SARIMA-RF combined model-based driving vehicle number prediction method, system, device, medium and program

Abstract

The invention discloses a running vehicle number prediction method, system, device, medium and program based on an SARIMA-RF combined model, and belongs to the technical field of railway freight volume prediction. The method comprises the following steps: acquiring the monthly data of the running condition of a railway international combined transport train to form a train running condition monthly data set; utilizing an SARIMA model to fit a running vehicle number sequence in the monthly data set of the running condition of the class, and obtaining a predicted value sequence and a residual sequence; fitting the residual error sequence by using an RF model to obtain a prediction result; and combining the predicted value sequence of the SARIMA model with the prediction result of the RF model to obtain a final prediction result of the SARIMA-RF combined model. According to the method, the accuracy of the prediction result of the running number of the international railway intermodal trains is improved, and the running uncertainty of railway transportation and the running uncertainty of the international railway intermodal trains are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of railway freight volume prediction, and in particular to a method, system, device, medium and program for predicting the number of running vehicles based on a SARIMA-RF combined model. Background Art

[0002] As an emerging railway freight mode, international railway transport trains have developed rapidly around the world in recent years with their large capacity, high speed, high stability, and green, low-carbon and environmentally friendly characteristics.

[0003] In the field of railway freight volume forecasting, researchers have explored a variety of forecasting methods, mainly including time series analysis and machine learning. Time series analysis builds mathematical models to fit and forecast historical data to reveal the trends and laws behind the data. For example, the FPSO grey Verhulst model, the GM (1,1) model and its residual correction method, and the improved grey Markov model have all improved the accuracy of railway freight volume forecasting to varying degrees.

[0004] Machine learning algorithms build prediction models by training a large amount of data and use the generalization ability of the model to predict new data. Machine learning algorithms such as random forest, GRU deep network, long short-term memory (LSTM) neural network, etc. have shown good performance in railway freight volume forecasting. In particular, LSTM neural network has high prediction accuracy in railway freight volume forecasting because it can capture long-term dependencies in time series.

[0005] However, although some progress has been made in the prediction of railway freight volume and the number of China-Europe trains, the existing single prediction models and combined prediction models still have some limitations. These models often fail to fully consider the linear and nonlinear characteristics and seasonal cyclical variation characteristics reflected in the time series of railway freight volume and the number of international railway combined transport trains. Linear characteristics are usually manifested as trends and cyclical changes in time series, while nonlinear characteristics may be caused by factors such as emergencies and policy adjustments. Seasonal cyclical variation characteristics reflect the cyclical fluctuations of time series between different seasons. Summary of the invention

[0006] The existing forecasting models for railway freight volume and the number of international railway transport trains in the prior art fail to fully consider the linear and nonlinear characteristics in the time series and the seasonal periodic variation characteristics, resulting in inaccurate forecasting results, which will bring unnecessary risks and uncertainties to the operation of railway transportation and international railway transport trains, affecting transportation efficiency and service quality. The present invention provides a method for forecasting the number of trains based on a SARIMA-RF combined model, which makes a preliminary forecast based on the SARIMA model, and then uses the RF model to predict its residual, and finally combines and corrects the two results to obtain a forecast result, thereby improving the accuracy of the forecast result and reducing the uncertainty of the operation of railway transportation and international railway transport trains.

[0007] In order to achieve the above object, the present invention provides the following technical solutions.

[0008] In a first aspect, the present invention provides a method for predicting the number of running vehicles based on a SARIMA-RF combined model, comprising: Obtain monthly data on the operation of international railway combined transport trains to form a monthly data set on the operation of trains; The SARIMA model is used to fit the train number sequence in the monthly data set of train operation to obtain the predicted value sequence and residual sequence; Use the RF model to fit the residual sequence and get the prediction results; The predicted value sequence of the SARIMA model is combined with the prediction results of the RF model to obtain the final prediction results of the SARIMA-RF combined model.

[0009] As a further improvement of the present invention, the obtaining of monthly data on the operation of international railway combined transport trains to form a monthly data set on the operation of trains includes: The ADF method is used to test the stability of the time series data in the monthly data set of train operation, and the AIC value of the test result is obtained; Select the parameter that best fits the data among the AIC values ​​to complete the SARIMA model construction.

[0010] As a further improvement of the present invention, the SARIMA model is used to fit the running train number sequence in the monthly data set of the train operation situation to obtain the predicted value sequence and the residual sequence, including: Extract the train number sequence from the monthly data set of train operation ; Fitting the running number series using SARIMA model The linear part , calculate the predicted value sequence And the residual sequence , specifically expressed as follows:

[0011] Where: is the sequence of running vehicle numbers; is the sequence of running vehicle numbers The linear part of the is the predicted value sequence; is the residual sequence.

[0012] As a further improvement of the present invention, the residual sequence It is necessary to perform maximum and minimum normalization processing to convert the residual sequence The different eigenvalues ​​in are transformed to the same scale.

[0013] As a further improvement of the present invention, the method of fitting the residual sequence using the RF model to obtain the prediction result includes: Use the RF model to fit the residual sequence and obtain the predicted residual data; Denormalize the predicted residual data to get the predicted results .

[0014] As a further improvement of the present invention, the predicted value sequence of the SARIMA model is combined with the predicted result of the RF model to obtain the final predicted result of the SARIMA-RF combined model, including: The predicted value sequence of the SARIMA model Prediction results of the RF model Combine them to get the final prediction result of the SARIMA-RF combined model , specifically expressed as follows:

[0015] In the formula, The final prediction result; Fitting the residual series of the SARIMA model to the RF model Then the prediction results can be obtained; is the predicted value sequence of the SARIMA model.

[0016] According to the prediction results of the final SARIMA-RF combined model , formulate plans for international railway combined transport trains.

[0017] In a second aspect, the present invention provides a vehicle running number prediction system based on a SARIMA-RF combined model, comprising: Data acquisition module: used to obtain monthly data on the operation of international railway combined transport trains and form a monthly data set on the operation of trains; SARIMA model processing module: used to use the SARIMA model to fit the train number sequence in the monthly data set of train operation, and obtain the predicted value sequence and residual sequence; RF model processing module: used to fit the residual sequence using the RF model to obtain the prediction results; Prediction result output module: used to combine the prediction value sequence of the SARIMA model with the prediction results of the RF model to obtain the final prediction results of the SARIMA-RF combined model.

[0018] In a third aspect, the present invention provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of a method for predicting the number of running vehicles based on a SARIMA-RF combined model when executing the computer program.

[0019] In a fourth aspect, the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for predicting the number of running vehicles based on the SARIMA-RF combined model are implemented.

[0020] In a fifth aspect, the present invention provides a computer program product, comprising computer instructions, which, when executed by a processor, implement the steps of a method for predicting the number of running vehicles based on a SARIMA-RF combined model.

[0021] Compared with the prior art, the present invention has the following beneficial effects: By combining the SARIMA model and the RF model, the present invention can more comprehensively capture the linear and nonlinear characteristics in the time series of railway freight volume and the number of railway international transport trains, as well as the seasonal periodic change characteristics. The SARIMA model is good at processing the linear part and seasonal changes of the time series, while the RF model can capture complex nonlinear relationships. The combination of the two makes the prediction results more accurate. Secondly, since the railway freight volume and the number of trains can be predicted more accurately, the present invention reduces the uncertainty of railway transportation and railway international transport trains. This helps railway transportation companies to better plan transportation resources, optimize train operation plans, and reduce idle or tense resources caused by inaccurate predictions. At the same time, accurate prediction results help railway transportation companies make reasonable transportation arrangements in advance and improve transportation efficiency. In addition, a stable train operation plan can also improve customer satisfaction, thereby improving the overall service quality. Therefore, the prediction method provided by the present invention can provide strong support for the decision-making of railway transportation companies, and formulate more scientific transportation strategies and market development plans to cope with market changes and challenges. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The drawings described herein are for illustrative purposes only and are not intended to limit the scope of the present disclosure in any way. In the drawings: Figure 1 A flowchart of a method for predicting the number of vehicles in operation based on the SARIMA-RF combined model is provided; Figure 2 A schematic diagram of the structure of a combined model of a method for predicting the number of running vehicles based on a SARIMA-RF combined model according to the present invention; Figure 3 A schematic diagram of the structure of the combined model of this embodiment of the present invention; Figure 4 In this embodiment of the present invention, a diagram showing the number of China-Europe trains running in Xi'an from January 2014 to December 2023; Figure 5 is the first-order difference sequence in the embodiment of the present invention as well as Comparison chart; Figure 6 is the first-order seasonal difference sequence in the embodiment of the present invention as well as Comparison chart; Figure 7 This is a schematic diagram of the residual test results of the model in an embodiment of the present invention; Figure 8 A comparison chart of the predicted values ​​and actual values ​​of each model in the embodiment of the present invention; Fig. 9 A point-by-point error comparison diagram of each model in an embodiment of the present invention; Fig.10 It is a structural schematic diagram of a running vehicle number prediction system based on the SARIMA-RF combined model of the present invention; Fig.11 Schematic diagram of an electronic device in an embodiment of the present invention. DETAILED DESCRIPTION

[0023] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the present invention will be clearly and completely described below in conjunction with the drawings in the present invention. The described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in the field without creative work should fall within the scope of protection of the present invention.

[0024] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which the present invention belongs. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0025] The existing forecasting models for railway freight volume and the number of international railway transport trains in the prior art fail to fully consider the linear and nonlinear characteristics of time series and the seasonal periodic variation characteristics, resulting in inaccurate forecasting results, which will bring unnecessary risks and uncertainties to the operation of railway transportation and international railway transport trains, affecting transportation efficiency and service quality. The present invention provides a method for predicting the number of trains in operation based on the SARIMA-RF combined model, such as Figure 1 As shown, the method includes: S101: Obtain monthly data on the operation of international railway combined transport trains to form a monthly data set on the operation of trains; S201: Use the SARIMA model to fit the train number sequence in the monthly data set of train operation to obtain the predicted value sequence and residual sequence; S301: Fit the residual sequence using the RF model to obtain the prediction result; S401: Combine the predicted value sequence of the SARIMA model with the prediction result of the RF model to obtain the final prediction result of the SARIMA-RF combined model.

[0026] The invention makes a preliminary prediction based on the SARIMA model, then uses the RF model to predict its residual, and finally combines and corrects the two results to obtain the prediction result, thereby improving the accuracy of the prediction result and reducing the uncertainty of running railway transportation and railway international combined transport trains.

[0027] The present invention is further explained below in conjunction with specific drawings.

[0028] Traditional time series prediction models take into account the linear features in time series, and the nonlinear features in machine learning models, such as Figure 2 As shown, the present invention constructs a SARIMA-RF combined model for the number of China-Europe trains based on residual correction.

[0029] The SARIMA-RF combined model based on residual correction is used to predict the number of international railway combined transport trains.

[0030] S1: Collect and organize monthly data on the operation of international railway combined transport trains to form a monthly data set on the operation of international railway combined transport trains; A white noise test is performed on the time series data in the monthly data set of train operation to ensure that the data is not random and has autocorrelation.

[0031] Use the ADF (Augmented Dickey-Fuller) method to test the stationarity of the original data time series data; If the original data is not stable, perform first-order difference to obtain a stable sequence, and perform a white noise test on the first-order difference sequence; calculate the autocorrelation coefficient of the first-order difference sequence ( ) and the partial autocorrelation coefficient ( ), determine the periodicity of the model; perform seasonal differentiation on the first-order difference sequence to eliminate the periodic trend, and calculate the autocorrelation coefficient of the seasonal difference sequence ( ) and the partial autocorrelation coefficient ( ).

[0032] Use the AIC (Akaike Information Criterion) criterion to select the optimal SARIMA model parameters . A SARIMA model was constructed, and normality test, DW test and white noise test of the model residual were performed to ensure the applicability of the model.

[0033] Among them, p is the autoregressive order; d is the number of non-seasonal differences; q is the moving average component order; P is the seasonal autoregressive order; D is the number of seasonal differences; Q is the seasonal moving average component order; S is the length of the seasonal cycle.

[0034] The monthly data set of train operation is divided into a training set and a validation set; The SARIMA model is trained with the training set and verified with the validation set. If the verification result is qualified, the construction of the SARIMA model is completed.

[0035] S2: Based on the monthly data set of train operation, the number of trains operated by international railway combined transport trains is obtained. ; Using SARIMA model to fit the train number series of international railway combined transport trains The linear part So as to calculate the predicted value sequence And the residual sequence :

[0036] Where: is the sequence of running vehicle numbers; is the sequence of running vehicle numbers The linear part of the is the predicted value sequence; is the residual sequence.

[0037] For the residual sequence Perform maximum and minimum normalization. Normalization can convert different eigenvalues ​​in the residual sequence to the same scale, making different features comparable. If the scale differences between data features are large, it may cause the model to converge more slowly during training. Through normalization, this scale difference can be reduced, thereby speeding up the convergence of the model and improving the accuracy of predictions. For data with time series characteristics such as time series data, normalization can reduce the impact of data fluctuations on the model and make the data more stable. This helps the model better capture trends and patterns in the data. Therefore, normalized data is easier to visualize and analyze. Since all data are converted to the same scale, it is easier to observe the distribution, trends, and patterns of the data. .

[0038] S3: Since the residual sequence of the SARIMA model prediction results contains the nonlinear characteristics of the time series of the number of trains running on international railway combined transport, considering this feature, it is necessary to use the RF model to predict the residual sequence. Perform the fitting.

[0039] RF model fitting residual sequence Then you can get the prediction results .

[0040] The residual data predicted by the RF model is normalized data and needs to be denormalized to obtain the final prediction result. .

[0041] S4: Combine the linear part of the SARIMA model prediction with the nonlinear part of the RF model prediction to calculate the final prediction result of the SARIMA-RF combined model :

[0042] In the formula, The final prediction result; Fitting the residual sequence to the RF model Then the prediction results can be obtained.

[0043] According to the prediction results of the final SARIMA-RF combined model , formulate relevant plans for international railway combined transport trains.

[0044] S5: Using mean square error , Root Mean Square Error , mean absolute error , mean absolute percentage error The four indicators are used as evaluation indicators for evaluating the prediction effect of the combined model and each single model, and are specifically expressed as follows:

[0045]

[0046]

[0047]

[0048] Where: Refers to the true value of the i-th sample; refers to the predicted value of the i-th sample; T refers to the number of samples.

[0049] The present invention is further explained below in conjunction with specific embodiments.

[0050] Example The SARIMA model in the SARIMA-RF combined model evolved from the ARMA model and is widely used in the prediction of periodic fluctuations. The SARIMA model can better fit the linear characteristics in the prediction time series, and can also characterize the periodic characteristics of the China-Europe Express train operation data, weakening the unsteady problem of the China-Europe Express train operation sequence data. The RF model is a machine learning algorithm proposed by combining the Bagging ensemble learning theory with the random subspace method.

[0051] In order to reduce the prediction error and improve the generalization ability of the model, the present invention combines the SARIMA model and the RF model based on the residual correction method, that is, based on the SARIMA model for preliminary prediction, the RF model is used to predict its residual, and finally the two results are combined and corrected. The process of combination and calculation is as follows Figure 3 shown.

[0052] Based on the fact that the traditional time series prediction model takes into account the linear characteristics in the time series and the nonlinear characteristics in the machine learning model, a SARIMA-RF combined model based on residual correction for the number of China-Europe trains running is constructed. The SARIMA-RF combined model based on residual correction is used to predict the number of China-Europe trains running in Xi'an. First, the SARIMA model is used to fit the number of trains running in Xi'an. The linear part So as to calculate the predicted value sequence And the residual sequence :

[0053] Since the residual sequence of the SARIMA model prediction results contains the nonlinear characteristics of the Xi'an China-Europe Express train number time series, considering this feature, it is necessary to use the RF model to predict the residual sequence. After fitting the residual sequence, the prediction result can be obtained. Finally, the linear part of the SARIMA model prediction and the nonlinear part of the RF model prediction are combined to calculate the final prediction result of the SARIMA-RF model. :

[0054] Mean square error , Root Mean Square Error , mean absolute error , mean absolute percentage error The four indicators are used as evaluation indicators for evaluating the prediction effect of the combined model and each single model, and are specifically expressed as follows:

[0055]

[0056]

[0057]

[0058] Where: Refers to the true value of the i-th sample; refers to the predicted value of the i-th sample; T refers to the number of samples.

[0059] like Figure 4 As shown, the forecast study is conducted based on the monthly data on the operation of the Xi'an "Chang'an" China-Europe freight train from January 2014 to December 2023 provided by the China Railway Bureau Group Co., Ltd.

[0060] The SARIMA model is used to make a preliminary forecast of the time series of monthly data on the number of trains running on the Xi'an "Chang'an" China-Europe freight train. Before the forecast, a white noise test is required for the time series of the number of trains running on the Xi'an "Chang'an" China-Europe freight train. The test statistic P-value is less than the selected significance level of 0.05, indicating that the time series is not random and has autocorrelation, that is, it is a non-white noise series.

[0061] The ADF (augmented dickey-fuller test) method determines whether the data is stationary by testing whether there is a unit root in the time series data. The ADF method is used to test the stationarity of the original time series of the monthly data of the number of China-Europe trains. Since the P-value representing the significance level is 0.956, which is greater than 0.05, the original time series is not stationary. In order to obtain a stationary series, it is necessary to perform a difference on the time series of the monthly data of the number of China-Europe trains to obtain the first-order difference series of the series. A white noise test is performed on the obtained first-order difference series. The obtained test statistic P-value is smaller than the significance level of 0.05, so the first-order difference series obtained by a difference is a non-white noise series. Finally, the ADF method is used again to test the series. The P-value representing the significance level is 0.007, which is less than 0.05, indicating that the series is a stationary series and a model can be constructed for the series.

[0062] Before selecting the relevant parameters of the SARIMA model, it is necessary to obtain the autocorrelation coefficient of the first-order difference sequence ( ) and the partial correlation coefficient ( ) to calculate, such as Figure 5 shown. Figure 5 (a) It appears as a tail, showing periodic oscillations, and its lag order is obviously not 0 at 3, 6, 9, and 12. Figure 5 (b) It also shows a tail. This shows that the time series of the number of China-Europe trains operated by Xi'an "Chang'an" is a periodic series.

[0063] In order to eliminate the cyclical trend of the time series of the number of China-Europe trains operated by Xi'an "Chang'an", the first-order seasonal difference with a period of 3 is first made to the series to obtain a seasonal difference series. The autocorrelation coefficient and partial correlation coefficient of the obtained seasonal difference series are calculated as follows: Figure 6 shown. Figure 6 It can be seen that the periodicity has been basically eliminated, indicating that the first-order seasonal difference with a period of 3 has basically eliminated the cyclical trend of the time series of the number of "Chang'an" China-Europe freight trains in Xi'an.

[0064] The AIC (akaike information criterion) value is a standard for measuring the goodness of statistical model fitting based on information theory. The first-order seasonal difference sequence needs to be tested for stationarity. The ADF method is used for stationarity test. The AIC value of the test result is -90.54. The ADF test results of the first-order seasonal difference sequence are shown in Table 1. Table 1 shows the AIC critical values ​​of the ADF test at 1%, 5% and 10% confidence levels. It can be seen that the AIC values ​​of the test results are all less than the critical values ​​at the three confidence levels, indicating that at the three confidence levels, the time series data have passed the stationarity test and the SARIMA model can be constructed based on the sequence.

[0065] Table 1 ADF test critical values ​​of first-order seasonal difference series

[0066] By analyzing the time series of the number of China-Europe trains operated by Xi'an "Chang'an", we can preliminarily determine the orders of p (autoregressive order), d (non-seasonal difference number), q (moving average component order), P (seasonal autoregressive order), D (seasonal difference number), Q (seasonal moving average component order), and S (seasonal cycle length) in the SARIMA model, which can be expressed as . Since the sequence obtained by taking the first-order difference of the time series of the number of China-Europe trains running in Xi'an is stationary, d is initially selected as 1; after taking the first-order seasonal difference of the sequence with a period of 3, the sequence has basically no periodic characteristics, so D is initially selected as 1 and S is initially selected as 3. Finally, after selecting different parameters for comparison, the parameter with the smallest AIC value is selected to best fit the data, and the final model is obtained as follows: .

[0067] After building the model, we first need to test the normality of the model residuals to evaluate the applicability of the model. The normal distribution histogram of the residuals is as follows: Figure 7 shown. Figure 7 It shows that the residual sequence is normally distributed.

[0068] The DW test is required for the constructed model. The result is 2.4 close to 2. The closer the value is to 2, the less autocorrelation there is. That is, the residual sequence has no obvious autocorrelation. The residual sequence needs to be tested for white noise to determine whether it is a white noise sequence. The P-value of the white noise test is greater than 0.05, indicating that the sequence is a white noise sequence. The model obtains the linear characteristics of the time series of the number of "Chang'an" China-Europe freight trains running in Xi'an, and is applicable to its changing trend.

[0069] Build-based The model is used for data prediction. The training data is selected from the first 108 months of the Xi'an "Chang'an" China-Europe freight train operation data set, that is, the data from January 2014 to December 2022, to predict the number of trains in operation in the next 12 months, and the monthly train operation quantity forecast results from January to December 2023 are obtained; finally, the results are compared and analyzed with the true values ​​to verify the effectiveness of the model.

[0070] exist Based on the model prediction results, the RF model is used to fit its prediction residual sequence to The model prediction results were corrected. The maximum and minimum normalization method was used to preliminarily process the residual data, and then the model was trained. Similarly, the training set selected the first 108 data in the residual sequence, and the test set selected the last 12 data. On the SPSS Pro platform, the RF model was used to predict the residuals for the next 12 months. The node splitting evaluation criteria of the RF model were selected. Indicators, the relevant parameters of other RF models are shown in Table 2.

[0071] Table 2 Parameters of RF model

[0072] The RF model is used to predict the residual. The residual data predicted by the RF model is normalized data, which needs to be denormalized to obtain the final residual prediction result.

[0073] Using the residual correction method, the prediction results of the SARIMA model and the residual prediction results of the RF model are combined, and finally the prediction values ​​of the SARIMA-RF combined model for the next 12 months are obtained.

[0074] In order to verify the prediction effect of the SARIMA-RF combined model, its results are compared with the actual number of trains of the "Chang'an" China-Europe Express in Xi'an from January to December 2023 and the prediction results of four single models: ARIMA, SARIMA, RF, and XGBoost. The results are as follows: Figure 8 shown.

[0075] Depend on Figure 8 It can be seen that compared with other models, the prediction results of the SARIMA-RF combined model are closer to the actual values, and the change trends are roughly the same, indicating that the combined model can well reflect the linear characteristics of the Xi'an "Chang'an" China-Europe freight train number series.

[0076] The error comparison of each model is as follows Fig. 9It can be seen that the error of the SARIMA-RF combined model is smaller, indicating that the RF model can reduce the error of the single model prediction after residual correction. The SARIMA-RF combined model combines the advantages of the SARIMA model and the RF model to obtain better prediction results. Table 3 calculates the evaluation index values ​​of the SARIMA-RF combined model and the other four single models.

[0077] Table 3 Comparison of evaluation indicators of each model

[0078] In Table 3, the four evaluation indicators of the SARIMA-RF model are all lower than those of other single models, so the prediction effect of the constructed SARIMA-RF combined model is better than that of other single models. For example, the SARIMA-RF model The value is 39.32% lower than that of the ARIMA model, 58.23% lower than that of the SARIMA model, 51.81% lower than that of the RF model, and 63.50% lower than that of the XGBoost model. It can be seen that compared with other single models, the prediction error of the SARIMA-RF combined model is significantly reduced, and the generalization ability of the model is also enhanced.

[0079] Based on the constructed SARIMA-RF forecasting model, the operation of Xi'an "Chang'an" China-Europe trains in 2024 and 2025 is predicted as shown in Table 4. It is estimated that the number of "Chang'an" China-Europe trains in 2024 will be 244,000 cars, equivalent to 6,100 trains, an increase of 13.40% over the number of 5,351 China-Europe trains in 2023; the number of China-Europe trains in 2025 will be 267,100 cars, equivalent to 6,513 trains.

[0080] Table 4 Forecast results of the number of China-Europe freight trains in 2024 and 2025

[0081] The seasonal periodic fluctuation characteristics of the number of China-Europe trains are incorporated into the study, and a prediction model combining the seasonal autoregressive moving average model (SARIMA) and the random forest (RF) is proposed. Taking the monthly data of China-Europe trains in Xi'an as an example, the linear features in the monthly data of China-Europe trains in Xi'an are first fitted by the SARIMA model, and then the residuals of the prediction results of the SARIMA model are corrected by the random forest model, so as to fit the nonlinear features in the monthly data of China-Europe trains in Xi'an. Finally, the SARIMA-RF model is compared with the ARIMA, SARIMA, RF and XGBoost models to verify the effectiveness of the SARIMA-RF combination model.

[0082] It can be seen that by combining the SARIMA model and the RF model, the present invention can more comprehensively capture the linear and nonlinear characteristics and seasonal periodic variation characteristics in the time series of the number of railway international combined transport trains. The SARIMA model is good at processing time series data with seasonal fluctuations, while the RF model can well capture the nonlinear relationship in the data. The combination of the two makes the prediction results more accurate and reduces the uncertainty of the prediction. Secondly, the more accurate prediction of the present invention helps the railway transportation department to plan the capacity in advance and arrange the train operation plan reasonably, thereby reducing the risk of excess or insufficient capacity caused by inaccurate prediction. This is of great significance for improving transportation efficiency and reducing operating costs. At the same time, accurate prediction helps the railway transportation department to better meet customer needs and improve the reliability and punctuality of transportation services. This is especially important for international combined transport trains, because on-time arrival is one of the key factors to ensure the smooth progress of international trade. The prediction method provided by the present invention can provide strong support for the decision-making of the railway transportation department. By understanding the changing trend of the number of trains in advance, the railway transportation department can formulate transportation strategies and optimize the layout of the transportation network more scientifically, thereby further improving the overall transportation efficiency and service level.

[0083] The second object of the present invention is to propose a vehicle running number prediction system based on the SARIMA-RF combined model. Fig.10 As shown, including: Data acquisition module 102: used to acquire monthly data on the operation of international railway combined transport trains to form a monthly data set on the operation of trains; SARIMA model processing module 202: used to fit the running train number sequence in the monthly data set of train operation using the SARIMA model to obtain the predicted value sequence and the residual sequence; RF model processing module 302: used to fit the residual sequence using the RF model to obtain a prediction result; Prediction result output module 402: used to combine the prediction value sequence of the SARIMA model with the prediction result of the RF model to obtain the final prediction result of the SARIMA-RF combination model.

[0084] like Fig.11 As shown, the third object of the present invention is to provide an electronic device, the electronic device comprising: a processor 501, a memory 502 and a display screen 503. Among them, the memory 502 and the display screen 503 are connected to the processor 501, such as through a bus 504. Optionally, the electronic device may also include a transceiver 505. It should be noted that in actual applications, the transceiver 505 is not limited to one, and the structure of the electronic device does not constitute a limitation on the embodiments of the present application.

[0085] Processor 501 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of this application. Processor 501 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0086] The bus 504 may include a path to transmit information between the above components. The bus 504 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. The bus 504 may be divided into an address bus, a data bus, a control bus, etc.

[0087] The memory 502 may be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto.

[0088] The memory 502 is used to store application code for executing the solution of the present application, and the execution is controlled by the processor 501. The processor 501 is used to execute the application code stored in the memory 502 to implement the contents shown in the above method embodiment.

[0089] Fig.11 The electronic device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0090] The fourth object of the present invention is to provide a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, wherein the computer program is stored thereon, and when the program is executed by a processor, the aforementioned Figures 1 to 3 The various processes of the method embodiment shown in the figure include, for example, a memory including instructions, and the instructions can be executed by a processor of an electronic device to complete the method.

[0091] A computer-readable storage medium may be a tangible device that holds and stores instructions used by an instruction execution device. A computer-readable storage medium may be, but is not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any combination thereof. Specifically, a computer-readable storage medium may be a portable computer disk, a hard disk, a USB flash drive, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a lectern random access memory (SRAM), a portable compact disk read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, an optical disk, a magnetic disk, a mechanical encoding device, and any combination thereof.

[0092] A fifth object of the present invention is to provide a computer program product, comprising computer instructions, which, when executed by a processor, implement the above Figures 1 to 3 The various processes of the method embodiment shown can achieve the same technical effect, and will not be described again here to avoid repetition.

[0093] Many embodiments and many applications beyond the examples provided will be apparent to those skilled in the art upon reading the above description. Therefore, the scope of the present teachings should not be determined with reference to the above description, but rather with reference to the preceding claims and the full scope of equivalents to which such claims are entitled. For the purpose of comprehensiveness, all articles and references, including disclosures of patent applications and publications, are incorporated herein by reference. The omission of any aspect of the subject matter disclosed herein in the preceding claims is not intended to be a disclaimer of such subject matter, nor should it be considered that the applicants did not consider such subject matter to be part of the disclosed inventive subject matter.

[0094] The above content is a further detailed description of the present invention. It cannot be determined that the specific implementation methods of the present invention are limited to these. For ordinary technicians in the technical field to which the present invention belongs, without departing from the concept of the present invention, they can also make several simple deductions or substitutions, which should be regarded as belonging to the protection scope of the present invention determined by the submitted claims.

Claims

1. A method for predicting the number of vehicles in operation based on the SARIMA-RF combined model, characterized in that: include: Obtain monthly data on the operation of international railway combined transport trains to form a monthly data set on the operation of trains; The SARIMA model is used to fit the train number sequence in the monthly data set of train operation to obtain the predicted value sequence and residual sequence; Use the RF model to fit the residual sequence and get the prediction results; The predicted value sequence of the SARIMA model is combined with the prediction results of the RF model to obtain the final prediction results of the SARIMA-RF combined model.

2. The method for predicting the number of vehicles in operation based on the SARIMA-RF combined model according to claim 1, characterized in that: The monthly data on the operation of international railway combined transport trains is obtained to form a monthly data set on the operation of international railway combined transport trains, including: The ADF method is used to test the stability of the time series data in the monthly data set of train operation, and the AIC value of the test result is obtained; Select the parameter that best fits the data among the AIC values ​​to complete the SARIMA model construction.

3. The method for predicting the number of vehicles in operation based on the SARIMA-RF combined model according to claim 1, characterized in that: The SARIMA model is used to fit the train number sequence in the monthly data set of train operation to obtain the predicted value sequence and the residual sequence, including: Extract the train number sequence from the monthly data set of train operation ; Fitting the running number series using SARIMA model The linear part , calculate the predicted value sequence And the residual sequence , specifically expressed as follows: Where: is the sequence of running vehicle numbers; is the sequence of running vehicle numbers The linear part of the is the predicted value sequence; is the residual sequence.

4. The method for predicting the number of vehicles in operation based on the SARIMA-RF combined model according to claim 3 is characterized in that: The residual sequence It is necessary to perform maximum and minimum normalization processing to convert the residual sequence The different eigenvalues ​​in are transformed to the same scale.

5. The method for predicting the number of running vehicles based on the SARIMA-RF combined model according to claim 1 is characterized in that: The method of fitting the residual sequence using the RF model to obtain the prediction result includes: Use the RF model to fit the residual sequence and obtain the predicted residual data; Denormalize the predicted residual data to get the predicted results .

6. The method for predicting the number of running vehicles based on the SARIMA-RF combined model according to claim 1, characterized in that: The prediction results of the SARIMA-RF combined model are obtained by combining the prediction value sequence of the SARIMA model with the prediction results of the RF model, including: The predicted value sequence of the SARIMA model Prediction results of the RF model Combine them to get the final prediction result of the SARIMA-RF combined model , specifically expressed as follows: In the formula, The final prediction result; Fitting the residual series of the SARIMA model to the RF model Then the prediction results can be obtained; is the predicted value sequence of the SARIMA model; According to the prediction results of the final SARIMA-RF combined model , formulate plans for international railway combined transport trains.

7. A vehicle running number prediction system based on the SARIMA-RF combined model, characterized in that: include: Data acquisition module: used to obtain monthly data on the operation of international railway combined transport trains and form a monthly data set on the operation of trains; SARIMA model processing module: used to use the SARIMA model to fit the train number sequence in the monthly data set of train operation, and obtain the predicted value sequence and residual sequence; RF model processing module: used to fit the residual sequence using the RF model to obtain the prediction results; Prediction result output module: used to combine the prediction value sequence of the SARIMA model with the prediction results of the RF model to obtain the final prediction results of the SARIMA-RF combined model.

8. An electronic device, characterized in that: The invention comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of a method for predicting the number of running vehicles based on a SARIMA-RF combined model as described in any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of a method for predicting the number of running vehicles based on a SARIMA-RF combined model as described in any one of claims 1 to 6 are implemented.

10. A computer program product, characterized in that The method comprises computer instructions, which, when executed by a processor, implement the steps of a method for predicting the number of running vehicles based on a SARIMA-RF combined model as described in any one of claims 1 to 6.

Images