Railway material demand prediction method and device and electronic equipment
By processing and analyzing the basic classification and influencing factors data of railway materials, and combining multiple models to build an adaptive adjustment mechanism, the problem of relying on experience in railway material demand forecasting has been solved, and more accurate demand forecasting and supply support have been achieved.
Patent Information
- Application Number
- CN202511042502.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-28
- Publication Date
- 2025-11-07
AI Technical Summary
In railway materials management, demand forecasting relies on personal experience and lacks scientific basis, resulting in large forecasting errors, material stockpiling or untimely supply, and affecting fund utilization and production operations.
By acquiring basic classification and influencing factor data of railway materials, data processing and time series pattern analysis are performed. Combined with models such as ARIMA, BP neural network, and LSTM, further subdivision and attribute feature analysis are conducted to construct an adaptive adjustment mechanism and optimize demand forecasting results.
It has improved the accuracy of railway material demand forecasting, ensured timely supply, reduced material backlog, and supported scientific decision-making.
Smart Images
Figure CN120912255A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of railway materials, in particular to a railway material demand prediction method, device and electronic equipment. BACKGROUND
[0002] In the traditional railway material management practice process, demand personnel mostly rely on personal experience for demand evaluation and prediction, lack of scientific basis, and the prediction error is large, which causes the consequences of material accumulation or supply not in time, and causes certain influence on railway fund application or production operation, therefore, a railway material demand prediction method, device and electronic equipment are urgently needed. SUMMARY
[0003] Therefore, it is necessary to provide a railway material demand prediction method, device and electronic equipment in view of the problems that in the traditional railway material management practice process, demand personnel mostly rely on personal experience for demand evaluation and prediction, lack of scientific basis, and the prediction error is large, which causes the consequences of material accumulation or supply not in time, and causes certain influence on railway fund application or production operation.
[0004] The present application provides a railway material demand prediction method, which comprises:
[0005] According to the basic classification and influence factors of railway materials, the actual delivery data of each category of materials of all or part of users in a set period, the adjustment data and the corresponding data of the influence factors are obtained, and the obtained data is processed to analyze the time sequence rule of consumption;
[0006] Based on the obtained data and time sequence rule, the railway materials are subdivided and attribute characteristic analyzed, and the demand quantity of future time period is predicted;
[0007] Based on the results of demand prediction, a mechanism for automatically selecting the optimal demand prediction result is constructed, and the error analysis result is outputted;
[0008] Based on the error analysis result, an adaptive adjustment mechanism of the selected demand prediction model is constructed.
[0009] In one of the embodiments, the data processing of the obtained data comprises:
[0010] The multiple delivery quantities of the same material on the same day are added and processed;
[0011] And / or, for the abnormal value exceeding the boundary value, the boundary value is taken as the value;
[0012] And / or, all the delivery quantities are normalized or de-normalized, and the specific normalization processing calculation formula is as follows:
[0013]
[0014] Wherein, X is the original value of the outbound quantity, X max , X min are the maximum value and the minimum value of the outbound quantity, respectively;
[0015] And / or, in response to the data after the outlier optimization and the inverse normalization processing being a decimal, performing an integer processing on the data.
[0016] In one of the embodiments, the demand quantity in the future time period is predicted, including:
[0017] In response to the attribute feature of the railway material being a replaceable part material, acquiring the applicable vehicle type and the state of each vehicle type, and determining the material demand time and quantity;
[0018] In response to the attribute feature of the railway material being a replaceable part material, identifying the material demand prediction result.
[0019] In one of the embodiments, in response to the attribute feature of the railway material being a replaceable part material, identifying the material demand prediction result, including:
[0020] Acquiring the attribute feature data of the railway replaceable part material;
[0021] Inputting the attribute feature data of the railway replaceable part material into a basic demand prediction model to output the demand quantity prediction result in the future time period, the basic demand prediction model being trained by using the historical demand sample of the replaceable part material, the applicable vehicle type sample, and the state sample of each vehicle type.
[0022] In one of the embodiments, based on the demand prediction results, a mechanism for automatically selecting the optimal demand prediction result is constructed, and an error analysis result is output, including:
[0023] Based on the demand quantity results in the future time period predicted by the ARIMA, BP neural network, LSTM, and RNN basic demand prediction models, analyzing the error indicators and the corresponding influence weights;
[0024] Taking the minimum comprehensive error value as the objective function, a mechanism for automatically selecting the most matched demand prediction scheme is constructed, and the evaluation indicators and the influence proportion with the largest error influence are output.
[0025] In one of the embodiments, based on the error analysis result, a self-adaptive adjustment mechanism of the selected demand prediction model is constructed, including:
[0026] Based on the error analysis result and the data accumulation, the basic law of the demand prediction error is analyzed;
[0027] In response to the most influential parameter, the influence law of the parameter can be analyzed, the parameter is adjusted, and the new prediction result after adjustment is selected as a reference;
[0028] The demand prediction result is adjusted by a fixed amount.
[0029] In one embodiment, the fixed amount adjustment of the demand prediction result comprises:
[0030] In response to the error proportion after multiple demand prediction calculations being higher than the actual value for positive and lower than the actual value for negative, an arithmetic mean or a weighted mean of the error proportion after multiple demand prediction calculations is calculated;
[0031] The arithmetic mean or the weighted mean is used as an adjustment amount for adjusting the demand prediction value.
[0032] The present application also provides a railway material demand prediction device, comprising:
[0033] The acquisition module is used to acquire the actual delivery data of each category of materials of all or part of users in a set period, adjustment data and corresponding data of influence factors according to the basic classification and influence factors of railway materials, and to process the acquired data to analyze the time sequence law of consumption;
[0034] The prediction module is used to subdivide and analyze the attribute characteristics of railway materials based on the acquired data and the time sequence law, and to predict the demand amount in the future time period;
[0035] The output module is used to construct a mechanism for automatically selecting the optimal demand prediction result based on a plurality of demand prediction results, and to output the error analysis result;
[0036] The construction module is used to construct an adaptive adjustment mechanism of the selected demand prediction model based on the error analysis result.
[0037] The present application also provides an electronic device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the railway material demand prediction method of any one of the above when executing the computer program.
[0038] The present application also provides a computer readable storage medium storing a computer program, wherein the computer program is executed by a processor to implement the railway material demand prediction method of any one of the above.
[0039] The railway material demand prediction method, device and electronic equipment, by obtaining the actual delivery data of each category of materials of all or part of users in a set period, adjustment data and data corresponding to influencing factors according to the basic classification and influencing factors of railway materials, facilitate the analysis and summary of historical laws, so as to analyze more subdivided attribute characteristics, comprehensively apply the relevant provisions of material management, various methods of demand prediction, carry out demand prediction and error evaluation, and construct a self-adaptive adjustment mechanism according to the error condition, fully guarantee the accuracy of demand prediction, and provide decision support for railway material demand. BRIEF DESCRIPTION OF DRAWINGS
[0040] In order to more clearly illustrate the technical solutions in the present application or prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.
[0041] Figure 1 A railway material demand prediction method flowchart in an embodiment;
[0042] Figure 2 A data processing flowchart for the obtained data in an embodiment;
[0043] Figure 3 A demand amount prediction flowchart for a future time period in an embodiment;
[0044] Figure 4 A railway material demand prediction result identification flowchart in an embodiment;
[0045] Figure 5 A mechanism flowchart for constructing an automatic selection of the optimal demand prediction result in an embodiment;
[0046] Figure 6 A self-adaptive adjustment mechanism flowchart for constructing a selected demand prediction model in an embodiment;
[0047] Figure 7 A demand prediction result fixed amount adjustment flowchart in an embodiment;
[0048] Figure 8 A railway material demand prediction device structure flowchart in an embodiment;
[0049] Figure 9 An internal structure diagram of an electronic device in an embodiment. DETAILED DESCRIPTION
[0050] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.
[0051] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Figures 1-9 The railway material demand prediction method, device and electronic equipment of the present application are described.
[0052] As shown in the drawings, Figure 1 In one embodiment, a railway material demand prediction method comprises the following steps:
[0053] In step S100, according to the basic classification and influencing factors of railway materials, the actual delivery data of each category of materials of all or part of users in a set period, the adjustment data and the corresponding data of influencing factors are obtained, and the obtained data is processed to analyze the time sequence rule of consumption.
[0054] According to the different regulations and replacement frequencies of relevant repair regulations and schedules, railway materials are divided into two categories: must-replace parts and occasional-replace parts. The must-replace parts are mainly subject to the regulations and schedules of repair, and the influencing factors mainly include vehicle type, repair cycle, running mileage, material list, etc. The influencing factors of occasional-replace parts mainly include running mileage, 100-kilometer failure rate, etc.
[0055] All users refer to all transport stations and sections of the railway and users at the same level; part of the users refer to the transport stations and sections of the railway bureau group company and users at the same level. The time period is determined according to the amount of data required by the method (in days as the dimension); if there are multiple methods for prediction comparison, the time period of the longest one is taken as the standard. The delivery (consumption) data and adjustment data are mainly collected from the railway material management system; the corresponding data of influencing factors are mainly collected from various professional systems and repair regulations.
[0056] In step S200, based on the obtained data and time sequence rule, the railway materials are subdivided and attribute characteristics are analyzed, and the demand amount of the future time period is predicted.
[0057] For the collected data, the data processing method includes addition, abnormal value optimization, normalization and reverse normalization, rounding, etc.; based on the processed data, the consumption time sequence rule is analyzed through mathematical statistics and related verification methods.
[0058] For the spare parts, the time sequence rule is analyzed again, and the material is classified and attributed. For example, the A material in the spare parts presents a basic rule of consuming 50 units every 10 (numerical value is an example) days, and is classified as a regular and quantitative material.
[0059] For the spare parts, through the application of several vehicle models, the current vehicle state (maintenance cycle, driving mileage, etc.), it can be inferred that when the several levels of repair process are needed, the material list (material type, quantity) corresponding to the repair process is determined, and the quantity of the same material is added to determine the material demand time and quantity.
[0060] For the spare parts, the demand prediction method includes traditional prediction method, intelligent prediction method, and combined prediction method, more than 10 specific methods, such as ARIMA, BP neural network, LSTM, RNN, etc.
[0061] Based on the above listed models, the basic demand prediction model can be constructed and coded; based on the re-subdivision and attribute feature analysis of railway materials, combined with the application conditions of the method, 2-3 kinds of demand prediction methods suitable for a certain material are selected, and the demand quantity in the future time period is predicted. The future time period and the division dimension (day, week, half month, month, etc.) can be adjusted according to user requirements.
[0062] Step S300, based on the results of demand prediction, an automatic selection mechanism of the optimal demand prediction result is constructed, and the error analysis result is output.
[0063] It is convenient to determine the optimal demand prediction result, and the accuracy of railway material demand prediction is improved.
[0064] Step S400, based on the error analysis result, a self-adaptive adjustment mechanism of the selected demand prediction model is constructed.
[0065] The railway material demand prediction method, by acquiring the actual delivery data of each category of materials of all or part of users in a set period, adjusting data and corresponding data of influencing factors according to the basic classification and influencing factors of railway materials, it is convenient to analyze and summarize the historical law, so as to analyze more subdivided attribute characteristics, by comprehensively applying the relevant regulations of material management, a variety of demand prediction methods, demand prediction and error evaluation, and according to the error situation, a self-adaptive adjustment mechanism is constructed, which fully guarantees the accuracy of demand prediction, and provides decision support for railway material demand.
[0066] In this embodiment, referring to Figure 2 The data obtained is processed, including the following steps:
[0067] Step S110, for the same item, the same day, the sum of the number of multiple outbound quantity processing.
[0068] Conveniently determine the same item, the same day, the specific quantity of multiple outbound.
[0069] Step S120, and / or, for the abnormal value exceeding the boundary value, according to the boundary value.
[0070] The abnormal value exceeding the boundary value, that is, the maximum value of the median and the average number of N (N generally takes 10, special case can be adjusted according to the actual) times the quantity value.
[0071] Step S130, and / or, for all the outbound quantity to do normalization processing or inverse normalization processing, the specific normalization processing formula is as follows:
[0072]
[0073] Wherein, X is the original value of the outbound quantity, X max , X min The maximum and minimum value of the outbound quantity.
[0074] Normalization processing is done for all the outbound quantity, which is convenient to eliminate the data dimension and reduce the calculation value order of magnitude. The inverse normalization is the inverse conversion of the normalization. Through the inverse operation conversion, it is convenient to return to the original value.
[0075] Step S140, and / or, in response to the data after the abnormal value optimization and inverse normalization processing is a decimal, the data is rounded.
[0076] Through a series of data processing, the data is more reasonable, and the data with large deviation is eliminated, so as to obtain better demand prediction result.
[0077] In this embodiment, see Figure 3 The demand of future time period is predicted, including the following steps:
[0078] Step S210, in response to the attribute characteristics of railway materials being the replaceable parts, the applicable vehicle type and the current state of each vehicle type are obtained, and the material demand time and quantity are determined.
[0079] For replaceable parts, through the applicable several vehicle types and the current state of each vehicle type (repair cycle, walking mileage, etc.), it can be inferred that when several levels of repair process are needed, the material list (material type, quantity) corresponding to the repair process is determined, and the same material quantity is summed up to determine the material demand time and quantity.
[0080] Step S220, in response to the attribute characteristics of railway materials being the replaceable parts, the material demand prediction result is identified.
[0081] For the odd replacement parts, the demand prediction method includes three categories: traditional prediction method, intelligent prediction method and combined prediction method, more than 10 specific methods, such as ARIMA, BP neural network, LSTM, RNN, etc. For example, ARIMA model can predict a result, BP neural network can predict a result, LSTM can predict a result, …, which is convenient for subsequent comparison and optimization of multiple model prediction results.
[0082] In this embodiment, referring to Figure 4 , in response to the attribute characteristics of the railway materials being odd replacement parts, the material demand prediction result is identified, including:
[0083] Step S221, attribute characteristic data of railway odd replacement parts is obtained.
[0084] Step S222, the attribute characteristic data of the railway odd replacement parts is input into the basic demand prediction model, and the demand quantity prediction result of the future time period is output. The basic demand prediction model is trained by using the historical demand sample of the odd replacement parts, the applicable vehicle sample and the state sample of the current vehicle as the training data.
[0085] Through the sequence training of ARIMA, BP neural network, LSTM, RNN and other basic demand prediction models, the demand quantity prediction result of the future time period is conveniently output.
[0086] In this embodiment, referring to Figure 5 , based on the results of demand prediction, a mechanism for automatically selecting the optimal demand prediction result is constructed, and the error analysis result is output, including the following steps:
[0087] Step S310, based on the demand quantity results of the future time period predicted by the ARIMA, BP neural network, LSTM and RNN basic demand prediction models, the error index and the corresponding influence weight are analyzed.
[0088] For example, in June 2025, the ARIMA model prediction result is 100 units, the BP neural network model prediction result is 150 units, the LSTM model prediction result is 180 units, …, for the above prediction results, the corresponding error indexes are: MAPE, MSE, R 2 , etc. (if any), each index has different influence degree on different materials, such as the weight influence of MAPE on some materials is MAPE, then the corresponding weight of MAPE is higher; the weight influence of some materials is R 2 , then the corresponding weight of R 2 is higher.
[0089] Step S320, a mechanism for automatically selecting the demand prediction scheme that best matches the demand prediction scheme is constructed with the minimum integrated error value as the objective function, and the evaluation index and the influence proportion of the error that has the greatest influence are output.
[0090] According to the different influence degrees of the error indicators on the prediction results of the materials, the corresponding weight proportions are distributed, such as through a certain method, it can be deduced that the weight proportions of the three error indicators are 4:4:3 or 1:2:7. Then, the prediction integrated error is obtained by multiplying the error value (if necessary, normalized) by the corresponding weight proportion.
[0091] In this embodiment, referring to Figure 6 , based on the error analysis result, an adaptive adjustment mechanism of the selected demand prediction model is constructed, including the following steps:
[0092] Step S410, based on the error analysis result and data accumulation, the basic law of demand prediction error is analyzed.
[0093] By analyzing the basic law of demand prediction error according to the error analysis result and data accumulation, the adaptive adjustment mechanism of the selected demand prediction model is constructed.
[0094] Step S420, in response to the parameter that has the greatest influence, the influence law of the parameter is analyzed, the parameter is adjusted, and the new prediction result after adjustment is selected as a reference.
[0095] Taking the BP application network as an example, if it can be found through a certain method that the parameter that has the greatest influence is the learning efficiency lr, the adjustment of the multiple parameter values of the learning efficiency and the corresponding prediction are focused on, and the optimal prediction result is selected.
[0096] Step S430, the demand prediction result is adjusted by a fixed amount.
[0097] For example, after 50 times of demand prediction calculation, the error proportions are calculated, and it is stipulated that the error proportions that are higher than the actual value are positive and the error proportions that are lower than the actual value are negative. The arithmetic mean or weighted mean of the error proportions is calculated, the arithmetic mean or weighted mean is used as the adjustment amount, and the demand prediction value is adjusted.
[0098] In this embodiment, referring to Figure 7 , the demand prediction result is adjusted by a fixed amount, including the following steps:
[0099] Step S431, in response to the error proportions that are higher than the actual value being positive and the error proportions that are lower than the actual value being negative after multiple times of demand prediction calculation, the arithmetic mean or weighted mean of the error proportions after multiple times of demand prediction calculation is calculated.
[0100] Step S432, the adjustment amount is the arithmetic mean or the weighted mean, and the adjustment of the demand prediction value is performed.
[0101] The demand prediction model can be dynamically adjusted according to the error analysis result, so as to improve the accuracy of the prediction result.
[0102] The railway material demand prediction device provided by the present application is described below. The railway material demand prediction device described below can be correspondingly referred to the railway material demand prediction method described above.
[0103] As shown in Figure 8 In one embodiment, a railway material demand prediction device includes an acquisition module 810, a prediction module 820, an output module 830, and a construction module 840.
[0104] The acquisition module 810 is configured to acquire, according to the basic classification and the influence factors of the railway materials, the actual delivery data of each category of materials of all or part of users in a set period, the adjustment data, and the data corresponding to the influence factors, and perform data processing on the acquired data to analyze the time sequence rule of consumption.
[0105] The prediction module 820 is configured to subdivide and analyze the attribute characteristics of the railway materials based on the acquired data and the time sequence rule, and predict the demand amount in a future time period.
[0106] The output module 830 is configured to construct a mechanism for automatically selecting an optimal demand prediction result based on a plurality of results of demand prediction, and output an error analysis result.
[0107] The construction module 840 is configured to construct an adaptive adjustment mechanism of the selected demand prediction model based on the error analysis result.
[0108] In this embodiment, the acquired data is processed, and specifically used for:
[0109] The multiple delivery quantities of the same material on the same day are added together;
[0110] And / or, for the abnormal values exceeding the boundary value, the boundary value is taken;
[0111] And / or, all the delivery quantities are normalized or de-normalized, and the specific normalization calculation formula is as follows:
[0112]
[0113] Wherein, X is the original value of the delivery quantity, X max , and X min are the maximum value and the minimum value of the delivery quantity, respectively.
[0114] And / or, in response to the data after the outlier optimization and the inverse normalization processing being a decimal number, performing an integer processing on the data.
[0115] In the embodiment, the demand quantity of the future time period is predicted, and the prediction is specifically used for:
[0116] In response to the attribute feature of the railway material being a spare part material, the applicable vehicle type and the state of each vehicle type are obtained, and the material demand time and quantity are determined.
[0117] In response to the attribute feature of the railway material being a spare part material, the material demand prediction result is identified.
[0118] In the embodiment, in response to the attribute feature of the railway material being a spare part material, the material demand prediction result is identified, and the prediction is specifically used for:
[0119] The attribute feature data of the railway spare part material is obtained.
[0120] The attribute feature data of the railway spare part material is input into a basic demand prediction model, and a demand quantity prediction result of a future time period is output, and the basic demand prediction model is trained by using a historical demand sample of the spare part material, an applicable vehicle type sample, and a state sample of each vehicle type as training data.
[0121] In the embodiment, the output module 830 is specifically used for:
[0122] Based on the demand quantity result of the future time period predicted by the ARIMA, the BP neural network, the LSTM, and the RNN basic demand prediction model, an error index and a corresponding influence weight are analyzed.
[0123] A mechanism for automatically selecting a most matched demand prediction scheme is constructed with a minimum comprehensive error value as an objective function, and an evaluation index and an influence proportion with the largest error influence are output.
[0124] In the embodiment, the construction module 840 is specifically used for:
[0125] Based on the error analysis result and data accumulation, a basic rule of demand prediction error is analyzed.
[0126] In response to the parameter with the largest influence being able to analyze the influence rule of the parameter, the parameter is adjusted, and a new prediction result after the adjustment is selected as a reference.
[0127] The demand prediction result is adjusted by a fixed amount.
[0128] In the embodiment, the demand prediction result is adjusted by a fixed amount, and the adjustment is specifically used for:
[0129] In response to the error proportion after the multiple demand prediction estimations being higher than the positive actual value and lower than the negative actual value, an arithmetic mean or a weighted mean of the error proportion after the multiple demand prediction estimations is calculated;
[0130] The adjustment amount is the arithmetic mean or the weighted mean.
[0131] The railway material demand prediction device, by obtaining the actual delivery data of each category of materials of all or part of users in a set period, adjustment data and corresponding data of influencing factors according to the basic classification and influencing factors of railway materials, facilitates the analysis and summary of historical laws, thereby analyzing more subdivided attribute characteristics, comprehensively applying material management related regulations and various methods of demand prediction, performing demand prediction and error evaluation, and constructing a self-adaptive adjustment mechanism according to the error situation, fully guaranteeing the accuracy of demand prediction and providing decision support for railway material demand.
[0132] Figure 9 An example of a schematic diagram of the physical structure of an electronic device, which can be a smart terminal, is shown in Figure 9 The electronic device includes a processor, a memory and a network interface connected by a system bus. The processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The network interface of the electronic device is used to communicate with external terminals through network connection. The computer program is executed by the processor to implement a railway material demand prediction method, which includes:
[0133] According to the basic classification and influencing factors of railway materials, the actual delivery data of each category of materials of all or part of users in a set period, adjustment data and corresponding data of influencing factors are obtained, and the obtained data is processed to analyze the time sequence law of consumption;
[0134] Based on the obtained data and time sequence law, the railway materials are subdivided and attribute characteristics are analyzed, and the demand quantity of the future time period is predicted;
[0135] Based on the results of demand prediction, a mechanism for automatically selecting the optimal demand prediction result is constructed, and an error analysis result is output;
[0136] Based on the error analysis result, a self-adaptive adjustment mechanism of the selected demand prediction model is constructed.
[0137] Those skilled in the art can understand, Figure 9The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the electronic device to which the scheme of the present application is applied. The specific electronic device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.
[0138] In another aspect, the present application also provides a computer storage medium storing a computer program, the computer program being executed by a processor to implement a railway material demand prediction method, the method comprising:
[0139] According to the basic classification and influencing factors of railway materials, actual delivery data, adjustment data and corresponding data of influencing factors of each category of materials of all or part of users in a set period are obtained, and the obtained data is processed to analyze the time sequence rule of consumption;
[0140] Based on the obtained data and time sequence rule, the railway materials are subdivided and attribute characteristic analyzed, and the demand quantity of the future time period is predicted;
[0141] Based on the results of demand prediction, a mechanism for automatically selecting the optimal demand prediction result is constructed, and an error analysis result is output;
[0142] Based on the error analysis result, an adaptive adjustment mechanism of the selected demand prediction model is constructed.
[0143] In another aspect, a computer program product or computer program is provided, the computer program product or computer program comprising computer instructions stored in a computer readable storage medium. The processor of the electronic device reads the computer instructions from the computer readable storage medium, and the processor implements a railway material demand prediction method when executing the computer instructions, the method comprising:
[0144] According to the basic classification and influencing factors of railway materials, actual delivery data, adjustment data and corresponding data of influencing factors of each category of materials of all or part of users in a set period are obtained, and the obtained data is processed to analyze the time sequence rule of consumption;
[0145] Based on the obtained data and time sequence rule, the railway materials are subdivided and attribute characteristic analyzed, and the demand quantity of the future time period is predicted;
[0146] Based on the results of demand prediction, a mechanism for automatically selecting the optimal demand prediction result is constructed, and an error analysis result is output;
[0147] Based on the error analysis result, an adaptive adjustment mechanism of the selected demand prediction model is constructed.
[0148] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiment methods. Any reference to memory, storage, database or other medium used in the embodiments provided by the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory.
[0149] By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.
[0150] The technical features of the above embodiments can be combined in any way. In order to make the description simple, not all possible combinations of the technical features in the above embodiments are described, but as long as the combination of the technical features does not exist, it should be considered as the scope of the present application.
[0151] The above embodiments only express several embodiments of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the present application. It should be noted that for those skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of the present application. Therefore, the scope of protection of the present application should be subject to the appended claims.
Claims
1. A method for forecasting demand for railway materials, characterized by, The method comprises: According to the basic classification and influencing factors of railway materials, the actual delivery data of each category of materials of all or part of users in a set period, the adjustment data and the corresponding data of the influencing factors are obtained, and the obtained data is processed to analyze the time sequence rule of consumption; Based on the obtained data and the time sequence rule, the railway materials are subdivided and the attribute characteristics are analyzed, and the demand quantity of the future time period is predicted; Based on the results of demand prediction, a mechanism for automatically selecting the optimal demand prediction result is constructed, and the error analysis result is output; Based on the error analysis result, an adaptive adjustment mechanism of the selected demand prediction model is constructed.
2. The method of claim 1, wherein, The data processing of the obtained data comprises: For the same material, the delivery quantities of the same day are added; And / or, for the abnormal values exceeding the boundary value, the boundary value is taken; And / or, all delivery quantities are normalized or denormalized, and the specific normalization calculation formula is as follows: Wherein, X is the original value of the quantity of warehouse out, X max , X min are the maximum value and the minimum value of the quantity of warehouse out, respectively; And / or, in response to the data after abnormal value optimization and denormalization being a decimal, the data is rounded.
3. The method of claim 2, wherein, The demand quantity of the future time period is predicted, comprising: In response to the attribute characteristics of the railway materials being spare parts, the applicable vehicle type and the current state of each vehicle type are obtained, and the material demand time and quantity are determined; In response to the attribute characteristics of the railway materials being occasional spare parts, the material demand prediction result is identified.
4. The method of claim 3, wherein, In response to the attribute characteristics of the railway materials being occasional spare parts, the material demand prediction result is identified, comprising: Obtain the attribute characteristic data of the railway occasional spare parts; The attribute characteristic data of the railway occasional spare parts is input into the basic demand prediction model, and the demand quantity prediction result of the future time period is output, wherein the basic demand prediction model is trained by using the historical demand sample of occasional spare parts, the applicable vehicle type sample and the current state sample of each vehicle type as training data.
5. The method of claim 4, wherein, Based on the demand prediction results, a mechanism for automatically selecting the optimal demand prediction result is constructed, and the error analysis result is output, comprising: Based on the demand quantity results of the future time period predicted by the ARIMA, BP neural network, LSTM and RNN basic demand prediction models, the error indicators and the corresponding influence weights are analyzed; Taking the minimum comprehensive error value as the objective function, a mechanism for automatically selecting the most matched demand prediction scheme is constructed, and the evaluation indicators with the largest error influence and the influence proportion are output.
6. The method of forecasting demand for railway materials according to claim 5, wherein, Based on the error analysis result, an adaptive adjustment mechanism of the selected demand prediction model is constructed, comprising: Based on the error analysis result and data accumulation, the basic rule of demand prediction error is analyzed; In response to the influence of the largest parameter being able to analyze the influence rule of the parameter, the parameter is adjusted, and the new prediction result after adjustment is selected as the reference; The demand prediction result is adjusted by a fixed amount.
7. The method of forecasting demand for railway materials according to claim 6, wherein, The demand prediction result is adjusted by a fixed amount, comprising: In response to the error proportion after multiple demand prediction calculations being higher than the actual value for positive and lower than the actual value for negative, the arithmetic mean or weighted mean of the error proportion after multiple demand prediction calculations is calculated; The adjustment of the demand prediction value is performed by using an arithmetic mean value or a weighted mean value as an adjustment amount.
8. A railway material demand forecasting device characterized by comprising: The method comprises the following steps: An acquisition module is configured to acquire, according to a basic classification and influence factors of railway materials, actual delivery data, adjustment data and corresponding data of influence factors of each category of materials of all or part of users in a set period, and perform data processing on the acquired data to analyze the time sequence rule of consumption; A prediction module is configured to perform subdivision and attribute feature analysis on the railway materials based on the acquired data and the time sequence rule, and predict the demand in a future time period; An output module is configured to construct a mechanism for automatically selecting an optimal demand prediction result based on a plurality of results of demand prediction, and output an error analysis result; A construction module is configured to construct an adaptive adjustment mechanism of the selected demand prediction model based on the error analysis result. 9.An electronic device comprising a memory and a processor, the memory storing a computer program, wherein, The processor executes the computer program to implement the steps of the method of any one of claims 1 to 7.
10. A computer readable storage medium storing a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method of any one of claims 1 to 7.