A real-time correction method for sintering raw material test data and raw material ratio

Abstract

The present invention provides a method for real-time correction of sintering raw material test data and raw material ratio, which belongs to the field of electronic data processing technology, including: obtaining historical raw material test data, historical sintering process data and historical finished product data within multiple historical specified time periods, determining multiple sintered ore composition targets; determining the important parameter vector and contribution algorithm of each sintered ore composition target; constructing a prediction model for each sintered ore composition target; determining the influencing parameter vector of each sintered ore composition target; determining current test correction data; obtaining multiple historical test correction data, determining feedback values ​​based on the current test correction data and multiple historical test correction data, and generating a correction report. The data quality and the scientificity of analysis and optimization can be improved, the scientificity and reliability of the prediction results can be ensured, the dynamic adjustment of theoretical and practical closed-loop feedback and the automatic correction and optimization can be realized, and the pre-positioning and intelligentization of quality control management can be realized.

Description

Technical Field

[0001] The invention relates to the technical field of electronic data processing, and in particular to a real-time correction method for sintering raw material test data and raw material ratio. Background Art

[0002] In the traditional sinter production process, the raw material test data and the adjustment of the raw material ratio often rely on the experience of the operator, which has problems such as poor real-time performance, delayed adjustment, and low accuracy. This method is difficult to deal with the impact of raw material fluctuations or changes in sintering process parameters on the quality of the finished product, and it is easy to cause the sintered ore composition target to deviate from the requirements, thereby affecting the blast furnace production efficiency. Existing technologies usually optimize the ratio through offline analysis of historical data or experimental verification, but lack support for dynamic adjustment during the production process, and the adjustment process cannot quantify the specific contribution of each parameter.

[0003] Therefore, the present invention provides a method for real-time correction of sintering raw material test data and raw material ratio. Summary of the invention

[0004] The present invention provides a real-time correction method for sintering raw material test data and raw material proportion. The method comprises the following steps: determining a plurality of sintering ore composition targets, determining an important parameter vector of each sintering ore composition target and a contribution algorithm of the important parameter vector to the corresponding sintering ore composition target, constructing a prediction model for each sintering ore composition target, determining an influencing parameter vector of each sintering ore composition target, collecting current raw material test data within a current specified time period, determining current test correction data, determining a feedback value, and generating a correction report. The method can improve data quality and the scientific nature of analysis and optimization, realize personalized and precise construction of a sintering ore composition target prediction model, ensure the scientific nature and reliability of the prediction result, realize dynamic adjustment of theoretical and practical closed-loop feedback as well as automatic correction and optimization, realize the pre-positioning and intelligentization of quality control management, and improve the stability and efficiency of sintering ore production.

[0005] The present invention provides a method for real-time correction of sintering raw material test data and raw material ratio, comprising:

[0006] 101: Acquire historical raw material test data, historical sintering process data, and historical finished product data within multiple historical specified time periods, and determine multiple sinter ore composition targets;

[0007] 102: Analyze historical raw material test data and historical sintering process data to determine the important parameter vector of each sintered ore composition target, and determine the contribution algorithm of the sintered ore composition target corresponding to the important parameter vector;

[0008] 103: Based on the contribution algorithm of each sinter composition target, historical raw material analysis data and historical sintering process data, a prediction model for each sinter composition target is constructed;

[0009] 104: Determine an influencing parameter vector of each sinter ore composition target based on a prediction model of each sinter ore composition target;

[0010] 105: Collect current raw material test data within the current specified time period, and determine current test correction data based on the current raw material test data and all prediction models;

[0011] 106: Acquire multiple historical test correction data, determine a feedback value based on the current test correction data and the multiple historical test correction data, and generate a correction report;

[0012] Among them, based on the contribution algorithm of each sinter composition target, historical raw material test data and historical sintering process data, a prediction model for each sinter composition target is constructed, including:

[0013] Inputting the important parameter vector of each sinter composition target into the prediction model of the corresponding sinter composition target;

[0014] The prediction models of all sintered ore composition targets are trained according to the historical raw material analysis data and the historical sintering process data, and the prediction models of the corresponding sintered ore composition targets are constructed based on the contribution algorithm of each sintered ore composition target.

[0015] According to a method for real-time correction of sintering raw material test data and raw material ratio provided by the present invention, historical raw material test data, historical sintering process data and historical finished product data within multiple historical specified time periods are obtained to determine multiple sintered ore composition targets, including:

[0016] Obtain historical raw material analysis sub-data, historical sintering process sub-data and historical finished product sub-data within each historical specified time period;

[0017] Determine historical raw material test data based on historical raw material test sub-data within all historical specified time periods after preprocessing, determine historical sintering process data based on historical sintering process sub-data within all historical specified time periods after preprocessing, and at the same time, determine historical finished product data based on historical finished product sub-data within all historical specified time periods after preprocessing;

[0018] Extract the finished product components whose appearance times in all historical finished product sub-data are lower than a preset threshold from the historical finished product data and mark them as discarded components;

[0019] All finished product components that are not marked as discarded components in the historical finished product data are determined as comprehensive sintered ore composition targets, wherein the comprehensive sintered ore composition target includes multiple sintered ore composition targets.

[0020] According to a real-time correction method for sintering raw material test data and raw material ratio provided by the present invention, historical raw material test data and historical sintering process data are analyzed to determine the important parameter vector of each sintered ore composition target, and a contribution algorithm of the sintered ore composition target corresponding to the important parameter vector is determined, including:

[0021] Based on the physical and chemical reactions in the sintering process, a first correlation relationship between each sintered ore composition target and the historical raw material analysis sub-data of each historical specified time period is analyzed, and at the same time, a second correlation relationship between each sintered ore composition target and the historical sintering process sub-data of each historical specified time period is analyzed;

[0022] Determine a first sub-important parameter vector of each sintered ore composition target based on the sintered raw material based on all parameters whose correlation exceeds a first preset correlation threshold value in a first correlation relationship between each sintered ore composition target and all historical raw material analysis sub-data of a historical specified time period;

[0023] Determine a second sub-important parameter vector of each sintered ore composition target based on the sintering process based on all parameters whose correlation exceeds a second preset correlation threshold value in a second correlation relationship between each sintered ore composition target and all historical raw material analysis sub-data of a historical specified time period;

[0024] Determine an important parameter vector for each sinter composition target based on a first sub-important parameter vector of the sinter raw material and a second sub-important parameter vector of the sinter process for each sinter composition target;

[0025] Select multiple algorithms and determine the important parameter vectors for each sinter composition target based on the fitting contribution of each algorithm;

[0026] The algorithm with the lowest fitting contribution is selected as the contribution algorithm corresponding to the sintered ore composition target.

[0027] According to a real-time correction method for sintering raw material test data and raw material ratio provided by the present invention, based on the prediction model of each sintered ore component target, the influencing parameter vector of each sintered ore component target is determined, including:

[0028] Based on the output results of the prediction model for each sinter composition target, determining the influence of each parameter in the important parameter vector on the corresponding sinter composition target;

[0029] Sorting the influence of all parameters in the important parameter vector of each sinter composition target on the sinter composition target from large to small, and determining the initial influencing parameter vector of each sinter composition target based on the sorting result;

[0030] Each parameter of the initial influencing parameter vector of each sintered ore composition target is judged, if the parameter is a state parameter, the parameter is replaced by the raw material feeding parameter of the parameter, if the parameter is an operation parameter, no replacement is performed;

[0031] Based on the initial influencing parameter vector after each sinter ore composition target is replaced, an influencing parameter vector for each sinter ore composition target is determined.

[0032] According to a method for real-time correction of sintering raw material test data and raw material ratio provided by the present invention, current raw material test data within a current specified time period is collected, and current test correction data is determined based on the current raw material test data and all prediction models, including:

[0033] Preprocessing the collected current raw material test data of the current specified time period, and determining the current parameter vector of each sinter ore composition target based on the preprocessed current raw material test data and the influencing parameter vector of each sinter ore composition target;

[0034] Inputting the current parameter vector of each sinter composition target into the corresponding prediction model, determining the actual sinter composition value of each sinter composition target, and determining the actual sinter vector based on the actual sinter composition values ​​of all sinter composition targets;

[0035] Determine the theoretical sinter composition value of each sinter composition target based on the current raw material test data and the raw material ratio, and determine the theoretical sinter vector based on the theoretical sinter composition values ​​of all sinter composition targets;

[0036] Determine the sinter difference matrix based on the accuracy requirements of all sinter composition targets and the theoretical sinter composition vector;

[0037] ;

[0038] in, represents the sinter difference matrix, represents the theoretical lower limit value of the first sintered ore composition target, the i-th sintered ore composition target, and the N1-th sintered ore composition target, represents the theoretical upper limit of the first sinter ore composition target, the i-th sinter ore composition target, and the N1-th sinter ore composition target, and N1 represents the number of sinter ore composition targets;

[0039] Determine whether each actual sintered ore component value in the actual sintered ore vector is between the lower limit value and the upper limit value of the corresponding sintered ore component target in the sintered ore difference matrix. If any actual sintered ore component value is not between the lower limit value and the upper limit value of the corresponding sintered ore component target, adjust the current raw material test data and determine the adjusted actual sintered ore vector; until all actual sintered ore component values ​​in the adjusted actual sintered ore vector are between the lower limit value and the upper limit value of the corresponding sintered ore component target in the sintered ore difference matrix, record the number of adjustments;

[0040] Determine the current raw material test data in which all actual sinter ore composition values ​​in the actual sinter ore vector are between the lower limit value and the upper limit value of the corresponding sinter ore composition target in the sinter ore difference matrix as the corrected raw material test data;

[0041] The correction value of each raw material is determined based on the corrected raw material assay data and the current raw material assay data, and the current assay correction data is determined based on the correction values ​​of all raw materials and the number of adjustments.

[0042] According to a method for real-time correction of sintering raw material test data and raw material ratio provided by the present invention, a feedback value is determined based on current test correction data, and feedback is pushed to the raw material source based on the feedback value, including:

[0043] Get historical test correction data for multiple historical specified time periods before the current specified time period;

[0044] Determining a feedback value based on current assay correction data for a current specified time period and historical assay correction data for a plurality of historical specified time periods;

[0045] If the feedback value exceeds the feedback threshold, feedback is determined based on the feedback value and the current assay correction data for the current specified time period, and the feedback is sent to the source of the raw materials;

[0046] Generates a correction report based on the feedback value, the influencing parameter vector, the number of adjustments, and the current assay correction data for the current specified time period.

[0047] According to a method for real-time correction of sintering raw material test data and raw material ratio provided by the present invention, a feedback value is determined based on current test correction data of a current specified time period and historical test correction data of multiple historical specified time periods, including:

[0048] ;

[0049] ;

[0050] in, Indicates the feedback value, Indicates the current feedback value of the current specified time period, It indicates the historical feedback value of all historical specified time periods, N2 indicates the number of raw materials in the current test correction data, Indicates the correction value of the jth raw material in the current assay correction data for the current specified time period, represents the weight of the correction value of the jth raw material, Indicates the number of adjustments for the current specified time period. Indicates the weight of the number of adjustments in the current specified time period. represents the weight of the absolute difference of the i-th sinter composition target, It represents the absolute difference between the actual sinter composition value and the theoretical sinter composition value of the i-th sinter composition target in the current specified time period based on the corrected raw material test data. represents the first amplification factor of the absolute difference, represents the second amplification factor of the absolute difference, represents the attenuation factor of the historical specified time period, T represents the number of historical specified time periods, Indicates the historical correction value of the jth raw material in the historical test correction data of the tth historical specified time period, The weight representing the number of adjustments in the t-th historical specified time period, Indicates the number of adjustments in the tth historical specified time period, It represents the absolute difference between the historical sinter composition value and the theoretical sinter composition value based on the historical corrected raw material analysis data for the i-th sinter composition target in the t-th historical specified time period. Represents the historical smoothing factor.

[0051] Compared with the prior art, the present invention has the following beneficial effects:

[0052] By determining multiple sinter ore composition targets, determining the important parameter vectors of each sinter ore composition target and the contribution algorithm of the important parameter vectors to the corresponding sinter ore composition target, building a prediction model for each sinter ore composition target, determining the influencing parameter vector of each sinter ore composition target, collecting the current raw material test data within the current specified time period, determining the current test correction data, determining the feedback value, and generating a correction report, it is possible to improve data quality and the scientific nature of analysis and optimization, realize the construction of a personalized and precise sinter ore composition target prediction model, ensure the scientific nature and reliability of the prediction results, realize dynamic adjustment of theoretical and practical closed-loop feedback as well as automated correction and optimization, realize the pre-positioning and intelligence of quality control management, and improve the stability and efficiency of sinter ore production. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0054] Figure 1 It is a flow chart of a method for real-time correction of sintering raw material test data and raw material ratio provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0056] Embodiment 1:

[0057] The embodiment of the present invention provides a method for real-time correction of sintering raw material test data and raw material ratio, such as Figure 1 As shown, including:

[0058] 101: Acquire historical raw material test data, historical sintering process data, and historical finished product data within multiple historical specified time periods, and determine multiple sinter ore composition targets;

[0059] 102: Analyze historical raw material test data and historical sintering process data to determine the important parameter vector of each sintered ore composition target, and determine the contribution algorithm of the sintered ore composition target corresponding to the important parameter vector;

[0060] 103: Based on the contribution algorithm of each sinter composition target, historical raw material analysis data and historical sintering process data, a prediction model for each sinter composition target is constructed;

[0061] 104: Determine an influencing parameter vector of each sinter ore composition target based on a prediction model of each sinter ore composition target;

[0062] 105: Collect current raw material test data within the current specified time period, and determine current test correction data based on the current raw material test data and all prediction models;

[0063] 106: Acquire multiple historical test correction data, determine a feedback value based on the current test correction data and the multiple historical test correction data, and generate a correction report;

[0064] Among them, based on the contribution algorithm of each sinter composition target, historical raw material test data and historical sintering process data, a prediction model for each sinter composition target is constructed, including:

[0065] Inputting the important parameter vector of each sinter composition target into the prediction model of the corresponding sinter composition target;

[0066] The prediction models of all sintered ore composition targets are trained according to the historical raw material analysis data and the historical sintering process data, and the prediction models of the corresponding sintered ore composition targets are constructed based on the contribution algorithm of each sintered ore composition target.

[0067] In this embodiment, the multiple historical specified time periods include the current time and a short period of time before.

[0068] In this embodiment, historical finished product data is analyzed to determine multiple sintered ore composition targets, that is, composition indicators that need to be focused on, such as iron content (Fe), basicity (CaO / SiO2), etc. in the sintered ore.

[0069] In this embodiment, for each sintered ore composition target (such as Fe content, basicity, etc.), important parameters that have a greater impact on it are extracted from historical raw material analysis data and sintering process data. These parameter sets are called important parameter vectors.

[0070] In this embodiment, historical raw material test data and sintering process data, as well as the relationship of the contribution algorithm, are used to construct a prediction model for each sintered ore composition target. Each prediction model learns the rules of each composition target by analyzing historical finished product data, and is able to predict the actual composition value of each sintered ore composition target when the current raw material test data is given.

[0071] In this embodiment, within the current specified time period, the current raw material test data is collected, the current test data is input into the prediction model, and the actual composition value of each sintered ore composition target is determined in combination with the important parameter vector. The current raw material test data is adjusted according to the actual composition value until the composition target requirements are met, thereby generating the final current test correction data.

[0072] In this embodiment, the important parameter vector of each sinter ore composition target is used as an input feature, and a prediction model is constructed separately for each sinter ore composition target using a selected contribution algorithm.

[0073] In this embodiment, the pre-processed historical raw material test data and the historical sintering process data are used as training data to train the prediction model of each sintered ore composition target.

[0074] The beneficial effects of the above technical solution are as follows: by determining multiple sintered ore composition targets, determining the important parameter vectors of each sintered ore composition target and the contribution algorithm of the important parameter vectors to the corresponding sintered ore composition target, constructing a prediction model for each sintered ore composition target, determining the influencing parameter vector of each sintered ore composition target, collecting the current raw material test data within the current specified time period, determining the current test correction data, determining the feedback value, and generating a correction report, it is possible to improve data quality and the scientific nature of analysis and optimization, realize the personalized and precise construction of a sintered ore composition target prediction model, ensure the scientific nature and reliability of the prediction results, realize dynamic adjustment of theoretical and practical closed-loop feedback as well as automated correction and optimization, realize the pre-positioning and intelligence of quality control management, and improve the stability and efficiency of sintered ore production.

[0075] Embodiment 2:

[0076] The embodiment of the present invention provides a method for real-time correction of sintering raw material test data and raw material ratio, which obtains historical raw material test data, historical sintering process data and historical finished product data within multiple historical specified time periods, and determines multiple sintered ore composition targets, including:

[0077] Obtain historical raw material analysis sub-data, historical sintering process sub-data and historical finished product sub-data within each historical specified time period;

[0078] Determine historical raw material test data based on historical raw material test sub-data within all historical specified time periods after preprocessing, determine historical sintering process data based on historical sintering process sub-data within all historical specified time periods after preprocessing, and at the same time, determine historical finished product data based on historical finished product sub-data within all historical specified time periods after preprocessing;

[0079] Extract the finished product components whose appearance times in all historical finished product sub-data are lower than a preset threshold from the historical finished product data and mark them as discarded components;

[0080] All finished product components that are not marked as discarded components in the historical finished product data are determined as comprehensive sintered ore composition targets, wherein the comprehensive sintered ore composition target includes multiple sintered ore composition targets.

[0081] In this embodiment, during multiple specified time periods, raw material analysis sub-data (chemical composition data of the raw materials in each time period), sintering process sub-data (process parameter data in each time period, such as temperature, pressure, etc.) and finished product sub-data (quality data of the finished sintered ore in each time period) are collected respectively.

[0082] In this embodiment, all collected sub-data are pre-processed (such as denoising, outlier processing, missing data completion, etc.), and the sub-data within all time periods are integrated to form historical raw material testing data, historical sintering process data and historical finished product data.

[0083] In this embodiment, ingredient information is extracted from historical finished product data, and the number of occurrences of each ingredient in all time periods is counted. If the frequency of occurrence of a certain ingredient is lower than a set threshold (i.e., it appears in a few time periods), it is marked as a "discarded ingredient."

[0084] In this embodiment, all finished product components that are not marked as discarded components are determined as "comprehensive sinter composition targets". These targets are composed of multiple specific sinter composition targets (such as Fe content, CaO / SiO2 ratio, etc.).

[0085] The beneficial effects of the above technical solution are: obtaining historical raw material analysis data, historical sintering process data and historical finished product data within multiple historical specified time periods, determining multiple sintered ore composition targets, improving data quality and analysis efficiency, and improving data basis for determining important parameter vectors for each sintered ore composition target.

[0086] Embodiment 3:

[0087] The embodiment of the present invention provides a real-time correction method for sintering raw material test data and raw material ratio, which analyzes historical raw material test data and historical sintering process data to determine the important parameter vector of each sintered ore composition target, and determines the contribution algorithm of the sintered ore composition target corresponding to the important parameter vector, including:

[0088] Based on the physical and chemical reactions in the sintering process, a first correlation relationship between each sintered ore composition target and the historical raw material analysis sub-data of each historical specified time period is analyzed, and at the same time, a second correlation relationship between each sintered ore composition target and the historical sintering process sub-data of each historical specified time period is analyzed;

[0089] Determine a first sub-important parameter vector of each sintered ore composition target based on the sintered raw material based on all parameters whose correlation exceeds a first preset correlation threshold value in a first correlation relationship between each sintered ore composition target and all historical raw material analysis sub-data of a historical specified time period;

[0090] Determine a second sub-important parameter vector of each sintered ore composition target based on the sintering process based on all parameters whose correlation exceeds a second preset correlation threshold value in a second correlation relationship between each sintered ore composition target and all historical raw material analysis sub-data of a historical specified time period;

[0091] Determine an important parameter vector for each sinter composition target based on a first sub-important parameter vector of the sinter raw material and a second sub-important parameter vector of the sinter process for each sinter composition target;

[0092] Select multiple algorithms and determine the important parameter vectors for each sinter composition target based on the fitting contribution of each algorithm;

[0093] The algorithm with the lowest fitting contribution is selected as the contribution algorithm corresponding to the sintered ore composition target.

[0094] In this embodiment, by analyzing the physical and chemical reactions in the sintering process, the correlation (such as correlation coefficient or other mathematical indicators) between each sintered ore composition target and the historical raw material analysis sub-data of each historical time period is calculated to evaluate the direct impact of the raw material composition on the sintered ore composition target.

[0095] In this embodiment, the correlation between the sintered ore composition target and the historical sintering process sub-data of each historical time period is analyzed to evaluate the indirect or direct effect of the process parameters on the sintered ore composition target.

[0096] In this embodiment, based on the first correlation relationship of all historical time periods, all raw material parameters (such as Fe and SiO2 content in the raw materials) whose correlation exceeds the first preset correlation threshold are screened out, and these parameters are combined into a first sub-important parameter vector, which represents the set of sintered ore composition targets directly affected by the raw material parameters.

[0097] In this embodiment, based on the second correlation relationship of all historical time periods, process parameters (such as sintering temperature, mixing time, etc.) whose correlation exceeds the second preset correlation threshold are screened out, and these parameters are combined into a second sub-important parameter vector, which represents the set of sintered ore composition targets affected by process conditions.

[0098] In this embodiment, the first sub-important parameter vector (related to raw materials) and the second sub-important parameter vector (related to process) are integrated to form a complete important parameter vector for each sintered ore composition target, comprehensively considering the impact of raw materials and process on the sintering target.

[0099] In this embodiment, multiple algorithms (such as linear regression, random forest, neural network, etc.) are selected to fit the important parameter vectors of each sintered ore composition target, and the fitting contribution (i.e., the explanatory power or prediction accuracy of each algorithm for the target parameter) is calculated.

[0100] In this embodiment, the algorithm with the lowest fitting contribution is selected from these algorithms as the contributing algorithm corresponding to the sintered ore composition target, so as to reflect the relationship between the parameter and the target more conservatively and stably.

[0101] In this embodiment, the important parameter vector of each sintered ore composition target is determined based on the fitting contribution of each algorithm, including determining the contribution of each parameter in the important parameter vector to the sintered ore composition target, and determining the fitting contribution of the sintered ore composition target according to the contribution of all parameters in the important parameter vector to the sintered ore composition target.

[0102] In this embodiment, the lowest fitting contribution means that the fitting contribution calculated by the algorithm is more consistent with the actual importance of each parameter in actual production, and can more accurately describe the impact of key parameters on the sintered ore composition target.

[0103] The beneficial effects of the above technical solution are: analyzing historical raw material analysis data and historical sintering process data to determine the important parameter vectors of each sintered ore composition target, and determining the contribution algorithm of the sintered ore composition target corresponding to the important parameter vectors, which can improve the scientific nature of analysis and optimization and ensure the robustness and applicability of the contribution algorithm.

[0104] Embodiment 4:

[0105] The embodiment of the present invention provides a method for real-time correction of sintering raw material test data and raw material ratio, which determines the influencing parameter vector of each sintering ore composition target based on the prediction model of each sintering ore composition target, including:

[0106] Based on the output results of the prediction model for each sinter composition target, determining the influence of each parameter in the important parameter vector on the corresponding sinter composition target;

[0107] Sorting the influence of all parameters in the important parameter vector of each sinter composition target on the sinter composition target from large to small, and determining the initial influencing parameter vector of each sinter composition target based on the sorting result;

[0108] Each parameter of the initial influencing parameter vector of each sintered ore composition target is judged, if the parameter is a state parameter, the parameter is replaced by the raw material feeding parameter of the parameter, if the parameter is an operation parameter, no replacement is performed;

[0109] Based on the initial influencing parameter vector after each sinter ore composition target is replaced, an influencing parameter vector for each sinter ore composition target is determined.

[0110] In this embodiment, a prediction model for each sintered ore composition target is used, and the model output results are used to calculate the influence of each parameter in the important parameter vector (such as Fe and SiO2 content in the raw materials or process parameters such as temperature and mixing time) on the target composition. The influence can be quantified by statistical methods (such as feature importance, sensitivity analysis or Shapley value, etc.), and the influence of each parameter on the target composition is expressed in numerical form.

[0111] In this embodiment, all parameters are sorted from large to small in terms of their influence on the target sintered ore composition, with priority given to key parameters that play a decisive role in the target composition. Based on the sorting results, the most important parameters are screened out to form an initial influencing parameter vector for each sintered ore composition target.

[0112] In this embodiment, the state parameters describe the characteristics or chemical composition of the sintering raw material components (such as Fe content, Al2O3 content, etc.). These parameters cannot be directly optimized, but can be indirectly adjusted through the corresponding raw material feeding parameters (such as the feeding amount of iron ore powder and limestone). Therefore, the state parameters need to be replaced by their corresponding raw material feeding parameters in the initial influencing parameter vector.

[0113] In this embodiment, the operating parameters describe the parameters of the sintering process conditions (such as sintering temperature, sintering speed, etc.), and these parameters can be adjusted directly, so there is no need to replace them.

[0114] In this embodiment, the substitution process ensures the operability of the optimization vector, the raw material composition is indirectly controlled by adjusting the feed amount, and the process parameters are directly used as influencing variables.

[0115] In this embodiment, each parameter of the initial influencing parameter vector of each sintered ore composition target is judged. If the parameter is a state parameter, the raw material feeding parameter of the parameter is used to replace the parameter. When there is one raw material feeding parameter of the parameter, replacement is performed. When there are multiple raw material feeding parameters of the parameter, the parameter in the initial influencing parameter vector is replaced with the first raw material feeding parameter, and the remaining raw material feeding parameters of the parameter are inserted between the first raw material feeding parameter after replacement and the next parameter of the parameter before replacement.

[0116] The beneficial effects of the above technical solution are as follows: based on the prediction model of each sintered ore composition target, the influencing parameter vector of each sintered ore composition target is determined, the actual operability of parameter adjustment can be enhanced, and the key parameters that have a great impact on the sintered ore composition target can be accurately screened out, providing an efficient way to achieve accurate optimization of the sintered ore composition target and stable quality.

[0117] Embodiment 5:

[0118] The embodiment of the present invention provides a method for real-time correction of sintering raw material test data and raw material ratio, which collects current raw material test data within a current specified time period, and determines current test correction data based on the current raw material test data and all prediction models, including:

[0119] Preprocessing the collected current raw material test data of the current specified time period, and determining the current parameter vector of each sinter ore composition target based on the preprocessed current raw material test data and the influencing parameter vector of each sinter ore composition target;

[0120] Inputting the current parameter vector of each sinter composition target into the corresponding prediction model, determining the actual sinter composition value of each sinter composition target, and determining the actual sinter vector based on the actual sinter composition values ​​of all sinter composition targets;

[0121] Determine the theoretical sinter composition value of each sinter composition target based on the current raw material test data and the raw material ratio, and determine the theoretical sinter vector based on the theoretical sinter composition values ​​of all sinter composition targets;

[0122] Determine the sinter difference matrix based on the accuracy requirements of all sinter composition targets and the theoretical sinter composition vector;

[0123] ;

[0124] in, represents the sinter difference matrix, represents the theoretical lower limit value of the first sintered ore composition target, the i-th sintered ore composition target, and the N1-th sintered ore composition target, represents the theoretical upper limit of the first sinter ore composition target, the i-th sinter ore composition target, and the N1-th sinter ore composition target, and N1 represents the number of sinter ore composition targets;

[0125] Determine whether each actual sintered ore component value in the actual sintered ore vector is between the lower limit value and the upper limit value of the corresponding sintered ore component target in the sintered ore difference matrix. If any actual sintered ore component value is not between the lower limit value and the upper limit value of the corresponding sintered ore component target, adjust the current raw material test data and determine the adjusted actual sintered ore vector; until all actual sintered ore component values ​​in the adjusted actual sintered ore vector are between the lower limit value and the upper limit value of the corresponding sintered ore component target in the sintered ore difference matrix, record the number of adjustments;

[0126] Determine the current raw material test data in which all actual sinter ore composition values ​​in the actual sinter ore vector are between the lower limit value and the upper limit value of the corresponding sinter ore composition target in the sinter ore difference matrix as the corrected raw material test data;

[0127] The correction value of each raw material is determined based on the corrected raw material assay data and the current raw material assay data, and the current assay correction data is determined based on the correction values ​​of all raw materials and the number of adjustments.

[0128] In this embodiment, the raw material test data collected in the specified time period is preprocessed to clean invalid data (such as missing values, abnormal values, etc.).

[0129] In this embodiment, the preprocessed data and the influencing parameter vector of each sintered ore component target are used to screen out the parameters related to each component target and determine the current parameter vector of each sintered ore component target.

[0130] In this embodiment, the current parameter vector of each sintered ore composition target is input into its corresponding prediction model, and the model outputs the actual sintered ore composition value of the target, summarizes the actual sintered ore composition values ​​of all targets, determines the overall actual sintered ore vector, and describes the comprehensive composition of the sintered ore under current conditions.

[0131] In this embodiment, based on the current raw material test data and raw material ratio, the theoretical sintered ore composition value of each component target is calculated to reflect the composition characteristics under the ideal ratio and parameters, and all theoretical composition values ​​are summarized to form a theoretical sintered ore vector, which describes the composition that the sintered ore should have under ideal conditions.

[0132] In this embodiment, a sinter difference matrix is ​​constructed by combining the accuracy requirement of the sinter composition target and the theoretical sinter vector.

[0133] In this embodiment, for each actual component value in the actual sintered ore vector, it is checked whether it falls within the upper and lower limits of the corresponding component target in the sintered ore difference matrix. If an actual component value exceeds the allowable range (unqualified), it is necessary to adjust the current raw material test data and recalculate the actual sintered ore component value until all actual component values ​​are qualified.

[0134] In this embodiment, the actual sintered ore vector is gradually optimized by adjusting the raw material test data (such as adjusting the component value or proportion of a certain raw material in the raw material test data with a large impact according to the prediction model). The actual sintered ore vector is recalculated after each adjustment until all component values ​​meet the requirements, and the number of adjustments is recorded.

[0135] In this embodiment, when all actual component values ​​in the actual sintered ore vector are within the upper and lower limits of the difference matrix, the raw material assay data at this time is determined to be the corrected raw material assay data.

[0136] In this embodiment, the corrected raw material test data is compared with the current raw material test data, and the correction value of each raw material is calculated to represent the deviation of the data before and after correction. Based on the correction values ​​of all raw materials and the number of adjustments, the final current test correction data is determined.

[0137] The beneficial effects of the above technical solution are: collecting the current raw material test data within the current specified time period, determining the current test correction data based on the current raw material test data and all prediction models, and realizing dynamic adjustment of the closed-loop feedback between theory and practice, automatic correction and optimization, and realizing efficient unification of theory and practice.

[0138] Embodiment 6:

[0139] The embodiment of the present invention provides a method for real-time correction of sintering raw material test data and raw material ratio, which determines a feedback value based on current test correction data, and pushes feedback to the raw material source based on the feedback value, including:

[0140] Get historical test correction data for multiple historical specified time periods before the current specified time period;

[0141] Determining a feedback value based on current assay correction data for a current specified time period and historical assay correction data for a plurality of historical specified time periods;

[0142] If the feedback value exceeds the feedback threshold, feedback is determined based on the feedback value and the current assay correction data for the current specified time period, and the feedback is sent to the source of the raw materials;

[0143] Generates a correction report based on the feedback value, the influencing parameter vector, the number of adjustments, and the current assay correction data for the current specified time period.

[0144] In this embodiment, historical assay correction data of multiple historical time periods before the current specified time period are collected.

[0145] In this embodiment, a reasonable feedback threshold is set to represent the acceptable range of the correction data deviation. If the feedback value exceeds the feedback threshold, it indicates that there is a large fluctuation or problem in the raw material quality or the test data.

[0146] In this embodiment, feedback opinions are generated for the raw material source according to the feedback value and the current test correction data. The content of the feedback opinions may include: quality deviation analysis of the raw material source (such as excessive Fe content or alkalinity in iron ore powder); abnormal conditions of data correction frequency or adjustment times; and recommended improvement measures (such as strengthening raw material screening and stabilizing supply quality).

[0147] In this embodiment, feedback is sent to the source of the raw materials so that the other party can adjust the quality of the raw materials in a timely manner to reduce fluctuations in subsequent sintering production.

[0148] The beneficial effects of the above technical solution are: the feedback value is determined based on the current test correction data, and feedback is pushed to the source of raw materials based on the feedback value. An active feedback mechanism can be determined, which realizes the pre-positioning and intelligentization of quality control management and improves the stability and efficiency of sintered ore production.

[0149] Embodiment 7:

[0150] The embodiment of the present invention provides a method for real-time correction of sintering raw material test data and raw material ratio, which determines a feedback value based on current test correction data of a current specified time period and historical test correction data of multiple historical specified time periods, including:

[0151] ;

[0152] ;

[0153] in, Indicates the feedback value, Indicates the current feedback value of the current specified time period, It indicates the historical feedback value of all historical specified time periods, N2 indicates the number of raw materials in the current test correction data, Indicates the correction value of the jth raw material in the current assay correction data for the current specified time period, represents the weight of the correction value of the jth raw material, Indicates the number of adjustments for the current specified time period. Indicates the weight of the number of adjustments in the current specified time period. represents the weight of the absolute difference of the i-th sinter composition target, It represents the absolute difference between the actual sinter composition value and the theoretical sinter composition value of the i-th sinter composition target in the current specified time period based on the corrected raw material test data. represents the first amplification factor of the absolute difference, represents the second amplification factor of the absolute difference, represents the attenuation factor of the historical specified time period, T represents the number of historical specified time periods, Indicates the historical correction value of the jth raw material in the historical test correction data of the tth historical specified time period, The weight representing the number of adjustments in the t-th historical specified time period, Indicates the number of adjustments in the tth historical specified time period, It represents the absolute difference between the historical sinter composition value and the theoretical sinter composition value based on the historical corrected raw material analysis data for the i-th sinter composition target in the t-th historical specified time period. Represents the historical smoothing factor.

[0154] In this embodiment, The larger it is, the more it weakens the abnormal contribution of the historical specified period to the feedback value.

[0155] The beneficial effects of the above technical solution are as follows: the feedback value is determined based on the current assay correction data of the current specified time period and the historical assay correction data of multiple historical specified time periods, which can improve the data quality and the scientific nature of the analysis and optimization, and realize the construction of a personalized and precise sintered ore composition target prediction model, ensuring the scientific nature and reliability of the prediction results, realizing the dynamic adjustment of theoretical and practical closed-loop feedback as well as automated correction and optimization, realizing the pre-positioning and intelligence of quality control management, and improving the stability and efficiency of sintered ore production.

[0156] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0157] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0158] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for real-time correction of sintering raw material test data and raw material ratio, characterized in that: include: 101: Acquire historical raw material test data, historical sintering process data, and historical finished product data within multiple historical specified time periods, and determine multiple sinter ore composition targets; 102: Analyze historical raw material test data and historical sintering process data to determine the important parameter vector of each sintered ore composition target, and determine the contribution algorithm of the sintered ore composition target corresponding to the important parameter vector; 103: Based on the contribution algorithm of each sinter composition target, historical raw material analysis data and historical sintering process data, a prediction model for each sinter composition target is constructed; 104: Determine an influencing parameter vector of each sinter ore composition target based on a prediction model of each sinter ore composition target; 105: Collecting current raw material test data within a current specified time period, and determining current test correction data based on the current raw material test data and all prediction models; 106: Acquire multiple historical test correction data, determine a feedback value based on the current test correction data and the multiple historical test correction data, and generate a correction report; Among them, based on the contribution algorithm of each sinter composition target, historical raw material test data and historical sintering process data, a prediction model for each sinter composition target is constructed, including: Inputting the important parameter vector of each sinter composition target into the prediction model of the corresponding sinter composition target; The prediction models of all sintered ore composition targets are trained according to the historical raw material analysis data and the historical sintering process data, and the prediction models of the corresponding sintered ore composition targets are constructed based on the contribution algorithm of each sintered ore composition target.

2. A method for real-time correction of sintering raw material test data and raw material ratio according to claim 1, characterized in that: Obtain historical raw material analysis data, historical sintering process data, and historical finished product data within multiple historical specified time periods to determine multiple sinter ore composition targets, including: Obtain historical raw material analysis sub-data, historical sintering process sub-data and historical finished product sub-data within each historical specified time period; Determine historical raw material test data based on historical raw material test sub-data within all historical specified time periods after preprocessing, determine historical sintering process data based on historical sintering process sub-data within all historical specified time periods after preprocessing, and at the same time, determine historical finished product data based on historical finished product sub-data within all historical specified time periods after preprocessing; Extract the finished product components whose appearance times in all historical finished product sub-data are lower than a preset threshold from the historical finished product data and mark them as discarded components; All finished product components that are not marked as discarded components in the historical finished product data are determined as comprehensive sintered ore composition targets, wherein the comprehensive sintered ore composition target includes multiple sintered ore composition targets.

3. The method for real-time correction of sintering raw material test data and raw material ratio according to claim 1, characterized in that: Analyze historical raw material test data and historical sintering process data to determine the important parameter vectors of each sinter composition target, and determine the contribution algorithm of the sinter composition target corresponding to the important parameter vectors, including: Based on the physical and chemical reactions in the sintering process, a first correlation relationship between each sintered ore composition target and the historical raw material analysis sub-data of each historical specified time period is analyzed, and at the same time, a second correlation relationship between each sintered ore composition target and the historical sintering process sub-data of each historical specified time period is analyzed; Determine a first sub-important parameter vector of each sintered ore composition target based on the sintered raw material based on all parameters whose correlation exceeds a first preset correlation threshold value in a first correlation relationship between each sintered ore composition target and all historical raw material analysis sub-data of a historical specified time period; Determine a second sub-important parameter vector of each sintered ore composition target based on the sintering process based on all parameters whose correlation exceeds a second preset correlation threshold value in a second correlation relationship between each sintered ore composition target and all historical raw material analysis sub-data of a historical specified time period; Determine an important parameter vector for each sinter composition target based on a first sub-important parameter vector of the sinter raw material and a second sub-important parameter vector of the sinter process for each sinter composition target; Select multiple algorithms and determine the important parameter vectors for each sinter composition target based on the fitting contribution of each algorithm; The algorithm with the lowest fitting contribution is selected as the contribution algorithm corresponding to the sintered ore composition target.

4. The method for real-time correction of sintering raw material test data and raw material ratio according to claim 1, characterized in that: Based on the prediction model of each sinter composition target, the influencing parameter vector of each sinter composition target is determined, including: Based on the output results of the prediction model for each sinter composition target, determining the influence of each parameter in the important parameter vector on the corresponding sinter composition target; Sorting the influence of all parameters in the important parameter vector of each sinter composition target on the sinter composition target from large to small, and determining the initial influencing parameter vector of each sinter composition target based on the sorting result; Each parameter of the initial influencing parameter vector of each sintered ore composition target is judged, if the parameter is a state parameter, the parameter is replaced by the raw material feeding parameter of the parameter, if the parameter is an operation parameter, no replacement is performed; Based on the initial influencing parameter vector after each sinter ore composition target is replaced, an influencing parameter vector for each sinter ore composition target is determined.

5. The method for real-time correction of sintering raw material test data and raw material ratio according to claim 1, characterized in that: Collect the current raw material test data within the current specified time period, and determine the current test correction data based on the current raw material test data and all prediction models, including: Preprocessing the collected current raw material test data of the current specified time period, and determining the current parameter vector of each sinter ore composition target based on the preprocessed current raw material test data and the influencing parameter vector of each sinter ore composition target; Inputting the current parameter vector of each sinter composition target into the corresponding prediction model, determining the actual sinter composition value of each sinter composition target, and determining the actual sinter vector based on the actual sinter composition values ​​of all sinter composition targets; Determine the theoretical sinter composition value of each sinter composition target based on the current raw material test data and the raw material ratio, and determine the theoretical sinter vector based on the theoretical sinter composition values ​​of all sinter composition targets; Determine the sinter difference matrix based on the accuracy requirements of all sinter composition targets and the theoretical sinter composition vector; ； in, represents the sinter difference matrix, represents the theoretical lower limit value of the first sintered ore composition target, the i-th sintered ore composition target, and the N1-th sintered ore composition target, represents the theoretical upper limit of the first sinter ore composition target, the i-th sinter ore composition target, and the N1-th sinter ore composition target, and N1 represents the number of sinter ore composition targets; Determine whether each actual sintered ore component value in the actual sintered ore vector is between the lower limit value and the upper limit value of the corresponding sintered ore component target in the sintered ore difference matrix. If any actual sintered ore component value is not between the lower limit value and the upper limit value of the corresponding sintered ore component target, adjust the current raw material test data and determine the adjusted actual sintered ore vector; until all actual sintered ore component values ​​in the adjusted actual sintered ore vector are between the lower limit value and the upper limit value of the corresponding sintered ore component target in the sintered ore difference matrix, record the number of adjustments; Determine the current raw material test data in which all actual sinter ore composition values ​​in the actual sinter ore vector are between the lower limit value and the upper limit value of the corresponding sinter ore composition target in the sinter ore difference matrix as the corrected raw material test data; The correction value of each raw material is determined based on the corrected raw material assay data and the current raw material assay data, and the current assay correction data is determined based on the correction values ​​of all raw materials and the number of adjustments.

6. A method for real-time correction of sintering raw material test data and raw material ratio according to claim 5, characterized in that: Determine the feedback value based on the current test correction data, and push feedback to the raw material source based on the feedback value, including: Get historical test correction data for multiple historical specified time periods before the current specified time period; Determining a feedback value based on current assay correction data for a current specified time period and historical assay correction data for a plurality of historical specified time periods; If the feedback value exceeds the feedback threshold, feedback is determined based on the feedback value and the current assay correction data for the current specified time period, and the feedback is sent to the source of the raw materials; Generates a correction report based on the feedback value, the influencing parameter vector, the number of adjustments, and the current assay correction data for the current specified time period.

7. A method for real-time correction of sintering raw material test data and raw material ratio according to claim 6, characterized in that: Determining a feedback value based on current assay correction data for a current specified time period and historical assay correction data for a plurality of historical specified time periods includes: ； ； in, Indicates the feedback value, Indicates the current feedback value of the current specified time period, It indicates the historical feedback value of all historical specified time periods, N2 indicates the number of raw materials in the current test correction data, Indicates the correction value of the jth raw material in the current assay correction data for the current specified time period, represents the weight of the correction value of the jth raw material, Indicates the number of adjustments for the current specified time period. Indicates the weight of the number of adjustments in the current specified time period. represents the weight of the absolute difference of the i-th sinter composition target, It represents the absolute difference between the actual sinter composition value and the theoretical sinter composition value of the i-th sinter composition target in the current specified time period based on the corrected raw material test data. represents the first amplification factor of the absolute difference, represents the second amplification factor of the absolute difference, represents the attenuation factor of the historical specified time period, T represents the number of historical specified time periods, Indicates the historical correction value of the jth raw material in the historical test correction data of the tth historical specified time period, The weight representing the number of adjustments in the t-th historical specified time period, Indicates the number of adjustments in the tth historical specified time period, It represents the absolute difference between the historical sinter composition value and the theoretical sinter composition value based on the historical corrected raw material analysis data for the i-th sinter composition target in the t-th historical specified time period. Represents the historical smoothing factor.

Images