Inventory management method for spare mechanical parts of heavy-duty car

Abstract

The invention discloses an inventory management method for spare mechanical parts of a heavy-duty car, which comprises the following steps of: identifying the importance degree of the spare mechanical parts by using a multi-criterion classification method, and determining important spare mechanical parts in the spare mechanical parts; preprocessing the historical demand quantity data to obtain important component historical demand quantity data of the important standby mechanical component; constructing a standby mechanical part demand quantity prediction model, training the standby mechanical part demand quantity prediction model, and outputting important part prediction demand quantity data through the trained standby mechanical part demand quantity prediction model; and according to the historical demand quantity data of the important parts and the predicted demand quantity data of the important parts, inventory simulation of optimal demand distribution fitting is carried out, and according to an inventory simulation result, the inventory quantity of the important standby mechanical parts is dynamically adjusted. According to the invention, the overall matching degree between demand prediction and inventory management can be improved, and the inventory management performance of standby mechanical parts is improved.

Description

Technical Field

[0001] The invention relates to the technical field of artificial intelligence, and in particular to an inventory management method for spare mechanical parts of heavy-duty vehicles. Background Art

[0002] In the context of my country's high attention to large-scale manufacturing and infrastructure, the importance of the heavy-duty vehicle industry that plays a basic transportation function has become increasingly prominent, and the effective management of spare mechanical parts for corresponding heavy-duty vehicles (such as trucks and construction machinery) is indispensable. Compared with ordinary automobile spare mechanical parts, due to the long-term and high-load operation characteristics of heavy-duty vehicles, the demand for their spare mechanical parts has greater volatility and intervals (often after integrating the demand on a monthly or quarterly basis, the demand still shows discontinuity). This characteristic seriously hinders the improvement of the performance of demand forecasting and inventory management of this type of spare mechanical parts. Therefore, many companies that mainly engage in heavy-duty automobile spare mechanical parts business currently have low order fulfillment rates, inventory backlogs, and low inventory turnover rates.

[0003] Traditional demand forecasting and inventory management of spare mechanical parts for heavy-duty vehicles include three sequential steps: (i) identification of key spare mechanical parts, (ii) parameter type or artificial intelligence-assisted demand forecasting, and (iii) inventory management of spare mechanical parts based on more accurate forecast results. For the identification of key spare mechanical parts, traditional methods often only consider a single dimension such as capital occupation, lack of systematicity in supply chain management and maintenance management, and easily lead to inconsistent identification of key parts by different systems and departments, and fail to achieve an organic combination of forecasting and inventory. For the demand forecasting and inventory management part, the above traditional solution emphasizes accurate forecasting and then builds an inventory strategy based on the forecasting results. However, this method lacks an overall match between demand forecasting and inventory management, resulting in low inventory management efficiency and high operating costs. Summary of the invention

[0004] In view of the above-mentioned deficiencies in the prior art, the present invention provides an inventory management method for spare mechanical parts of heavy-duty vehicles, which solves the problem of relatively low overall matching between demand forecasting and inventory management in the prior art.

[0005] In order to achieve the above-mentioned invention object, the technical solution adopted by the present invention is: an inventory management method for spare mechanical parts of heavy-duty vehicles, comprising the following steps:

[0006] S1. Obtain historical demand data for spare mechanical parts, use a multi-criteria classification method to identify the importance of spare mechanical parts, and determine the important spare mechanical parts among the spare mechanical parts;

[0007] S2. Preprocess the historical demand data to obtain important parts historical demand data of important spare mechanical parts;

[0008] S3, construct a spare mechanical parts demand prediction model, and use the historical demand data of important parts to train the spare mechanical parts demand prediction model to obtain a trained spare mechanical parts demand prediction model, and output the important parts predicted demand data through the trained spare mechanical parts demand prediction model;

[0009] S4. Based on the historical demand data and predicted demand data of important components, an inventory simulation is performed to fit the best demand distribution. Based on the inventory simulation results, the inventory of important spare mechanical parts is dynamically adjusted.

[0010] The beneficial effect of the above scheme is: the present invention takes the best matching inventory strategy as the starting point, and incorporates external data such as component economic data into the prediction model, thereby enhancing the model's sensitivity to external factors, thereby guiding the heavy-duty vehicle spare mechanical parts demand prediction modeling and data-driven inventory simulation under the best demand distribution fitting, etc., thereby realizing the organic integration of heavy-duty vehicle spare mechanical parts demand prediction and inventory management, and effectively guiding the inventory management practice of this type of spare mechanical parts.

[0011] Furthermore, step S1 specifically includes:

[0012] S11. Obtain historical demand data for spare mechanical parts, select a multi-dimensional classification criterion, and determine the classification dimensions of spare mechanical parts;

[0013] The classification dimensions include the unit price of spare mechanical parts, sales volume of spare mechanical parts, order lead time, maintenance period of spare mechanical parts, inventory inspection period of spare mechanical parts, importance of spare mechanical parts in equipment and degree of discontinuity of demand for spare mechanical parts;

[0014] S12. According to the classification dimensions, the historical demand data is normalized, the classification weight of each dimension in the classification dimension is determined, and the comprehensive score of each spare mechanical component is determined by applying the hierarchical analysis method;

[0015] S13. Cluster the comprehensive scores of the spare mechanical parts using a clustering method based on an average algorithm, and determine the important spare mechanical parts among the spare mechanical parts based on the classification results.

[0016] The beneficial effects of the above further solutions are: it can significantly improve the comprehensiveness and consistency of data, enhance the accuracy of demand forecasting, optimize inventory management, reduce operating costs, and support more scientific decision-making.

[0017] Furthermore, step S2 specifically includes:

[0018] S21, obtaining a preset cycle length;

[0019] S22. Integrate historical demand data according to a preset cycle length;

[0020] S23. Determine the historical demand data of important spare mechanical parts from the integration results.

[0021] The beneficial effect of the above further scheme is: by integrating the historical demand data according to the preset cycle length and determining the historical demand data of important spare mechanical parts therefrom, the standardization and consistency of the data can be significantly improved, the accuracy of demand forecasting can be enhanced, inventory management can be optimized, operating costs can be reduced, and more scientific decision-making can be supported.

[0022] Furthermore, in step S3, the historical demand data of important components and the predicted demand data of important components both include K periods, and the predicted demand data of important components in the kth period is obtained by predicting the historical demand data of important components in the kth period using the spare mechanical component demand prediction model.

[0023] The beneficial effect of the above further scheme is that by comparing and analyzing the historical demand data and predicted demand data of important components, the accuracy of demand forecasting can be significantly improved, inventory management can be optimized, operating costs can be reduced, and service levels can be improved.

[0024] Furthermore, the inventory simulation method includes negative binomial distribution, gamma distribution, Poisson distribution and normal distribution. Step S4 specifically includes:

[0025] S41, determining the demand variability corresponding to the historical demand data of the important components according to the historical demand data of the important components, the predicted demand data of the important components and the smoothing parameter according to a preset formula;

[0026] S42, performing inventory simulation for the best demand distribution fitting for demand variability, and obtaining simulation results of negative binomial distribution, gamma distribution, Poisson distribution and normal distribution respectively;

[0027] S43. According to the simulation results of negative binomial distribution, gamma distribution, Poisson distribution and normal distribution, the inventory of important spare mechanical parts is adjusted according to the predicted demand.

[0028] Furthermore, in step S41, the preset formula is:

[0029]

[0030] in, represents the demand variability of the important spare mechanical component s at the kth period of the node in period t, α represents the smoothing parameter, represents the demand variability of spare mechanical component s at the kth period of node t-1, represents the historical demand data of important spare mechanical parts s in the kth period of the node in the t-1 period, Represents the important component forecast demand data of the important spare mechanical component s in the kth period at the node t-1 time period.

[0031] Furthermore, in step S42, an inventory simulation is performed for the best demand distribution fitting of demand variability, and the formula used is:

[0032]

[0033] in, represents the simulation results of negative binomial distribution, represents the simulation results of the gamma distribution, represents the simulation results of Poisson distribution, represents the simulation results of normal distribution, represents the negative binomial distribution function of demand over k periods, represents the gamma distribution function of demand within k periods, represents the Poisson distribution function of demand within k periods, represents the safety stock coefficient under the standard normal distribution, with both the mean and unit variance being zero. Level represents the customer service level, which refers to the probability of no out-of-stock during the replenishment cycle. P 1,NBD represents the probability of the next order from the negative binomial distribution, P 2,NBD represents the probability of two orders under the negative binomial distribution, P 1,Gamma represents the concentration of ordering behavior under the gamma distribution, P 2,Gamma represents the volatility of ordering behavior under the gamma distribution, represents the forecast demand data of important spare mechanical parts s at the kth period of the node in time period t, It represents the demand variability of spare mechanical component s in the kth cycle of node in period t.

[0034] The beneficial effect of the above further solution is that through the application of the above formula, more refined and efficient inventory management can be achieved in actual operation

[0035] Furthermore, the method further comprises:

[0036] The mean absolute scaled error is used to evaluate the forecast model for spare mechanical parts demand.

[0037] The beneficial effect of the above further scheme is that using mean absolute scaled error as an evaluation metric not only provides an intuitive and reliable performance measure, but also helps optimize models and business processes in practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 The present invention is a flowchart of an inventory management method for spare mechanical parts of heavy-duty vehicles. DETAILED DESCRIPTION

[0039] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0040] like Figure 1 As shown, a method for inventory management of spare mechanical parts for heavy-duty vehicles comprises the following steps:

[0041] S1. Obtain historical demand data for spare mechanical parts, use a multi-criteria classification method to identify the importance of spare mechanical parts, and determine the important spare mechanical parts among the spare mechanical parts.

[0042] In this embodiment, step S1 specifically includes:

[0043] S11. Obtain historical demand data for spare mechanical parts, select a multi-dimensional classification criterion, and determine the classification dimensions of spare mechanical parts;

[0044] The classification dimensions include the unit price of spare mechanical parts, sales volume of spare mechanical parts, order lead time, maintenance period of spare mechanical parts, inventory inspection period of spare mechanical parts, importance of spare mechanical parts in equipment and degree of discontinuity of demand for spare mechanical parts;

[0045] S12. According to the classification dimensions, the historical demand data is normalized, the classification weight of each dimension in the classification dimension is determined, and the comprehensive score of each spare mechanical component is determined by applying the hierarchical analysis method;

[0046] S13. Cluster the comprehensive scores of the spare mechanical parts using a clustering method based on an average algorithm, and determine the important spare mechanical parts among the spare mechanical parts based on the classification results.

[0047] For example, for a large number of spare mechanical parts s∈S to be classified, select multi-dimensional classification criteria i∈I, and the classification dimensions may include spare mechanical parts unit price, spare mechanical parts sales volume, order lead time, spare mechanical parts maintenance period, spare mechanical parts inventory inspection period, spare mechanical parts importance in equipment, spare mechanical parts demand discontinuity, etc. is (i∈I,s∈S) is normalized to obtain Determine the classification weight w of each classification criterion i (i∈I), the comprehensive score P of each spare mechanical component is obtained by applying the hierarchical analysis method s (s∈S). Then, a clustering method based on an average algorithm is used to cluster the comprehensive scores of spare mechanical parts, and the clustering level is determined (the clustering level can include three categories: important spare mechanical parts, conventional spare mechanical parts, and unimportant spare mechanical parts). Through clustering, the importance of spare mechanical parts can be quickly identified, and the important spare mechanical parts among the spare mechanical parts can be determined.

[0048] S2. Preprocess the historical demand data to obtain important component historical demand data of important spare mechanical components.

[0049] In this embodiment, step S2 specifically includes:

[0050] S21, obtaining a preset cycle length;

[0051] S22. Integrate historical demand data according to a preset cycle length;

[0052] S23. Determine the historical demand data of important spare mechanical parts from the integration results.

[0053] For example, the arrival time of orders for spare mechanical parts for heavy-duty vehicles is irregular, so the preset cycle length (week, month, quarter, etc.) can be determined based on the order lead time and inspection period specifically agreed upon by the enterprise, and the historical demand data of various spare mechanical parts can be integrated according to the preset cycle length.

[0054] For example, the sum of the order lead time L and the inventory review period R can be regarded as a parameter k (k = 1, 2, ..., K). It can be the demand for spare mechanical parts s in the tth period (depending on the specific situation, t can be defined as week, month, quarter, etc.). It can be the demand for spare mechanical component s within k cycles starting from t.

[0055] Optionally, this embodiment selects important spare mechanical components from all spare mechanical components for specific analysis. This embodiment can also be used to analyze non-important spare mechanical components, or to analyze all spare mechanical components, without specific limitation here.

[0056] S3. Construct a spare mechanical parts demand prediction model, and use the historical demand data of important parts to train the spare mechanical parts demand prediction model to obtain a trained spare mechanical parts demand prediction model, and output the important parts predicted demand data through the trained spare mechanical parts demand prediction model.

[0057] In this embodiment, the historical demand data of important components and the predicted demand data of important components both include K periods, and the predicted demand data of important components in the kth period is obtained by predicting the historical demand data of important components in the kth period using the spare mechanical component demand prediction model.

[0058] Exemplarily, the spare mechanical parts demand forecasting model can adaptively generate model input features according to the characteristics of demand time series data, and includes rolling training, verification and testing processes. During the entire rolling training process, the size of the training sample of the spare mechanical parts demand forecasting model will increase as the training progresses. The verification process aims to use a global optimization algorithm to adaptively adjust the main hyperparameters of the model used for testing. The testing process is mainly used to evaluate the effectiveness of the spare mechanical parts demand forecasting model.

[0059] For example, the forecasting process of the spare mechanical parts demand forecasting model is a collaborative forecasting of important spare mechanical parts, which aims to use time series demand data Predict demand data for k periods starting from T+1 The prediction process is dynamically rolled with the update cycle. In addition, in the prediction process, different k values ​​can be set to obtain the K-point prediction value of each important spare mechanical component in each cycle, that is,

[0060] S4. Based on the historical demand data and predicted demand data of important components, an inventory simulation is performed to fit the best demand distribution. Based on the inventory simulation results, the inventory of important spare mechanical parts is dynamically adjusted.

[0061] In this embodiment, the inventory simulation method includes negative binomial distribution, gamma distribution, Poisson distribution and normal distribution. Step S4 specifically includes:

[0062] S41, determining the demand variability corresponding to the historical demand data of the important components according to the historical demand data of the important components, the predicted demand data of the important components and the smoothing parameter according to a preset formula;

[0063] S42, performing inventory simulation for the best demand distribution fitting for demand variability, and obtaining simulation results of negative binomial distribution, gamma distribution, Poisson distribution and normal distribution respectively;

[0064] S43. According to the simulation results of negative binomial distribution, gamma distribution, Poisson distribution and normal distribution, the inventory of important spare mechanical parts is adjusted according to the predicted demand.

[0065] In this embodiment, in step S41, the preset formula is:

[0066]

[0067] in, represents the demand variability of the important spare mechanical component s at the kth period of the node in period t, α represents the smoothing parameter, represents the demand variability of spare mechanical component s at the kth period of node t-1, represents the historical demand data of important spare mechanical parts s in the kth period of the node in the t-1 period, Represents the important component forecast demand data of the important spare mechanical component s in the kth period at the node t-1 time period.

[0068] In this embodiment, in step S42, the inventory simulation for best demand distribution fitting of demand variability is performed, and the formula used is:

[0069]

[0070] in, represents the simulation results of negative binomial distribution, represents the simulation results of the gamma distribution, represents the simulation results of Poisson distribution, represents the simulation results of normal distribution, represents the negative binomial distribution function of demand over k periods, represents the gamma distribution function of demand within k periods, represents the Poisson distribution function of demand within k periods, represents the safety stock coefficient under the standard normal distribution, with both the mean and unit variance being zero. Level represents the customer service level, which refers to the probability of no out-of-stock during the replenishment cycle. P 1,NBD represents the probability of the next order from the negative binomial distribution, P 2,NBD represents the probability of two orders under the negative binomial distribution, P 1,Gamma represents the concentration of ordering behavior under the gamma distribution, P 2,Gamma represents the volatility of ordering behavior under the gamma distribution, represents the forecast demand data of important spare mechanical parts s at the kth period of the node in time period t, It represents the demand variability of spare mechanical component s in the kth cycle of node in period t.

[0071] In this embodiment, when the demand for spare mechanical parts s follows a normal distribution, To calculate its order point quantity, An order point quantity that meets a specified service level (CSL) is determined based on the predicted demand mean and demand variance, which is suitable for spare mechanical parts with relatively symmetrical demand distribution.

[0072] When the demand for spare mechanical parts has a high variance, the negative binomial distribution (NBD) It is more adaptable to the discreteness and uncertainty of demand, and is suitable for spare mechanical parts with large demand fluctuations or "long tail" demand.

[0073] When the demand for spare mechanical parts has a large deviation but the demand mean is relatively stable, To calculate the order point quantity, it is suitable for spare mechanical parts with positive skew and asymmetric distribution of demand. and They are respectively applicable to the calculation of order point quantities for spare mechanical parts with different demand distribution characteristics.

[0074] Optionally, the method further comprises: evaluating the spare mechanical parts demand prediction model using mean absolute scaled error.

[0075] For example, the formula for evaluating the spare mechanical parts demand prediction model using the mean absolute scaled error may be:

[0076]

[0077] Among them, α represents the smoothing parameter, represents the historical demand data of important spare mechanical parts s in the kth (k=L+R) cycle of the node in the t-1 period, represents the forecast demand data of important spare mechanical parts s in the kth period of the node in the t-1 period, represents the average absolute scaled error of the important spare mechanical component s in the kth cycle of the node in the t-1 period, It shows the average absolute scaled error of the important spare mechanical component s in the kth cycle at the node in period t.

[0078] In this embodiment, the improved Syntetos-Boylan approximation method, the basic support vector machine method with parameter optimization, and the artificial neural network method with parameter optimization can also be used as comparative methods. In addition, the trade-off curve between holding cost and undelivered cost and the trade-off curve between holding cost and service rate level can be used as quantitative analysis indicators to evaluate inventory management performance.

[0079] In a specific embodiment, compared with the improved Syntetos-Boylan approximation method, the basic support vector machine method with parameter optimization, and the artificial neural network with parameter optimization, the prediction error under the adaptive univariate support vector machine method decreased by 5.9%, 3.5%, and 11.3%. When the target customer service rate level is 95%, the inventory management cost (holding cost + out-of-stock cost) corresponding to the inventory management method of heavy-duty vehicle spare mechanical parts proposed by the present invention is reduced by 12.4% (compared with the inventory simulation based on the improved Syntetos-Boylan approximation method), 8.9% (compared with the inventory simulation based on the basic support vector machine method with parameter optimization) and 17.2% (compared with the inventory simulation based on the artificial neural network method with parameter optimization). In addition, the present invention proposes an organic coupling of spare mechanical parts demand forecasting and spare mechanical parts inventory management, which has a modeling guidance effect on spare mechanical parts inventory management using data-driven and machine learning.

[0080] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the invention.

Claims

1. A method for inventory management of spare mechanical parts for heavy vehicles, characterized in that: The method comprises: S1. Obtain historical demand data of spare mechanical parts, use a multi-criteria classification method to identify the importance of the spare mechanical parts, and determine the important spare mechanical parts among the spare mechanical parts; S2. Preprocessing the historical demand data to obtain important component historical demand data of the important spare mechanical components; S3, constructing a spare mechanical parts demand prediction model, and using the historical demand data of important parts to train the spare mechanical parts demand prediction model to obtain a trained spare mechanical parts demand prediction model, and outputting important parts predicted demand data through the trained spare mechanical parts demand prediction model; S4. Perform inventory simulation for optimal demand distribution fitting based on the historical demand data of the important components and the predicted demand data of the important components, and dynamically adjust the inventory of the important spare mechanical components based on the inventory simulation results.

2. The method according to claim 1, characterized in that The step S1 specifically includes: S11, obtaining historical demand data of spare mechanical parts, selecting a multi-dimensional classification criterion, and determining the classification dimensions of the spare mechanical parts; The classification dimensions include the unit price of spare mechanical parts, sales volume of spare mechanical parts, order lead time, maintenance period of spare mechanical parts, inventory inspection period of spare mechanical parts, importance of spare mechanical parts in equipment and degree of discontinuity of demand for spare mechanical parts; S12, normalizing the historical demand data according to the classification dimensions, determining the classification weight of each dimension in the classification dimensions, and applying the hierarchical analysis method to determine the comprehensive score of each of the spare mechanical components; S13. Clustering the comprehensive scores of the spare mechanical components using a clustering method based on an average algorithm, and determining important spare mechanical components among the spare mechanical components according to the classification results.

3. The method according to claim 1, characterized in that The step S2 specifically includes: S21, obtaining a preset cycle length; S22, integrating the historical demand data according to a preset cycle length; S23. Determine the important component historical demand data of the important spare mechanical components from the integration results.

4. The method according to claim 3, characterized in that In step S3, the important component historical demand data and the important component predicted demand data both include K periods, and the important component predicted demand data of the kth period is obtained by predicting the important component historical demand data of the kth period using the spare mechanical component demand prediction model.

5. The method according to claim 1, characterized in that The inventory simulation method includes negative binomial distribution, gamma distribution, Poisson distribution and normal distribution, and the step S4 specifically includes: S41, determining the demand variability corresponding to the historical demand data of the important components according to the historical demand data of the important components, the predicted demand data of the important components and a smoothing parameter according to a preset formula; S42, performing inventory simulation for optimal demand distribution fitting on the demand variability, and obtaining simulation results of the negative binomial distribution, the gamma distribution, the Poisson distribution, and the normal distribution respectively; S43. According to the simulation results of the negative binomial distribution, the gamma distribution, the Poisson distribution and the normal distribution, the inventory of the important spare mechanical parts is adjusted according to the predicted demand.

6. The method according to claim 5, characterized in that In step S41, the preset formula is: in, represents the demand variability of the important spare mechanical component s at the kth period of the node in period t, α represents the smoothing parameter, represents the demand variability of spare mechanical component s at the kth period of node t-1, represents the historical demand data of important spare mechanical parts s in the kth period of the node in the t-1 period, Represents the important component forecast demand data of the important spare mechanical component s in the kth period at the node t-1 time period.

7. The method according to claim 5, characterized in that In step S42, the inventory simulation for performing the best demand distribution fitting for the demand variability uses the formula: in, represents the simulation results of negative binomial distribution, represents the simulation results of the gamma distribution, represents the simulation results of Poisson distribution, represents the simulation results of normal distribution, represents the negative binomial distribution function of demand over k periods, represents the gamma distribution function of demand within k periods, represents the Poisson distribution function of demand within k periods, represents the safety stock coefficient under the standard normal distribution, with both the mean and unit variance being zero. Level represents the customer service level, which refers to the probability of no out-of-stock during the replenishment cycle. P 1,NBD represents the probability of the next order from the negative binomial distribution, P 2,NBD represents the probability of two orders under the negative binomial distribution, P 1,Gamma represents the concentration of ordering behavior under the gamma distribution, P 2,Gamma represents the volatility of ordering behavior under the gamma distribution, represents the forecast demand data of important spare mechanical parts s at the kth period of the node in time period t, It represents the demand variability of spare mechanical component s in the kth cycle of node in period t.

8. The method according to claim 1, characterized in that The method further comprises: The mean absolute scaled error is used to evaluate the spare mechanical parts demand prediction model.

Images