A method, system, device, and medium for predicting railway freight volume.

Abstract

This invention discloses a method, system, equipment, and medium for predicting railway freight volume, relating to the field of railway freight technology. The method includes: acquiring target data, which includes target historical railway freight volume data and target historical railway freight-related data. The target historical railway freight-related data consists of target historical data corresponding to multiple target influencing factors related to railway freight volume, and the target influencing factors are factors that have a direct or indirect impact on railway freight volume; inputting the target data into a pre-constructed target railway freight volume prediction model to obtain the predicted railway freight volume value corresponding to the target data. This invention solves the problem that existing railway freight volume prediction methods do not consider the impact of influencing factors related to railway freight volume on railway freight volume and the relationships between these factors, resulting in low prediction accuracy and limiting and lagging operational decisions for railway freight enterprises.

Description

Technical Field

[0001] This invention relates to the field of railway freight technology, and specifically to a method, system, equipment, and medium for predicting railway freight volume. Background Technology

[0002] Railway freight demand is a crucial component of overall social freight demand, and accurate forecasting of this demand provides a reliable basis for business decisions. With the continuous advancement of railway freight informatization, railway freight operations and information flows are gradually synchronizing, and the massive amounts of freight data possess immense application value.

[0003] Traditional railway freight volume forecasting methods are mainly divided into two categories: qualitative forecasting and quantitative forecasting. Qualitative forecasting relies primarily on the experience and professional knowledge of railway staff at all levels to predict freight volume. While this method is relatively flexible, it is overly subjective and has low reliability. Quantitative forecasting, on the other hand, collects historical and related railway data and builds different models based on this data to predict future trends in railway freight volume. Existing forecasting models for railway freight volume include time series forecasting, regression forecasting, grey forecasting, artificial neural network forecasting, fractal theory, rough set theory, and corresponding improved models. These models mainly input all influencing factors related to railway freight volume as variables into the constructed model, without considering the impact of these factors on railway freight volume (i.e., whether each influencing factor directly or indirectly affects railway freight volume) and the relationships between these factors, resulting in low accuracy in predicting railway freight volume. Summary of the Invention

[0004] The technical problem this invention aims to solve is that existing railway freight volume forecasting methods do not consider the impact of relevant factors on railway freight volume and the relationships between these factors, resulting in low forecast accuracy and limiting and lagging operational decisions for railway freight companies. To address this problem, this invention provides a method, system, equipment, and medium for forecasting railway freight volume.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0006] A method for forecasting railway freight volume includes:

[0007] Step S1: Obtain target data, which includes target historical railway freight volume data and target historical railway freight related data. The target historical railway freight related data are the target historical data corresponding to each of multiple target influencing factors related to railway freight volume. The target influencing factors are factors that have a direct or indirect impact on railway freight volume.

[0008] Step S2: Input the target data into the pre-built target railway freight volume prediction model to obtain the railway freight volume prediction value corresponding to the target data;

[0009] The target railway freight volume prediction model is obtained through the following steps:

[0010] Obtain raw data, which includes first historical railway freight volume data and first historical railway freight related data. The first historical railway freight related data consists of the original historical data corresponding to each of the multiple factors related to railway freight volume.

[0011] Based on the raw data, target influencing factors and corresponding target historical data are determined. The target influencing factors include direct influencing factors, secondary influencing factors, and tertiary influencing factors. The direct influencing factors are primary influencing factors. The target historical data corresponding to the tertiary influencing factors affect the target historical data corresponding to the secondary influencing factors. The target historical data corresponding to the secondary influencing factors affect the target historical data corresponding to the direct influencing factors. The target historical data corresponding to the direct influencing factors affect railway freight volume.

[0012] Based on the target historical data corresponding to the third-level influencing factors and the target historical data corresponding to the second-level influencing factors, a second-level influencing factor prediction model is constructed. The second-level influencing factor prediction model is used to determine the first prediction result corresponding to the second-level influencing factors.

[0013] Based on the first prediction result corresponding to the secondary influencing factor and the target historical data corresponding to the direct influencing factor, a prediction model for the direct influencing factor is constructed. The prediction model for the direct influencing factor is used to determine the second prediction result corresponding to the direct influencing factor.

[0014] Based on the second prediction results corresponding to the direct influencing factors and the first historical railway freight volume data, a railway freight volume prediction model is constructed, which is used to predict railway freight volume.

[0015] The target railway freight volume prediction model includes the secondary influencing factor prediction model, the direct influencing factor prediction model, and the railway freight volume prediction model.

[0016] The beneficial effects of this invention are as follows: Target data for predicting railway freight volume is obtained based on the target influencing factors (i.e., factors that have a direct impact on railway freight volume and factors that have an indirect impact on railway freight volume). This target data is then input into a pre-constructed target railway freight volume prediction model to obtain the predicted railway freight volume value corresponding to the target data. This improves the accuracy of railway freight volume prediction and solves the problem that existing railway freight volume prediction methods do not consider the impact of influencing factors related to railway freight volume on railway freight volume and the relationships between these factors, resulting in low prediction accuracy.

[0017] Based on the above technical solution, the present invention can be further improved as follows.

[0018] Furthermore, determining the target influencing factors and the corresponding historical data based on the original data includes:

[0019] Step A1: Perform data cleaning on the first historical railway freight volume data to obtain the second historical railway freight volume data; perform data cleaning on the first historical railway freight-related data to obtain the processed railway freight-related data, wherein the processed railway freight-related data consists of target historical data corresponding to multiple factors related to railway freight volume.

[0020] Step A2: According to the data category, the second historical railway freight volume data and the processed railway freight related data are grouped to obtain preprocessed data. The preprocessed data includes the second historical railway freight volume data and data related to factors to be determined. The data related to factors to be determined are the target historical data corresponding to each of the multiple factors to be determined. The factors to be determined are factors that have a direct impact on railway freight volume or factors that have an indirect impact on railway freight volume.

[0021] Step A3: Based on the preprocessed data, determine the target influencing factor. The plurality of factors to be determined include the target influencing factor. The target historical data corresponding to the target influencing factor is the target historical data corresponding to each of the plurality of factors to be determined.

[0022] The beneficial effects of adopting the above-mentioned further scheme are: by cleaning the acquired raw data, duplicate information can be deleted, existing errors can be corrected, and the data can be kept intact, consistent and valid; by grouping the data after cleaning, it is easier to determine the target influencing factors affecting railway freight volume.

[0023] Furthermore, in step A3, determining the target influencing factors based on the preprocessed data includes:

[0024] Step A3.1: For each of the factors to be determined, the grey relational degree of the factor to be determined is determined based on the target historical data corresponding to the factor to be determined and the second historical railway freight volume data. The grey relational degree characterizes the degree of correlation between the factor to be determined and the railway freight volume.

[0025] Step A3.2: For each of the factors to be determined, if the gray correlation degree of the factor to be determined is greater than the preset correlation degree threshold, then proceed to step A3.3.

[0026] Step A3.3: Based on the target historical data corresponding to the factor to be determined and the second historical railway freight volume data, determine the correlation coefficient of the factor to be determined, wherein the correlation coefficient characterizes the degree of linear correlation between the factor to be determined and the railway freight volume;

[0027] Step A3.4: Determine whether the factor to be determined is a target influencing factor based on the correlation coefficient of the factor to be determined and the preset correlation threshold.

[0028] The beneficial effects of adopting the above-mentioned further scheme are: by calculating the grey relational degree of each of the factors to be determined, the relevant data of the factors to be determined contained in the preprocessed data can be simplified; by calculating the correlation coefficient of the factors to be determined, the target influencing factors can be screened out, laying the foundation for the subsequent establishment of a target railway freight volume prediction model for predicting railway freight volume and improving the prediction accuracy of railway freight volume.

[0029] Furthermore, the railway freight volume prediction model includes a first transmission prediction model and a second transmission prediction model. The first transmission prediction model is used to predict railway freight volume based on industry development, and the second transmission prediction model is used to predict railway freight volume based on economic development.

[0030] The method further includes:

[0031] Based on the second historical railway freight volume data, the first transmission prediction model, and the second transmission prediction model, a combined prediction model is constructed. The combined prediction model is used to determine the first weight corresponding to the first transmission prediction model and the second weight corresponding to the second transmission prediction model.

[0032] The beneficial effect of adopting the above-mentioned further scheme is that by assigning corresponding weights to the first transmission prediction model and the second transmission prediction model, and predicting railway freight volume based on the weighted first transmission prediction model and the weighted second transmission prediction model, the accuracy of railway freight volume prediction can be further improved.

[0033] Furthermore, the step of constructing a combined prediction model based on the second historical railway freight volume data, the first transmission prediction model, and the second transmission prediction model includes:

[0034] For the first transmission prediction model, based on the second historical railway freight volume data, the railway freight volume corresponding to multiple times to be predicted is predicted to obtain multiple first railway freight volume estimates corresponding to the first transmission prediction model. The second historical railway freight volume data includes the actual values ​​of railway freight volume corresponding to multiple historical times. Each time to be predicted represents the historical time corresponding to an actual value of railway freight volume included in the second historical railway freight volume data. Each time to be predicted corresponds to a first railway freight volume estimate.

[0035] For the second transmission prediction model, based on the second historical railway freight volume data, the railway freight volume corresponding to multiple times to be predicted is predicted to obtain multiple second railway freight volume estimates corresponding to the second transmission prediction model, and each time to be predicted corresponds to a second railway freight volume estimate.

[0036] For each time to be predicted, the first prediction relative error corresponding to the first transmission prediction model at the time to be predicted is determined based on the first railway freight volume estimate corresponding to the time to be predicted.

[0037] For each time to be predicted, the second prediction relative error corresponding to the second transmission prediction model at the time to be predicted is determined based on the second railway freight volume estimate corresponding to the time to be predicted.

[0038] A combined prediction model is constructed based on the second historical railway freight volume data, multiple first railway freight volume estimates, multiple second railway freight volume estimates, multiple first prediction relative errors, and multiple second prediction relative errors.

[0039] The beneficial effect of adopting the above-mentioned further scheme is that by assigning corresponding weights to the first transmission prediction model and the second transmission prediction model, and predicting railway freight volume based on the weighted first transmission prediction model and the weighted second transmission prediction model, the accuracy of railway freight volume prediction can be further improved.

[0040] Furthermore, step S2 includes:

[0041] Based on the target historical data corresponding to the third-level influencing factors in the target data and the target historical data corresponding to the second-level influencing factors in the target data, the first prediction result corresponding to the second-level influencing factors in the target data is obtained through the second-level influencing factor prediction model;

[0042] Based on the first prediction result corresponding to the secondary influencing factors in the target data and the target historical data corresponding to the direct influencing factors in the target data, the second prediction result corresponding to the direct influencing factors in the target data is obtained through the direct influencing factor prediction model.

[0043] Based on the second prediction result corresponding to the direct influencing factors in the target data, the first transmission prediction model, the second transmission prediction model, the first weight, and the second weight, the predicted value of railway freight volume corresponding to the target data is determined.

[0044] The beneficial effect of adopting the above-mentioned further solution is that by inputting the target data into a pre-built target railway freight volume prediction model to predict railway freight volume, the accuracy of railway freight volume prediction is improved.

[0045] To address the aforementioned technical problems, the present invention also provides a system for predicting railway freight volume, comprising:

[0046] The data acquisition module is used to acquire target data, which includes target historical railway freight volume data and target historical railway freight related data. The target historical railway freight related data are the target historical data corresponding to each of multiple target influencing factors related to railway freight volume. The target influencing factors are factors that have a direct or indirect impact on railway freight volume.

[0047] The railway freight volume prediction model construction module is used to construct a target railway freight volume prediction model based on the target historical data corresponding to the target influencing factors. The target railway freight volume prediction model includes a secondary influencing factor prediction model, a direct influencing factor prediction model, and a railway freight volume prediction model.

[0048] The railway freight volume prediction module is used to input the target data into a pre-built target railway freight volume prediction model to obtain the railway freight volume prediction value corresponding to the target data.

[0049] The railway freight volume prediction model construction module includes:

[0050] The raw data acquisition submodule is used to acquire raw data, which includes first historical railway freight volume data and first historical railway freight related data. The first historical railway freight related data consists of the raw historical data corresponding to each of the multiple factors related to railway freight volume.

[0051] The target influencing factors and data determination submodule is used to determine the target influencing factors and the target historical data corresponding to the target influencing factors based on the original data. The target influencing factors include direct influencing factors, secondary influencing factors and tertiary influencing factors. The target historical data corresponding to the tertiary influencing factors affects the target historical data corresponding to the secondary influencing factors. The target historical data corresponding to the secondary influencing factors affects the target historical data corresponding to the direct influencing factors. The target historical data corresponding to the direct influencing factors affects railway freight volume.

[0052] The secondary influencing factor prediction model construction submodule is used to construct a secondary influencing factor prediction model based on the target historical data corresponding to the tertiary influencing factors and the target historical data corresponding to the secondary influencing factors. The secondary influencing factor prediction model is used to determine the first prediction result corresponding to the secondary influencing factors.

[0053] The direct influencing factor prediction model construction submodule is used to construct a direct influencing factor prediction model based on the first prediction result corresponding to the secondary influencing factor and the target historical data corresponding to the direct influencing factor. The direct influencing factor prediction model is used to determine the second prediction result corresponding to the direct influencing factor.

[0054] The railway freight volume prediction model construction submodule is used to construct a railway freight volume prediction model based on the second prediction result corresponding to the direct influencing factor and the first historical railway freight volume data. The railway freight volume prediction model is used to predict railway freight volume.

[0055] To address the aforementioned technical problems, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method for predicting railway freight volume as described above.

[0056] To address the aforementioned technical problems, the present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method for predicting railway freight volume as described above. Attached Figure Description

[0057] Figure 1 This is a flowchart illustrating the method for predicting railway freight volume in this invention.

[0058] Figure 2 This is a schematic diagram illustrating the relationships between the various factors included in the target influencing factors of this invention;

[0059] Figure 3 This is a schematic diagram of the structure of the artificial neural network model in this invention;

[0060] Figure 4 This is a schematic diagram of the system for predicting railway freight volume in this invention. Detailed Implementation

[0061] The principles and features of the present invention are described below. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0062] Example 1

[0063] like Figure 1 As shown, this embodiment provides a method for predicting railway freight volume, including:

[0064] Step S1: Obtain target data, which includes target historical railway freight volume data and target historical railway freight related data. The target historical railway freight related data are the target historical data corresponding to each of multiple target influencing factors related to railway freight volume. The target influencing factors are factors that have a direct or indirect impact on railway freight volume.

[0065] Step S2: Input the target data into the pre-built target railway freight volume prediction model to obtain the railway freight volume prediction value corresponding to the target data.

[0066] Currently, railway freight volume is the sum of freight volumes from 28 categories. Among them, coal, metallurgical materials (including coke, metallic ores, steel, non-ferrous metals, and non-metallic ores), and containers are the main categories of railway transportation, accounting for more than 85% of my country's railway freight volume. Therefore, this method, based on the Pareto principle, focuses on coal, metallurgical materials, and containers, while also considering other categories. It analyzes the transmission effect of industry development—category freight volume—railway freight volume from the perspective of industry development—category freight volume—railway freight volume. Furthermore, it analyzes the impact of economic and social development on various modes of transportation competition and the impact of various modes of transportation competition on railway freight volume from the perspective of economic development—transportation competition mode freight volume—railway freight volume, thereby determining the target influencing factors.

[0067] The target railway freight volume prediction model is obtained through the following steps:

[0068] The process involves acquiring raw data, which includes first historical railway freight volume data and first historical railway freight-related data. The first historical railway freight-related data consists of raw historical data corresponding to multiple factors related to railway freight volume. In this embodiment, the first historical railway freight volume data includes total railway freight volume and freight volume of major categories. The first historical railway freight-related data includes macroeconomic data, market supply data, and competitive environment data. The macroeconomic data includes GDP, industrial added value, total retail sales of consumer goods, fixed asset investment, real estate development investment, and total import and export volume. The market supply data includes the output of products from upstream and downstream industries of the railway, the production and sales volume of railway freight enterprises, and the railway market share. The competitive environment data includes the freight volume of various modes of transportation.

[0069] Based on the raw data, target influencing factors and corresponding target historical data are determined. The target influencing factors include direct influencing factors, secondary influencing factors, and tertiary influencing factors. The direct influencing factors are primary influencing factors. The target historical data corresponding to the tertiary influencing factors affect the target historical data corresponding to the secondary influencing factors. The target historical data corresponding to the secondary influencing factors affect the target historical data corresponding to the direct influencing factors. The target historical data corresponding to the direct influencing factors affect railway freight volume.

[0070] Based on the target historical data corresponding to the third-level influencing factors and the target historical data corresponding to the second-level influencing factors, a second-level influencing factor prediction model is constructed. The second-level influencing factor prediction model is used to determine the first prediction result corresponding to the second-level influencing factors.

[0071] Based on the first prediction result corresponding to the secondary influencing factor and the target historical data corresponding to the direct influencing factor, a prediction model for the direct influencing factor is constructed. The prediction model for the direct influencing factor is used to determine the second prediction result corresponding to the direct influencing factor.

[0072] Based on the second prediction results corresponding to the direct influencing factors and the first historical railway freight volume data, a railway freight volume prediction model is constructed. The railway freight volume prediction model is used to predict railway freight volume. The railway freight volume prediction model includes a first transmission prediction model and a second transmission prediction model. The first transmission prediction model is used to predict railway freight volume based on industry development, and the second transmission prediction model is used to predict railway freight volume based on economic development.

[0073] The target railway freight volume prediction model includes the secondary influencing factor prediction model, the direct influencing factor prediction model, and the railway freight volume prediction model.

[0074] The step of determining the target influencing factors and the corresponding historical data of the target influencing factors based on the original data includes:

[0075] Step A1 involves cleaning the first historical railway freight volume data to obtain the second historical railway freight volume data; cleaning the first historical railway freight-related data to obtain processed railway freight-related data, wherein the processed railway freight-related data consists of target historical data corresponding to multiple factors related to railway freight volume; in this embodiment, cleaning the first historical railway freight volume data and the first historical railway freight-related data includes determining whether there are missing values ​​in the data; if there are missing values, the estimated values ​​of continuous variables are calculated using time series methods to fill them, and the estimated values ​​of discrete variables are calculated using the hot calorimetry method to fill them, and the form and dimensions of the data are unified according to the data characteristics;

[0076] Step A2: Based on the data category, the second historical railway freight volume data and the processed railway freight-related data are grouped to obtain preprocessed data. The preprocessed data includes the second historical railway freight volume data and data related to factors to be determined. The data related to factors to be determined consists of target historical data corresponding to multiple factors to be determined. The factors to be determined are factors that have a direct or indirect impact on railway freight volume. In this embodiment, the binning method is used to classify and summarize the second historical railway freight volume data and the processed railway freight-related data under multiple application scenarios of total volume, category, and channel.

[0077] Step A3: Based on the preprocessed data, determine the target influencing factor. The plurality of factors to be determined include the target influencing factor. The target historical data corresponding to the target influencing factor is the target historical data corresponding to each of the plurality of factors to be determined.

[0078] In step A3, determining the target influencing factors based on the preprocessed data includes:

[0079] Step A3.1: For each of the factors to be determined, the grey relational degree of the factor to be determined is determined based on the target historical data corresponding to the factor to be determined and the second historical railway freight volume data. The grey relational degree characterizes the degree of correlation between the factor to be determined and the railway freight volume.

[0080] Step A3.2: For each of the to-be-determined factors, if the grey correlation degree of the to-be-determined factor is greater than the preset correlation degree threshold, then execute Step A3.3; wherein, for the value of the correlation degree threshold, it is generally considered in the art that a factor with a corresponding correlation degree R greater than 0.9 (i.e., a factor related to railway freight volume) is very closely related to railway freight volume, a factor with a corresponding correlation degree R satisfying 0.7 < R ≤ 0.9 has a good correlation with railway freight volume and can be used as a key factor, a factor with a corresponding correlation degree R satisfying 0.6 ≤ R ≤ 0.7 has a general correlation with railway freight volume and can be selectively determined as a key factor according to actual research needs, and a factor with a corresponding correlation degree R less than 0.6 is a weakly correlated factor and is not used as a key factor. To ensure that the selected factors are comprehensive and can better reflect the influence on railway freight volume, in this embodiment, the value of the correlation degree threshold is 0.7;

[0081] Step A3.3: Determine the correlation coefficient of the to-be-determined factor according to the target historical data corresponding to the to-be-determined factor and the second historical railway freight volume data, and the correlation coefficient characterizes the linear correlation degree between the to-be-determined factor and railway freight volume;

[0082] Step A3.4: Determine whether the to-be-determined factor is a target influencing factor according to the correlation coefficient of the to-be-determined factor and the preset correlation degree threshold. Among them, for the value of the correlation degree threshold, it is generally considered in the art that a factor with a corresponding correlation coefficient C greater than 0.8 has a very strong correlation with railway freight volume, a factor with a corresponding correlation coefficient C satisfying 0.6 < C ≤ 0.8 has a strong correlation with railway freight volume, a factor with a corresponding correlation coefficient C satisfying 0.4 < C ≤ 0.6 has a medium correlation with railway freight volume, a factor with a corresponding correlation coefficient C satisfying 0.2 < C ≤ 0.4 has a weak correlation with railway freight volume; a factor with a corresponding correlation coefficient C satisfying 0 ≤ C ≤ 0.2 has a relatively weak correlation with railway freight volume. To ensure that the selected factors are comprehensive and can better reflect the influence on railway freight volume, in this embodiment, the value of the correlation degree threshold is 0.6, and the to-be-determined factor with a corresponding correlation coefficient greater than 0.6 is determined as a target influencing factor.

[0083] This method can obtain the importance degree of each factor on railway freight volume by analyzing the strength of the relationship between railway freight volume and each factor. In this embodiment, to calculate the grey correlation degree of each to-be-determined factor, first extract a reference sequence (i.e., multiple second historical railway freight volume data) and a set of comparison sequences from the preprocessed data. Each comparison sequence in the set of comparison sequences corresponds to a set of target historical data corresponding to a to-be-determined factor at multiple moments. According to the reference sequence and the set of comparison sequences, a first data sequence is formed through the first formula, and the first formula is:

[0084]

[0085] Where (X0,X1,X2,…,X) M () represents the first data sequence, t represents the amount of data contained in the reference sequence / each comparison sequence, M represents the total number of factors to be determined, X0 represents the reference sequence, X0(l) represents the data corresponding to time l in the reference sequence, where 1≤l≤t; (X1,X2,…,X) M ) represents the set of comparison sequences, X h X represents the h-th comparison sequence in the set of comparison sequences. h (l) represents the comparison sequence X h The data corresponding to time l, where 1 ≤ h ≤ M;

[0086] Then, for each comparison sequence, the correlation coefficient between each data point in the comparison sequence and each data point in the reference sequence at time l is calculated using a second formula, which is:

[0087]

[0088] Where, ζ h (l) represents X0(l) at time l in the reference sequence X0 and the comparison sequence X h X corresponding to time l h The relative difference between (l) (i.e., X0(l) and X) h (l) correlation coefficient), ρ is the resolution coefficient, representing For the correlation coefficient ζ h The degree of influence of (l) is usually ρ = 0.5;

[0089] Finally, for each of the factors to be determined, the grey relational degree of the factor to be determined is calculated using the third formula, which is:

[0090]

[0091] Where, r oh This represents the grey relational degree of the factor to be determined corresponding to the h-th comparison sequence.

[0092] This embodiment uses Pearson correlation coefficient analysis to determine the correlation coefficient of the factor to be determined. The specific method is as follows:

[0093] Factors whose corresponding grey relational degree is greater than the relational degree threshold are taken as target factors to be determined. Based on the comparison sequence set, a second data sequence is constructed using a fourth formula, wherein the fourth formula is:

[0094]

[0095] Where (X1,X2,…,X) M' ) represents the second data sequence, M' represents the total number of the target factors to be determined, and X h' Let X represent the h'-th comparison sequence in the set of comparison sequences. h' (l) represents the comparison sequence X h' The data corresponding to time l, where 1 ≤ h' ≤ M', each of the comparison sequences X h' A set of historical data corresponding to a target factor to be determined at multiple times;

[0096] For each of the target factors to be determined, the correlation coefficient of the target factors to be determined is calculated using the fifth formula, which is:

[0097]

[0098] Where r represents the comparison sequence X h' Corresponding target factors to be determined and comparison sequence X j The correlation coefficient between the corresponding target factors to be determined, COV(X) h' ,X j ) is the comparison sequence X h' Compared with the comparison sequence X j Covariance between them, VarX h' For the comparison sequence X h' The variance between the data points, VarX j For the comparison sequence X j The variance between the data points, where h'≠j and 1≤j≤N'; in this embodiment, if the comparison sequence X h' The corresponding target factors to be determined and the comparison sequence X j If the correlation coefficient between the corresponding target factors to be determined is greater than 0.6, then the comparison sequence X... h' The corresponding target factors to be determined and the comparison sequence X j The corresponding factors that need to be determined for the target are all factors that affect the target.

[0099] In this embodiment, based on the preprocessed data and according to the data change pattern, the relevant data of the factors to be determined are first simplified using the grey relational analysis method (corresponding to steps A3.1 and A3.2 above). Then, the target influencing factors and their corresponding target historical data are determined using the Pearson correlation coefficient analysis method (corresponding to steps A3.3 and A3.4 above). Finally, the category of each target influencing factor is determined using the principal component analysis method combined with manual judgment. The target influencing factors are divided into tertiary influencing factors, secondary influencing factors, and direct influencing factors.

[0100] like Figure 2 As shown, in this embodiment, the secondary influencing factors come in three forms: one is having two or more key influencing factors (i.e., the target historical data corresponding to the secondary influencing factor is affected by the target historical data corresponding to two or more of the tertiary influencing factors), another is having one key influencing factor, and the third is having no key influencing factor. Among these, there are six secondary influencing factors with two or more key influencing factors, namely, thermal power generation X. 11 Total retail sales of consumer goods X 31 Total import and export volume X 32 Port container throughput X 33 Express delivery volume X 54 and port cargo throughput X 61 There are a total of 6 secondary influencing factors with one key influencing factor, namely, coal import volume X. 13 Iron ore imports X 22 Steel production X 23 Grain output X 41 Fertilizer production X 42 The added value of the secondary industry X 63 There are four secondary influencing factors that have no key influencing factors: coal production X. 12 Iron ore production X 21 Total highway mileage X 52 And the production of freight vehicles X 53 .

[0101] The secondary influencing factor prediction model is obtained through the following steps:

[0102] Based on the preprocessed data, the target historical data (hereinafter referred to as the third-level impact data) corresponding to the third-level impact factors and the target historical data (hereinafter referred to as the second-level impact data) corresponding to the second-level impact factors are obtained;

[0103] For each of the secondary influencing factors, based on the target historical data and the tertiary influencing data corresponding to the secondary influencing factor, a first historical dataset corresponding to the secondary influencing factor is constructed according to the sixth formula, which is:

[0104]

[0105] Among them, (x im ,x ime () represents the first historical dataset corresponding to the im-th second influencing factor, x im Let x represent the set of target historical data corresponding to the im-th second influencing factor from time 1 to time t. im (l) represents the target historical data corresponding to the imth second influencing factor at time l, x ime (e is a variable, 1≤e≤u) represents the set of target historical data corresponding to the e-th third influencing factor from time 1 to time t, x ime (l) represents the target historical data corresponding to the e-th third influencing factor at time l, and u represents the total number of the three-level influencing factors;

[0106] Secondary influencing factors having two or more key influencing factors are designated as Category A secondary influencing factors. For each Category A secondary influencing factor, a first prediction model is constructed based on the first historical dataset corresponding to the Category A secondary influencing factor. The first prediction model is as follows:

[0107]

[0108] in, This represents the first prediction result corresponding to the a-th Class A secondary influencing factor, E represents the total number of the tertiary influencing factors, β0 is a constant term, and ε is an error term; β ime Assuming other influencing factors remain constant, a change of 1 unit in the target historical data corresponding to the e-th level 3 influencing factor can cause x to... ia The change in;

[0109] To ensure that the loss function of the first prediction model is minimized, this method uses the least squares method to solve for β, where β = (β0, β1, ..., β2). e ) T ,get:

[0110] β=(x ime T x ime ) -1 x ime T x ia

[0111] The secondary influencing factors having a key influencing factor are designated as Category B secondary influencing factors. For each Category B secondary influencing factor, a second prediction model is constructed based on the first historical dataset corresponding to the Category B secondary influencing factor. The second prediction model is as follows:

[0112]

[0113]

[0114] in, x represents the first prediction result corresponding to the b-th type B secondary influencing factor. im (d) represents the target historical data corresponding to the b-th type B secondary influencing factor at time d, x ime (d) represents the e-th level 3 impact data at time d;

[0115] For the secondary influencing factors that have no key influencing factors, a third prediction model is constructed based on the data published in national policy documents for the secondary influencing factors that have no key influencing factors.

[0116] Therefore, the secondary influencing factor prediction model includes the first prediction model, the second prediction model, and the third prediction model.

[0117] The direct influencing factor prediction model is obtained through the following steps:

[0118] Based on the preprocessed data, obtain the target historical data (hereinafter referred to as direct impact data) corresponding to the secondary impact data and the direct impact factors;

[0119] For each of the directly influencing factors, based on the target historical data and the secondary impact data corresponding to the directly influencing factor, a second historical dataset corresponding to the directly influencing factor is constructed using the seventh formula, which is:

[0120]

[0121] Among them, (x i ,x iw ) represents the second historical dataset corresponding to the i-th direct influencing factor, x i Let x represent the set of target historical data corresponding to the i-th direct influencing factor from time 1 to time t. i (l) represents the i-th directly influential data at time l, x iw (w is a variable, 1≤w≤m) represents the set of target historical data corresponding to the w-th second influencing factor from time 1 to time t, xiw (l) represents the w-th secondary influence data at time l; m represents the total number of the secondary influence factors;

[0122] The secondary impact data is used as training data to train an artificial neural network model to obtain a direct impact factor prediction model. This method correlates the training data with the variables to be output (i.e., the direct impact factors) to calculate the actual output (i.e., the second prediction result corresponding to the direct impact factors), thereby accurately predicting the nonlinear changes in railway transportation demand.

[0123] In this embodiment, the structure of the artificial neural network model is as follows: Figure 3 As shown, the artificial neural network model includes an input layer, a hidden layer, and an output layer, with the input vector X input to the input layer. i =(X i1 ,X i2 ,…,X im ), where m represents the number of neurons in the input layer, and each element in the input vector X corresponds to a first prediction result for one of the secondary influencing factors; the neurons in the hidden layer Z = (Z i1 Z i2 ,…,Z iq ), where q represents the number of neurons in the hidden layer; the output vector X of the output layer i In the structure of the artificial neural network model, the values ​​of m and q are determined based on the second-order influence data. The neurons in the hidden layer are fully connected to the input layer, and the input of the hidden layer is represented as follows:

[0124] Z(1)=ω(1)X i +b(1)

[0125] Where Z(1) is a linear combination of input values ​​in the hidden layer, ω(1) is the weight from the input layer to the hidden layer, and b(1) is the bias from the input layer to the hidden layer;

[0126] In this embodiment, the sigmoid function is used to define the output of the hidden layer, and the output of the hidden layer is expressed as follows:

[0127]

[0128] Similarly, the input to the output layer is represented as:

[0129] Z(2)=ω(2)f(Z(1))+b(2)

[0130] Where Z(2) is a linear combination of the output values ​​in the hidden layer, ω(2) is the weight from the hidden layer to the output layer, and b(2) is the bias from the hidden layer to the output layer;

[0131] The output of the output layer is represented as follows:

[0132]

[0133] The final artificial neural network model is as follows:

[0134] x i =f(ω(2)(f(ω(1)X) i +b(1)))+b(2)).

[0135] The railway freight volume prediction model is obtained through the following steps:

[0136] Based on the directly impacted data, a third historical dataset is constructed using the eighth formula, which is:

[0137]

[0138] Among them, (y,x) i Let y represent the third historical dataset, y represent the set of second historical railway freight volume data corresponding to each time from time 1 to time t, y(l) represent the second historical railway freight volume data corresponding to time 1, and x represent the second historical railway freight volume data corresponding to time t. i Let x represent the set of target historical data corresponding to the i-th direct influencing factor from time 1 to time t. i (l) represents the i-th directly influential data at time l;

[0139] Based on the transmission effect of industry development—various product categories—railway freight volume, a first transmission prediction model is constructed using the third historical dataset. The first transmission prediction model is as follows:

[0140]

[0141] in, This represents the first estimated railway freight volume obtained through the first transmission prediction model. This represents the second prediction result corresponding to the i-th direct influencing factor obtained through the direct influencing factor prediction model. The first transmission prediction model is used to predict the railway freight volume corresponding to the direct influencing factor determined by the transmission effect of industry development - various categories - railway freight volume (i.e., the first railway freight volume estimate). In this embodiment, the direct influencing factors determined by the transmission effect of industry development - various categories - railway freight volume are coal freight volume X1, metallurgical material freight volume X2, container freight volume X3, and other category freight volume X4.

[0142] Based on the transmission effect of economic development, transportation competition mode volume, and railway freight volume, a second transmission prediction model is constructed using the third historical dataset. The second transmission prediction model is as follows:

[0143]

[0144]

[0145] in, This represents the estimated second railway freight volume obtained through the second transmission prediction model. The second prediction result is the i-th direct influencing factor at time d. The second transmission prediction model is used to predict the railway freight volume (i.e. the second railway freight volume estimate) corresponding to the direct influencing factor determined by the transmission effect of economic development, transportation competition mode volume, and railway freight volume. In this embodiment, the direct influencing factors determined by the transmission effect of economic development, transportation competition mode volume, and railway freight volume are road freight volume X5 and water freight volume X6, respectively.

[0146] The railway freight volume forecast is determined based on the first railway freight volume estimate obtained by the first transmission forecast model and the second railway freight volume estimate obtained by the second transmission forecast model.

[0147] This method employs a multi-model combination approach for prediction. First, based on the three-level impact data, the two-level impact factor prediction model is used to predict the two-level impact factors, resulting in a first prediction result for each two-level impact factor. Then, based on the first prediction result, the direct impact factor prediction model is used to predict the direct impact factors, resulting in a second prediction result for each direct impact factor. Finally, based on the second prediction result and historical railway freight volume (i.e., the second historical railway freight volume data), the railway freight volume prediction model is used to predict the railway freight volume.

[0148] To further improve the accuracy of railway freight volume forecasting, the method further includes:

[0149] Based on the second historical railway freight volume data, the first transmission prediction model, and the second transmission prediction model, a combined prediction model is constructed. The combined prediction model is used to determine the first weight corresponding to the first transmission prediction model and the second weight corresponding to the second transmission prediction model.

[0150] The step of constructing a combined prediction model based on the second historical railway freight volume data, the first transmission prediction model, and the second transmission prediction model includes:

[0151] For the first transmission prediction model, based on the second historical railway freight volume data, the railway freight volume corresponding to multiple times to be predicted is predicted to obtain multiple first railway freight volume estimates corresponding to the first transmission prediction model. The second historical railway freight volume data includes the actual values ​​of railway freight volume corresponding to multiple historical times. Each time to be predicted represents the historical time corresponding to an actual value of railway freight volume included in the second historical railway freight volume data. Each time to be predicted corresponds to a first railway freight volume estimate.

[0152] For the second transmission prediction model, based on the second historical railway freight volume data, the railway freight volume corresponding to multiple times to be predicted is predicted to obtain multiple second railway freight volume estimates corresponding to the second transmission prediction model, and each time to be predicted corresponds to a second railway freight volume estimate.

[0153] For each time to be predicted, the first prediction relative error corresponding to the first transmission prediction model at the time to be predicted is determined based on the first railway freight volume estimate corresponding to the time to be predicted.

[0154] For each time to be predicted, the second prediction relative error corresponding to the second transmission prediction model at the time to be predicted is determined based on the second railway freight volume estimate corresponding to the time to be predicted.

[0155] A combined prediction model is constructed based on the second historical railway freight volume data, multiple first railway freight volume estimates, multiple second railway freight volume estimates, multiple first prediction relative errors, and multiple second prediction relative errors.

[0156] In this embodiment, the combined prediction model is obtained through the following steps:

[0157] Assuming a historical time point N, based on the second historical railway freight volume data, the railway freight volume corresponding to time N is predicted using the j-th (j=1,2) transmission prediction model, thus obtaining the estimated railway freight volume corresponding to this model. (When j=1, the j-th transmission prediction model is the first transmission prediction model, and the estimated railway freight volume corresponding to the j-th transmission prediction model is...) The first railway freight volume estimate obtained through the first transmission prediction model. When j=2, the j-th transmission prediction model is the second transmission prediction model, and the estimated railway freight volume corresponding to the j-th transmission prediction model is... The second railway freight volume estimate obtained through the second transmission prediction model. );

[0158] Based on the estimated railway freight volume corresponding to the j-th transmission prediction model The predictive effectiveness of the j-th transmission prediction model can be determined using the ninth formula, which is:

[0159]

[0160] Among them, U j A represents the predictive effectiveness of the j-th transmission prediction model. jn =1-|e jn |,A jn Let Q be the prediction accuracy of the j-th propagation prediction model at time n. n Let Q be the discrete probability distribution of the j-th transmission prediction model at time n (if the prior information of the weighting coefficients for the prediction accuracy of the j-th transmission prediction model cannot be determined, then Q...). n = 1 / N, where Q of the first and second transmission prediction models at time n is... n (same), e jn Let E(A) be the relative error of the j-th propagation prediction model at time n (i.e., the first relative error of the first propagation prediction model at time n or the second relative error of the second propagation prediction model at time n). j Let A be the prediction accuracy sequence of the j-th transmission prediction model. j The mathematical expectation, σ(A) j Let A be the prediction accuracy sequence of the j-th transmission prediction model. j The standard deviation, where A j =[A j1 A j2 ,…,A jN ];

[0161] Assume A n The combined prediction accuracy at time n is defined as follows: the combined prediction involves combining the first and second transmission prediction models to predict railway freight volume. The combined prediction accuracy is determined using the tenth formula, which is:

[0162]

[0163] Among them, e n Let y be the relative error of the combination prediction at time n. n Let n be the railway freight volume at time n. For the combined prediction of time n, l represents the estimated railway freight volume. j denoted as the weight of the j-th transmission prediction model.

[0164] The effectiveness M of the combined prediction at time N is expressed as:

[0165]

[0166] Wherein, E(A) is the expected value of the prediction accuracy sequence A combining the first transmission prediction model and the second transmission prediction model, and σ(A) is the standard deviation of the prediction accuracy sequence A;

[0167] The combined prediction model is:

[0168] maxM(l1,l2)=E(A)(1-σ(A))

[0169]

[0170] Step S2 includes:

[0171] Based on the target historical data corresponding to the third-level influencing factors in the target data and the target historical data corresponding to the second-level influencing factors in the target data, the first prediction result corresponding to the second-level influencing factors in the target data is obtained through the second-level influencing factor prediction model;

[0172] Based on the first prediction result corresponding to the secondary influencing factors in the target data and the target historical data corresponding to the direct influencing factors in the target data, the second prediction result corresponding to the direct influencing factors in the target data is obtained through the direct influencing factor prediction model.

[0173] Based on the second prediction result corresponding to the direct influencing factors in the target data, the first transmission prediction model, the second transmission prediction model, the first weight, and the second weight, the predicted value of railway freight volume corresponding to the target data is determined.

[0174] In this embodiment, the formula for calculating the predicted railway freight volume corresponding to the target data is as follows:

[0175]

[0176] in, The dimensionless predicted value of railway freight volume corresponding to the target data is specifically the predicted value of railway freight volume for the target data. The dimensionless value of l when j=1 j This represents the first weight corresponding to the first transmission prediction model. This represents the first railway freight volume estimate obtained by predicting the target data through the first transmission prediction model; when j=2, l j This represents the second weight corresponding to the second transmission prediction model. This indicates the second railway freight volume estimate obtained by predicting the target data through the second transmission prediction model.

[0177] Example 2

[0178] Similar in principle to the railway freight volume prediction method described in Embodiment 1 above, this embodiment provides a railway freight volume prediction system, such as... Figure 4 As shown, the system includes:

[0179] The data acquisition module is used to acquire target data, which includes target historical railway freight volume data and target historical railway freight related data. The target historical railway freight related data are the target historical data corresponding to each of multiple target influencing factors related to railway freight volume. The target influencing factors are factors that have a direct or indirect impact on railway freight volume.

[0180] The railway freight volume prediction model construction module is used to construct a target railway freight volume prediction model based on the target historical data corresponding to the target influencing factors. The target railway freight volume prediction model includes a secondary influencing factor prediction model, a direct influencing factor prediction model, and a railway freight volume prediction model.

[0181] The railway freight volume prediction module is used to input the target data into a pre-built target railway freight volume prediction model to obtain the railway freight volume prediction value corresponding to the target data.

[0182] The railway freight volume prediction model construction module includes:

[0183] The raw data acquisition submodule is used to acquire raw data, which includes first historical railway freight volume data and first historical railway freight related data. The first historical railway freight related data consists of the raw historical data corresponding to each of the multiple factors related to railway freight volume.

[0184] The target influencing factors and data determination submodule is used to determine the target influencing factors and the target historical data corresponding to the target influencing factors based on the original data. The target influencing factors include direct influencing factors, secondary influencing factors and tertiary influencing factors. The direct influencing factors are primary influencing factors. The target historical data corresponding to the tertiary influencing factors affect the target historical data corresponding to the secondary influencing factors. The target historical data corresponding to the secondary influencing factors affect the target historical data corresponding to the direct influencing factors. The target historical data corresponding to the direct influencing factors affect railway freight volume.

[0185] The secondary influencing factor prediction model construction submodule is used to construct a secondary influencing factor prediction model based on the target historical data corresponding to the tertiary influencing factors and the target historical data corresponding to the secondary influencing factors. The secondary influencing factor prediction model is used to determine the first prediction result corresponding to the secondary influencing factors.

[0186] The direct influencing factor prediction model construction submodule is used to construct a direct influencing factor prediction model based on the first prediction result corresponding to the secondary influencing factor and the target historical data corresponding to the direct influencing factor. The direct influencing factor prediction model is used to determine the second prediction result corresponding to the direct influencing factor.

[0187] The railway freight volume prediction model construction submodule is used to construct a railway freight volume prediction model based on the second prediction result corresponding to the direct influencing factor and the first historical railway freight volume data. The railway freight volume prediction model is used to predict railway freight volume.

[0188] The target influencing factors and data determination submodule includes:

[0189] The data cleaning unit is used to perform data cleaning processing on the first historical railway freight volume data to obtain the second historical railway freight volume data; and to perform data cleaning processing on the first historical railway freight-related data to obtain the processed railway freight-related data, wherein the processed railway freight-related data are the target historical data corresponding to each of the multiple factors related to railway freight volume.

[0190] The grouping processing unit is used to group the second historical railway freight volume data and the processed railway freight-related data according to the data category to obtain preprocessed data. The preprocessed data includes the second historical railway freight volume data and data related to factors to be determined. The data related to factors to be determined are target historical data corresponding to multiple factors to be determined. The factors to be determined are factors that have a direct impact on railway freight volume or factors that have an indirect impact on railway freight volume.

[0191] The target influencing factor and data determination unit is used to determine the target influencing factor based on the preprocessed data. The plurality of factors to be determined include the target influencing factor, and the target historical data corresponding to the target influencing factor is the target historical data corresponding to each of the plurality of factors to be determined.

[0192] The target influencing factors and data determination unit includes:

[0193] The grey relational degree determination subunit is used to determine the grey relational degree of each factor to be determined based on the target historical data and the second historical railway freight volume data corresponding to the factor to be determined. The grey relational degree characterizes the degree of correlation between the factor to be determined and the railway freight volume.

[0194] The factor simplification subunit is used to execute the correlation coefficient determination subunit if the gray correlation degree of each factor to be determined is greater than the preset correlation degree threshold.

[0195] The correlation coefficient determination subunit is used to determine the correlation coefficient of the factor to be determined based on the target historical data corresponding to the factor to be determined and the second historical railway freight volume data. The correlation coefficient characterizes the degree of linear correlation between the factor to be determined and the railway freight volume.

[0196] The target influencing factor determination subunit is used to determine whether the factor to be determined is a target influencing factor based on the correlation coefficient of the factor to be determined and a preset correlation threshold.

[0197] The railway freight volume prediction model includes a first transmission prediction model and a second transmission prediction model. The first transmission prediction model is used to predict railway freight volume based on industry development, and the second transmission prediction model is used to predict railway freight volume based on economic development.

[0198] The system also includes a weight setting submodule, which is used for:

[0199] Based on the second historical railway freight volume data, the first transmission prediction model, and the second transmission prediction model, a combined prediction model is constructed. The combined prediction model is used to determine the first weight corresponding to the first transmission prediction model and the second weight corresponding to the second transmission prediction model.

[0200] Specifically, the weight setting submodule, when constructing a combined prediction model based on the second historical railway freight volume data, the first transmission prediction model, and the second transmission prediction model, is used for:

[0201] For the first transmission prediction model, based on the second historical railway freight volume data, the railway freight volume corresponding to multiple times to be predicted is predicted to obtain multiple first railway freight volume estimates corresponding to the first transmission prediction model. The second historical railway freight volume data includes the actual values ​​of railway freight volume corresponding to multiple historical times. Each time to be predicted represents the historical time corresponding to an actual value of railway freight volume included in the second historical railway freight volume data. Each time to be predicted corresponds to a first railway freight volume estimate.

[0202] For the second transmission prediction model, based on the second historical railway freight volume data, the railway freight volume corresponding to multiple times to be predicted is predicted to obtain multiple second railway freight volume estimates corresponding to the second transmission prediction model, and each time to be predicted corresponds to a second railway freight volume estimate.

[0203] For each time to be predicted, the first prediction relative error corresponding to the first transmission prediction model at the time to be predicted is determined based on the first railway freight volume estimate corresponding to the time to be predicted.

[0204] For each time to be predicted, the second prediction relative error corresponding to the second transmission prediction model at the time to be predicted is determined based on the second railway freight volume estimate corresponding to the time to be predicted.

[0205] A combined prediction model is constructed based on the second historical railway freight volume data, multiple first railway freight volume estimates, multiple second railway freight volume estimates, multiple first prediction relative errors, and multiple second prediction relative errors.

[0206] The railway freight volume prediction module includes:

[0207] The first prediction result determination submodule is used to obtain the first prediction result corresponding to the second-level influencing factors in the target data based on the target historical data corresponding to the third-level influencing factors in the target data and the target historical data corresponding to the second-level influencing factors in the target data, through the second-level influencing factor prediction model.

[0208] The second prediction result determination submodule is used to obtain the second prediction result corresponding to the direct influencing factor in the target data based on the first prediction result corresponding to the secondary influencing factor in the target data and the target historical data corresponding to the direct influencing factor in the target data through the direct influencing factor prediction model.

[0209] The railway freight volume forecast determination submodule is used to determine the railway freight volume forecast value corresponding to the target data based on the second forecast result corresponding to the direct influencing factors in the target data, the first transmission forecast model, the second transmission forecast model, the first weight, and the second weight.

[0210] Example 3

[0211] To address the aforementioned technical problems, this embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the railway freight volume prediction method as described in Embodiment 1.

[0212] Example 4

[0213] To address the aforementioned technical problems, this embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the railway freight volume prediction method as described in Embodiment 1.

[0214] In the description of this invention, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0215] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0216] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for forecasting railway freight traffic volume, characterized by, Comprise: Step S1, obtaining target data, the target data comprising target historical railway freight volume data and target historical railway freight related data, the target historical railway freight related data being target historical data corresponding to each of a plurality of target influencing factors related to railway freight volume, the target influencing factors being factors having direct or indirect influence on railway freight volume; Step S2, inputting the target historical railway freight related data in the target data into a pre-constructed target railway freight volume prediction model to obtain a railway freight volume prediction value corresponding to the target data; Wherein, the target railway freight volume prediction model is obtained by the following steps: Obtaining original data, the original data comprising first historical railway freight volume data and first historical railway freight related data, the first historical railway freight related data being original historical data corresponding to each of a plurality of factors related to railway freight volume; According to the original data, determining target influencing factors and target historical data corresponding to the target influencing factors, the target influencing factors comprising direct influencing factors, secondary influencing factors and tertiary influencing factors, the direct influencing factors being primary influencing factors, the target historical data corresponding to the tertiary influencing factors affecting the target historical data corresponding to the secondary influencing factors, the target historical data corresponding to the secondary influencing factors affecting the target historical data corresponding to the direct influencing factors, the target historical data corresponding to the direct influencing factors affecting railway freight volume; According to the target historical data corresponding to the tertiary influencing factors and the target historical data corresponding to the secondary influencing factors, constructing a secondary influencing factor prediction model, the secondary influencing factor prediction model being used to determine a first prediction result corresponding to the secondary influencing factors; According to the first prediction result corresponding to the secondary influencing factors and the target historical data corresponding to the direct influencing factors, constructing a direct influencing factor prediction model, the direct influencing factor prediction model being used to determine a second prediction result corresponding to the direct influencing factors; According to the second prediction result corresponding to the direct influencing factors and the first historical railway freight volume data, constructing a railway freight volume prediction model, the railway freight volume prediction model being used to predict railway freight volume; The target railway freight volume prediction model comprises the secondary influencing factor prediction model, the direct influencing factor prediction model and the railway freight volume prediction model.

2. The method of claim 1, wherein, According to the original data, determining target influencing factors and target historical data corresponding to the target influencing factors, comprising: Step A1, performing data cleaning processing on the first historical railway freight volume data to obtain second historical railway freight volume data; performing data cleaning processing on the first historical railway freight related data to obtain processed railway freight related data, the processed railway freight related data being target historical data corresponding to each of a plurality of factors related to railway freight volume; Step A2, grouping the second historical railway freight volume data and the processed railway freight related data according to data categories to obtain preprocessed data, the preprocessed data including the second historical railway freight volume data and to-be-determined factor related data, the to-be-determined factor related data being target historical data corresponding to each of a plurality of to-be-determined factors, the to-be-determined factor being a factor directly affecting railway freight volume or a factor indirectly affecting railway freight volume; Step A3, determining the target influence factor according to the preprocessed data, the plurality of to-be-determined factors including the target influence factor, the target historical data corresponding to the target influence factor being the target historical data corresponding to each of the plurality of to-be-determined factors.

3. The method of claim 2, wherein, In the step A3, the determining the target influence factor according to the preprocessed data includes: Step A3.1, for each to-be-determined factor, determining a grey correlation degree of the to-be-determined factor according to the target historical data corresponding to the to-be-determined factor and the second historical railway freight volume data, the grey correlation degree representing a correlation degree between the to-be-determined factor and railway freight volume; Step A3.2, for each to-be-determined factor, if the grey correlation degree of the to-be-determined factor is greater than a preset correlation degree threshold, performing step A3.3; Step A3.3, determining a correlation coefficient of the to-be-determined factor according to the target historical data corresponding to the to-be-determined factor and the second historical railway freight volume data, the correlation coefficient representing a linear correlation degree between the to-be-determined factor and railway freight volume; Step A3.4, determining whether the to-be-determined factor is a target influence factor according to the correlation coefficient of the to-be-determined factor and a preset correlation degree threshold.

4. The method of claim 2, wherein, The railway freight volume prediction model includes a first conduction prediction model and a second conduction prediction model, the first conduction prediction model being used to predict railway freight volume according to industry development, and the second conduction prediction model being used to predict railway freight volume according to economic development; The method further includes: constructing a combined prediction model according to the second historical railway freight volume data, the first conduction prediction model and the second conduction prediction model, the combined prediction model being used to determine a first weight corresponding to the first conduction prediction model and a second weight corresponding to the second conduction prediction model.

5. The method of claim 4, wherein, The constructing a combined prediction model according to the second historical railway freight volume data, the first conduction prediction model and the second conduction prediction model includes: for the first conduction prediction model, predicting railway freight volume corresponding to a plurality of to-be-predicted time points according to the second historical railway freight volume data to obtain a plurality of first railway freight volume estimated values corresponding to the first conduction prediction model, wherein the second historical railway freight volume data contains railway freight volume actual values corresponding to a plurality of historical time points, each to-be-predicted time point representing a historical time point corresponding to one railway freight volume actual value contained in the second historical railway freight volume data, and each to-be-predicted time point corresponds to one first railway freight volume estimated value; For the second conduction prediction model, according to the second historical railway freight volume data, railway freight volumes corresponding to a plurality of the to-be-predicted time points are predicted, a plurality of second railway freight volume estimated values corresponding to the second conduction prediction model are obtained, and each to-be-predicted time point corresponds to a second railway freight volume estimated value; For each to-be-predicted time point, according to the first railway freight volume estimated value corresponding to the to-be-predicted time point, a first prediction relative error corresponding to the first conduction prediction model at the to-be-predicted time point is determined; For each to-be-predicted time point, according to the second railway freight volume estimated value corresponding to the to-be-predicted time point, a second prediction relative error corresponding to the second conduction prediction model at the to-be-predicted time point is determined; According to the second historical railway freight volume data, a plurality of the first railway freight volume estimated values, a plurality of the second railway freight volume estimated values, a plurality of the first prediction relative errors and a plurality of the second prediction relative errors, a combined prediction model is constructed.

6. The method according to claim 4 or 5, characterized in that, The step S2 comprises: According to the target historical data corresponding to the third influencing factor in the target data and the target historical data corresponding to the second influencing factor in the target data, a first prediction result corresponding to the second influencing factor in the target data is obtained through the second influencing factor prediction model; According to the first prediction result corresponding to the second influencing factor in the target data and the target historical data corresponding to the direct influencing factor in the target data, a second prediction result corresponding to the direct influencing factor in the target data is obtained through the direct influencing factor prediction model; According to the second prediction result corresponding to the direct influencing factor in the target data, the first conduction prediction model, the second conduction prediction model, the first weight and the second weight, a railway freight volume prediction value corresponding to the target data is determined.

7. A system for forecasting railway freight volumes, characterized by, Comprise: A data acquisition module is configured to acquire target data, wherein the target data comprises target historical railway freight volume data and target historical railway freight related data, the target historical railway freight related data comprises target historical data corresponding to a plurality of target influencing factors related to railway freight volume, and the target influencing factors are factors having direct or indirect influence on railway freight volume; A railway freight volume prediction model construction module is configured to construct a target railway freight volume prediction model according to target historical data corresponding to the target influencing factors, wherein the target railway freight volume prediction model comprises a second influencing factor prediction model, a direct influencing factor prediction model and a railway freight volume prediction model; A railway freight volume prediction module is configured to input target historical railway freight related data in the target data into a pre-constructed target railway freight volume prediction model to obtain a railway freight volume prediction value corresponding to the target data; The railway freight volume prediction model construction module comprises: An original data acquisition submodule is configured to acquire original data, wherein the original data includes first historical railway freight volume data and first historical railway freight related data, and the first historical railway freight related data is original historical data corresponding to multiple factors related to railway freight volume; A target influence factor and data determination submodule is configured to determine target influence factors and target historical data corresponding to the target influence factors according to the original data, wherein the target influence factors include direct influence factors, secondary influence factors and tertiary influence factors, the target historical data corresponding to the tertiary influence factors affect the target historical data corresponding to the secondary influence factors, the target historical data corresponding to the secondary influence factors affect the target historical data corresponding to the direct influence factors, and the target historical data corresponding to the direct influence factors affect railway freight volume; A secondary influence factor prediction model construction submodule is configured to construct a secondary influence factor prediction model according to the target historical data corresponding to the tertiary influence factors and the target historical data corresponding to the secondary influence factors, wherein the secondary influence factor prediction model is configured to determine a first prediction result corresponding to the secondary influence factors; A direct influence factor prediction model construction submodule is configured to construct a direct influence factor prediction model according to the first prediction result corresponding to the secondary influence factors and the target historical data corresponding to the direct influence factors, wherein the direct influence factor prediction model is configured to determine a second prediction result corresponding to the direct influence factors; A railway freight volume prediction model construction submodule is configured to construct a railway freight volume prediction model according to the second prediction result corresponding to the direct influence factors and the first historical railway freight volume data, wherein the railway freight volume prediction model is configured to predict railway freight volume.

8. An electronic device, comprising: A computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the prediction method for railway freight volume.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the prediction method for railway freight volume.

Images