Natural gas consumption prediction method and device and knowledge graph construction method and device

Abstract

The invention discloses a natural gas consumption prediction method and device and a knowledge graph construction method and device. The method comprises the following steps: acquiring initial historical data of a target production enterprise in a plurality of sampling periods within a preset time period, and correcting the initial historical data of each sampling period to obtain corresponding corrected historical data; the corrected historical data comprises historical steel yield production data and corresponding historical natural gas consumption data; according to the corrected historical data of each sampling period, constructing a producer function corresponding to the target production enterprise; the producer function represents the natural gas consumption required for producing the steel product; and based on an output result of a pre-constructed steel yield prediction model of the target production enterprise, solving the producer function, and taking a solving result as a predicted value of the natural gas consumption of the target production enterprise. And the natural gas consumption required by steel product production of a production enterprise can be predicted more accurately.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of new energy technology, in particular to a method and device for predicting the consumption of natural gas and constructing a knowledge graph. BACKGROUND

[0002] The steel industry has a very large demand for energy, and is a high energy-consuming industry. Its main energy supply includes coal, electricity, natural gas, etc. With the global environment deteriorating, it is urgent to replace heavy-pollution energy sources such as coal and oil with clean energy. The use of natural gas, a relatively clean and low-carbon energy source, in the steel industry is increasing. Therefore, each production enterprise in the steel industry has a very large demand for natural gas. Detailed analysis of the gas (natural gas) characteristics of these production enterprises can help optimize natural gas supply, improve service levels, and meet enterprise demand, which is crucial for natural gas suppliers and can ensure sustainable cooperation. Therefore, accurate demand prediction for gas for production enterprise customers is beneficial to ensuring gas supply and demand, and is of great significance for the optimal scheduling of pipe networks.

[0003] In the prior art, the use of gas quota analysis method or the use of time series model is mainly used to predict the demand for natural gas of production enterprises. However, the analysis method used in the prior art has analysis errors in principle, and the actual steel production of production enterprises is not considered, so the accuracy of the results is low. SUMMARY

[0004] The steel industry is an industrial industry mainly engaged in the production activities of black metal mineral mining and selection and black metal smelting and processing, including mineral mining of metal iron, chromium, manganese, etc., iron smelting industry, steel industry, steel processing industry, ferroalloy smelting industry, steel wire and product industry, etc. It is one of the important raw material industries in the country. The upstream of the steel industry carries non-ferrous metal, electricity and coal industry, and the downstream connects machinery, real estate, home appliances, light industry, automobile, shipbuilding and other industries. The steel industry is a high energy-consuming industry with a very large demand for energy, and its main energy supply includes coal, electricity, natural gas, etc. With the global environment deteriorating, it is urgent to replace heavy-pollution energy sources such as coal and oil with clean energy. Natural gas, as a clean energy source, is increasingly attracting attention. Therefore, with the increasing pressure of energy saving and emission reduction and the continuous promotion of a low-carbon society, the use of natural gas, a relatively clean and low-carbon energy source, in the steel industry is increasing. In the prior art, the demand for gas of production enterprise customers is generally predicted by using gas quota analysis method or using time series model. However, due to the influence of gas fluctuation factors of production enterprise customers, the accuracy of the prediction of the demand for natural gas of production enterprise customers is low.

[0005] In view of the above problems, the present application is proposed in order to provide a natural gas consumption prediction method and device which overcomes the above problems or at least partially solves the above problems.

[0006] In a first aspect, the embodiments of the present application provide a natural gas consumption prediction method, comprising:

[0007] Obtaining initial historical data of a target production enterprise in a plurality of sampling periods within a preset time period, correcting the initial historical data of each sampling period to obtain corresponding corrected historical data; the corrected historical data includes historical steel production data and corresponding historical natural gas consumption data;

[0008] According to the corrected historical data of each sampling period, a producer function corresponding to the target production enterprise is constructed; the producer function represents the natural gas consumption required for producing steel products;

[0009] Based on the output result of the steel production prediction model of the target production enterprise constructed in advance, the producer function is solved, and the solving result is taken as the prediction value of the natural gas consumption of the target production enterprise.

[0010] In one embodiment, the producer function corresponding to the target production enterprise is constructed according to the corrected historical data of each sampling period, comprising:

[0011] Defining a producer function formula:

[0012]

[0013] In the above formula, y i represents the historical steel production data of the target production enterprise, in units of sampling periods; G i represents the historical natural gas consumption data of the target production enterprise; A i represents the technical level of the target production enterprise; γ i represents the substitution elasticity parameter of natural gas; α i represents the degree of dependence of the target production enterprise on natural gas; Constant represents a preset constant.

[0014] The historical steel production data and the corresponding historical natural gas consumption data of each sampling period are substituted into the above producer function formula to solve the coefficients A i , α i , γ i and Constant; the coefficients A i , α i , γ iand Constant into the above producer function formula, to obtain a producer function, the producer function being a function of gas consumption of natural gas with respect to variable y i .

[0015] In one embodiment, the historical steel production data includes historical steel production data or historical crude steel production data; the historical natural gas consumption data includes historical crude steel-natural gas consumption data or historical steel material-natural gas consumption data, the historical steel material-natural gas consumption data corresponding to the historical steel production data, and the historical crude steel-natural gas consumption data corresponding to the historical crude steel production data;

[0016] The historical steel production data and the corresponding historical natural gas consumption data of each sampling period are substituted into the above producer function formula to solve coefficients A i , α i , γ i , and Constant; and the solved coefficients A i , α i , γ i , and Constant are substituted into the above producer function formula, including:

[0017] The historical steel production data and the historical steel material-natural gas consumption data of each sampling period are substituted into the producer function formula to solve coefficients A i , α i , γ i , and Constant; and the solved coefficients A i , α i , γ i , and Constant are substituted into the above producer function formula to obtain a steel material producer function formula;

[0018] Or

[0019] The historical crude steel production data and the historical crude steel-natural gas consumption data of each sampling period are substituted into the producer function formula to solve coefficients A i , α i , γ i , and Constant; and the solved coefficients A i , α i , γ i , and Constant are substituted into the above producer function formula to obtain a crude steel producer function formula;

[0020] Correspondingly, the obtaining of the producer function includes

[0021] Determine the fitting degrees of the steel material producer function formula and the crude steel producer function formula respectively, and determine the producer function formula according to the comparison result of the fitting degrees.

[0022] In one embodiment, the determining of the fitting degrees of the steel material producer function formula and the crude steel producer function formula respectively, and the determining of the producer function formula according to the comparison result of the fitting degrees, comprises:

[0023] If the fitting degree of the steel material producer function formula is not lower than the fitting degree of the crude steel producer function formula, the steel material producer function formula is determined as the producer function formula, and correspondingly, the historical steel material production data is used as the historical steel production data, and the historical steel material-natural gas consumption data is used as the historical natural gas consumption data.

[0024] If the fitting degree of the steel material producer function formula is lower than the fitting degree of the crude steel producer function formula, the crude steel producer function formula is determined as the producer function formula, and correspondingly, the historical crude steel production data is used as the historical steel production data, and the historical crude steel-natural gas consumption data is used as the historical natural gas consumption data.

[0025] In one embodiment, the steel production prediction model corresponding to the target production enterprise is constructed by the following way:

[0026] Based on the historical steel production data, a comprehensive influence factor combination corresponding to the historical steel production data is determined from a preset influence factor set, and the influence factor set comprises at least one influence factor; the influence factor represents a factor determined based on an upstream and downstream industrial chain of the target production enterprise and influencing the steel production of the target production enterprise.

[0027] The historical steel production data and the numerical values of each influence factor in the corresponding comprehensive influence factor combination are used as sample data to obtain a sample data set.

[0028] The sample data set is input into a preset steel production prediction model for training to obtain a trained steel production prediction model, and the steel production prediction model can obtain a steel production prediction value of the target production enterprise based on the numerical values of several input influence factors.

[0029] In one embodiment, the determining of the comprehensive influence factor combination corresponding to the historical steel production data from the preset influence factor set comprises:

[0030] For each factor in the set of factors, a fitting degree of each factor to the historical steel production data is calculated according to the numerical value of each factor and the historical steel production data, and a single-factor influence value of each factor is determined according to the fitting degree of each factor to the historical steel production data, the single-factor influence value representing an influence degree of the factor on the target production enterprise's steel production;

[0031] The factors in the set of factors are sorted according to the single-factor influence values in descending order, and a preset number of factors from the maximum value are selected and reserved to obtain a seed factor combination.

[0032] Based on the historical steel production data and the numerical values of the factors in the second set of factors, a comprehensive factor combination is determined.

[0033] In one embodiment, the calculation of the fitting degree of each factor to the historical steel production data according to the numerical value of each factor and the historical steel production data, and the determination of the single-factor influence value of each factor according to the fitting degree of each factor to the historical steel production data, comprises:

[0034] The numerical values of the factors and the historical steel production data are fitted respectively by using a relationship fitting algorithm, and each fitting result is taken as a single-factor influence value of the corresponding factor.

[0035] In one embodiment, the calculation of the fitting degree of each factor to the historical steel production data according to the numerical value of each factor and the historical steel production data, and the determination of the single-factor influence value of each factor according to the fitting degree of each factor to the historical steel production data, comprises:

[0036] For each factor, at least one relationship fitting algorithm is used to fit the numerical value of the factor and the historical steel production data respectively, and each fitting result is taken as a single-factor influence value of the factor corresponding to each relationship fitting algorithm.

[0037] In the single-factor influence values of the factors corresponding to various relationship fitting algorithms, the single-factor influence value with the largest numerical value is selected as the single-factor influence value of the factor.

[0038] In one embodiment, the determination of the comprehensive factor combination based on the historical steel production data and the numerical values of the factors in the seed factor combination comprises:

[0039] matching each of the factors in the factor set except any factor in the seed factor combination with the seed factor combination respectively to obtain a plurality of candidate factor combinations;

[0040] for each candidate factor combination, determining a combined factor influence value of the candidate factor combination based on the historical steel production data by using a relationship fitting algorithm, the combined factor influence value of the candidate factor combination representing an influence degree of each factor in the candidate factor combination on the target production enterprise steel production;

[0041] comparing the influence value of the combination with the highest combined factor influence value in the candidate factor combinations with the influence value of the seed factor combination, and updating the factor combination according to the comparison result;

[0042] repeating the step of updating the factor combination until each factor in the factor set is traversed, and stopping iteration.

[0043] In an embodiment, the determining of the combined factor influence value of the candidate factor combination based on the historical steel production data by using the relationship fitting algorithm comprises:

[0044] fitting the historical steel production data with the influence value corresponding to each factor in the candidate factor combination respectively by using the relationship fitting algorithm to obtain a second fitting value corresponding to each factor in the candidate factor combination, and selecting the maximum second fitting value as the combined factor influence value of the candidate factor combination.

[0045] In an embodiment, the initial historical data comprises initial production data and corresponding initial natural gas consumption data.

[0046] The correction processing of the initial historical data of each sampling period to obtain corresponding corrected historical data comprises:

[0047] obtaining a steel product production reduction of the target production enterprise during the running abnormal event in the preset time period;

[0048] summing the steel product production reduction and the initial production data as historical steel production data;

[0049] summing the steel product production reduction and the initial natural gas consumption data as historical natural gas consumption data.

[0050] In a second aspect, an embodiment of the present application provides a knowledge graph construction method, comprising:

[0051] obtain a producer function and a steel production prediction model of a target production enterprise; the producer function and the steel production prediction model are obtained by the aforementioned method for predicting the natural gas consumption;

[0052] construct an ontology relationship model of the producer function and the steel production prediction model;

[0053] extract entity, attribute and relationship information from historical production data of the target production enterprise;

[0054] construct a knowledge graph based on the obtained ontology relationship model and the extracted entity, attribute and relationship information; the knowledge graph represents the relationship between the predicted steel production value and the corresponding predicted natural gas consumption value of the target production enterprise.

[0055] In a third aspect, an embodiment of the present application provides a device for predicting natural gas consumption, comprising:

[0056] an obtaining module, configured to obtain initial historical data of a target production enterprise in a plurality of sampling periods in a preset time period, and correct the initial historical data in each sampling period to obtain corresponding corrected historical data;

[0057] a constructing module, configured to construct a producer function corresponding to the target production enterprise according to the corrected historical data in each sampling period; the corrected historical data includes historical steel production data and corresponding historical natural gas consumption data; the producer function represents the natural gas consumption required for producing steel products;

[0058] a solving module, configured to solve the producer function based on an output result of a steel production prediction model corresponding to the target production enterprise, which is constructed in advance, and take the solving result as a predicted value of the natural gas consumption of the target production enterprise.

[0059] In a fourth aspect, an embodiment of the present application provides a computer storage medium, which stores computer executable instructions, and the computer executable instructions are executed by a processor to implement the aforementioned method for predicting natural gas consumption.

[0060] In a fourth aspect, an embodiment of the present application provides a terminal device, which comprises a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor implements the aforementioned method for predicting natural gas consumption when executing the program.

[0061] The above technical solution provided by the embodiment of the present application has at least the following beneficial effects:

[0062] The method for predicting the gas consumption of natural gas provided by the embodiment of the present application collects initial historical data of a target production enterprise in a plurality of sampling periods in a preset time period, and corrects the initial historical data of each sampling period to obtain the corrected initial historical data of each sampling period, that is, the historical steel production data of each sampling period and the corresponding historical natural gas consumption data. Based on the historical steel production data of each sampling period and the corresponding historical natural gas consumption data, a producer function capable of representing the natural gas consumption required by the target production enterprise for producing steel products is constructed, and the producer function is solved according to the output result of the steel production prediction model of the target production enterprise constructed in advance, and the result of the solving is the predicted natural gas consumption. Compared with the gas consumption quota analysis method used in the prior art, the prediction result of the method for predicting the natural gas consumption provided by the embodiment of the present application is more accurate, because the steel production value and the required natural gas consumption value are not in a linear relationship due to the gas consumption fluctuation factor. The embodiment of the present application changes the prediction method in principle, and in addition, the steel production prediction value of the production enterprise is different due to the differences in production capacity, technical level, upstream and downstream demand of the enterprise chain and other factors of the production enterprise. The embodiment of the present application takes the predicted steel production value of the production enterprise as a calculation factor for predicting the corresponding natural gas consumption, so that the predicted natural gas consumption value calculated is more accurate.

[0063] Other features and advantages of the present application will be set forth in the following description, and in part will become apparent to those skilled in the art from the description, or can be learned by practice of the present application. The objects and other advantages of the present application can be realized and attained by means of the instrumentalities particularly pointed out in the written description and claims hereof as well as the appended drawings.

[0064] The technical solutions of the present application will be further described in detail below with the help of the drawings and examples. BRIEF DESCRIPTION OF DRAWINGS

[0065] The accompanying drawings are included to provide a further understanding of the present application, and constitute a part of the specification, illustrate embodiments of the present application, and are used to explain the present application, and do not constitute a limitation on the present application. In the drawings:

[0066] Figure 1 The flowchart of the method for predicting the natural gas consumption in the embodiment one of the present application;

[0067] Figure 2 The flowchart of the normalization processing method in the embodiment one of the present application;

[0068] Figure 3 The flowchart of the construction method of the producer function in the embodiment one of the present application;

[0069] Figure 4It is a chain influencing factor analysis schematic diagram in the embodiment one of the present application.

[0070] Figure 5 It is an example of the method flow chart of the steel production prediction model construction in the embodiment one of the present application.

[0071] Figure 6 It is a numerical source schematic diagram of the influencing factor in the embodiment of the present application.

[0072] Figure 7 It is one of the method flow charts of determining the comprehensive influencing factor combination in the embodiment of the present application.

[0073] Figure 8 It is the second of the method flow charts of determining the comprehensive influencing factor combination in the embodiment of the present application.

[0074] Figure 9 It is the flow chart of the construction method of the knowledge graph in the embodiment two of the present application.

[0075] Figure 10 It is a structure schematic diagram of the natural gas consumption prediction device in the embodiment of the present application. DETAILED DESCRIPTION

[0076] Exemplary embodiments of the present disclosure will be described below in greater detail with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be accurately conveyed to those skilled in the art.

[0077] In order to solve the problem that the natural gas consumption cannot be accurately predicted in the prior art, the embodiment of the present application provides a natural gas consumption prediction method and a knowledge graph construction method and device.

[0078] Embodiment one

[0079] The embodiment one of the present application provides a natural gas consumption prediction method, the flow of which is shown in the figure as follows, including the following steps: Figure 1

[0080] Step S101: obtaining initial historical data of a target production enterprise in a plurality of sampling periods in a preset time period, correcting the initial historical data of each sampling period to obtain corresponding corrected historical data; the corrected historical data includes historical steel production data and corresponding historical natural gas consumption data;

[0081] Step S102: constructing a producer function corresponding to the target production enterprise according to the corrected historical data of each sampling period; the producer function represents the natural gas consumption required for producing steel products; ​

[0082] Step S103: based on the output result of the pre-constructed steel production enterprise steel production prediction model, the producer function is solved, and the solving result is taken as the predicted value of the target production enterprise natural gas consumption.

[0083] The inventors of the present application consider that due to the fluctuation factors of gas consumption (i.e. using natural gas), the steel production and the required natural gas consumption are not in a linear relationship. In addition to changing the existing prediction method in principle, the scale benefit of each production enterprise customer in the steel industry is obvious, and the demand for natural gas of each production enterprise customer is large. The annual natural gas consumption of some production enterprises exceeds 100 million cubic meters. Therefore, it is particularly important to analyze the gas consumption characteristics of these large steel production enterprise customers in detail. The present application divides the natural gas consumption prediction problem of the production enterprise into the relationship between the steel production and the corresponding natural gas consumption and the steel production demand prediction problem of the production enterprise, so as to solve the problem that the original prediction of the natural gas consumption demand and the production relationship is not accurate and cannot be predicted according to the market changes, improve the prediction accuracy and market adaptability of the natural gas consumption demand of the steel production enterprise, and thus improve the natural gas consumption demand prediction effect of the production enterprise.

[0084] The initial historical data includes initial production data and corresponding initial natural gas consumption data. The initial production data refers to the steel production, and the corresponding initial natural gas consumption data refers to the natural gas consumption required by the steel production.

[0085] In some optional embodiments, in order to eliminate the influence of different dimensions and units of data, the initial production data and the corresponding initial natural gas consumption data of each sampling period need to be normalized. Taking the initial production data as an example, as shown in FIG. 2, the normalization can be realized by the following steps: Figure 2

[0086] Step S201: obtaining the maximum initial production data value and the minimum initial production data in each sampling period;

[0087] Step S202: calculating the difference between the maximum initial production data and the minimum initial production data to obtain a normalized difference value;

[0088] Step S203: taking the ratio of the difference between the initial production data of each sampling period and the minimum initial production data to the normalized difference value as the corresponding initial production data of each sampling period.

[0089] The initial natural gas consumption data corresponding to the initial production data is normalized in the same way as the normalized initial production data, which is not described herein.

[0090] ​In some optional embodiments, the initial historical data of each sampling period is processed to obtain corresponding modified historical data, and the modified historical data includes historical steel production data and corresponding historical natural gas consumption data. For example, the initial historical data can be processed in the following manner:

[0091] (1) obtaining the steel product reduction of the target production enterprise during the running abnormal event in the preset time period;

[0092] (2) taking the sum of the steel product reduction and the initial production data as the historical steel production data;

[0093] (3) taking the sum of the steel product reduction and the initial natural gas consumption data as the historical natural gas consumption data.

[0094] The running abnormal event generally refers to the commissioning, maintenance, equipment failure and the like of the steel production enterprise. For the historical production data of the steel production enterprise customer in the preset time period, due to the influence of some abnormal events, the running states of the steel production enterprise customer in each preset time period are different. Therefore, in order to ensure the consistency of the data of the steel production enterprise customer in each preset time period, the obtained data needs to be cleaned and processed to remove or modify the data corresponding to the running abnormal event period of the production enterprise, so as to obtain the corresponding relationship between the production and the natural gas consumption. For example, the crude steel reduction and the steel reduction caused by the running abnormal event are supplemented to the initial crude steel production and the initial steel production, and the reduced natural gas consumption caused by the running abnormal event is supplemented to the initial historical natural gas consumption data, so as to ensure the accuracy of the predicted value of the natural gas consumption.

[0095] The "preset time period" in the embodiment is at least 2 years before the current time. Generally, the historical steel production data and the corresponding historical natural gas consumption data in the past 5 years before the current time are obtained. The sampling period can be in units of months or in units of quarters, and the embodiment of the present application does not limit this. For example, the historical steel production data and the corresponding historical natural gas consumption data of each month in the past 5 years of the target production enterprise before the current time are obtained.

[0096] In some optional embodiments, in step S102, the producer function corresponding to the target production enterprise is constructed according to the modified historical data of each sampling period, as shown in the following formula: Figure 3

[0097] Step 1021: defining the producer function formula based on the Douglas function:

[0098]

[0099] ​In formula (1), y i represents historical steel production data of a target production enterprise in a sampling period; G i represents historical natural gas consumption data of the target production enterprise; A i represents the technical level of the target production enterprise; γ i represents the substitution elasticity parameter of natural gas; α i represents the degree of dependence of the target production enterprise on natural gas; and Constant represents a preset constant.

[0100] Step 1022: historical steel production data and corresponding historical natural gas consumption data in each sampling period are substituted into the above producer function formula to solve coefficients A i , α i , γ i , and Constant; and the solved coefficients A i , α i , γ i , and Constant are substituted into the above producer function formula to obtain a producer function, which is a function of natural gas consumption with respect to variable y i .

[0101] In fact, steel products generally include steel and crude steel, that is, historical steel production data includes historical steel production data or historical crude steel production data; historical natural gas consumption data includes historical crude steel-natural gas consumption data or historical steel-natural gas consumption data, historical steel-natural gas consumption data corresponds to historical steel production data, and historical crude steel-natural gas consumption data corresponds to historical crude steel production data.

[0102] Correspondingly, historical steel production data and historical steel-natural gas consumption data in each sampling period are substituted into the producer function formula to solve coefficients A i , α i , γ i , and Constant; and the solved coefficients A i , α i , γ i , and Constant are substituted into the above producer function formula to obtain a steel producer function formula;

[0103] Or

[0104] historical crude steel production data and historical crude steel-natural gas consumption data in each sampling period are substituted into the producer function formula to solve coefficients A i , α i , γ i , and Constant; and the solved coefficients A ii , αi , γ i and Constant into the above producer function formula, a crude steel producer function formula is obtained;

[0105] Correspondingly, the producer function is obtained, including:

[0106] The fitting degrees of the steel material producer function formula and the crude steel producer function formula are determined respectively, and according to the comparison result of the fitting degrees, the producer function formula is determined. Specifically, the producer function formula can be determined by the following manner:

[0107] If the fitting degree of the steel material producer function formula is not lower than the fitting degree of the crude steel producer function formula, the steel material producer function formula is determined as the producer function formula, and correspondingly, the historical steel material production data is taken as the historical steel production data, and the historical steel material-natural gas consumption data is taken as the historical natural gas consumption data.

[0108] If the fitting degree of the steel material producer function formula is lower than the fitting degree of the crude steel producer function formula, the crude steel producer function formula is determined as the producer function formula, and correspondingly, the historical crude steel production data is taken as the historical steel production data, and the historical crude steel-natural gas consumption data is taken as the historical natural gas consumption data.

[0109] Based on the historical production data of the target production enterprise and the corresponding historical natural gas consumption data, the fitting operation is performed according to the production function of the target production enterprise, so as to obtain the enterprise technical level, the substitution elasticity parameter, the dependence degree of the steel production enterprise customer on the natural gas and the preset constant of the target production enterprise. Therefore, the natural gas consumption data prediction can be performed based on the production function of the target production enterprise and the production data of the target production enterprise, and the accuracy of the prediction result is increased.

[0110] In some optional embodiments, based on the production process of the target production enterprise, the relationship between the competitors and the upstream and downstream industrial chains, the industrial chain influencing factors that may affect the steel output (production data) of the target production enterprise are obtained. For example, the industrial chain influencing factors can be raw material price, coal price, iron ore price, scrap steel price, product price, US dollar index, steel export index, etc., such as Figure 4The shown industrial chain influencing factor analysis, the crude steel output is affected by the raw material price (cost) and the finished product price (income), wherein, the raw material price includes but is not limited to: coke price, natural gas price, iron ore price, scrap steel price and power price, the finished product price includes but is not limited to: bearing steel price, steel price. Among them, the scrap steel price is affected by the supply of upstream industry (customer), the upstream industry supply includes but is not limited to: domestic steel output, scrap steel import and export volume, natural gas supply, domestic iron ore output, iron ore import volume, the iron ore price is affected by the competing enterprises, including but not limited to: the bearing steel output of competing enterprises, special steel output, steel PMI inventory, dollar index, downstream industry (customer) demand affects special steel export amount, bearing steel domestic self-use amount, bearing steel export amount, raw material price, finished product price and dollar index affect short process profit; The demand of downstream industry is affected by the demand of automobile industry bearing steel, the demand of mechanical processing industry steel, the demand of national automobile consumption.

[0111] Further, the data table of each industrial chain influencing factor and the actual steel output of the target production enterprise can be established, and the relationship between the steel output and the industrial chain influencing factor is established, and the steel output prediction model corresponding to the target production enterprise is constructed based on the relationship between the steel output and the industrial chain influencing factor, and the steel output prediction model corresponding to the target production enterprise is constructed based on the relationship between the steel output and the industrial chain influencing factor. Figure 5 As shown, the steel output prediction model corresponding to the target production enterprise is constructed by the following way:

[0112] Step S401: based on the historical steel output production data, the comprehensive influencing factor combination corresponding to the historical steel output production data is determined from the preset influencing factor set, the influencing factor set includes at least one influencing factor; the influencing factor represents the factor determined based on the upstream and downstream industrial chain of the target production enterprise, which affects the steel output of the target production enterprise;

[0113] The influencing factor set is selected from the influencing factor pool, and the influencing factor pool refers to the union of the industrial chain influencing factors that may affect the steel output of each production enterprise (including the target production enterprise and other steel production enterprises). Since there are many industrial chain influencing factors in the influencing factor pool, and some of them may have little or even no effect on the steel output of the target production enterprise, it is necessary to preliminarily determine the influencing factors related to the target production enterprise from the influencing factor pool according to the upstream and downstream industrial chain of the target production enterprise, and eliminate the irrelevant influencing factors, which can effectively reduce the influence of irrelevant factors on the steel output prediction and improve the accuracy of the steel output prediction result.

[0114] Further, it is also necessary to determine the comprehensive influence factor combination corresponding to the historical steel production data from the set of influence factors. The industry chain influence factors in the comprehensive influence factor combination have a greater impact on the steel production of the target production enterprise. In other words, by eliminating the industry chain influence factors that have a small impact on the steel production of the target production enterprise, the accuracy of the output result of the constructed steel production prediction model can be improved.

[0115] Step S402: Taking the historical steel production data and the numerical values of each influence factor in the corresponding comprehensive influence factor combination as sample data, a sample data set is obtained.

[0116] The numerical values of the influence factors can be directly obtained from the target production enterprise or can be obtained from other public sources. For reference, Figure 6 the data source diagram is shown in Figure 6 which includes project number, data type, data name, frequency, region and data source. For example, the data type includes but is not limited to price data, industry supply and demand data, steel plant operation data, inventory data, LNG price data, alternative energy data, coking coal data and other derivatives. The embodiments of the present application do not repeat here.

[0117] Step S403: Inputting the sample data set into the preset steel production prediction model for training to obtain a trained steel production prediction model. The steel production prediction model can obtain the steel production prediction value of the target production enterprise based on the numerical values of the input several influence factors.

[0118] In some optional embodiments, the step S401 of determining the comprehensive influence factor combination corresponding to the historical steel production data can be realized by the following manner, for reference Figure 7 .

[0119] Step S501: For each influence factor in the set of influence factors, the fitting degree of each influence factor and the historical steel production data is calculated according to the numerical values of each influence factor and the historical steel production data. The single-factor influence value of the influence factor is determined according to the fitting degree of each influence factor and the historical steel production data. The single-factor influence value represents the influence degree of the influence factor on the steel production of the target production enterprise.

[0120] There are various ways to determine the single-factor influence value of the influence factor, for example:

[0121] Method one:

[0122] The relationship fitting algorithm is used to fit the numerical values of each influence factor and the historical steel production data respectively. Each fitting result is taken as the single-factor influence value of the corresponding influence factor.

[0123] The second mode is as follows:

[0124] For each influencing factor, at least one relationship fitting algorithm is used to fit the numerical value of the influencing factor and the historical steel production data, and each fitting result is taken as a single-factor influence value of the influencing factor corresponding to each relationship fitting algorithm.

[0125] Among the single-factor influence values of the influencing factors corresponding to various relationship fitting algorithms, the single-factor influence value with the largest value is selected as the single-factor influence value of the influencing factor.

[0126] In the second mode, for each influencing factor, a linear relationship fitting algorithm, a polynomial relationship fitting algorithm, an exponential relationship fitting algorithm, a power function relationship fitting algorithm, and a logarithmic relationship fitting algorithm are used to fit the numerical value of the industrial chain influencing factor and the corresponding historical steel production data, to obtain a fitting value (denoted as R2) of the industrial chain influencing factor for each fitting algorithm, and the R2 with the highest value is selected as the single-factor influence value of the industrial chain influencing factor.

[0127] For each industrial chain influencing factor, single-factor regression analysis is performed, that is, single-variable regression analysis is performed on the industrial chain influencing factor and the historical steel production data, the industrial chain influencing factor and the corresponding steel production data are fitted by linear relationship, polynomial relationship, exponential relationship, power function relationship, and logarithmic relationship, the corresponding fitting R2 values are calculated, the highest fitting R2 value is taken as the fitting value of the industrial chain influencing factor, and the function form (i.e., the fitting algorithm) corresponding to the highest fitting R2 value is taken as the function form used in combination regression. After the fitting and calculation of all industrial chain influencing factors are completed, based on the sorting of the fitting R2 values of the industrial chain influencing factors, the higher the fitting R2 value, the higher the explainability of the industrial chain influencing factor to the steel production of the target production enterprise.

[0128] Step S502: The influencing factors in the influencing factor set are sorted in descending order of single-factor influence value, a preset number of influencing factors starting from the maximum value are selected and retained, and a seed influencing factor combination is obtained.

[0129] The influencing factors in the seed influencing factor combination are the influencing factors that represent the determined several influencing factors with the largest influence on the steel production of the target production enterprise.

[0130] Specifically, the several influencing factors with the highest R2 values are selected from the influencing factor set as elements in the seed influencing factor combination.

[0131] The seed influencing factor combination is used as a starting point for the combination regression analysis of the industrial chain influencing factors, and the remaining influencing factors in the influencing factor set are sequentially added to the seed influencing factor combination to form a new influencing factor combination, and the new influencing factor combination is used as a starting point for the combination regression analysis of the industrial chain influencing factors.The impact factor corresponding to the highest R2 value can also be taken as a seed impact factor, and accordingly, there is one element in the obtained seed impact factor combination. It should be noted that if the R2 values of two or more impact factors are the highest, the impact factors corresponding to the highest R2 values are all seed impact factors, and accordingly, there are at least two elements in the obtained seed impact factor combination.

[0132] Both the above two manners can determine the seed impact factor combination, and different manners can be selected according to actual conditions, and the embodiments of the present application do not limit this.

[0133] Step S503: determining a comprehensive impact factor combination based on historical steel production data, the numerical values of the impact factors in the impact factor set, and the numerical values of the impact factors in the seed impact factor combination.

[0134] In some optional embodiments, the comprehensive impact factor combination can be determined by the following manner with reference to FIG. 5. Figure 8

[0135] Step S601: matching any impact factor in the impact factor set except any impact factor in the seed impact factor combination with the seed impact factor combination respectively to obtain a plurality of alternative impact factor combinations.

[0136] It should be noted that after obtaining the plurality of alternative impact factor combinations, in order to improve the accuracy of the prediction result, the factors in each alternative impact factor combination are subjected to a multicollinearity risk test (multicollinearity optimization) respectively, that is, the alternative impact factor combination obtained by combining the seed industrial chain impact factor and each industrial chain impact factor is subjected to a multicollinearity risk test, and if the industrial chain impact factor and the seed industrial chain impact factor in the combination have a high degree of multicollinearity risk, the alternative impact factor combination is abandoned.

[0137] Step S602: for each alternative impact factor combination, determining a combined factor impact value of the alternative impact factor combination based on historical steel production data by using a relationship fitting algorithm; the combined factor impact value of the alternative impact factor combination represents the influence degree of each impact factor in the alternative impact factor combination on the target production enterprise steel production.

[0138] The historical steel production data are fitted with the impact values corresponding to each impact factor in the alternative impact factor combination by using the relationship fitting algorithm to obtain a second fitting value corresponding to each impact factor in the alternative impact factor combination, and the largest second fitting value is selected as the combined factor impact value of the alternative impact factor combination.

[0139] ​For each alternative influence factor combination, the linear relationship fitting algorithm, the polynomial relationship fitting algorithm, the exponential relationship fitting algorithm, the power function relationship fitting algorithm, and the logarithmic relationship fitting algorithm are used to fit the influence factors in the alternative influence factor combination and the historical steel production data corresponding to the influence factors in the alternative influence factor combination, to obtain the fitting R2 value of the influence factors in the alternative influence factor combination for each fitting algorithm, and the highest fitting R2 value is selected as the fitting R2 value of the alternative influence factor combination.

[0140] Step S603: Comparing the influence value of the combination with the highest influence value of the combination factors in the alternative influence factor combination with the influence value of the seed influence factor combination, and updating the seed influence factor combination according to the comparison result.

[0141] Step S604: Repeating the step of updating the seed influence factor combination until each influence factor in the second set of influence factors is traversed, and stopping iteration.

[0142] After all the industrial chain influence factors in the influence factor set are traversed, the fitting R2 values are sorted. If the fitting R2 value is higher than the fitting R2 value of the last iteration, the combination influence factor with the highest fitting R2 value is selected as the new seed, and iteration is performed again. If the fitting R2 value of the new combination influence factor is not better than the last iteration combination, the seed combination obtained last time is selected as the input combination influence factor of the steel production enterprise customer's steel production influence factor (industrial chain important influence factor), and the iteration is ended.

[0143] Embodiment Two

[0144] Embodiment Two of the present application provides a knowledge graph construction method, the flowchart of which is shown in Figure 9 The method comprises the following steps:

[0145] Step S701: Obtain the producer function and the steel production prediction model of the target production enterprise; the producer function and the steel production prediction model are obtained by the aforementioned natural gas consumption prediction method;

[0146] Step S702: Construct the ontology relationship model of the producer function and the steel production prediction model;

[0147] Step S703: Extract entity, attribute and relationship information from the historical production data of the target production enterprise;

[0148] Step S704: Based on the obtained ontology relationship model and the extracted entity, attribute and relationship information, construct a knowledge graph; the knowledge graph represents the relationship between the steel production prediction value and the corresponding natural gas consumption prediction value of the target production enterprise.

[0149] Based on the same inventive concept, the embodiment of the present application also provides a natural gas consumption prediction device, the structure of the device is shown in Figure 10 as follows, comprising:

[0150] The acquisition module 101 is configured to acquire initial historical data of the target production enterprise in a plurality of sampling periods within a preset time period, correct the initial historical data of each sampling period, and obtain corresponding corrected historical data.

[0151] The construction module 102 is configured to construct a producer function corresponding to the target production enterprise according to the corrected historical data of each sampling period; the corrected historical data includes historical steel production data and corresponding historical natural gas consumption data; and the producer function represents the natural gas consumption required for producing steel products.

[0152] The solving module 103 is configured to solve the producer function based on the output result of the pre-constructed steel production prediction model corresponding to the target production enterprise, and take the solving result as the prediction value of the natural gas consumption of the target production enterprise.

[0153] Regarding the natural gas consumption prediction device in the above embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment related to the method, and will not be described in detail here.

[0154] The following describes a natural gas consumption prediction method with a specific example:

[0155] A company was founded in 1957 and is a major supplier of high-end equipment manufacturing materials at home and abroad. The project was completed and put into operation on October 29, 2020, using a "scrap shearing → electric furnace preliminary refining → refining → continuous casting → rolling → finishing" short process production process, with an annual production capacity of 2 million tons of steel and 192 million tons of materials. A company is the first full-scrap short-process green low-carbon special steel enterprise in China, which does not contain the ironmaking link, so it will not produce coke oven gas, and the energy consumption in the whole process is only electricity and natural gas. The plan is to stabilize gas consumption of 160 million cubic meters per year after reaching the design rated gas consumption of 60 cubic meters per ton (the actual measured natural gas consumption per unit is 53 cubic meters per ton of steel). From the historical gas consumption data, Shigang Company adopts a continuous uninterrupted production process throughout the year, but the daily average gas consumption fluctuates greatly, and the gas demand prediction is difficult. Based on the above analysis, the natural gas consumption prediction method provided by the embodiment of the present application is used to predict the gas demand of A company, including the following steps:

[0156] (1) Construct the relationship between the natural gas demand of A company and the steel production of Shigang Company (producer function). The steel production can be represented by the crude steel production and the steel production, and the producer function includes: crude steel production function and steel production function, which are calculated by fitting respectively, and the better index is selected for calculation.

[0157] Crude steel production function: G is the demand for natural gas, y i is the output of crude steel. The production function interpretability index R2=0.97.

[0158] Steel production function: G is the demand for natural gas, x i is the output of steel. The production function interpretability index R2=0.77.

[0159] After comparison, the crude steel output index construction producer function effect is better, so choose the crude steel output as the intermediate variable of prediction.

[0160] (2) Analysis of factors affecting the steel industry chain.

[0161] According to the upstream and downstream links of A company, find the factors that may affect the crude steel output of Shigang company, including iron ore price, crude steel price, US dollar index, etc. The main data sources are Wind database, Metallurgical Industry Information Center, Mysteel steel network, Rich data, and Zhuochuang website. The output of crude steel (billet) is provided by Shigang company of He Steel Group.

[0162] (3) Single factor regression analysis. For influencing factors, single variable regression is carried out, North China steel price index, coking coal price index, key enterprise seamless steel pipe sales volume, bearing steel price index, steel export price index have high correlation, and other steel data have low correlation.

[0163] (4) Iterative combination regression analysis, get the combination of comprehensive influencing factors.

[0164] (5) Multiple collinearity optimization. After measurement, bearing steel price index is highly correlated with pig iron price index. The final model will have the problem of multiple collinearity. In the final model, bearing steel price index has included the influence of pig iron price. For example, bearing steel price index also includes the influence of North China steel price. By further adjusting the variable, remove the pig iron price, keep the bearing steel price index, the multiple collinearity is eliminated, R2 is improved to 89%, and the final effect is improved.

[0165] Evaluation score of industry chain influencing factors:

[0166] By comparing the correlation ranking of different industry chain influencing factors in the algorithm, the top five influence factors with the highest correlation are obtained, and then the steel output prediction model is constructed. And use adaptive weight weighting for each group of influence factors to adapt to the multiple prediction scenarios of steel output prediction model.

[0167] Steel output prediction model:

[0168] Steel production = 0.9799 + 0.0003 * (bearing steel price) + 3.693e-05 * (short-process profit in the steel industry) - 4.221e^(-05) * (steel industry PMI inventory) + 0.1004 * (North China steel price index) + 1[e]^(-04) * (coking coal price index) - 4.7014 * (environmental policy reduction effect) + ε (impact)

[0169] The final crude steel production forecast generated by the steel production prediction model, compared with the actual steel production and price, showed an interpretability index (R²) of 0.89 and an accuracy rate of 95.2%.

[0170] (7) Natural gas demand forecasting model.

[0171] The numerical values ​​of each influencing factor in the comprehensive influencing factor combination are substituted into the natural gas demand forecasting model for calculation. For example, the XGBoost model can be selected for forecasting.

[0172] Based on the same inventive concept, embodiments of the present invention also provide a computer storage medium storing computer-executable instructions, which, when executed by a processor, implement the aforementioned method for predicting natural gas consumption.

[0173] Based on the same inventive concept, embodiments of the present invention also provide a terminal device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the aforementioned method for predicting natural gas consumption.

[0174] Unless otherwise specifically stated, terms such as processing, calculation, operation, determination, display, etc., may refer to the actions and / or processes of one or more processing or computing systems or similar devices that represent the manipulation and conversion of data representing physical (e.g., electronic) quantities within the registers or memory of the processing system into other data similarly representing physical quantities within the memory, registers, or other such information storage, transmission, or display devices of the processing system. Information and signals can be represented using any of a variety of different techniques and methods. For example, data, instructions, commands, information, signals, bits, symbols, and chips mentioned throughout the above description can be represented by voltage, current, electromagnetic waves, magnetic fields or particles, light fields or particles, or any combination thereof.

[0175] It should be understood that the specific order or hierarchy of steps in the disclosed process is an example of an exemplary method. Based on design preferences, it should be understood that the specific order or hierarchy of steps in the process may be rearranged without departing from the scope of this disclosure. The appended method claims provide elements of various steps in an exemplary order and are not intended to limit the scope to the specific order or hierarchy described.

[0176] In the detailed description herein, various features are grouped together in single embodiments for the purpose of streamlining the disclosure. Such disclosure should not, of course, be interpreted as reflecting a necessity of multiple features in execution of the claimed embodiments. On the contrary, the disclosure contemplates that of few of the disclosed features might be implemented without others. Thus, the claims following the detailed description are hereby expressly incorporated into this detailed description, with each claim standing on its own as a separate preferred embodiment.

[0177] Those skilled in the art will further appreciate that the various illustrative logical blocks, modules, circuits, and algorithm steps described in connection with the embodiments disclosed herein can be implemented as electronic hardware, computer software, or combinations of both. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, circuits, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the overall system. Skilled artisans can implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present disclosure.

[0178] The steps of a method or algorithm described in connection with the embodiments disclosed herein can be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module can reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor such that the processor can read information from, and write information to, the storage medium. In the alternative, the storage medium can be integral to the processor. The processor and the storage medium can reside in an ASIC. The ASIC can reside in a user terminal. In the alternative, the processor and the storage medium can reside as discrete components in a user terminal.

[0179] For a software implementation, the techniques described herein can be implemented with modules (e.g., procedures, functions, and so on) that perform the functions described herein. The software codes can be stored in memory units and executed by processors. The memory unit can be implemented within the processor or external to the processor, in which case it can be communicatively coupled to the processor via various means as is well known in the art.

[0180] The above description includes examples of one or more embodiments. Of course, not all possible combinations of components or methods described above can be claimed as an embodiment, but one of ordinary skill in the art will recognize that many such further combinations and permutations of the embodiments described are possible. Accordingly, the described embodiments are intended to embrace all such alterations, modifications and variations which fall within the scope of the appended claims. Additionally, where the description or the claims recite "a", "an" or a "the" one or more of elements, this does not exclude multiple numbers for this element. Further, where the description or the claims recite that elements are each independently recited, this does not exclude multiple elements from being the same. Further, where the description or the claims recite "comprising", "containing", "including" or "having" this does not exclude other elements. Further, where the description or the claims recite "or" this does not exclude "and". Further, where the description or the claims recite "about" or "substantially" this does not exclude other elements.

Claims

1. A method for predicting natural gas consumption, characterized in that, include: The initial historical data of the target production enterprise for several sampling periods within a preset time period is obtained, and the initial historical data for each sampling period is corrected to obtain the corresponding corrected historical data; the corrected historical data includes historical steel production data and corresponding historical natural gas consumption data. Based on the corrected historical data of each sampling period, a producer function corresponding to the target production enterprise is constructed; the producer function represents the amount of natural gas required to produce steel products. Based on the output of the pre-built steel production prediction model of the target production enterprise, the producer function is solved, and the solution is used as the predicted value of the natural gas consumption of the target production enterprise.

2. The method as described in claim 1, characterized in that, The step of constructing the producer function corresponding to the target production enterprise based on the corrected historical data of each sampling period includes: Define the producer function formula: In the above formula, y i This represents the historical steel production data of the target manufacturer, in units of sampling periods; G i This represents the historical natural gas consumption data of the target production enterprise; A i This indicates the technological level of the target manufacturing enterprise; γ i The substitution elasticity parameter for natural gas; α i This indicates the target production company's dependence on natural gas; Constant represents a preset constant. Substitute the historical steel production data and corresponding historical natural gas consumption data for each sampling period into the producer function formula above to solve for the coefficient A. i α i γ i and Constant; the coefficients A obtained from the solution i α i γ i Substituting Constant into the above producer function formula, we obtain the producer function, which is the natural gas consumption with respect to the variable y. i The function.

3. The method as described in claim 2, characterized in that, The historical steel production data includes historical steel product production data or historical crude steel production data; the historical natural gas consumption data includes historical crude steel-natural gas consumption data or historical steel-natural gas consumption data, wherein the historical steel-natural gas consumption data corresponds to the historical steel production data, and the historical crude steel-natural gas consumption data corresponds to the historical crude steel production data. The historical steel production data and corresponding historical natural gas consumption data for each sampling period are substituted into the producer function formula above to solve for coefficient A. i α i γ i and Constant; the coefficients A obtained from the solution i α i γ i Substituting Constant into the producer function formula above, we obtain the producer function, which includes: Substitute the historical steel production data and the historical steel-natural gas consumption data from each sampling period into the producer function formula to solve for coefficient A. i α i γ i And Constant, the coefficients A obtained from the solution i α i γ i Substituting Constant into the above producer function formula, we obtain the steel producer function formula; or Substitute the historical crude steel production data and the historical crude steel-natural gas consumption data for each sampling period into the producer function formula to solve for coefficient A. i α i γ i And Constant, the coefficients A obtained from the solution i α i γ i Substituting Constant into the above producer function formula, we obtain the crude steel producer function formula; Accordingly, the producer function includes The goodness of fit between the steel producer function formula and the crude steel producer function formula is determined respectively. Based on the comparison results of the goodness of fit, the producer function formula is determined.

4. The method as described in claim 3, characterized in that, The step of determining the goodness of fit between the steel producer function formula and the crude steel producer function formula, and determining the producer function formula based on the comparison results of the goodness of fit, includes: If the goodness of fit of the steel producer function formula is not lower than that of the crude steel producer function formula, then the steel producer function formula is determined as the producer function formula. Accordingly, the historical steel production data is used as the historical steel production data, and the historical steel-natural gas consumption data is used as the historical natural gas consumption data. If the goodness of fit of the steel producer function formula is lower than that of the crude steel producer function formula, then the crude steel producer function formula is determined as the producer function formula. Accordingly, the historical crude steel production data is used as the historical steel production data, and the historical crude steel-natural gas consumption data is used as the historical natural gas consumption data.

5. The method as described in claim 4, characterized in that, The steel production prediction model corresponding to the target production enterprise is constructed in the following manner: Based on the historical steel production data, a comprehensive combination of influencing factors corresponding to the historical steel production data is determined from a preset set of influencing factors. The set of influencing factors includes at least one influencing factor. The influencing factors represent the factors affecting the steel production of the target production enterprise based on the upstream and downstream industrial chains of the target production enterprise. Historical steel production data, along with the values ​​of each influencing factor in the corresponding combination of comprehensive influencing factors, are used as sample data to obtain a sample dataset. The sample dataset is input into a preset steel production prediction model for training, resulting in a trained steel production prediction model. This model can predict the steel production of the target production enterprise based on the values ​​of several input influencing factors.

6. The method as described in claim 5, characterized in that, Based on the historical steel production data, the method of determining a comprehensive combination of influencing factors corresponding to the historical steel production data from a preset set of influencing factors includes: For each influencing factor in the set of influencing factors, the goodness of fit between each influencing factor and the historical steel production data is calculated based on the value of each influencing factor and the historical steel production data. Based on the goodness of fit between each influencing factor and the historical steel production data, the single-factor influence value of the influencing factor is determined. The single-factor influence value characterizes the degree of influence of the influencing factor on the steel production of the target production enterprise. The influencing factors in the set of influencing factors are sorted in descending order of their single-factor influence values. A preset number of influencing factors, starting from the maximum value, are selected and retained to obtain the seed influencing factor combination. Based on the historical steel production data, the values ​​of each influencing factor in the influencing factor set, and the values ​​of each influencing factor in the seed influencing factor combination, a comprehensive influencing factor combination is determined.

7. The method as described in claim 6, characterized in that, The process involves calculating the goodness of fit between each influencing factor and the historical steel production data, based on the numerical values ​​of each influencing factor and the historical steel production data. Then, based on the goodness of fit between each influencing factor and the historical steel production data, the single-factor influence value of each influencing factor is determined, including: A relationship fitting algorithm was used to fit the values ​​of each influencing factor and the historical steel production data, and the fitting results were used as the single-factor influence values ​​of the corresponding influencing factors.

8. The method as described in claim 6, characterized in that, The process involves calculating the goodness of fit between each influencing factor and the historical steel production data, based on the numerical values ​​of each influencing factor and the historical steel production data. Then, based on the goodness of fit between each influencing factor and the historical steel production data, the single-factor influence value of each influencing factor is determined, including: For each influencing factor, at least one relationship fitting algorithm is used to fit the numerical values ​​of the influencing factors and the historical steel production data, and each fitting result is used as the single-factor influence value of the influencing factor corresponding to each relationship fitting algorithm. Among the single-factor influence values ​​of the influencing factors corresponding to various relationship fitting algorithms, the single-factor influence value with the largest value is selected as the single-factor influence value of the influencing factor.

9. The method as described in claim 7 or 8, characterized in that, The determination of the comprehensive influencing factor combination based on the historical steel production data and the values ​​of each influencing factor in the seed influencing factor combination includes: Each influencing factor in the influencing factor set, excluding any influencing factor in the seed influencing factor combination, is matched with the seed influencing factor combination to obtain multiple alternative influencing factor combinations. For each candidate combination of influencing factors, based on the historical steel production data, a relationship fitting algorithm is used to determine the combined factor influence value of the candidate combination of influencing factors; the combined factor influence value of the candidate combination of influencing factors characterizes the degree of influence of each influencing factor in the candidate combination on the steel production of the target production enterprise; Compare the influence value of the combination with the highest influence value among the candidate combinations of influencing factors with the influence value of the seed combination of influencing factors, and update the seed combination of influencing factors based on the comparison results; Repeat the step of updating the combination of influencing factors until every influencing factor in the set of influencing factors has been traversed, then stop the iteration.

10. The method as described in claim 9, characterized in that, Based on the historical steel production data, a relationship fitting algorithm is used to determine the combined factor influence value of the candidate influencing factor combinations, including: A relationship fitting algorithm is used to fit the historical steel production data with the influence values ​​of each influencing factor in the candidate influencing factor combination to obtain the second fitting value of each influencing factor in the candidate influencing factor combination. The largest second fitting value is selected as the combined factor influence value of the candidate influencing factor combination.

11. The method as described in claim 1, characterized in that, The initial historical data includes initial production data and corresponding initial natural gas consumption data; The step of correcting the initial historical data for each sampling period to obtain the corresponding corrected historical data includes: Obtain the reduction in steel product output of the target production enterprise during the period of abnormal operation events within the preset time period; The sum of the reduction in steel production and the initial production data is used as the historical steel production data. The sum of the reduction in steel production and the initial natural gas consumption data is used as historical natural gas consumption data.

12. A method for constructing a knowledge graph, characterized in that, include: Obtain the producer function and steel production prediction model of the target production enterprise; the producer function and steel production prediction model are obtained by the natural gas consumption prediction method described in any one of claims 1-11; Construct an ontology relationship model between the producer function and the steel production prediction model; Extract entity, attribute, and relationship information from the historical production data of the target manufacturing enterprise; Based on the obtained ontology relationship model and the extracted entity, attribute and relationship information, a knowledge graph is constructed. The knowledge graph represents the relationship between the predicted steel production of the target production enterprise and the corresponding predicted natural gas consumption.

13. A device for predicting natural gas consumption, characterized in that, include: The acquisition module is used to acquire the initial historical data of the target production enterprise for several sampling periods within a preset time period, and to perform correction processing on the initial historical data of each sampling period to obtain the corresponding corrected historical data. A construction module is used to construct the producer function corresponding to the target production enterprise based on the corrected historical data of each sampling period; the corrected historical data includes historical steel production data and corresponding historical natural gas consumption data; the producer function represents the amount of natural gas required to produce steel products; The solution module is used to solve the producer function based on the output of the pre-built steel production prediction model corresponding to the target production enterprise, and use the solution result as the predicted value of the natural gas consumption of the target production enterprise.

14. A computer storage medium, characterized in that, The computer storage medium stores computer-executable instructions, which, when executed by a processor, implement the natural gas consumption prediction method according to any one of claims 1-11.

15. A terminal device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the natural gas consumption prediction method according to any one of claims 1-11.

Images