SACL process parameters are the process settings for the production path designed in step S2.

[0087] Step 4 specifically includes the following steps:

[0088] S4: Substitute the parameters retrieved in step S3 into the performance prediction model for non-oriented silicon steel to predict the performance of the planned non-oriented silicon steel on different production line combinations.

[0089] Specifically, ensure that the non-oriented silicon steel performance prediction model used has been validated and has high prediction accuracy. Substitute the various parameters retrieved in step S3 (such as steelmaking composition data, steelmaking process parameters, hot rolling process parameters, normalizing process parameters, cold rolling process parameters, etc.) into the prediction model, call the non-oriented silicon steel performance prediction model, predict the performance of the planned non-oriented silicon steel on different production line combinations, analyze the performance prediction results output by the model, compare the performance prediction results under different production line combinations, evaluate the impact of each production line combination on the performance of silicon steel, and the performance prediction model can quickly provide silicon steel performance prediction results under different production line combinations, which helps to quickly determine the optimal production line combination.

[0090] Step 5 specifically includes the following steps:

[0091] S5: Based on the contract performance requirements and internal control requirements, compare and select multiple paths according to the path recommendation rules based on the forecast results of different production lines.

[0092] Specifically, the performance requirements in the contract are interpreted in detail, and key indicators such as the quality standards, production efficiency, and cost control that the product needs to achieve are clarified. Based on the internal control system, the relevant requirements for production process, production line management, and path selection are sorted out. Based on the contract performance requirements and internal control requirements, the forecast results of different production lines are compared and analyzed in multiple paths.

[0093] Generally, the magnetic properties of non-oriented silicon steel have a higher priority than the mechanical properties. Among the magnetic properties of silicon steel, iron loss has the highest priority. Therefore, in this embodiment, the weight of iron loss is set to 4, the weight of magnetic induction is set to 3, and the weight of yield strength is set to 2. When the predicted performance is qualified, the corresponding performance weight value is taken; otherwise, it is 0. The sum of the three properties is denoted as N, which is used to represent the performance qualification of the production line.

[0094] If different performance characteristics are being considered, performance weights can be designed based on the number and importance of each characteristic. In practical applications, the performance characteristics being considered are not limited to iron loss, magnetic induction, and yield strength. For example, in certain special application scenarios, other mechanical properties of silicon steel, such as tensile strength, elongation, hardness, and impact toughness, may also be considered. In such cases, it is necessary to add corresponding performance indicators according to specific requirements and assign appropriate weights to each indicator.

[0095] For example, in some cases, tensile strength can also be included in the performance evaluation of non-oriented silicon steel. According to the order of importance of the indicators, the weight of iron loss can be set to 7, the weight of magnetic induction to 5, the weight of yield strength to 3, and the weight of tensile strength to 1.

[0096] In other applications, the importance of the same performance metrics may change. In some cases, magnetic induction may become more critical than iron loss; while in others, yield strength or other mechanical properties may become the primary consideration. Therefore, the weighting settings should reflect this variation.

[0097] For example, in some cases, the magnetic induction weight is set to 4, the iron loss to 3, and the yield strength weight to 2.

[0098] In conclusion, setting performance weights is a flexible process that needs to be adjusted according to specific requirements and scenarios.

[0099] The path recommendation rule specifically includes the following steps: When recommending paths, four types of path recommendations are made (the recommendation type can be adjusted), namely, the path with the best performance (lowest iron loss in this embodiment), the most commonly used path (highest output in this embodiment), the path with the lowest cost (lowest average processing cost of subsequent processes in this embodiment), and the path with the highest cost-effectiveness (highest Ce value in this embodiment). The formula for calculating the cost-effectiveness Ce is as follows:

[0100] Ce = f(P_xn, yield, cost);

[0101] Where P_xn represents the predicted value of the key performance indicator, yield represents the historical output of the production line, and cost represents the average processing cost of the subsequent process.

[0102] Specifically, route selection includes the following steps:

[0103] S51: Calculate the pass rate N of the three types of performance for each production line, and judge the pass rate N;

[0104] S52: If there is a path with N=9, that is, a path whose performance in all three prediction methods is satisfactory, then recommend several types of paths normally according to the recommendation requirements.

[0105] S53: If N≠9, meaning there are one or more performance characteristics that are predicted as unqualified on all paths, then the path with the highest qualification N value should be recommended.

[0106] S54: Regarding the path recommendation in step S53, if there are multiple calculated paths, the path with the best performance prediction is selected for recommendation; that is, the path with the highest iron loss pass rate, the highest magnetic induction pass rate, and the highest yield pass rate are selected in sequence for recommendation; ensuring that the recommended path is closest to the performance requirements.

[0107] S55: After performing the path recommendation step in S54, if the recommended path is a single path, then output that path as the optimal path; if there are still multiple recommended paths, then compare them in the order of importance of performance, yield, and cost.

[0108] When recommending the best-performing path, if multiple paths have the same and optimal critical performance values, the path with the highest output will be recommended as the optimal path.

[0109] If multiple paths have the same and lowest cost when recommending the lowest-cost path, then the path with the best critical performance is selected as the optimal path recommendation.

[0110] The determination of whether the predicted performance of non-oriented silicon steel is qualified specifically includes: judging by calculating the probability that the predicted performance value meets the upper and lower limits of the performance requirements, and selecting a probability of ±2σ for judgment. The specific formula for calculating the pass rate is as follows:

[0111] CDF_Y=CDF('NORMAL',(Y_MAX-P_Y) / Y_STD)-CDF('NORMAL',(Y_MIN-P_Y) / Y_STD);

[0112] Where Y represents the performance index; Y_MAX represents the maximum performance value, Y_MIN represents the maximum performance value, P_Y represents the predicted performance value, and Y_STD represents the standard deviation of the predicted performance value.

[0113] Step 6 specifically includes the following steps:

[0114] S6: Based on the comparison results in S4, recommend the production path with the best performance, highest output, lowest cost, or highest cost-effectiveness. Select the appropriate path for production based on the actual situation.

[0115] Specifically, the process involves obtaining product performance indicators for each production line combination, evaluating the performance advantages and disadvantages of each combination, and determining the optimal production line combination. The model is used to predict the expected output of each production line combination, and the output data is analyzed to determine the production line combination with the highest output. Production costs are calculated based on factors such as production processes and material consumption for each production line combination. The costs of different production line combinations are compared to determine the lowest-cost combination. Combining performance, output, and cost data, the cost-effectiveness of each production line combination is calculated, evaluated, and the production line combination with the highest cost-effectiveness is determined. Based on the above analysis and evaluation, a production path with optimal performance, highest output, lowest cost, or highest cost-effectiveness is recommended. If no effective path is recommended, production can also proceed according to the original planned production path.

[0116] Example 1

[0117] like Figure 2 As shown, this embodiment involves the entire production process of non-oriented silicon steel products in a steel plant, and selects steel coil 1 produced by a certain production line as an example. The specific steps include:

[0118] Historical production data for non-oriented silicon steel was collected, including production line, equipment, composition, process, and testing data. After preprocessing the raw data (such as weight reduction, noise reduction, and handling missing values), correlation analysis and feature selection were performed. Machine learning methods were then used to model various performance prediction models.

[0119] In this embodiment, the performance prediction model is as follows:

[0120] TS=f1 (Si, Al, Mn, P, Ti, S, C, N, FRN, FT, CT, NOF, SF, HS, SS, Speed, Thick);

[0121] CG=f2(Si, Al, Mn, P, Ti, S, C, N, FRN, FT, CT, NOF, SF, HS, SS, Speed, Thick);

[0122] YP=f3(Si, Al, Mn, P, Ti, S, C, N, FRN, FT, CT, NOF, SF, HS, SS, Speed, Thick);

[0123] Where TS represents iron loss, CG represents magnetic flux density, and YP represents yield strength;

[0124] Si represents silicon content, Al represents aluminum content, Mn represents manganese content, P represents phosphorus content, Ti represents titanium content, S represents sulfur content, C represents carbon content, and N represents nitrogen content.

[0125] FRN represents the exit temperature, FT represents the finishing rolling temperature, CT represents the coiling temperature, NOF represents the furnace temperature of the normalizing annealing heating section, SF represents the furnace temperature of the normalizing annealing soaking section, HS represents the furnace temperature of the high-temperature zone of the heating section of the SACL continuous annealing furnace, SS represents the furnace temperature of the soaking section of the SACL continuous annealing furnace, Speed ​​represents the annealing speed, and Thick represents the SACL exit thickness.

[0126] The actual content of the steelmaking chemical composition of the selected non-oriented silicon steel coil 1 is as follows:

[0127] Si_ACT AL_ACT Mn_ACT P_ACT 2.806％ 1.0392％ a1％ b1％ Ti_ACT S_ACT C_ACT N_ACT c1％ d1％ e1％ f1％

[0128] The design values ​​for the hot rolling process of steel coil 1 on each hot rolling production line are as follows:

[0129] hot rolling production line FRN_AIM FT_AIM CT_AIM H01 T_frn_aim1 T_ft_aiml T_ct_aim1 H02 T_frn_aim2 T_ft_aim2 T_ct_aim2

[0130] Steel coil 1 undergoes normalizing. The normalizing process design values ​​for steel coil 1 are as follows:

[0131] Normalizing unit NOF_AIM SF_AIM C01 T_nof_aiml T_sf_aiml C02 T_nof_aim2 T_sf_aim2

[0132] The SACL process design values ​​for steel coil 1 are as follows:

[0133] SACL units HS_AIM SS_AIM S01 T_hs_aim1 T_ss_aim1 S02 T_hs_aim2 T_ss_aim2

[0134] The contract performance requirements for steel coil 1 are as follows:

[0135] Iron loss TS TS≤2.30 Magnetic CG CG≥1.62 Yield strength YS YS≥360

[0136] The internal control requirements for steel coil 1 are as follows:

[0137] TS_AIM 2.17 TS_MIN 2.02 TS_MAX 2.28 CG_AIM 1.67 CG_MIN 1.65 CG_MAX 1.69 YS_AIM 390 YS_MIN 370 YS_MAX 410

[0138] The target thickness of finished steel coil 1 is 0.30 mm, and the target speed of the central section of SACL is v1 m / min. Internal control requirements are within the contract scope, therefore only internal control requirements need to be considered. Since cold rolling parameters are not used in this example model, the cold rolling process is not recommended, and production can proceed by default.

[0139] 1) Hot rolling and subsequent process path design: Before the hot rolling schedule, design the process path combination based on the subsequent process unit / production line.

[0140] The hot rolling process includes two production lines, H01 and H02; the normalizing process includes two units, C01 and C02; and the SACL process includes three units, S01, S02 and S03.

[0141] The three processes can be combined to form the following 12 paths. The annual output of steel coil 1 corresponding to these 12 paths and the estimated average post-processing cost for each path are shown in the table below (the values ​​are for comparison and explanation purposes only and are not actual values):

[0142] Serial Number Path combination Production / ton Average post-processing cost per thousand yuan 1 H01-C01-S01 40 8.8 2 H01-C01-S02 140 8.7 3 H01-C01-S03 0 9.1 4 H01-C02-S01 50 8.9 5 H01-C02-S02 70 8.8 6 H01-C02-S03 0 9.2 7 H02-C01-S01 28 8.9 8 H02-C01-S02 80 8.8 9 H02-C01-S03 0 9.2 10 H02-C02-S01 32 9.0 11 H02-C02-S02 180 8.9 12 H02-C02-S03 0 9.3

[0143] As can be seen from the table above, for some reason, the output of this grade is 0 for all paths through SACL unit S03. Therefore, these paths are considered unsuitable for the production of this steel coil and are not predicted.

[0144] The actual values ​​of steelmaking composition, target values ​​for hot rolling, normalizing, SACL processes, and finished product thickness were input into the performance prediction model for non-oriented silicon steel. The target values ​​for iron loss, magnetic induction, and yield strength were calculated. The prediction results of the model are as follows:

[0145] Serial Number Path combination P_TS P_CG P_YS STD_TS STD_CG STD_YS 1 H01-C01-S01 2.226366 1.662513 396.31 0.030201 0.002230 7.25223 2 H01-C01-S02 2.154322 1.675427 397.32 0.028702 0.001296 7.13246 3 H01-C01-S03 -- -- -- -- -- -- 4 H01-C02-S01 2.187673 1.657789 389.34 0.033582 0.002212 7.19292 5 H01-C02-S02 2.154322 1.673428 394.38 0.031188 0.001771 7.14257 6 H01-C02-S03 -- -- -- -- -- -- 7 H02C01-S01 2.193657 1.668948 381.28 0.034725 0.002524 7.42165 8 H02-C01-S02 2.213894 1.665625 396.22 0.033668 0.003138 7.33287 9 H02-C01-S03 -- -- -- -- -- -- 10 H02-C02-S01 2.267658 1.665435 387.26 0.045124 0.002514 7.31324 11 H02-C02-S02 2.202195 1.665651 392.31 0.042593 0.002524 7.27236 12 H02-C02-S03 -- -- -- -- -- --

[0146] Where P_TS is the predicted value of iron loss, P_CG is the predicted value of magnetic induction, and P_YS is the predicted value of yield strength;

[0147] STD_TS is the standard deviation of iron loss, STD_CG is the standard deviation of magnetic flux density, and STD_YS is the standard deviation of yield strength.

[0148] The predicted values ​​of iron loss, magnetic induction, and yield strength obtained by the model calculation all meet the internal control requirements. However, since there may be some deviation between the model prediction value and the actual value of the performance, when using the predicted value of the performance to determine whether the performance is qualified, we do not directly judge whether the predicted performance is qualified based on the upper and lower limits. Instead, we judge by calculating the probability that the predicted value meets the upper and lower limits of the performance requirements, and choose the probability of ±2σ (i.e., 0.9545) to judge whether the predicted performance is qualified.

[0149] Substituting the performance requirements and predicted performance values ​​into the pass rate formula, the pass rates of the predicted performance indicators for different paths are calculated as shown in the table below:

[0150] Serial Number Path combination CDF_TS CDF_CG CDF_YS 1 H01-C01-S01 0.9621 0.9999 0.9703 2 H01-C01S02 0.9999 0.9999 0.9622 3 H01-C01-S03 -- -- -- 4 H01-C02-S01 0.9970 0.9997 0.9943 5 H01-C02-S02 0.9999 0.9999 0.9853 6 H01-C02-S03 -- -- -- 7 H02-C01-S01 0.9935 0.9999 0.9356 8 H02-C01-S02 0.9752 0.9999 0.9697 9 H02-C01-S03 -- -- -- 10 H02-C02-S01 0.6077 0.9999 0.9899 11 H02-C02S02 0.9661 0.9999 0.9914 12 H02-C02-S03 -- -- --

[0151] When the CDF_Y value is greater than 0.9545, the predicted performance is considered acceptable. Therefore, it can be seen that in the path combinations with serial numbers 7 and 10, one of the predicted performances does not meet the requirements (the yield pass rate CDF_YS of path 7 is 0.9356 < 0.9545; the iron loss pass rate CDF_TS of path 10 is 0.6077 < 0.9545), and their pass rates N are 4+3 = 7 and 3+2 = 5, respectively.

[0152] For the remaining 6 paths, the pass rate (CDF_Y) of all three predicted performance indicators is less than 0.9545, meaning that the predicted performance values ​​are all qualified, and the pass rate (N) is 4+3+2=9. According to the path recommendation rules, path recommendation is made among the 6 paths that are qualified in all three aspects.

[0153] Optimal Performance Path: According to the path recommendation rules, the importance of the three performance indicators is generally considered to be in the order of iron loss > magnetic induction > yield strength. Therefore, comparing the iron loss of several qualified paths, we can see that the iron loss of paths numbered 2 and 5 is the smallest among all paths. Therefore, comparing their magnetic induction, the magnetic induction of path number 2 is slightly greater than that of path number 5. Therefore, path number 2 is recommended, that is, the hot rolling + normalizing + SACL path of H01-C01-S02 is the optimal performance path.

[0154] The highest output path: According to the path recommendation rules, the highest output path is the path with the highest historical output among the qualified paths. Comparing the output information of several qualified paths, it can be seen that the path with the number 11 has the highest historical output. Therefore, the path with the number 11 is recommended, that is, the hot rolling + normalizing + SACL path with H02-C02-S02 is the highest output path.

[0155] Lowest cost path: According to the path recommendation rules, the lowest cost path is the path with the lowest average subsequent processing cost among the qualified paths. Comparing the average subsequent processing cost information of several qualified paths, it can be seen that the path with the number 2 has the lowest average subsequent processing cost. Therefore, the path with the number 2, that is, the path of hot rolling + normalizing + SACL with the value of H01-C01-S02, is recommended as the lowest cost path.

[0156] The most cost-effective path: Substitute the value of each parameter into the cost-effectiveness formula to calculate the cost-effectiveness Ce value for each qualified path:

[0157] Serial Number Path combination P_TS Yield Cost Ce 1 H01-C01-S01 2.226366 40 8.8 9.28 2 H01-C01-S02 2.154322 140 8.7 22.43 3 H01-C01-S03 -- -- -- -- 4 H01-C02-S01 2.187673 50 8.9 12.64 5 H01-C02-S02 2.154322 70 8.8 18.14 6 H01-C02-S03 -- -- -- -- 7 H02-C01-S01 -- -- -- -- 8 H02-C01-S02 2.213894 80 8.8 15.73 9 H02-C01-S03 -- -- -- -- 10 H02-C02-S01 -- -- -- -- 11 H02-C02-S02 2.202195 180 8.9 6.89 12 H02-C02--S03 -- -- -- --

[0158] According to the path recommendation rules, the path with the highest cost-effectiveness is the path with the highest cost-effectiveness Ce value among the qualified paths. Calculating and comparing the cost-effectiveness Ce of several qualified paths, we can see that the path with the number 2 has the highest cost-effectiveness Ce value. Therefore, the path with the number 2, that is, the path of hot rolling + normalizing + SACL with the value of H01-C01-S02, is recommended as the path with the highest cost-effectiveness.

[0159] 2) The normalizing and subsequent process route design, the actual hot rolling mill is H01. After the hot rolling process is completed, the actual hot rolling temperature is retrieved, as shown in the table below:

[0160] Furnace temperature Final rolling temperature winding temperature frn_r1 ft_r1 ct_r1

[0161] Before the normalizing and rolling schedule is finalized, process path combinations are designed based on the downstream mills / production lines. Normalizing includes two mills, C01 and C02; SACL includes three mills, S01, S02, and S03. Therefore, the two processes can be combined to create the following six paths. The table below shows the annual output of the grade corresponding to coil 1 along these six paths and the estimated average downstream processing cost for each path:

[0162] Serial Number Path combination Production / ton Average post-processing cost per thousand yuan 1 C01-S01 40 6.6 2 C01-S02 140 6.5 3 C01-S03 0 6.9 4 C02-S01 50 6.7 5 C02-S02 70 6.6 6 C02-S03 0 7.0

[0163] Similarly, based on the path combination of SACL unit S03, it is considered unsuitable for the production of this steel coil, and therefore no prediction is made for it. The actual values ​​of steelmaking composition, hot rolling process, normalizing process target value, SACL process target value, and finished product thickness target value are input into the non-oriented silicon steel performance prediction model to calculate the target values ​​of iron loss, magnetic induction, and yield strength. The model's prediction results are as follows:

[0164] Serial Number Path combination P_TS P_CG P_YS STD_TS STD_CG STD_YS 1 C01--S01 2.225421 1.662231 381.44 0.030201 0.002230 7.25223 2 C01-S02 2.158794 1.672487 398.23 0.028702 0.001296 7.13246 3 C01-S03 -- -- -- -- -- -- 4 C02-S01 2.186893 1.658829 381.67 0.033582 0.002212 7.19292 5 C02-S02 2.161287 1.673428 398.89 0.031188 0.001771 7.14257 6 C02-S03 -- -- -- -- -- --

[0165] Substituting the performance requirements and predicted performance values, the pass rates of the predicted performance indicators for different paths are calculated as shown in the table below:

[0166] Serial Number Path combination CDF_TS CDF_CG CDF_YS 1 C01-S01 0.9646 0.9999 0.9426 2 C01-S02 0.9999 0.9999 0.9505 3 C01-S03 -- -- -- 4 C02-S01 0.9972 0.9999 0.9476 5 C02-S02 0.9999 0.9999 0.9400 6 C02-S03 -- -- --

[0167] As can be seen, the CDF_YS calculated for all four paths is less than 0.9545, meaning the predicted yield strength does not meet the requirements. However, other performance indicators are within the required range. Therefore, according to the path recommendation rule, the pass rate N for all four paths is calculated to be 4+3=7, meaning the current maximum pass rate is 7. At this point, the non-compliance performance, i.e., yield strength, is used as the key performance indicator to recommend the optimal path.

[0168] According to the path recommendation rules, in order to ensure that the performance indicators meet production requirements as much as possible, the path with the highest yield strength qualification rate among the four paths is recommended, which is path number 2, i.e., the path with normalization + SACL of C01-S02. The optimal path is then automatically issued with a process adjustment prompt signal to the relevant process technicians to modify and adjust the process, further ensuring that the yield strength meets the qualification requirements.

[0169] 3) The SACL process path design actually passes through the C01 hot rolling mill. After the normalizing process is completed, the actual normalizing furnace temperature is retrieved, as shown in the table below:

[0170] Normalizing heating section temperature Normalizing temperature of the homogenization zone nof_r1 sf_r1

[0171] Before the SACL rolling schedule, the process path combinations are designed based on the downstream mills / production lines. SACL includes three mills: S01, S02, and S03; therefore, there are only three paths per process. The following table shows the average downstream processing cost estimated for each path and the annual output of the grade corresponding to coil 1 on these three paths:

[0172] Serial Number Path combination Production / ton Average post-processing cost per thousand yuan 1 S01 40 3.5 2 S02 140 3.4 3 S03 0 3.8

[0173] Similarly, based on the path combination of SACL unit S03, it is considered unsuitable for the production of this steel coil, and no prediction is made for it.

[0174] The actual values ​​of steelmaking composition, hot rolling process, normalizing process, SACL process target value, and finished product thickness target value were input into the performance prediction model for non-oriented silicon steel. The target values ​​of iron loss, magnetic induction, and yield strength were calculated. The prediction results of the model are as follows:

[0175] Serial Number Path combination P_TS P_CG P_YS STD_TS STD_CG STD_YS 1 S01 2.242381 1.664982 385.78 0.030201 0.002230 7.25223 2 S02 2.158794 1.675422 399.89 0.028702 0.001296 7.13246 3 S03 -- -- -- -- -- --

[0176] Substituting the performance requirements and predicted performance values, the pass rates of the predicted performance indicators for different paths are calculated as shown in the table below;

[0177] Serial Number Path combination CDF_TS CDF_CG CDF_YS 1 S01 0.8935 0.9999 0.9847 2 S02 0.9999 0.9218 0.9218 3 S03 -- -- --

[0178] As can be seen, there is a certain performance qualification rate in both paths that does not meet the requirements. For path number 1, CDF_TS is less than 0.9545, which means that the predicted iron loss value does not meet the requirements; for path number 2, CDF_YS is less than 0.9545, which means that the predicted yield strength value does not meet the requirements.

[0179] Based on the path recommendation rules, the pass / fail N values ​​for the three performance characteristics of the two paths are calculated, and the following can be obtained:

[0180] The magnetic field and yield strength of the path with serial number 1 meet the requirements, and the pass rate N = 3 + 2 = 5;

[0181] The iron loss and magnetic flux density of the path with serial number 2 meet the requirements, and the pass rate N = 4 + 3 = 7.

[0182] The path with sequence number 2 has a higher qualification rate than the path with sequence number 1. Therefore, the path with the highest qualification rate, i.e., the path with sequence number 1, which is the path with SACL unit S02, is recommended as the optimal path.

[0183] In the description of this invention, it should be noted that the terms "upper," "lower," "inner," "outer," "left," and "right," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of this invention is in use, or the orientation or positional relationship commonly understood by those skilled in the art. These terms are used only for the convenience of describing this invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention. Furthermore, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance. In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, terms such as "set" and "connect" should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium; it can be a connection within two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances. In this document, the terms “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, which includes not only the elements listed but also other elements not expressly listed.

[0184] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A method for designing a production path for non-oriented silicon steel, characterized in that, Specifically, the following steps are included: Step 1: Collect historical production data of non-oriented silicon steel, process the acquired historical production data, and establish a performance prediction model for non-oriented silicon steel based on the processed data. Step 2: When a certain production process is completed, design the path combination for subsequent processes; Step 3: Based on the production path combination designed in Step 2, extract the process performance of completed processes and the process setting data of different production lines for uncompleted processes; Step 4: Substitute the process parameters extracted in Step 3 into the non-oriented silicon steel performance prediction model in Step 1. Step 5: Predict the magnetic and mechanical properties of non-oriented silicon steel under various path combinations using the non-oriented silicon steel performance prediction model, compare the multiple performance prediction values ​​of multiple paths, and output the comparison results. Step 6: Based on the comparison results of different paths obtained in Step 5, make a judgment according to the preset conditions and select the optimal path for production.

2. The method for designing a production path for non-oriented silicon steel according to claim 1, characterized in that, Step 1 specifically includes the following steps: S1: Collect historical production data of non-oriented silicon steel, analyze the data, and preprocess the data. After the data preprocessing is completed, establish a performance prediction model for non-oriented silicon steel for different production line combinations. The data preprocessing includes data deduplication, noise reduction, and handling of missing values. After preprocessing, correlation analysis and feature selection are performed. A performance prediction model for non-oriented silicon steel is established using machine learning methods. The performance predicted by the silicon steel model includes the magnetic and mechanical properties of non-oriented silicon steel.

3. The method for designing a production path for non-oriented silicon steel according to claim 1, characterized in that, Step 2 specifically includes the following steps: S2: Before the production plan for hot rolling / normalizing annealing / cold rolling / SACL annealing is issued, design different combinations of production paths for subsequent processes. Based on the production requirements and process characteristics of hot rolling, normalizing annealing, cold rolling, and SACL annealing, different combinations of production paths are designed for subsequent processes.

4. The method for designing a production path for non-oriented silicon steel according to claim 1, characterized in that, Step 3 specifically includes the following steps: S3: Based on the production path combination designed in S2, retrieve the steelmaking composition data, steelmaking process parameters, hot rolling process parameters, normalizing process parameters, cold rolling process parameters, SACL process parameters, contract performance requirements, internal control requirements, specification parameters, historical output, and average downstream processing cost under different production line combinations.

5. The method for designing a production path for non-oriented silicon steel according to claim 1, characterized in that, Step 4 specifically includes the following steps: S4: Substitute the parameters retrieved in step S3 into the non-oriented silicon steel performance prediction model, call the non-oriented silicon steel performance prediction model, and predict the performance of the planned non-oriented silicon steel on different production line combinations.

6. The method for designing a production path for non-oriented silicon steel according to claim 1, characterized in that, Step 5 specifically includes the following steps: S5: Based on the contract performance requirements and internal control requirements, compare and select multiple paths according to the path recommendation rules based on the forecast results of different production lines.

7. The method for designing a production path for non-oriented silicon steel according to claim 6, characterized in that, In the path recommendation rules of step S5, it is generally assumed that the magnetic properties of non-oriented silicon steel have a higher priority than the mechanical properties. Among the magnetic properties of silicon steel, iron loss has the highest priority. If the focus is on three properties, namely iron loss, magnetic induction, and yield strength, then the weight of iron loss is set to 4, the weight of magnetic induction is set to 3, and the weight of yield strength is set to 2. When the predicted performance is qualified, the corresponding performance weight value is taken; otherwise, it is 0. The sum of the three properties is denoted as N, which is used to represent the performance qualification of the production line. If different performances are being focused on, the performance weights can be designed according to the number and importance of the performances. In some application scenarios, the performance that needs to be considered may also include tensile strength, elongation, etc. In this case, other performance indicators can also be included in the performance consideration of non-oriented silicon steel according to specific needs. The specific indicator weights need to be redesigned according to the needs. In other application scenarios, the importance of the same performance indicators may change. For example, when magnetic induction becomes the primary consideration, the order of importance of the indicators can be adjusted, setting the weight of magnetic induction to 4, iron loss to 3, and yield strength to 2.

8. The method for designing a production path for non-oriented silicon steel according to claim 6, characterized in that, The path recommendation rules in step S5 specifically include the following steps: When recommending paths, multiple types of path recommendations are performed, including optimal performance recommendation, most frequently used recommendation, lowest cost recommendation, and highest cost-effectiveness recommendation. The cost-effectiveness Ce calculation formula is shown below: Ce = f(P_xn, yield, cost); Where P_xn represents the predicted value of the key performance indicator, yield represents the historical output of the production line, and cost represents the average processing cost of the subsequent process.

9. The method for designing a production path for non-oriented silicon steel according to claim 6, characterized in that, Step S5, path selection, specifically includes the following steps: S51: Calculate the pass rate N of the three types of performance for each production line, and judge the pass rate N; S52: If there is a path with N=9, that is, a path whose performance in all three prediction methods is satisfactory, then recommend several types of paths normally according to the recommendation requirements. S53: If N≠9, meaning there are one or more performance characteristics that are predicted as unqualified on all paths, then the path with the highest qualification N value should be recommended. S54: Regarding the path recommendation in step S53, if there are multiple calculated paths, the path with the best performance prediction is selected for recommendation; that is, the path with the highest iron loss pass rate, the highest magnetic induction pass rate, and the highest yield pass rate are selected in sequence for recommendation; ensuring that the recommended path is closest to the performance requirements. S55: After performing the path recommendation step in S54, if the recommended path is a single path, then output that path as the optimal path; if there are still multiple recommended paths, then compare them in order of importance of performance, output, and cost. When recommending the best-performing path, if multiple paths have the same and optimal critical performance values, the path with the highest output will be recommended as the optimal path. If multiple paths have the same and lowest cost when recommending the lowest-cost path, then the path with the best critical performance is selected as the optimal path recommendation.

10. The method for designing a production path for non-oriented silicon steel according to claim 9, characterized in that, The pass rate of path selection is determined by calculating the probability that the predicted performance value meets the upper and lower limits of the performance requirements, and a probability of ±2σ is selected for judgment. The pass rate calculation formula is as follows: CDF_Y=CDF('NORMAL',(Y_MAX-P_Y) / Y_STD)-CDF('NORMAL',(Y_MIN-P_Y) / Y_STD); Where Y represents the performance index; Y_MAX represents the maximum performance value, Y_MIN represents the maximum performance value, P_Y represents the predicted performance value, and Y_STD represents the standard deviation of the predicted performance value.

Images