Splice part quantity prediction method and device

Abstract

The application provides a kind of exchange quantity prediction method and device, it is related to motor train unit maintenance technical field, the method comprises: according to historical exchange maintenance data and preset time series analysis model, obtain predicted exchange rate;According to historical exchange maintenance data, historical passenger data and the vector machine model trained, obtain predicted maintenance quantity;According to the predicted exchange rate and the predicted maintenance quantity, obtain predicted exchange quantity.The application can improve the accuracy of predicted exchange quantity, so as to accurately determine the inventory exchange quantity according to the predicted exchange quantity, thereby reducing the probability of additional consumption of inventory resources and cost caused by insufficient inventory exchange quantity when the motor train is maintained due to lack of parts or excessive inventory exchange quantity.

Description

Technical Field

[0001] This invention relates to the field of EMU maintenance technology, and in particular to a method and device for predicting the number of replacement parts. Background Technology

[0002] Replaceable parts refer to EMU (Electric Multiple Unit) components whose replacement needs are determined based on the actual operating conditions of the EMU and according to preset standards, such as washers and plugs. Since the replacement of these parts is determined by the actual operating conditions of the EMU, sufficient inventory of various replacement parts needs to be prepared in advance to meet the unpredictable replacement requirements. Therefore, predicting the quantity of replacement parts is crucial. However, existing methods for predicting the quantity of replacement parts often involve averaging multiple historical values. These methods have low accuracy, leading to a high probability that the prepared inventory of replacement parts will be insufficient to meet replacement needs or excessive, resulting in additional consumption of inventory resources and costs, which is detrimental to the normal operation of the EMU. Summary of the Invention

[0003] One objective of this invention is to provide a method for predicting the quantity of replacement parts, addressing the problem that existing methods for predicting the quantity of replacement parts have low accuracy. This leads to a high probability that the quantity of replacement parts prepared based on the predicted quantity is insufficient to meet replacement needs or excessive, resulting in additional consumption of inventory resources and costs, which is detrimental to the normal operation of high-speed trains. Another objective of this invention is to provide a device for predicting the quantity of replacement parts. A further objective of this invention is to provide a computer device. A final objective of this invention is to provide a readable medium.

[0004] To achieve the above objectives, one aspect of the present invention discloses a method for predicting the quantity of interchangeable parts, the method comprising:

[0005] Based on historical maintenance data of coupler replacement parts and a preset time series analysis model, the predicted coupler replacement rate is obtained;

[0006] Based on historical replacement parts maintenance data, historical passenger traffic data, and a trained vector machine model, the predicted maintenance volume is obtained.

[0007] The predicted number of replacement parts is obtained based on the predicted replacement rate and the predicted maintenance quantity.

[0008] Optionally, the step of obtaining the predicted replacement rate based on historical replacement component maintenance data and a preset time series analysis model includes:

[0009] Based on the historical replacement component maintenance data, the historical replacement component maintenance quantity and the historical replacement component non-conforming quantity are obtained.

[0010] The historical replacement rate is obtained based on the historical maintenance volume of replacement parts and the historical non-conforming maintenance volume of replacement parts.

[0011] Based on the historical even commutation rate and the preset time series analysis model, the predicted even commutation rate is obtained.

[0012] Optionally, obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model includes:

[0013] The historical even commutation rates are preprocessed to obtain the preprocessed even commutation rates.

[0014] Based on the preprocessed even commutation rate and the preset time series analysis model, the predicted even commutation rate is obtained.

[0015] Optional, further including:

[0016] Before obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model, it is determined whether the average value of the historical even commutation rate is within the preset even commutation rate range. If not, the average value of the historical even commutation rate is used as the predicted even commutation rate.

[0017] Optional, further including:

[0018] Before obtaining the predicted even switching rate based on the historical even switching rate and the preset time series analysis model, the historical even switching rate is input into the anomaly detection algorithm to obtain the anomaly score corresponding to each historical even switching rate value.

[0019] Based on the anomaly score and the preset anomaly score threshold, the anomaly score is obtained from the anomaly score.

[0020] The abnormal historical even slew rate value is obtained from the anomaly score.

[0021] The historical even swapping rate is corrected based on the abnormal historical even swapping rate value.

[0022] Optional, further including:

[0023] Before correcting the historical even switching rate based on the abnormal historical even switching rate value, the normal range of the historical even switching rate is obtained based on the average value and the standard deviation of the historical even switching rate.

[0024] Determine whether the abnormal historical even switching rate value is within the normal range of the historical even switching rate. If it is, then the step of correcting the historical even switching rate based on the abnormal historical even switching rate value is not performed.

[0025] Optional, further including:

[0026] Before obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model, the preset critical parameters and the historical even commutation rate are input into the stationarity test algorithm to obtain the stationarity type of the historical even commutation rate.

[0027] Determine whether the stationarity type is stationary; if not, perform stationarization processing on the historical even commutation rate according to the stationarity type to obtain the stationary even commutation rate.

[0028] Based on the stationary even commutation rate and the preset time series analysis model, the predicted even commutation rate is obtained.

[0029] Optionally, the step of stabilizing the historical even commutation rate according to the stationarity type to obtain a stationary even commutation rate includes:

[0030] Determine whether the stationarity type is a non-stationary term without an intercept;

[0031] If so, differential stationarization is performed on the historical even commutation rates to obtain stationary even commutation rates;

[0032] If not, perform linear fitting on the historical even commutation rates to de-trend and obtain stable even commutation rates.

[0033] Optional, further including:

[0034] Before obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model, the autoregressive order and the moving average order of the time series analysis model are selected from the preset set of autoregressive orders and the preset set of moving average orders based on the historical even commutation rate.

[0035] The time series analysis model is constructed based on the autoregressive order and the moving average order.

[0036] Optionally, the step of selecting the autoregressive order and moving average order of the time series analysis model from a preset set of autoregressive orders and a preset set of moving average orders based on the historical even commutation rate includes:

[0037] The historical even commutation rate is input into a preset parameter estimation function required for constructing a time series analysis model to obtain an estimated value;

[0038] Based on the estimated values, an information content function is constructed;

[0039] By cross-grouping the autoregressive order elements in the set of autoregressive orders and the moving average order elements in the set of moving average orders, multiple time series analysis model order combinations are obtained.

[0040] Each time series analysis model order combination is input into the information content function to obtain multiple corresponding information contents;

[0041] The autoregression order and the moving average order are obtained by combining the time series analysis model orders corresponding to the minimum information content in the information content.

[0042] Optional, further including:

[0043] The time series analysis model includes residuals;

[0044] Before obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model, it is determined whether the residual of the time series analysis model is white noise. If not, the time series analysis model is reconstructed.

[0045] Optionally, when the stationarity type is not stationary, obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model includes:

[0046] The historical even commutation rate is input into the time series analysis model to obtain the initial even commutation rate;

[0047] The initial even commutation rate is subjected to a reverse restoration process corresponding to the stabilization process to obtain the predicted even commutation rate.

[0048] Optional, further including:

[0049] After obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model, the preset confidence level and the historical even commutation rate are input into the time series analysis model to obtain the even commutation rate confidence interval.

[0050] Determine whether the current even commutation rate is within the confidence interval of the even commutation rate. If not, issue an alarm to the staff.

[0051] Optional, further including:

[0052] Before obtaining the predicted maintenance quantity based on historical replacement component maintenance data, historical passenger transport data, and the trained vector machine model, the historical replacement component maintenance quantity is obtained based on the historical replacement component maintenance data.

[0053] Based on the aforementioned historical passenger traffic data, the historical passenger turnover was obtained;

[0054] Based on the current maintenance volume of the replacement parts, the historical maintenance volume of the replacement parts, and the historical passenger turnover, the key parameter penalty coefficients and kernel function coefficients of the vector machine model are selected from the preset set of key parameter penalty coefficients and the preset set of kernel function coefficients.

[0055] The vector machine model is constructed based on the key parameters, penalty coefficient and kernel function coefficient.

[0056] The vector machine model is trained using the current maintenance volume of the replacement parts, the historical maintenance volume of the replacement parts, and the historical passenger turnover as training samples.

[0057] Optionally, the step of selecting the key parameter penalty coefficients and kernel function coefficients of the vector machine model from a preset set of key parameter penalty coefficients and a preset set of kernel function coefficients based on the current maintenance volume of replacement parts, the historical maintenance volume of replacement parts, and the historical passenger turnover includes:

[0058] Cross-grouping the key parameter penalty coefficient elements in the key parameter penalty coefficient set and the kernel function coefficient elements in the kernel function coefficient set yields multiple vector machine model parameter combinations.

[0059] Construct a corresponding test model based on each combination of vector machine model parameters;

[0060] The historical replacement component maintenance volume and the historical passenger turnover volume are input into each of the test models to obtain the corresponding test model output values;

[0061] The key parameter penalty coefficient and kernel function coefficient are obtained by combining the vector machine model parameters corresponding to the output value that is closest to the current maintenance quantity of the replacement component in the output value of the test model.

[0062] Optionally, obtaining the predicted number of coupler replacement parts based on the predicted coupler replacement rate and the predicted maintenance quantity includes:

[0063] Multiply each predicted couple switching rate value in the predicted couple switching rate by the corresponding predicted maintenance quantity value in the predicted maintenance quantity to obtain multiple corresponding couple switching component quantity components.

[0064] The predicted number of even-changing components is obtained based on the component number of even-changing components.

[0065] Optional, further including:

[0066] After obtaining the predicted number of coupler replacement parts based on the predicted coupler replacement rate and the predicted maintenance quantity, the maximum component and the minimum component of the number of coupler replacement parts are obtained based on the predicted number of coupler replacement parts.

[0067] Multiply the maximum component of the number of the coupler by a preset confidence level to obtain the standard upper limit of the number of couplers.

[0068] Multiply the minimum component of the number of the coupler by a preset confidence level to obtain the standard lower limit of the number of couplers.

[0069] Determine whether the current number of replacement parts is greater than the upper limit of the replacement part quantity standard or less than the lower limit of the replacement part quantity standard. If so, issue an alarm to the staff.

[0070] To achieve the above objectives, another aspect of the present invention discloses a device for predicting the quantity of replacement parts, comprising:

[0071] The coupler switching rate prediction module is used to obtain the predicted coupler switching rate based on historical coupler switching component maintenance data and a preset time series analysis model.

[0072] The maintenance quantity prediction module is used to obtain the predicted maintenance quantity based on historical replacement component maintenance data, historical passenger traffic data, and a trained vector machine model.

[0073] The component replacement quantity prediction module is used to obtain the predicted component replacement quantity based on the predicted component replacement rate and the predicted maintenance quantity.

[0074] The present invention also discloses a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method described above.

[0075] The present invention also discloses a computer-readable medium having a computer program stored thereon, which, when executed by a processor, implements the method described above.

[0076] The present invention provides a method and apparatus for predicting the number of replacement parts. By using historical replacement part maintenance data and a preset time series analysis model, a predicted replacement rate is obtained. This method, based on a time series analysis model suitable for predicting future variables from historical data, improves the accuracy of the predicted replacement rate, thereby enhancing the accuracy of the predicted number of replacement parts in subsequent steps. By using historical replacement part maintenance data, historical passenger transport data, and a trained vector machine model, a predicted maintenance quantity is obtained. This method combines the impact of actual EMU passenger load on replacement part replacement and historical replacement part maintenance data, using a vector machine model to predict the maintenance quantity, thus improving the accuracy of the predicted maintenance quantity and consequently enhancing the accuracy of the predicted number of replacement parts in subsequent steps. By using the predicted replacement rate and the predicted maintenance quantity to determine the predicted number of replacement parts based on maintenance-related principles, the accuracy of the predicted number of replacement parts is further improved. In summary, the present invention can improve the accuracy of predicting the number of replacement parts, thereby enabling accurate determination of the inventory of replacement parts based on the predicted number of replacement parts, and thus reducing the probability of insufficient inventory of replacement parts leading to missing parts during train maintenance or excessive inventory of replacement parts leading to additional consumption of inventory resources and costs. Attached Figure Description

[0077] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0078] Figure 1 A flowchart illustrating a method for predicting the number of interchangeable parts according to an embodiment of the present invention is shown.

[0079] Figure 2 A schematic diagram illustrating an optional step in obtaining the predicted even commutation rate according to an embodiment of the present invention is shown.

[0080] Figure 3 A schematic diagram illustrating another optional step in obtaining the predicted even commutation rate according to an embodiment of the present invention is shown;

[0081] Figure 4 The diagram illustrates an optional step in determining the key parameters, penalty coefficient and kernel function coefficient, according to an embodiment of the present invention.

[0082] Figure 5 A schematic diagram illustrating an optional step in obtaining the predicted number of coupling components according to an embodiment of the present invention is shown.

[0083] Figure 6 A schematic diagram of a component quantity prediction device according to an embodiment of the present invention is shown.

[0084] Figure 7 A schematic diagram of a computer device suitable for implementing embodiments of the present invention is shown. Detailed Implementation

[0085] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0086] The terms "first," "second," etc., used in this document do not specifically refer to any order or sequence, nor are they intended to limit the invention; they are merely used to distinguish elements or operations described using the same technical terms.

[0087] The terms “include,” “including,” “have,” “contain,” etc., used in this article are all open-ended terms, meaning that they include but are not limited to.

[0088] The term "and / or" as used herein includes any or all of the things mentioned.

[0089] It should be noted that the acquisition, storage, use, and processing of data in the technical solution of this invention all comply with the relevant provisions of national laws and regulations.

[0090] This invention discloses a method for predicting the number of interchangeable parts, such as... Figure 1 As shown, the method specifically includes the following steps:

[0091] S101: Based on historical maintenance data of coupler replacement parts and a preset time series analysis model, the predicted coupler replacement rate is obtained.

[0092] S102: Based on historical replacement component maintenance data, historical passenger traffic data, and a trained vector machine model, the predicted maintenance quantity is obtained.

[0093] S103: Based on the predicted coupler replacement rate and the predicted maintenance quantity, the predicted number of coupler replacement parts is obtained.

[0094] The present invention provides a method and apparatus for predicting the number of replacement parts. By using historical replacement part maintenance data and a preset time series analysis model, a predicted replacement rate is obtained. This method, based on a time series analysis model suitable for predicting future variables from historical data, improves the accuracy of the predicted replacement rate, thereby enhancing the accuracy of the predicted number of replacement parts in subsequent steps. By using historical replacement part maintenance data, historical passenger transport data, and a trained vector machine model, a predicted maintenance quantity is obtained. This method combines the impact of actual EMU passenger load on replacement part replacement and historical replacement part maintenance data, using a vector machine model to predict the maintenance quantity, thus improving the accuracy of the predicted maintenance quantity and consequently enhancing the accuracy of the predicted number of replacement parts in subsequent steps. By using the predicted replacement rate and the predicted maintenance quantity to determine the predicted number of replacement parts based on maintenance-related principles, the accuracy of the predicted number of replacement parts is further improved. In summary, the present invention can improve the accuracy of predicting the number of replacement parts, thereby enabling accurate determination of the inventory of replacement parts based on the predicted number of replacement parts, and thus reducing the probability of insufficient inventory of replacement parts leading to missing parts during train maintenance or excessive inventory of replacement parts leading to additional consumption of inventory resources and costs.

[0095] In one alternative implementation, such as Figure 2 As shown, the step of obtaining the predicted replacement rate based on historical replacement component maintenance data and a preset time series analysis model includes the following steps:

[0096] S201: Based on the historical replacement component maintenance data, obtain the historical replacement component maintenance quantity and the historical replacement component non-conforming quantity.

[0097] S202: Based on the historical replacement parts maintenance quantity and the historical replacement parts non-conforming quantity, the historical replacement rate is obtained.

[0098] S203: Based on the historical even commutation rate and the preset time series analysis model, the predicted even commutation rate is obtained.

[0099] For example, the historical replacement component maintenance data includes, but is not limited to, the historical replacement component maintenance quantity, the historical replacement component non-conforming quantity, and the historical replacement component maintenance time. Therefore, the historical replacement component maintenance quantity and the historical replacement component non-conforming quantity can be directly obtained from the historical replacement component maintenance data.

[0100] For example, the historical maintenance volume of replacement parts includes, but is not limited to, the maintenance volume values ​​of replacement parts at multiple points in time or time periods in history. For instance, the historical maintenance volume of replacement parts can be, but is not limited to, the total maintenance volume of replacement parts for each month in the past 3 months, the total maintenance volume of replacement parts for each month in the past 5 months, or the total maintenance volume of replacement parts for each week in the past 7 weeks. It should be noted that the historical maintenance volume of replacement parts can be determined by those skilled in the art based on actual circumstances; the above description is merely an example and does not constitute a limitation.

[0101] For example, the historical non-conforming quantity of replacement parts (i.e., the quantity of replacement parts that need to be replaced after maintenance) includes, but is not limited to, the values ​​of non-conforming replacement parts at multiple points in time or time periods in history. For instance, the historical non-conforming quantity of replacement parts can be, but is not limited to, the total number of non-conforming replacement parts for each month in the past 3 months, the total number of historical non-conforming replacement parts for each month in the past 5 months, or the total number of historical non-conforming replacement parts for each week in the past 7 weeks. It should be noted that the historical non-conforming quantity of replacement parts can be determined by those skilled in the art based on actual circumstances; the above description is merely an example and does not constitute a limitation.

[0102] For example, obtaining the historical replacement rate based on the historical replacement component maintenance volume and the historical replacement component non-conformance volume can be, but is not limited to, dividing the historical replacement component non-conformance volume by the historical replacement component maintenance volume. Specifically, for the same point in time or time period, the replacement component maintenance volume value and the replacement component non-conformance volume value in the historical replacement component maintenance volume can be calculated to obtain the historical replacement rate value for the aforementioned point in time or time period. For example, if the replacement component maintenance volume value in April 2021 is 60, and the historical replacement component non-conformance volume value in April 2021 is 15, then the replacement rate value for April 2021 can be obtained as 0.25. The historical replacement rate is specifically, but is not limited to, a sequence including multiple replacement rate values, where each replacement rate value in the sequence is the replacement rate value for a certain time period or point in time in history. For example, the historical replacement rate has the following examples:

[0103] The values ​​are: {April 2021: 0.25, May 2021: 0.45, June 2021: 0.1, July 2021: 0.2} or {0.25, 0.45, 0.1, 0.2}, etc., where it is assumed that April 2021 to July 2021 refers to the past four months. It should be noted that the specific content and format of the historical even-number exchange rate can be determined by those skilled in the art based on actual circumstances. The above description is merely an example and does not constitute a limitation.

[0104] For example, obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model includes, but is not limited to, inputting the historical even commutation rate into the time series analysis model to obtain the predicted even commutation rate. The predicted even commutation rate can be a set of predicted even commutation rate values ​​for a future point in time, multiple points in time, or multiple time periods. It should be noted that the specific content and format of the predicted even commutation rate can be determined by those skilled in the art based on actual circumstances; the above description is merely an example and does not constitute a limitation.

[0105] By following the steps above, the historical replacement rate can be determined based on the actual maintenance situation and maintenance-related principles in history, thereby improving the accuracy of the obtained historical replacement rate and making it more reflective of the actual maintenance situation. This, in turn, improves the accuracy of the predicted replacement rate obtained in subsequent steps, and consequently improves the accuracy of the predicted number of replacement parts in subsequent steps.

[0106] In one alternative implementation, such as Figure 3 As shown, the step of obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model includes the following steps:

[0107] S301: Perform data preprocessing on the historical even switching rate to obtain the preprocessed even switching rate.

[0108] S302: Based on the preprocessed even commutation rate and the preset time series analysis model, the predicted even commutation rate is obtained.

[0109] For example, the data preprocessing includes, but is not limited to, filling in missing values ​​in the historical even commutation rate sequence using linear interpolation or spline interpolation. It should be noted that the specific implementation of the data preprocessing can be determined by those skilled in the art based on actual circumstances; the above description is merely illustrative and does not constitute a limitation.

[0110] For example, obtaining the predicted even commutation rate based on the preprocessed even commutation rate and the preset time series analysis model includes, but is not limited to, inputting the preprocessed even commutation rate into the time series analysis model to obtain the predicted even commutation rate. The predicted even commutation rate can be a set of predicted even commutation rate values ​​for a future point in time, multiple points in time, or multiple time periods. It should be noted that the specific content and format of the predicted even commutation rate can be determined by those skilled in the art based on actual circumstances; the above description is merely an example and does not constitute a limitation.

[0111] Through the above steps, problematic data in the historical even-change rate can be repaired in the form of preprocessing, thereby reducing the probability of errors in subsequent steps and improving the accuracy of the obtained predicted even-change rate, thus improving the accuracy of the predicted number of even-change components.

[0112] In an optional implementation, it further includes:

[0113] Before obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model, it is determined whether the average value of the historical even commutation rate is within the preset even commutation rate range. If not, the average value of the historical even commutation rate is used as the predicted even commutation rate.

[0114] For example, the average of the historical even commutation rates can be, but is not limited to, the average of the even commutation rate values ​​at various time points or time periods in the historical even commutation rate sequence. For example, the following historical even commutation rates are given:

[0115] {September 2021: 0.25, October 2021: 0.75, November 2021: 0.5}

[0116] The average historical even scalar rate is (0.25 + 0.75 + 0.5) / 3 = 0.5.

[0117] For example, the preset even slew rate range can be, but is not limited to, [0.05, 0.95]. It should be noted that the preset even slew rate range can be determined by those skilled in the art based on actual circumstances; the above description is merely an example and does not constitute a limitation.

[0118] For example, the use of the average of the historical even commutation rates as the predicted even commutation rate can be exemplified by the following:

[0119] If the predicted even variability rates for June and July 2020 are to be predicted, and based on actual circumstances and needs, those skilled in the art determine that the historical even variability rate is calculated using the set of historical even variability rate values ​​for the four months preceding the predicted month, then the predicted even variability rate includes the predicted even variability rate values ​​for June 2020 and July 2020. Specifically, the predicted even variability rate value for June 2020 is the average of the even variability rate values ​​from February to May 2020, and the predicted even variability rate value for July 2020 is the average of the even variability rate values ​​from March to June 2020. Alternatively, the predicted even variability rate values ​​for both June and July 2020 can be the average of the even variability rate values ​​from February to May 2020.

[0120] It should be noted that the specific implementation method of using the average of the historical even switching rates as the predicted even switching rates can be determined by those skilled in the art based on the actual situation. The above description is only an example and does not constitute a limitation.

[0121] If the average historical replacement rate is not within the preset replacement rate range, it indicates that the past maintenance of replacement parts was in a relatively extreme situation (the replacement rate was too high or too low). Therefore, in this case, it is not necessary to obtain the predicted replacement rate based on the historical replacement rate and the preset time series analysis model. Using the average historical replacement rate as the predicted replacement rate can also ensure that the predicted replacement rate has high accuracy and saves the additional resources and time required for subsequent steps to determine the predicted replacement rate. This increases the speed of determining the predicted replacement rate and, consequently, the speed of obtaining the predicted number of replacement parts.

[0122] In an optional implementation, it further includes:

[0123] Before obtaining the predicted even switching rate based on the historical even switching rate and the preset time series analysis model, the historical even switching rate is input into the anomaly detection algorithm to obtain the anomaly score corresponding to each historical even switching rate value.

[0124] Based on the anomaly score and the preset anomaly score threshold, the anomaly score is obtained from the anomaly score.

[0125] The abnormal historical even slew rate value is obtained from the anomaly score.

[0126] The historical even swapping rate is corrected based on the abnormal historical even swapping rate value.

[0127] For example, the anomaly detection algorithm can be, but is not limited to, the Isolation Forest algorithm. It should be noted that the selection of the anomaly detection algorithm can be determined by those skilled in the art based on the actual situation; the above description is merely illustrative and does not constitute a limitation.

[0128] For example, the anomaly threshold can be, but is not limited to, 0.95 or 0.92. It should be noted that the anomaly threshold can be determined by those skilled in the art based on actual circumstances; the above description is merely an example and does not constitute a limitation.

[0129] For example, obtaining the abnormal score from the abnormality score based on the abnormality score and a preset abnormality score threshold can be, but is not limited to, using the score greater than the abnormality score threshold as the abnormal score.

[0130] For example, obtaining the abnormal historical even switching rate value in the historical even switching rate based on the abnormal score can be, but is not limited to, obtaining the abnormal historical even switching rate value corresponding to the abnormal score in the historical even switching rate.

[0131] For example, the step of correcting the historical even commutation rate based on the abnormal historical even commutation rate value can be, but is not limited to, removing abnormal historical even commutation rate values ​​from the sequence of historical even commutation rates or using repair methods such as spline interpolation.

[0132] By taking the above steps, the probability of values ​​with obvious errors or large deviations from historical maintenance conditions in the historical replacement rate can be reduced, thereby improving the accuracy of the historical replacement rate and, consequently, improving the accuracy of the predicted number of replacement parts in subsequent steps.

[0133] In a preferred embodiment, the method further includes feeding back the abnormal historical replacement rate value to the staff, so that the staff can focus on inspecting and analyzing the replacement parts within the time period corresponding to the abnormal historical replacement rate value, in order to determine the reason for the high failure rate of the replacement parts within that time period, and thereby improve the production and processing methods of the replacement parts based on the reason, which is more conducive to improving the quality of the replacement parts.

[0134] In an optional implementation, it further includes:

[0135] Before correcting the historical even switching rate based on the abnormal historical even switching rate value, the normal range of the historical even switching rate is obtained based on the average value and the standard deviation of the historical even switching rate.

[0136] Determine whether the abnormal historical even switching rate value is within the normal range of the historical even switching rate. If it is, then the step of correcting the historical even switching rate based on the abnormal historical even switching rate value is not performed.

[0137] For example, the standard deviation of the historical even variability can be, but is not limited to, the standard deviation of the numerical sequence composed of the individual historical even variability values.

[0138] For example, the normal range of historical even commutation rates can be, but is not limited to, [the average historical even commutation rate minus three times the standard deviation of historical even commutation rates, or the average historical even commutation rate plus three times the standard deviation of historical even commutation rates]. It should be noted that the specific implementation of obtaining the normal range of historical even commutation rates based on the average and standard deviation of historical even commutation rates can be determined by those skilled in the art based on actual circumstances. The above description is merely an example and does not constitute a limitation.

[0139] When the abnormal historical commutation rate value is within the normal range of the historical commutation rate, it means that the abnormal historical commutation rate value has almost no impact on the accuracy of the historical commutation rate, and therefore has almost no impact on the accuracy of the predicted commutation rate. In this case, if the historical commutation rate is corrected based on the abnormal historical commutation rate value, it may actually reduce the accuracy of the historical commutation rate. Therefore, not performing the step of correcting the historical commutation rate based on the abnormal historical commutation rate value can indirectly improve the accuracy of the historical commutation rate, thereby indirectly improving the accuracy of the predicted commutation rate, and further indirectly improving the accuracy of the predicted number of commutation components.

[0140] In an optional implementation, it further includes:

[0141] Before obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model, the preset critical parameters and the historical even commutation rate are input into the stationarity test algorithm to obtain the stationarity type of the historical even commutation rate.

[0142] Determine whether the stationarity type is stationary; if not, perform stationarization processing on the historical even commutation rate according to the stationarity type to obtain the stationary even commutation rate.

[0143] Based on the stationary even commutation rate and the preset time series analysis model, the predicted even commutation rate is obtained.

[0144] For example, the preset critical parameter is related to the confidence interval of the sequence of predicted numbers of replacement parts expected by the worker. For instance, the critical parameter includes, but is not limited to, the critical value or critical probability corresponding to the confidence interval of the sequence of predicted numbers of replacement parts expected by the worker. The critical value can be obtained by looking up a relevant confidence interval table. For example, if the confidence interval is 95%, the critical value can be 2.33; if the confidence interval is 99%, the critical value can be 2.58. The critical probability can also be obtained by looking up a relevant confidence interval table. It should be noted that the critical parameter can be determined by those skilled in the art based on actual circumstances; the above description is merely an example and does not constitute a limitation.

[0145] For example, the stationarity test algorithm can be, but is not limited to, the ADF algorithm (Unit Root Test). The ADF algorithm takes the historical even commutation rate (sequence) as input and outputs an ADF statistic and an ADF test value as intermediate results during intermediate processing. Then, the ADF algorithm determines whether the ADF statistic is greater than a preset critical value and whether the ADF test value is greater than a preset critical probability. If the ADF statistic is greater than the preset critical value, or the ADF test value is greater than the preset critical probability, then the stationarity of the historical even commutation rate is determined to be non-stationary; otherwise, it is stationary. It should be noted that the selection of the stationarity test algorithm can be determined by those skilled in the art based on the actual situation, and the operating mechanism of the ADF algorithm is prior art. The above description is merely an example and does not constitute a limitation.

[0146] For example, the predicted even commutation rate can be obtained by inputting the stationary even commutation rate into the preset time series analysis model, but is not limited to.

[0147] Through the above steps, the stationarity of the historical even commutation rate can be verified, and when the historical even commutation rate is not stationary, the series of historical even commutation rates can be stationary, thereby ensuring that the historical even commutation rates used in calculating the predicted even commutation rate are stationary. Since the time series analysis model has certain requirements for the stationarity of the input, the above steps can reduce the possibility of errors when obtaining the predicted even commutation rate based on the stationary even commutation rate and the preset time series analysis model, thereby improving the accuracy of the obtained predicted even commutation rate.

[0148] In an optional implementation, the step of stabilizing the historical even commutation rate according to the stationarity type to obtain a stationary even commutation rate includes:

[0149] Determine whether the stationarity type is a non-stationary term without an intercept;

[0150] If so, differential stationarization is performed on the historical even commutation rates to obtain stationary even commutation rates;

[0151] If not, perform linear fitting on the historical even commutation rates to de-trend and obtain stable even commutation rates.

[0152] For example, the types of stationarity include, but are not limited to, stationary, non-stationary without intercept, and non-stationary with intercept, wherein the types of stationarity are determined by a stationarity test algorithm.

[0153] For example, the differential stationarization process performed on the historical even commutation rates to obtain stationary even commutation rates is a conventional technique in the art and will not be elaborated further here. The differential stationarization process can involve performing a differential transformation on the sequence of historical even commutation rates to obtain a sequence of stationary even commutation rates.

[0154] For example, the method of performing linear fitting to de-trend and stationarize the historical even commutation rates to obtain stationary even commutation rates is a conventional technique in the art and will not be elaborated here. Specifically, the linear fitting to de-trend and stationarize process can involve performing a first-order linear fit (e.g., using, but not limited to, least squares fitting) on ​​the sequence of historical even commutation rates to obtain a trend sequence, and then subtracting the trend sequence from the sequence of historical even commutation rates to obtain the sequence of stationary even commutation rates.

[0155] Through the above steps, when the historical even commutation rate sequence is non-stationary, different corresponding stationarization processes can be applied to different types of non-stationary historical even commutation rate sequences based on existing mathematical principles. This improves the stationarity and accuracy of the obtained stationary even commutation rate sequence, thereby improving the accuracy of the predicted even commutation rate obtained in subsequent steps, and ultimately improving the accuracy of the number of even commutation components.

[0156] In an optional implementation, it further includes:

[0157] Before obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model, the autoregressive order and the moving average order of the time series analysis model are selected from the preset set of autoregressive orders and the preset set of moving average orders based on the historical even commutation rate.

[0158] The time series analysis model is constructed based on the autoregressive order and the moving average order.

[0159] For example, the time series analysis model can be, but is not limited to, an ARMA model (Autoregressive Moving Average) or an ARIMA model (Differential Integrated Moving Average Autoregressive Model). It should be noted that the selection of the time series analysis model can be determined by those skilled in the art based on the actual situation; the above description is merely illustrative and does not constitute a limitation.

[0160] For example, the autoregressive order and the moving average order are key parameters of the time series analysis model, which can greatly affect the properties of the time series analysis model. Constructing a time series analysis model based on the autoregressive order and the moving average order is a conventional technique in this field and will not be elaborated upon here.

[0161] By following the steps above, the determined autoregressive order and moving average order can be made to match the historical even commutation rate, thereby making the constructed time series analysis model more suitable for calculating and predicting the even commutation rate, and thus improving the accuracy and speed of calculating and predicting the even commutation rate.

[0162] In an optional implementation, the step of selecting the autoregressive order and moving average order of the time series analysis model from a preset set of autoregressive orders and a preset set of moving average orders based on the historical even commutation rate includes:

[0163] The historical even commutation rate is input into a preset parameter estimation function required for constructing a time series analysis model to obtain an estimated value;

[0164] Based on the estimated values, an information content function is constructed;

[0165] By cross-grouping the autoregressive order elements in the set of autoregressive orders and the moving average order elements in the set of moving average orders, multiple time series analysis model order combinations are obtained.

[0166] Each time series analysis model order combination is input into the information content function to obtain multiple corresponding information contents;

[0167] The autoregression order and the moving average order are obtained by combining the time series analysis model orders corresponding to the minimum information content in the information content.

[0168] For example, the parameter estimation function can be, but is not limited to, the maximum likelihood estimation function inherent in the framework of the time series analysis model.

[0169] For example, the information content function can be, but is not limited to, the AIC function (Akaike Information Content Function) or the BIC function (Bayesian Information Content Function). The estimated value is a component parameter of the information content function.

[0170] For example, the AIC function can be, but is not limited to:

[0171] AIC = -2ln(L) + 2k

[0172] The BIC function can be, but is not limited to:

[0173] BIC = -2ln(L) + k·ln(n)

[0174] Where AIC and BIC represent information content, L represents the estimated value, and k is the sum of the autoregressive order and the moving average order (when the historical even commutation sequence contains an intercept term) or the sum of the autoregressive order and the moving average order plus 1 (when the historical even commutation sequence does not contain an intercept term). n is the sample size obtained from the number of elements in the estimated value or the historical even commutation sequence.

[0175] For example, the preset set of autoregressive orders can be, but is not limited to, {1, 2, 3}; the preset set of moving average orders can be, but is not limited to, {1, 2, 3}. It should be noted that the set of autoregressive orders and the set of moving average orders can be determined by those skilled in the art based on actual circumstances. The above description is merely an example and does not constitute a limitation.

[0176] For example, corresponding to the example above, the cross-grouping of the autoregressive order elements in the set of autoregressive orders and the moving average order elements in the set of moving average orders to obtain multiple time series analysis model order combinations can be exemplified by the following:

[0177] The cross-matching specifically involves combining each element in the set of autoregressive orders with each element in the set of moving average orders to obtain multiple time series analysis model order combinations. For example, corresponding to the example above, the resulting multiple event series analysis model order combinations include {autoregressive order: 1, moving average order: 1}, {autoregressive order: 1, moving average order: 2}, {autoregressive order: 1, moving average order: 3}, {autoregressive order: 2, moving average order: 1}, {autoregressive order: 2, moving average order: 2}, {autoregressive order: 2, moving average order: 3}, {autoregressive order: 3, moving average order: 1}, {autoregressive order: 3, moving average order: 2}, and {autoregressive order: 3, moving average order: 3}.

[0178] For example, if the time series analysis model order combination corresponding to the minimum information content is {autoregressive order: 2, moving average order: 1}, then the obtained autoregressive order is 2 and the moving average order is 1.

[0179] It should be noted that the process of cross-grouping the autoregressive order elements in the set of autoregressive orders and the moving average order elements in the set of moving average orders yields multiple combinations of time series analysis model orders. Each combination of time series analysis model orders is input into the information content function to obtain multiple corresponding information contents. The specific implementation of the autoregressive order and the moving average order is obtained based on the time series analysis model order combination corresponding to the minimum information content. This can be determined by those skilled in the art based on the actual situation. The above description is merely an example and does not constitute a limitation.

[0180] By following the steps described above, the accuracy of the obtained autoregressive order and moving average order can be improved based on relevant mathematical and computer science principles. This makes the determined autoregressive order and moving average order more consistent with the historical even commutation rate, further making the constructed time series analysis model more suitable for calculating and predicting even commutation rate, thereby further improving the accuracy and speed of calculating and predicting even commutation rate.

[0181] In an optional implementation, it further includes:

[0182] The time series analysis model includes residuals;

[0183] Before obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model, it is determined whether the residual of the time series analysis model is white noise. If not, the time series analysis model is reconstructed.

[0184] For example, the residual is a residual sequence, and the residual sequence is a parameter that comes with the time series analysis model.

[0185] For example, determining whether the residuals of the time series analysis model are white noise includes, but is not limited to:

[0186] The residual sequence of the residuals is standardized to obtain a standard residual sequence; wherein the standardization method includes, but is not limited to, normalization method, regularization method and normalization method.

[0187] A QQ plot (normal quantile plot) is plotted based on the standard residual sequence. The distribution of the QQ plot is determined, but not limited to, manually, whether it approximates a curve. If it does, the residual sequence conforms to a normal distribution, the residuals are non-autocorrelated, and the residuals are considered white noise. If not, the residual sequence does not conform to a normal distribution, the residuals of the time series analysis model are autocorrelated, and the residuals are not considered white noise.

[0188] Alternatively, the standard residual sequence can be input into the LB test algorithm (mixed test algorithm) to obtain the LB test value. If the LB test value is greater than the critical value corresponding to the preset confidence interval, it indicates that the residual is autocorrelated, and the residual is determined not to be white noise. If the LB test value is less than or equal to the critical value corresponding to the preset confidence interval, it indicates that the residual is not autocorrelated, and the residual is determined to be white noise.

[0189] If the residuals of the time series analysis model are not white noise, it indicates that the time series analysis model does not conform to the definition of a time series analysis model. Using this model to predict even commutation rates may result in inaccurate predicted even commutation rates. Therefore, the above steps can make the time series analysis model more compliant with the specifications, thereby improving the accuracy of the predicted even commutation rates and ultimately improving the accuracy of the predicted number of even commutated components.

[0190] In an optional implementation, when the stationarity type is not stationary, obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model includes:

[0191] The historical even commutation rate is input into the time series analysis model to obtain the initial even commutation rate;

[0192] The initial even commutation rate is subjected to a reverse restoration process corresponding to the stabilization process to obtain the predicted even commutation rate.

[0193] For example, if the stabilization process is a linear fitting detrending stabilization process, then the inverse restoration process can be, but is not limited to, superimposing the trend sequence onto the initial result even variability sequence to obtain the predicted even variability sequence; if the stabilization process is a difference stabilization process, then the inverse restoration process can be, but is not limited to, performing an inverse difference transformation on the initial result even variability sequence to obtain the predicted even variability sequence. It should be noted that the specific implementation of performing the inverse restoration process corresponding to the stabilization process on the initial result even variability to obtain the predicted even variability can be determined by those skilled in the art based on the actual situation. The above description is merely an example and does not constitute a limitation.

[0194] Considering the potential distortion of historical commutation rates caused by existing stabilization techniques, which in turn may lead to distortion of the predicted commutation rates, the above steps, through reverse restoration processing, reduce the distortion of the predicted commutation rates, improve the accuracy of the predicted commutation rates, and thus improve the accuracy of the predicted number of commutation components.

[0195] In an optional implementation, it further includes:

[0196] After obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model, the preset confidence level and the historical even commutation rate are input into the time series analysis model to obtain the even commutation rate confidence interval.

[0197] Determine whether the current even commutation rate is within the confidence interval of the even commutation rate. If not, issue an alarm to the staff.

[0198] For example, the confidence level can be, but is not limited to, 0.95, 0.93, or 0.96. It should be noted that the confidence level can be determined by those skilled in the art based on the actual situation; the above description is merely an example and does not constitute a limitation.

[0199] For example, the even commutation rate confidence interval corresponds to the confidence level and the historical even commutation rate. For instance, if the confidence level is 0.95, the even commutation rate confidence interval is a 95% confidence interval; if the confidence level is 0.93, the even commutation rate confidence interval is a 93% confidence interval. It should be noted that the specific implementation method of inputting the preset confidence level and the historical even commutation rate into the time series analysis model to obtain the even commutation rate confidence interval can be determined by those skilled in the art based on the actual situation. The above description is merely an example and does not constitute a limitation.

[0200] For example, the alarm to staff can be, but is not limited to, sending alarm information to staff. For example, the alarm information can be, but is not limited to, "The replacement rate this month is abnormal. Please pay close attention to the quality of replacement parts or abnormal maintenance data."

[0201] By following the above steps, when the current commutation rate is not within the confidence interval of the commutation rate, it can be determined that there is an abnormality in the current maintenance situation or the maintenance quality of the commutation components, and this can be reported to the staff so that the staff can handle the abnormality, which in turn helps to ensure the normal operation of the EMU.

[0202] In an optional implementation, it further includes:

[0203] Before obtaining the predicted maintenance quantity based on historical replacement component maintenance data, historical passenger transport data, and the trained vector machine model, the historical replacement component maintenance quantity is obtained based on the historical replacement component maintenance data.

[0204] Based on the aforementioned historical passenger traffic data, the historical passenger turnover was obtained;

[0205] Based on the current maintenance volume of the replacement parts, the historical maintenance volume of the replacement parts, and the historical passenger turnover, the key parameter penalty coefficients and kernel function coefficients of the vector machine model are selected from the preset set of key parameter penalty coefficients and the preset set of kernel function coefficients.

[0206] The vector machine model is constructed based on the key parameters, penalty coefficient and kernel function coefficient.

[0207] The vector machine model is trained using the current maintenance volume of the replacement parts, the historical maintenance volume of the replacement parts, and the historical passenger turnover as training samples.

[0208] For example, the historical replacement component maintenance data includes, but is not limited to, historical replacement component maintenance quantity and historical replacement component maintenance time, so the historical replacement component maintenance quantity can be directly obtained from the historical replacement component maintenance data. The historical replacement component maintenance quantity includes, but is not limited to, the historical replacement component maintenance quantity values ​​at a single point in time, a time period, or multiple points in time or time periods in history, and can be represented by a sequence, but is not limited to. For example, the historical replacement component maintenance quantity can be, but is not limited to, {January 2019: 150, February 2019: 145, March 2019: 155, April 2019: 160, May 2019: 140} or {150, 145, 155, 160, 140}, etc. It should be noted that the content and format of the historical replacement component maintenance quantity can be determined by those skilled in the art based on the actual situation; the above description is merely an example and does not constitute a limitation.

[0209] For example, the historical passenger transport data includes historical passenger turnover, so the historical passenger turnover can be directly obtained from the historical passenger transport data. The historical passenger transport data can be downloaded from sources such as, but not limited to, the official website of the Ministry of Transport. The historical passenger turnover includes, but is not limited to, historical passenger turnover values ​​at a single point in time, a period of time, or multiple points in time or periods of time, and can be represented, but is not limited to, as a sequence.

[0210] For example, the vector machine model can be, but is not limited to, an SVR model (Support Vector Regression model). It should be noted that the selection of the vector machine model can be determined by those skilled in the art based on the actual situation; the above description is merely illustrative and does not constitute a limitation.

[0211] For example, the key parameters, penalty coefficient and kernel function coefficient, are critical parameters of the vector machine model and can significantly influence its properties. Constructing a vector machine model based on these key parameters is a conventional technique in this field and will not be elaborated upon here.

[0212] In a preferred embodiment, the step of training the vector machine model using the current maintenance volume of replacement parts, the historical maintenance volume of replacement parts, and the historical passenger turnover as training samples includes, but is not limited to:

[0213] The vector machine model is trained using the current maintenance volume of the replacement parts as the standard output and the historical maintenance volume of the replacement parts and the historical passenger turnover as the standard input.

[0214] For example, the current maintenance quantity of the spare parts can be, but is not limited to, the maintenance quantity of the spare parts in the current month or the current week, or it can be, but is not limited to, the maintenance quantity of the spare parts in a selected month or a selected week.

[0215] The maintenance time corresponding to the historical maintenance volume of the spare parts must be earlier than the maintenance time corresponding to the current maintenance volume of the spare parts. For example, if the maintenance volume of the spare parts in April 2021 is selected as the current maintenance volume, the sequence of historical maintenance volumes can include the maintenance volumes in March 2021, February 2021, etc. It should be noted that the historical and current maintenance volumes of the spare parts can be determined by those skilled in the art based on the actual situation. The above description is merely an example and does not constitute a limitation.

[0216] For example, the corresponding time for the historical passenger turnover must be earlier than the maintenance time corresponding to the current occasional replacement maintenance volume.

[0217] Preferably, the ratio of the number of elements in the sequence of historical replacement parts maintenance volume to the number of elements in the sequence of historical passenger turnover volume can be, but is not limited to, 5:1.

[0218] For example, the training samples may include a combination of multiple sets of current replacement component maintenance quantities, historical replacement component maintenance quantities, and historical passenger turnover quantities.

[0219] Through the above steps, the determined key parameter penalty coefficient and kernel function coefficient can be aligned with the historical maintenance volume of replacement parts and the historical passenger turnover, thereby making the constructed vector machine model more suitable for calculating and predicting maintenance volumes, and thus improving the accuracy and speed of calculating and predicting maintenance volumes. By using the current maintenance volume of replacement parts, the historical maintenance volume of replacement parts, and the historical passenger turnover as training samples to train the vector machine model, the accuracy and speed of the vector machine model in calculating and predicting maintenance volumes can be further improved.

[0220] In one alternative implementation, such as Figure 4 As shown, the step of selecting the key parameter penalty coefficients and kernel function coefficients of the vector machine model from a preset set of key parameter penalty coefficients and a preset set of kernel function coefficients based on the current maintenance volume of replacement parts, the historical maintenance volume of replacement parts, and the historical passenger turnover includes the following steps:

[0221] S401: Cross-group the key parameter penalty coefficient elements in the key parameter penalty coefficient set and the kernel function coefficient elements in the kernel function coefficient set to obtain multiple vector machine model parameter combinations.

[0222] S402: Construct the corresponding test model based on the parameter combination of each vector machine model.

[0223] S403: Input the historical replacement component maintenance quantity and the historical passenger turnover quantity into each of the test models to obtain the corresponding test model output value.

[0224] S404: Based on the vector machine model parameter combination corresponding to the output value that is closest to the current maintenance quantity of the replacement component in the output value of the test model, obtain the penalty coefficient and kernel function coefficient of the key parameters.

[0225] For example, the preset key parameter penalty coefficient set may have elements ranging from, but not limited to, [2]. -10 ,2 10 The step size between elements can be, but is not limited to, 2. 0.5 The preset set of kernel function coefficients may have elements ranging from, but not limited to, [2]. -10 ,2 10 The step size between elements can be, but is not limited to, 2. 0.5 It should be noted that the numerical range and number of elements of the set of key parameter penalty coefficients and the set of kernel function coefficients can be determined by those skilled in the art based on the actual situation. The above description is only an example and does not constitute a limitation.

[0226] For example, the principle of cross-grouping the key parameter penalty coefficient elements in the key parameter penalty coefficient set and the kernel function coefficient elements in the kernel function coefficient set to obtain multiple vector machine model parameter combinations is similar to the principle of cross-grouping the autoregressive order elements in the autoregressive order set and the moving average order elements in the moving average order set to obtain multiple time series analysis model order combinations in the embodiments of the present invention, and will not be repeated here.

[0227] For example, constructing a corresponding test model based on each vector machine model parameter combination is a conventional technique in this field and will not be elaborated further here. The test model is also a vector machine model.

[0228] For example, if the current maintenance quantity of the replacement part is 160, and the output value of test model A is 60, the output value of test model B is 100, the output value of test model C is 200, and the output value of test model D is 161, then it can be determined that the output value of test model D is closest to the current maintenance quantity of the replacement part. The vector machine model parameter combination corresponding to test model D (i.e., the vector machine model parameter combination corresponding to the output value of the test model that is closest to the current maintenance quantity of the replacement part) is {key parameter penalty coefficient: 2√2, kernel function coefficient: 2√2}, then the key parameter penalty coefficient is 2√2, and the kernel function coefficient is 2√2.

[0229] For example, the kernel function can be, but is not limited to, a Gaussian kernel function (corresponding to the SVR model, which is included in the structural framework of the SVR model).

[0230] Through the above steps, based on relevant mathematical and computer science principles, the accuracy of the obtained key parameter penalty coefficients and kernel function coefficients can be improved. This makes the determined key parameter penalty coefficients and kernel function coefficients more consistent with the historical maintenance quantity of the replacement parts, further making the constructed vector machine model more suitable for calculating and predicting maintenance quantity, and thus further improving the accuracy and speed of calculating and predicting maintenance quantity.

[0231] In a preferred embodiment, the step of obtaining the predicted maintenance quantity based on historical replacement component maintenance data, historical passenger traffic data, and a trained vector machine model includes:

[0232] The historical maintenance volume of replacement parts and the historical passenger turnover are input into the vector machine model to obtain the predicted maintenance volume.

[0233] Preferably, the ratio of the number of elements in the sequence of historical replacement parts maintenance volume to the number of elements in the historical passenger turnover volume can be, but is not limited to, 5:1. The time point or time period corresponding to the historical passenger turnover volume needs to be the same as the time point or time period corresponding to the historical replacement parts maintenance volume. For example, if the historical replacement parts maintenance volume is {January 2021: 100, February 2021: 101, March 2021: 102, April 2021: 103, May 2021: 104}, then the corresponding historical passenger turnover volume can be {May 2021: 1,000,000}. It should be noted that the specific format, content, and correspondence of the historical replacement parts maintenance volume and historical passenger turnover volume can be determined by those skilled in the art based on actual circumstances. The above description is merely an example and does not constitute a limitation.

[0234] For example, the predicted maintenance quantity can be, but is not limited to, a sequence including one or more predicted maintenance quantity values.

[0235] For example, if the historical maintenance volume of occasional replacement parts is {January 2021: 100, February 2021: 101, March 2021: 102, April 2021: 103, May 2021: 104} and the historical passenger turnover is {May 2021: 1,000,000}, then the predicted maintenance volume for June 2021 is 103; if the historical maintenance volume of occasional replacement parts is {February 2021: 101, March 2021: 102, April 2021: 103, May 2021: 104, June 2021: 110} and the historical passenger turnover is {June 2021: 10,000,10}, then the predicted maintenance volume for July 2021 is 108. The predicted maintenance quantity can be, but is not limited to, {June 2021: 103, July 2021: 108}, {June 2021: 103}, or {July 2021: 108}, etc. It should be noted that the content and format of the predicted maintenance quantity can be determined by those skilled in the art based on actual circumstances; the above description is merely an example and does not constitute a limitation.

[0236] It should be noted that the specific implementation method of inputting the historical replacement component maintenance quantity and the historical passenger turnover into the vector machine model to obtain the predicted maintenance quantity can be determined by those skilled in the art based on the actual situation. The above description is only an example and does not constitute a limitation.

[0237] Inputting the historical replacement component maintenance volume and the historical passenger turnover volume into the vector machine model to obtain the predicted maintenance volume is the step required to calculate the predicted maintenance volume using the vector machine model.

[0238] In one alternative implementation, such as Figure 5 As shown, the step of obtaining the predicted number of coupler replacement parts based on the predicted coupler replacement rate and the predicted maintenance quantity includes the following steps:

[0239] S501: Multiply each predicted couple conversion rate value in the predicted couple conversion rate by the corresponding predicted maintenance quantity value in the predicted maintenance quantity to obtain multiple corresponding couple conversion component quantity components.

[0240] S502: Based on the number components of the even-changing components, the predicted number of even-changing components is obtained.

[0241] For example, if the predicted replacement rate sequence is {May 2022: 0.5, June 2022: 0.6, July 2022: 0.7}, and the predicted maintenance quantity sequence is {May 2022: 100, June 2022: 200, July 2022: 300}, then the following multiple replacement component quantity components can be obtained:

[0242] {May 2022: 50}, {June 2022: 120}, {July 2022: 210}.

[0243] The “corresponding” in the corresponding predicted maintenance quantity value and the corresponding number of replacement parts means that the time of the predicted maintenance quantity value, the time of the predicted replacement rate value, and the time of the number of replacement parts must be the same. For example, the above times are all May 2022.

[0244] It should be noted that the format and content of the predicted switching rate and the predicted maintenance quantity can be determined by those skilled in the art based on the actual situation. The above description is only an example and does not constitute a limitation.

[0245] For example, obtaining the predicted number of even-changing components based on the even-changing component quantity components can be, but is not limited to, aggregating all the even-changing component quantity components to obtain a sequence, which is the predicted number of even-changing components. Corresponding to the above example, the number of even-changing components could be {May 2022: 50, June 2022: 120, July 2022: 210}.

[0246] It should be noted that the format and content of the predicted number of replacement parts can be determined by those skilled in the art based on the actual situation. The above description is only an example and does not constitute a limitation.

[0247] It should be noted that the specific implementation of steps S501 and S502 can be determined by those skilled in the art based on the actual situation. The above description is only an example and does not constitute a limitation.

[0248] By following the steps described above, the accuracy of the predicted number of even-changing components can be further improved based on relevant mathematical principles.

[0249] In an optional implementation, it further includes:

[0250] After obtaining the predicted number of coupler replacement parts based on the predicted coupler replacement rate and the predicted maintenance quantity, the maximum component and the minimum component of the number of coupler replacement parts are obtained based on the predicted number of coupler replacement parts.

[0251] Multiply the maximum component of the number of the coupler by a preset confidence level to obtain the standard upper limit of the number of couplers.

[0252] Multiply the minimum component of the number of the coupler by a preset confidence level to obtain the standard lower limit of the number of couplers.

[0253] Determine whether the current number of replacement parts is greater than the upper limit of the replacement part quantity standard or less than the lower limit of the replacement part quantity standard. If so, issue an alarm to the staff.

[0254] For example, if the number of replacement parts is {May 2022: 50, June 2022: 120, July 2022: 210}, then the maximum number of replacement parts is 210 and the minimum number of replacement parts is 50.

[0255] For example, the preset confidence level can be, but is not limited to, 0.95, 0.93, or 0.97. It should be noted that the value of the confidence level can be determined by those skilled in the art based on the actual situation; the above description is merely an example and does not constitute a limitation.

[0256] When the current number of replacement parts exceeds the upper limit of the replacement part quantity standard or is less than the lower limit of the replacement part quantity standard, it indicates that there is a problem with the quality of the replacement parts or a false alarm in the number of replacement parts. Therefore, by following the above steps, when the current number of replacement parts exceeds the upper limit of the replacement part quantity standard or is less than the lower limit of the replacement part quantity standard, the staff can be notified in a timely manner so that the staff can carry out relevant repairs, which is more conducive to the normal operation of the EMU.

[0257] Based on the same principle, this invention discloses a device 600 for predicting the number of interchangeable parts, such as... Figure 6 As shown, the device 600 for predicting the number of replacement parts includes:

[0258] The coupler switching rate prediction module 601 is used to obtain the predicted coupler switching rate based on historical coupler maintenance data and a preset time series analysis model.

[0259] The maintenance quantity prediction module 602 is used to obtain the predicted maintenance quantity based on historical replacement component maintenance data, historical passenger traffic data, and a trained vector machine model.

[0260] The component replacement quantity prediction module 603 is used to obtain the predicted component replacement quantity based on the predicted component replacement rate and the predicted maintenance quantity.

[0261] In an optional implementation, the even commutation rate prediction module 601 is used for:

[0262] Based on the historical replacement component maintenance data, the historical replacement component maintenance quantity and the historical replacement component non-conforming quantity are obtained.

[0263] The historical replacement rate is obtained based on the historical maintenance volume of replacement parts and the historical non-conforming maintenance volume of replacement parts.

[0264] Based on the historical even commutation rate and the preset time series analysis model, the predicted even commutation rate is obtained.

[0265] In an optional implementation, the even commutation rate prediction module 601 is used for:

[0266] The historical even commutation rates are preprocessed to obtain the preprocessed even commutation rates.

[0267] Based on the preprocessed even commutation rate and the preset time series analysis model, the predicted even commutation rate is obtained.

[0268] In an optional implementation, a secondary even commutation rate prediction module is further included, for:

[0269] Before obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model, it is determined whether the average value of the historical even commutation rate is within the preset even commutation rate range. If not, the average value of the historical even commutation rate is used as the predicted even commutation rate.

[0270] In an optional implementation, a historical even commutation rate correction module is also included, for:

[0271] Before obtaining the predicted even switching rate based on the historical even switching rate and the preset time series analysis model, the historical even switching rate is input into the anomaly detection algorithm to obtain the anomaly score corresponding to each historical even switching rate value.

[0272] Based on the anomaly score and the preset anomaly score threshold, the anomaly score is obtained from the anomaly score.

[0273] The abnormal historical even slew rate value is obtained from the anomaly score.

[0274] The historical even swapping rate is corrected based on the abnormal historical even swapping rate value.

[0275] In an optional implementation, a historical even commutation rate correction and verification module is further included, for:

[0276] Before correcting the historical even switching rate based on the abnormal historical even switching rate value, the normal range of the historical even switching rate is obtained based on the average value and the standard deviation of the historical even switching rate.

[0277] Determine whether the abnormal historical even switching rate value is within the normal range of the historical even switching rate. If it is, then the step of correcting the historical even switching rate based on the abnormal historical even switching rate value is not performed.

[0278] In an optional implementation, a smoothing processing module is also included, for:

[0279] Before obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model, the preset critical parameters and the historical even commutation rate are input into the stationarity test algorithm to obtain the stationarity type of the historical even commutation rate.

[0280] Determine whether the stationarity type is stationary; if not, perform stationarization processing on the historical even commutation rate according to the stationarity type to obtain the stationary even commutation rate.

[0281] Based on the stationary even commutation rate and the preset time series analysis model, the predicted even commutation rate is obtained.

[0282] In an optional implementation, the stabilization processing module is used to:

[0283] Determine whether the stationarity type is a non-stationary term without an intercept;

[0284] If so, differential stationarization is performed on the historical even commutation rates to obtain stationary even commutation rates;

[0285] If not, perform linear fitting on the historical even commutation rates to de-trend and obtain stable even commutation rates.

[0286] In an optional implementation, a time series analysis model order determination module is also included, for:

[0287] Before obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model, the autoregressive order and the moving average order of the time series analysis model are selected from the preset set of autoregressive orders and the preset set of moving average orders based on the historical even commutation rate.

[0288] The time series analysis model is constructed based on the autoregressive order and the moving average order.

[0289] In an optional implementation, the time series analysis model order determination module is used for:

[0290] The historical even commutation rate is input into a preset parameter estimation function required for constructing a time series analysis model to obtain an estimated value;

[0291] Based on the estimated values, an information content function is constructed;

[0292] By cross-grouping the autoregressive order elements in the set of autoregressive orders and the moving average order elements in the set of moving average orders, multiple time series analysis model order combinations are obtained.

[0293] Each time series analysis model order combination is input into the information content function to obtain multiple corresponding information contents;

[0294] The autoregression order and the moving average order are obtained by combining the time series analysis model orders corresponding to the minimum information content in the information content.

[0295] In an optional implementation, a time series analysis model residual verification module is also included, for:

[0296] The time series analysis model includes residuals;

[0297] Before obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model, it is determined whether the residual of the time series analysis model is white noise. If not, the time series analysis model is reconstructed.

[0298] In an optional implementation, a stabilization and restoration module is also included, for:

[0299] The historical even commutation rate is input into the time series analysis model to obtain the initial even commutation rate;

[0300] The initial even commutation rate is subjected to a reverse restoration process corresponding to the stabilization process to obtain the predicted even commutation rate.

[0301] In an optional implementation, a current even commutation rate confidence verification module is also included, for:

[0302] After obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model, the preset confidence level and the historical even commutation rate are input into the time series analysis model to obtain the even commutation rate confidence interval.

[0303] Determine whether the current even commutation rate is within the confidence interval of the even commutation rate. If not, issue an alarm to the staff.

[0304] In an optional implementation, a vector machine model building module is also included, for:

[0305] Before obtaining the predicted maintenance quantity based on historical replacement component maintenance data, historical passenger transport data, and the trained vector machine model, the historical replacement component maintenance quantity is obtained based on the historical replacement component maintenance data.

[0306] Based on the aforementioned historical passenger traffic data, the historical passenger turnover was obtained;

[0307] Based on the current maintenance volume of the replacement parts, the historical maintenance volume of the replacement parts, and the historical passenger turnover, the key parameter penalty coefficients and kernel function coefficients of the vector machine model are selected from the preset set of key parameter penalty coefficients and the preset set of kernel function coefficients.

[0308] The vector machine model is constructed based on the key parameters, penalty coefficient and kernel function coefficient.

[0309] The vector machine model is trained using the current maintenance volume of the replacement parts, the historical maintenance volume of the replacement parts, and the historical passenger turnover as training samples.

[0310] In an optional implementation, the vector machine model building module is used for:

[0311] Cross-grouping the key parameter penalty coefficient elements in the key parameter penalty coefficient set and the kernel function coefficient elements in the kernel function coefficient set yields multiple vector machine model parameter combinations.

[0312] Construct a corresponding test model based on each combination of vector machine model parameters;

[0313] The historical replacement component maintenance volume and the historical passenger turnover volume are input into each of the test models to obtain the corresponding test model output values;

[0314] The key parameter penalty coefficient and kernel function coefficient are obtained by combining the vector machine model parameters corresponding to the output value that is closest to the current maintenance quantity of the replacement component in the output value of the test model.

[0315] In an optional implementation, the replacement component quantity prediction module 603 is used for:

[0316] Multiply each predicted couple switching rate value in the predicted couple switching rate by the corresponding predicted maintenance quantity value in the predicted maintenance quantity to obtain multiple corresponding couple switching component quantity components.

[0317] The predicted number of even-changing components is obtained based on the component number of even-changing components.

[0318] In an optional implementation, a replacement part quantity verification module is also included, for:

[0319] After obtaining the predicted number of coupler replacement parts based on the predicted coupler replacement rate and the predicted maintenance quantity, the maximum component and the minimum component of the number of coupler replacement parts are obtained based on the predicted number of coupler replacement parts.

[0320] Multiply the maximum component of the number of the coupler by a preset confidence level to obtain the standard upper limit of the number of couplers.

[0321] Multiply the minimum component of the number of the coupler by a preset confidence level to obtain the standard lower limit of the number of couplers.

[0322] Determine whether the current number of replacement parts is greater than the upper limit of the replacement part quantity standard or less than the lower limit of the replacement part quantity standard. If so, issue an alarm to the staff.

[0323] Since the principle of the replacement component quantity prediction device 600 is similar to the above method, the implementation of this replacement component quantity prediction device 600 can refer to the implementation of the above method, and will not be repeated here.

[0324] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer device, specifically, a computer device can be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or any combination of these devices.

[0325] In a typical example, a computer device specifically includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the method described above.

[0326] The following is for reference. Figure 7 It shows a schematic diagram of the structure of a computer device 700 suitable for implementing the embodiments of this application.

[0327] like Figure 7 As shown, the computer device 700 includes a central processing unit (CPU) 701, which can perform various appropriate tasks and processes based on programs stored in read-only memory (ROM) 702 or programs loaded from storage section 708 into random access memory (RAM) 703. The RAM 703 also stores various programs and data required for the operation of the system 700. The CPU 701, ROM 702, and RAM 703 are interconnected via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0328] The following components are connected to the I / O interface 705: an input section 706 including a keyboard, mouse, etc.; an output section 707 including a cathode ray tube (CRT), liquid crystal feedback (LCD), etc., and speakers, etc.; a storage section 708 including a hard disk, etc.; and a communication section 709 including a network interface card such as a LAN card, modem, etc. The communication section 709 performs communication processing via a network such as the Internet. A drive 710 is also connected to the I / O interface 705 as needed. A removable medium 711, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed on the drive 710 as needed so that computer programs read from it can be installed in the storage section 708 as needed.

[0329] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program tangibly embodied on a machine-readable medium, the computer program including program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 709, and / or installed from removable medium 711.

[0330] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0331] For ease of description, the above devices are described separately by function as various units. Of course, in implementing this application, the functions of each unit can be implemented in one or more software and / or hardware.

[0332] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0333] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0334] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0335] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0336] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0337] This application can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0338] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0339] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for predicting the quantity of evenly replaced parts, characterized in that, include: Based on historical maintenance data of coupler replacement parts and a preset time series analysis model, the predicted coupler replacement rate is obtained; Based on historical replacement parts maintenance data, historical passenger traffic data, and a trained vector machine model, the predicted maintenance volume is obtained. Based on the predicted coupler replacement rate and the predicted maintenance quantity, the predicted number of coupler replacement parts is obtained. The method further includes: before obtaining the predicted maintenance quantity based on historical replacement component maintenance data, historical passenger traffic data, and the trained vector machine model, obtaining the historical replacement component maintenance quantity based on the historical replacement component maintenance data; obtaining the historical passenger traffic volume based on the historical passenger traffic data; selecting the key parameter penalty coefficient and kernel function coefficient of the vector machine model from a preset set of key parameter penalty coefficients and a preset set of kernel function coefficients based on the current replacement component maintenance quantity, the historical replacement component maintenance quantity, and the historical passenger traffic volume; constructing the vector machine model based on the key parameter penalty coefficient and kernel function coefficient; training the vector machine model using the current replacement component maintenance quantity, the historical replacement component maintenance quantity, and the historical passenger traffic volume as training samples; wherein, the historical passenger traffic volume includes historical passenger traffic volume values ​​for a single time point or time period and multiple time points or time periods in history, and the corresponding time of the historical passenger traffic volume must be earlier than the maintenance time corresponding to the current replacement component maintenance quantity; The step of obtaining the predicted number of coupler components based on the predicted coupler switching rate and the predicted maintenance quantity includes: multiplying each predicted coupler switching rate value in the predicted coupler switching rate by the corresponding predicted maintenance quantity value in the predicted maintenance quantity to obtain multiple corresponding coupler component quantity components; and obtaining the predicted number of coupler components based on the coupler component quantity components.

2. The method according to claim 1, characterized in that, The step of obtaining the predicted replacement rate based on historical replacement component maintenance data and a preset time series analysis model includes: Based on the historical replacement component maintenance data, the historical replacement component maintenance quantity and the historical replacement component non-conforming quantity are obtained. The historical replacement rate is obtained based on the historical maintenance volume of replacement parts and the historical non-conforming maintenance volume of replacement parts. Based on the historical even commutation rate and the preset time series analysis model, the predicted even commutation rate is obtained.

3. The method according to claim 2, characterized in that, The step of obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model includes: The historical even commutation rates are preprocessed to obtain the preprocessed even commutation rates. Based on the preprocessed even commutation rate and the preset time series analysis model, the predicted even commutation rate is obtained.

4. The method according to claim 2, characterized in that, Further includes: Before obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model, it is determined whether the average value of the historical even commutation rate is within the preset even commutation rate range. If not, the average value of the historical even commutation rate is used as the predicted even commutation rate.

5. The method according to claim 2, characterized in that, Further includes: Before obtaining the predicted even switching rate based on the historical even switching rate and the preset time series analysis model, the historical even switching rate is input into the anomaly detection algorithm to obtain the anomaly score corresponding to each historical even switching rate value. Based on the anomaly score and the preset anomaly score threshold, the anomaly score is obtained from the anomaly score. The abnormal historical even slew rate value is obtained from the anomaly score. The historical even swapping rate is corrected based on the abnormal historical even swapping rate value.

6. The method according to claim 5, characterized in that, Further includes: Before correcting the historical even switching rate based on the abnormal historical even switching rate value, the normal range of the historical even switching rate is obtained based on the average value and the standard deviation of the historical even switching rate. Determine whether the abnormal historical even switching rate value is within the normal range of the historical even switching rate. If it is, then the step of correcting the historical even switching rate based on the abnormal historical even switching rate value is not performed.

7. The method according to claim 2, characterized in that, Further includes: Before obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model, the preset critical parameters and the historical even commutation rate are input into the stationarity test algorithm to obtain the stationarity type of the historical even commutation rate. Determine whether the stationarity type is stationary; if not, perform stationarization processing on the historical even commutation rate according to the stationarity type to obtain the stationary even commutation rate. Based on the stationary even commutation rate and the preset time series analysis model, the predicted even commutation rate is obtained.

8. The method according to claim 7, characterized in that, The process of stabilizing the historical even commutation rate according to the stationarity type to obtain the stationary even commutation rate includes: Determine whether the stationarity type is a non-stationary term without an intercept; If so, differential stationarization is performed on the historical even commutation rates to obtain stationary even commutation rates; If not, perform linear fitting on the historical even commutation rates to de-trend and obtain stable even commutation rates.

9. The method according to claim 2, characterized in that, Further includes: Before obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model, the autoregressive order and the moving average order of the time series analysis model are selected from the preset set of autoregressive orders and the preset set of moving average orders based on the historical even commutation rate. The time series analysis model is constructed based on the autoregressive order and the moving average order.

10. The method according to claim 9, characterized in that, The step of selecting the autoregressive order and moving average order of the time series analysis model from a preset set of autoregressive orders and a preset set of moving average orders based on the historical even commutation rate includes: The historical even commutation rate is input into a preset parameter estimation function required for constructing a time series analysis model to obtain an estimated value; Based on the estimated values, an information content function is constructed; By cross-grouping the autoregressive order elements in the set of autoregressive orders and the moving average order elements in the set of moving average orders, multiple time series analysis model order combinations are obtained. Each time series analysis model order combination is input into the information content function to obtain multiple corresponding information contents; The autoregression order and the moving average order are obtained by combining the time series analysis model orders corresponding to the minimum information content in the information content.

11. The method according to claim 2, characterized in that, Further includes: The time series analysis model includes residuals; Before obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model, it is determined whether the residual of the time series analysis model is white noise. If not, the time series analysis model is reconstructed.

12. The method according to claim 7, characterized in that, When the stationarity type is not stationary, the step of obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model includes: The historical even commutation rate is input into the time series analysis model to obtain the initial even commutation rate; The initial even commutation rate is subjected to a reverse restoration process corresponding to the stabilization process to obtain the predicted even commutation rate.

13. The method according to claim 2, characterized in that, Further includes: After obtaining the predicted even commutation rate based on the historical even commutation rate and the preset time series analysis model, the preset confidence level and the historical even commutation rate are input into the time series analysis model to obtain the even commutation rate confidence interval. Determine whether the current even commutation rate is within the confidence interval of the even commutation rate. If not, issue an alarm to the staff.

14. The method according to claim 1, characterized in that, The step of selecting the key parameter penalty coefficients and kernel function coefficients of the vector machine model from a preset set of key parameter penalty coefficients and a preset set of kernel function coefficients based on the current maintenance volume of the replacement parts, the historical maintenance volume of the replacement parts, and the historical passenger turnover includes: Cross-grouping the key parameter penalty coefficient elements in the key parameter penalty coefficient set and the kernel function coefficient elements in the kernel function coefficient set yields multiple vector machine model parameter combinations. Construct a corresponding test model based on each combination of vector machine model parameters; The historical replacement component maintenance volume and the historical passenger turnover volume are input into each of the test models to obtain the corresponding test model output values; The key parameter penalty coefficient and kernel function coefficient are obtained by combining the vector machine model parameters corresponding to the output value that is closest to the current maintenance quantity of the replacement component in the output value of the test model.

15. The method according to claim 1, characterized in that, Further includes: After obtaining the predicted number of coupler replacement parts based on the predicted coupler replacement rate and the predicted maintenance quantity, the maximum component and the minimum component of the number of coupler replacement parts are obtained based on the predicted number of coupler replacement parts. Multiply the maximum component of the number of the coupler by a preset confidence level to obtain the standard upper limit of the number of couplers. Multiply the minimum component of the number of the coupler by a preset confidence level to obtain the standard lower limit of the number of couplers. Determine whether the current number of replacement parts is greater than the upper limit of the replacement part quantity standard or less than the lower limit of the replacement part quantity standard. If so, issue an alarm to the staff.

16. A device for predicting the quantity of interchangeable parts, characterized in that, include: The coupler switching rate prediction module is used to obtain the predicted coupler switching rate based on historical coupler switching component maintenance data and a preset time series analysis model. The maintenance quantity prediction module is used to obtain the predicted maintenance quantity based on historical replacement component maintenance data, historical passenger traffic data, and a trained vector machine model. The component replacement quantity prediction module is used to obtain the predicted component replacement quantity based on the predicted component replacement rate and the predicted maintenance quantity. The module for predicting the quantity of replacement parts is further configured to: obtain the historical replacement part maintenance quantity based on the historical replacement part maintenance data; obtain the historical passenger turnover based on the historical passenger traffic data; select the key parameter penalty coefficient and kernel function coefficient of the vector machine model from a preset set of key parameter penalty coefficients and a preset set of kernel function coefficients based on the current replacement part maintenance quantity, the historical replacement part maintenance quantity, and the historical passenger turnover; construct the vector machine model based on the key parameter penalty coefficient and kernel function coefficient; and train the vector machine model using the current replacement part maintenance quantity, the historical replacement part maintenance quantity, and the historical passenger turnover as training samples. The historical passenger turnover includes historical passenger turnover values ​​at a single point in time or time period, as well as at multiple points in time or time periods, and the corresponding time of the historical passenger turnover must be earlier than the maintenance time corresponding to the current replacement part maintenance quantity. The component quantity prediction module is further used for: Multiply each predicted couple switching rate value in the predicted couple switching rate by the corresponding predicted maintenance quantity value in the predicted maintenance quantity to obtain multiple corresponding couple switching component quantity components. The predicted number of even-changing components is obtained based on the component number of even-changing components.

17. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1-15.

18. A computer-readable medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-15.

Images