A dynamic demand forecasting method based on Markov chains and periodicity prediction

Abstract

This invention provides a dynamic demand forecasting method based on Markov chains and periodic forecasting. It acquires and preprocesses supply chain demand data, extracts periodic fluctuation characteristics of the demand data through time-domain change analysis to determine the forecasting period, and predicts future trends based on the Markov transition matrix and the forecasting period. Demand fluctuation events are identified based on material acquisition records during supply chain operation, and their correlation characteristics are extracted to adjust future trends, resulting in a future demand forecast for the supply chain. The future demand forecast is analyzed to obtain the expected demand at future time points in the supply chain, and combined with real-time inventory levels, notification messages are sent to related parties in the supply chain. This method enables global and dynamic demand forecasting of the entire supply chain based on actual conditions during supply chain operation, providing timely and accurate early warning notifications to related parties.

Description

Technical Field

[0001] This invention relates to the field of supply chain, and more particularly to a dynamic demand forecasting method based on Markov chains and periodic forecasting. Background Technology

[0002] Supply chain and warehousing scenarios require consideration of market demand to ensure stable operation, and market demand information has a significant impact on supply chain production and warehousing planning. Given the significant volatility of market demand, traditional demand forecasting models struggle to adapt to the complex and ever-changing nature of the market. Their fixed forecasting periods fail to accommodate the actual demand intervals for different products, and they lack sufficient quantitative consideration of key factors such as product price and lifecycle, and their forecasting strategies lack differentiated processing. Therefore, existing demand forecasting methods cannot provide comprehensive and dynamic demand forecasting for the entire supply chain based on actual conditions during supply chain operations, and cannot provide timely and accurate early warnings to related ends of the supply chain, thus reducing the overall stability and sustainability of the supply chain. Summary of the Invention

[0003] The purpose of this invention is to provide a dynamic demand forecasting method based on Markov chains and periodic forecasting. This method acquires and preprocesses supply chain demand data, extracts the periodic fluctuation characteristics of the demand data through time-domain change analysis to determine the forecasting period, and predicts future trends based on the Markov transition matrix and the forecasting period. It identifies demand fluctuation events based on material acquisition records during supply chain operation and extracts the correlation characteristics of these events to adjust future trends, thus obtaining the future demand forecast results for the supply chain. Analyzing these future demand forecast results yields the expected demand volume of the supply chain at future time points. Combined with the real-time inventory levels of the supply chain, notification messages are sent to related parties in the supply chain. This allows for global and dynamic demand forecasting of the entire supply chain based on actual conditions during operation, providing timely and accurate early warning notifications to related parties and improving the operational stability and sustainability of the entire supply chain.

[0004] This invention is achieved through the following technical solution:

[0005] A dynamic demand forecasting method based on Markov chains and periodicity prediction includes:

[0006] Acquire and preprocess supply chain demand data, perform time-domain variation analysis on the demand data, extract the periodic fluctuation characteristics of the demand data, and thereby determine the forecast period;

[0007] Based on the Markov transition matrix and the prediction period, predict future trends;

[0008] Based on the material acquisition records during the operation of the supply chain, demand fluctuation events during the operation period are identified, and correlation features of the demand fluctuation events are extracted. Based on the correlation features, the future trend is adjusted to obtain the future demand forecast results of the supply chain.

[0009] The expected demand of the supply chain at future time points is obtained by analyzing the future demand forecast results; based on the expected demand and the real-time inventory of the supply chain, a notification message is sent to the relevant ends of the supply chain.

[0010] Optionally, demand data from the supply chain may be acquired and preprocessed, including:

[0011] Monitor the task execution progress of each business node under the supply chain, determine whether the business node has completed the task, and obtain the logistics logs corresponding to the completed tasks of all business nodes; identify and extract the data types of the logistics logs to obtain the demand data of the supply chain.

[0012] The required data is preprocessed by cleaning up erroneous data and merging duplicate data.

[0013] Optionally, time-domain variation analysis is performed on the demand data to extract the periodic fluctuation characteristics of the demand data, thereby determining the prediction period, including:

[0014] Based on the generation time of all sub-data under the demand data, perform time-domain change analysis on the demand data to obtain all demand events corresponding to the generation time interval of the demand data.

[0015] Obtain the time interval between every two adjacent demand events in the time domain, average all time intervals, and determine the prediction period.

[0016] Optionally, based on the Markov transition matrix and the prediction period, predicting future trends includes:

[0017] Based on the Markov transition matrix, the demand data is segmented within the prediction period to predict future trends.

[0018] Optionally, based on the Markov transition matrix, the demand data is segmented within the forecast period to predict future trends, including:

[0019] Based on the Markov transition matrix, the changing trend of the demand data is segmented within the prediction period to obtain intervals that are in an upward trend, intervals that are in a stable trend, and intervals that are in a downward trend.

[0020] Based on the sample size and mean slope of each of the intervals in an upward trend, a stable trend, and a downward trend, the expected slope value of the demand data is obtained; based on the expected slope value, the future trend is predicted.

[0021] Optionally, based on the material acquisition records during the operation of the supply chain, demand fluctuation events during the operation are determined, including:

[0022] Material demand change data is extracted from the records of materials during the operation of the supply chain. The material demand change data is divided into several sub-data of material demand change by a preset time window. Any two material demand change sub-data that are adjacent in time are compared to identify demand fluctuation events during the operation of the supply chain.

[0023] Optionally, the correlation features of the demand fluctuation events are extracted, including:

[0024] Based on the occurrence time of the demand fluctuation event, the temporal relationship characteristics between the occurrence time and the time of the holiday are obtained;

[0025] Obtain the material price data corresponding to the occurrence time, and determine the price relationship characteristics based on the material price data;

[0026] The time-domain relationship features and the price relationship features are used as the correlation features of the demand fluctuation event.

[0027] Optionally, based on the correlation characteristics, the future change trend is adjusted to obtain the future demand forecast result of the supply chain, including:

[0028] Based on the time-domain relationship characteristics and the price relationship characteristics, a first adjustment factor and a second adjustment factor are generated respectively;

[0029] The trend of change is adjusted based on the first adjustment factor and the second adjustment factor to obtain the future demand forecast result of the supply chain.

[0030] Optionally, the analysis of the future demand forecast results yields the expected demand of the supply chain at future time points, including:

[0031] Based on the production and market launch cycle of the materials related to the supply chain, determine the future time nodes corresponding to the materials during the operation of the supply chain;

[0032] Based on the future time points, the future demand forecast results are extracted to obtain the expected demand of the supply chain at the future time points.

[0033] Optionally, based on the expected demand and the real-time inventory levels of the supply chain, a notification message is sent to the relevant parties in the supply chain, including:

[0034] Obtain the material quantity difference between the expected demand and the real-time inventory of the supply chain, and send a notification message to the relevant end of the supply chain based on the material quantity difference; wherein, the notification message includes the additional material demand and the demand supply time limit.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] This application provides a dynamic demand forecasting method based on Markov chains and periodic forecasting. It acquires and preprocesses supply chain demand data, extracts periodic fluctuation characteristics of the demand data through time-domain change analysis to determine the forecasting period, and predicts future trends based on the Markov transition matrix and the forecasting period. Based on material acquisition records during supply chain operation, it identifies demand fluctuation events and extracts their correlation characteristics to adjust future trends, resulting in a future demand forecast for the supply chain. Analyzing the future demand forecast results, it obtains the expected demand at future time points in the supply chain and, combined with real-time inventory levels, sends notification messages to related parties in the supply chain. This method enables global and dynamic demand forecasting of the entire supply chain based on actual conditions during operation, providing timely and accurate early warnings to related parties and improving the overall stability and sustainability of the supply chain. Attached Figure Description

[0037] 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. Wherein:

[0038] Figure 1 This is a flowchart illustrating a dynamic demand forecasting method based on Markov chains and periodic prediction provided by the present invention.

[0039] Figure 2 It is the process of acquiring and preprocessing demand data.

[0040] Figure 3 It is the process of predicting future trends. Detailed Implementation

[0041] To make the above-mentioned objectives, features, and advantages of this application more apparent and understandable, the specific embodiments of this application will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of this application. Furthermore, it should be noted that, for ease of description, only the parts relevant to this application are shown in the accompanying drawings, not the entire structure. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this application.

[0042] The terms “comprising” and “having”, and any variations thereof, used in this application are intended to cover non-exclusive inclusion. For example, a process, method, product, or apparatus that includes a series of steps or units is not limited to the steps or units listed, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus.

[0043] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0044] Please see Figure 1 As shown, an embodiment of this application provides a dynamic demand forecasting method based on Markov chains and periodicity prediction. This dynamic demand forecasting method based on Markov chains and periodicity prediction includes:

[0045] Acquire and preprocess supply chain demand data, perform time-domain variation analysis on demand data, extract periodic fluctuation characteristics of demand data, and thus determine the forecast period;

[0046] Predict future trends based on Markov transition matrix and prediction period;

[0047] Based on the material acquisition records during the operation of the supply chain, demand fluctuation events during the operation period are identified, and the correlation characteristics of these demand fluctuation events are extracted. Based on the correlation characteristics, future trends are adjusted to obtain the future demand forecast results of the supply chain.

[0048] Analyze future demand forecasts to obtain the expected demand for the supply chain at future time points; based on the expected demand and the real-time inventory of the supply chain, send notification messages to the relevant ends of the supply chain.

[0049] The beneficial effects of the above embodiments are as follows: This dynamic demand forecasting method based on Markov chains and periodic forecasting acquires and preprocesses supply chain demand data, extracts the periodic fluctuation characteristics of demand data through time-domain change analysis to determine the forecasting period, predicts future trends based on Markov transition matrices and the forecasting period, identifies demand fluctuation events based on material acquisition records during supply chain operation, extracts the correlation characteristics of these events, adjusts future trends accordingly, and obtains future demand forecasting results for the supply chain. Analyzing these future demand forecasting results yields the expected demand volume of the supply chain at future time points, and combining this with real-time inventory levels, sends notification messages to related parties within the supply chain. This enables global and dynamic demand forecasting of the entire supply chain based on actual conditions during supply chain operation, providing timely and accurate early warning notifications to related parties, and improving the operational stability and sustainability of the entire supply chain.

[0050] In another embodiment, acquiring and preprocessing supply chain demand data includes:

[0051] Monitor the task execution progress of each business node in the supply chain, determine whether the business node has completed the task, and obtain the logistics logs corresponding to the completed tasks of all business nodes; identify and extract the data types of the logistics logs to obtain the supply chain demand data;

[0052] Perform error cleaning and duplicate data merging preprocessing on the required data.

[0053] Please see Figure 2 The supply chain comprises several business nodes, each responsible for different types of operations such as procurement, transportation, and warehousing. Each business node generates a logistics log during the execution of its tasks. To ensure the logistics logs completely record the operational process of each business node, each node is independently monitored to obtain its real-time task execution progress, thus determining whether the current task has been completed. If the current task has been completed, the corresponding logistics log for that task is retrieved; otherwise, the log is not retrieved. The logistics logs are then processed for data type identification and extraction to obtain supply chain demand data, which may include, but is not limited to, the quantity of materials required by the supply chain. Considering the potential for data errors in the logistics logs generated by the business nodes, error cleaning and duplicate data merging preprocessing are performed to improve the accuracy of the demand data. This reduces the error rate of the demand data and provides a reliable data source for subsequent trend prediction.

[0054] In another embodiment, time-domain variation analysis is performed on the demand data to extract the periodic fluctuation characteristics of the demand data, thereby determining the forecast period, including:

[0055] Based on the generation time of all sub-data under the demand data, perform time-domain change analysis on the demand data to obtain all demand events corresponding to the generation time interval of the demand data;

[0056] Obtain the time interval between every two adjacent demand events in the time domain, average all time intervals, and determine the prediction period.

[0057] Demand data encompasses demand changes throughout the entire supply chain operation. Given the long time span of supply chain operations, the trends in demand data vary across different time periods, leading to different demand events occurring in different timeframes. For example, minimum demand may occur in one period, while maximum demand may occur in the next. These different types of demand events reflect the dynamic changes in demand during supply chain operations. To accurately determine the cyclical fluctuations in demand during supply chain operations, a time-domain variation analysis is first performed on the demand data based on the generation time of all its sub-data. This yields all demand events corresponding to the generation time interval of the demand data, including events corresponding to minimum and maximum demand. Next, the time interval between every two adjacent demand events in the time domain is obtained, and all time intervals are averaged to determine the forecast period. This establishes a reliable and effective periodic time interval for predicting future demand trends.

[0058] In another embodiment, based on the Markov transition matrix and the prediction period, future trends are predicted, including:

[0059] Based on the Markov transition matrix, demand data is segmented within the forecast period to predict future trends.

[0060] In practical applications, Markov transition matrices are used, with the aforementioned forecast period as a benchmark, to segment and forecast demand data, predicting future trends in the supply chain and thus the future demand for materials. The use of Markov transition matrices for future trend forecasting is a conventional technique in this field and will not be described in detail here.

[0061] In another embodiment, based on the Markov transition matrix, the demand data is segmented within the forecast period to predict future trends, including:

[0062] Based on the Markov transition matrix, the trend of demand data is segmented within the prediction period to obtain the intervals of upward trend, stable trend, and downward trend.

[0063] Based on the sample size and mean slope of each of the intervals in an upward trend, a stable trend, and a downward trend, the expected slope value of the demand data is obtained; based on the expected slope value, the future trend is predicted.

[0064] Please see Figure 3 By using the Markov transition matrix to segment the demand data within the aforementioned forecast period, the intervals showing an upward trend are obtained. inc Interval in a stable trend stable Interval in a downward trend dec Then, obtain the sample size N for each of the following intervals: the interval in an upward trend, the interval in a stable trend, and the interval in a downward trend. inc N stable N dec and the mean slope μ inc μ stable μ dec ER is used to calculate the expected slope of the demand data, where ER = *μ inc + *μ stable + *μ dec , where n is the total number of samples. Based on the expected slope value, the future trend is predicted, ensuring that the future trend within the predicted period exhibits a basically linear change.

[0065] In another embodiment, determining demand fluctuation events during the operation period based on material acquisition records during supply chain operation includes:

[0066] Extract material demand change data from material acquisition records during supply chain operation, divide the material demand change data into preset time windows to obtain several material demand change sub-data, and compare any two material demand change sub-data that are adjacent in time to identify demand fluctuation events during supply chain operation.

[0067] During the actual operation of the supply chain, the demand for materials fluctuates across different time periods due to factors such as holiday effects and material prices. To accurately determine the impact of holiday effects and material prices on demand fluctuations during supply chain operation, we first extract data on changes in material demand from records during the supply chain operation. This data is then divided into several sub-data points corresponding to pre-defined time windows. Next, we compare any two adjacent sub-data points. If the rate of change in material demand between two adjacent time windows exceeds a pre-defined threshold, a demand fluctuation event is identified; otherwise, no demand fluctuation event is identified. This provides a reliable basis for subsequently determining the impact of holiday effects and material prices on supply chain demand changes.

[0068] In another embodiment, the correlation features of demand fluctuation events are extracted, including:

[0069] Based on the occurrence time of demand fluctuation events, the temporal relationship characteristics between the occurrence time and the time of holidays are obtained;

[0070] Obtain the material price data corresponding to the time of occurrence, and determine the price relationship characteristics based on the material price data;

[0071] The temporal relationship characteristics and price relationship characteristics are used as the correlation characteristics of demand fluctuation events.

[0072] To determine the impact of holiday effects and material prices on supply chain demand fluctuations, this study obtains the temporal relationship characteristics between the occurrence time of demand fluctuation events and the timing of holidays, and determines price relationship characteristics based on material price data. These characteristics serve as the correlation features of demand fluctuation events, providing a reliable basis for subsequently determining adjustment factors related to holiday effects and material prices.

[0073] In another embodiment, based on correlation characteristics, future trends are adjusted to obtain future demand forecasts for the supply chain, including:

[0074] Based on the time-domain relationship characteristics and price relationship characteristics, the first adjustment factor and the second adjustment factor are generated respectively;

[0075] Based on the first and second adjustment factors, the trend of change is adjusted to obtain the future demand forecast results for the supply chain.

[0076] To determine the adjustment factor for the impact of holiday effects on supply chain demand changes, the first adjustment factor related to holiday effects can be determined using the following formula. , ,in This represents the average demand value before the holiday. This represents the average demand during holidays. The preset threshold coefficient is used; then, according to the following formula, the adjustment and correction of the expected slope value r of the predicted future trend due to the holiday effect is determined, and the adjusted slope value is obtained. ,Right now To adjust for the impact of holiday effects and material price-related changes on supply chain demand, a second adjustment factor related to material prices was determined. Then, based on the formula below, determine the second adjustment and correction of the expected slope value r of the predicted future trend of material price changes, and obtain the adjusted slope value. ,Right now .

[0077] In another embodiment, the analysis of future demand forecasts yields the expected demand of the supply chain at future points in time, including:

[0078] Based on the production and launch cycle of materials in the supply chain, determine the future time nodes corresponding to the materials during the operation of the supply chain;

[0079] Based on future time points, the results of future demand forecasts are extracted to obtain the expected demand of the supply chain at those future time points.

[0080] Using the above method, based on the production and market launch cycle of materials in the supply chain, the time interval corresponding to the entire process of IoT-related processes from the start to the completion of production is determined. The start and completion dates within this time interval are then used as future time points for the materials during the supply chain operation. Based on these future time points, the expected demand for each corresponding time point is extracted from future demand forecasts, providing a reliable basis for subsequent notification messages.

[0081] In another embodiment, a notification message is sent to relevant parties in the supply chain based on the expected demand and the real-time inventory levels, including:

[0082] Obtain the material quantity difference between the expected demand and the real-time inventory of the supply chain. Based on the material quantity difference, send a notification message to the relevant parties in the supply chain. The notification message includes the additional material demand and the demand supply time limit.

[0083] Through the above process, the material quantity difference between the expected demand and the real-time inventory of the supply chain is obtained. The material quantity difference is compared with the preset difference range. If the material quantity difference is not within the preset difference range, a notification message is sent to the relevant end of the supply chain. This enables timely information on additional material demand and supply time limits to be sent to the relevant end of the supply chain, ensuring the continuous and stable operation of the supply chain.

[0084] In summary, this dynamic demand forecasting method based on Markov chains and periodic forecasting acquires and preprocesses supply chain demand data. It extracts the periodic fluctuation characteristics of the demand data through time-domain change analysis to determine the forecasting period. Based on the Markov transition matrix and the forecasting period, it predicts future trends. According to material acquisition records during supply chain operation, it identifies demand fluctuation events and extracts their correlation characteristics to adjust future trends, resulting in a future demand forecast for the supply chain. Analyzing the future demand forecast results, it obtains the expected demand at future time points in the supply chain. Combined with real-time inventory levels, it sends notification messages to related parties in the supply chain. This method enables global and dynamic demand forecasting of the entire supply chain based on actual conditions during operation, providing timely and accurate early warnings to related parties and improving the overall stability and sustainability of the supply chain.

[0085] The above is only one specific embodiment of the present invention, and any improvements made based on the concept of the present invention shall be considered within the scope of protection of the present invention.

Claims

1. A dynamic demand forecasting method based on Markov chains and periodicity prediction, characterized in that, include: Acquire and preprocess supply chain demand data, perform time-domain variation analysis on the demand data, extract the periodic fluctuation characteristics of the demand data, and thereby determine the forecast period; Based on the Markov transition matrix, the changing trend of the demand data is segmented within the prediction period to obtain intervals that are in an upward trend, intervals that are in a stable trend, and intervals that are in a downward trend. Based on the sample size and mean slope of each of the above intervals, the expected slope value of the demand data is obtained. Based on the expected slope value, the future changing trend is predicted. Based on the material acquisition records during the operation of the supply chain, demand fluctuation events during the operation period are determined, and the temporal and price relationship features of the demand fluctuation events are extracted as the correlation features of the demand fluctuation events. Based on the time-domain relationship characteristics and the price relationship characteristics, a first adjustment factor and a second adjustment factor are generated respectively; based on the first adjustment factor and the second adjustment factor, the change trend is adjusted to obtain the future demand forecast result of the supply chain; The expected demand of the supply chain at future time points is obtained by analyzing the future demand forecast results; based on the expected demand and the real-time inventory of the supply chain, a notification message is sent to the relevant ends of the supply chain.

2. The dynamic demand forecasting method based on Markov chains and periodic forecasting as described in claim 1, characterized in that: Acquiring and preprocessing supply chain demand data, including: Monitor the task execution progress of each business node under the supply chain, determine whether the business node has completed the task, and obtain the logistics logs corresponding to the completed tasks of all business nodes; identify and extract the data types of the logistics logs to obtain the demand data of the supply chain. The required data is preprocessed by cleaning up erroneous data and merging duplicate data.

3. The dynamic demand forecasting method based on Markov chains and periodic forecasting as described in claim 2, characterized in that: Perform time-domain variation analysis on the demand data to extract the periodic fluctuation characteristics of the demand data, thereby determining the prediction period, including: Based on the generation time of all sub-data under the demand data, perform time-domain change analysis on the demand data to obtain all demand events corresponding to the generation time interval of the demand data. Obtain the time interval between every two adjacent demand events in the time domain, average all time intervals, and determine the prediction period.

4. The dynamic demand forecasting method based on Markov chains and periodic forecasting as described in claim 1, characterized in that: Based on the material acquisition records during the operation of the supply chain, determine the demand fluctuation events during the operation, including: Material demand change data is extracted from the records of materials during the operation of the supply chain. The material demand change data is divided into several sub-data of material demand change by a preset time window. Any two material demand change sub-data that are adjacent in time are compared to identify demand fluctuation events during the operation of the supply chain.

5. The dynamic demand forecasting method based on Markov chains and periodic prediction as described in claim 1, characterized in that: The step of extracting the temporal and price relationship features of the demand fluctuation events as correlation features of the demand fluctuation events includes: Based on the occurrence time of the demand fluctuation event, the temporal relationship characteristics between the occurrence time and the time of the holiday are obtained; Obtain the material price data corresponding to the occurrence time, and determine the price relationship characteristics based on the material price data; The time-domain relationship features and the price relationship features are used as the correlation features of the demand fluctuation event.

6. The dynamic demand forecasting method based on Markov chains and periodic forecasting as described in claim 1, characterized in that: Analysis of the future demand forecast results yields the expected demand of the supply chain at future time points, including: Based on the production and market launch cycle of the materials related to the supply chain, determine the future time nodes corresponding to the materials during the operation of the supply chain; Based on the future time points, the future demand forecast results are extracted to obtain the expected demand of the supply chain at the future time points.

7. The dynamic demand forecasting method based on Markov chains and periodic forecasting as described in claim 6, characterized in that: Based on the expected demand and the real-time inventory levels of the supply chain, a notification message is sent to the relevant parties in the supply chain, including: Obtain the material quantity difference between the expected demand and the real-time inventory of the supply chain, and send a notification message to the relevant end of the supply chain based on the material quantity difference; wherein, the notification message includes the additional material demand and the demand supply time limit.

Images