A method and system for collaborative management of sheet stock based on home furnishing orders

Abstract

The present invention discloses a method and system for collaborative management of sheet stock based on home orders, which relates to the technical field of inventory management. The method includes collecting order data and inventory data for data preprocessing; calculating the moving average order and autoregressive order based on the order data through the autocorrelation value and correction coefficient, determining the differencing order through time series data testing, and constructing an ARIMA model for order data prediction. The method of the present invention determines the moving average order q and autoregressive order p by combining the ACF and PACF graphs, and automatically optimizes the parameter combination through the AIC / BIC criterion, achieving the accuracy of model parameter selection, multi-criterion collaboration, and avoiding the limitations of a single evaluation method. By combining the ARIMA model with the random forest model, the method comprehensively captures the linear and nonlinear characteristics of order data, enhancing the accuracy and robustness of prediction.

Description

Technical Field

[0001] The present invention relates to the technical field of inventory management, and particularly to a method and system for collaborative management of sheet material inventory based on home orders. Background Art

[0002] In recent years, with the rapid development of the e-commerce and home customization industries, the demand for refined inventory management by enterprises has gradually increased. As an important raw material in home manufacturing, the inventory management of sheet materials directly affects the supply chain efficiency and enterprise operation costs. Traditional inventory management methods mainly rely on empirical judgment or simple inventory quantity calculation based on static inventory models, and it is difficult to effectively cope with the non-linear fluctuations and sudden increases in order demands. Although the application of time series analysis technology has played a certain role in inventory forecasting.

[0003] Although the existing technologies have made certain progress in inventory management, there are still multiple deficiencies. First, traditional methods have high requirements for the stationarity of time series, resulting in difficult accurate prediction of non-stationary series with complex fluctuations; second, a single prediction model cannot capture both short-term fluctuations and long-term trends in order demands simultaneously, and it is difficult to cope with complex market environments; third, there is a lack of a multi-model collaborative optimization mechanism, resulting in a significant deviation between inventory forecasting and actual demands. In addition, existing inventory management methods usually use a globally fixed time range for forecasting, ignoring the dynamic impact of historical patterns in different time periods on future orders, and it is difficult to achieve a balance between timeliness and accuracy. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a method for collaborative management of sheet material inventory based on home orders to solve the problems that traditional methods have high requirements for the stationarity of time series, resulting in difficult accurate prediction of non-stationary series with complex fluctuations; second, a single prediction model cannot capture both short-term fluctuations and long-term trends in order demands simultaneously, and it is difficult to cope with complex market environments; third, there is a lack of a multi-model collaborative optimization mechanism, resulting in a significant deviation between inventory forecasting and actual demands. In addition, existing inventory management methods usually use a globally fixed time range for forecasting, ignoring the dynamic impact of historical patterns in different time periods on future orders, and it is difficult to achieve a balance between timeliness and accuracy.

[0006] To solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a method for collaborative management of sheet material inventory based on home orders, which includes:

[0008] Collect order data and inventory data for data preprocessing;

[0009] Calculate the moving average order and autoregressive order based on the order data through the autocorrelation value and the correction coefficient, determine the differencing order through time series data testing, and construct an ARIMA model for order data prediction;

[0010] Construct a random forest model for non-linear order data prediction, and perform dynamic time alignment paths based on the order data prediction sequence and historical data. Segment the time interval based on the time range of the minimum alignment path, and screen the time range according to the correlation of the segmented time intervals;

[0011] Based on the screened time range, conduct inventory collaboration, determine the optimal order quantity and inventory value, perform data storage and backup, and generate security logs.

[0012] As a preferred solution of the method for collaborative management of sheet stock based on home orders of the present invention, wherein: the preprocessing of order data and inventory data refers to obtaining order data, order time, and customer information from the enterprise management system based on home orders, and obtaining the inventory data of the warehouse, including inventory quantity and inventory cost data;

[0013] Preprocess the collected data.

[0014] As a preferred solution of the method for collaborative management of sheet stock based on home orders of the present invention, wherein: calculating the moving average order and autoregressive order based on the order data through the autocorrelation value and the correction coefficient, calculating the autocorrelation value based on the order data, expressed as:

[0015]

[0016] where ACF(k) represents the autocorrelation value at the lag order k, N represents the total number of data points, represents the average order quantity, k represents the lag order, D t represents the order quantity at time t in the time series, D t-k represents the order quantity at time t-k;

[0017] Draw an ACF graph based on the autocorrelation value, where the horizontal axis represents the lag order k and the vertical axis represents the autocorrelation value. Determine the value as the moving average order q according to the truncation position of the lag order on the horizontal axis of the ACF graph;

[0018] Define and calculate the recursive coefficient based on the autocorrelation value, expressed as:

[0019]

[0020] where a i,k represents the correction coefficient of the influence of the i-th order lag on the current lag, p kDenote the k-th autocorrelation value as p k-j Denote the (k - j)-th autocorrelation value, i as the iteration variable of the current correction coefficient, and j as the intermediate variable for recursive calculation;

[0021] Based on the correction coefficient and the autocorrelation value, calculate the partial autocorrelation function, expressed as:

[0022]

[0023] where PACF(k) represents the partial autocorrelation value of the sequence at the lag order k, and p k-i Denote the (k - i)-th autocorrelation value;

[0024] Draw the PACF graph according to the partial autocorrelation values, where the horizontal axis is the lag order k and the vertical axis is the partial autocorrelation value. Determine the autoregressive order p according to the truncation position of the lag order on the horizontal axis of the PACF graph.

[0025] As a preferred solution of the method for collaborative management of sheet stock based on home orders according to the present invention, wherein: determining the difference order through time series data test, constructing an ARIMA model for order data prediction, and performing difference operation on the data based on the order data, expressed as:

[0026] ΔD t = D t - D t-1 ;

[0027] where ΔD t represents the difference sequence value at time t, and D t and D t-1 respectively represent the original order quantities at time t and time t - 1;

[0028] According to ΔD t form time series data, and introduce ADF for time series data test according to the lag order z of the ADF test;

[0029] Input the time series data formed according to ΔD t and determine the unit root coefficient value φ by the least squares method, and calculate the t-statistic according to the unit root coefficient value φ, expressed as:

[0030]

[0031] where t' represents the t-statistic and SE(φ) represents the standard error of the unit root coefficient;

[0032] Use the t-distribution table to determine the corresponding p-value. Based on historical data, determine the significance level value. If the p-value is less than or equal to the significance level value, it is determined that the time series data is stable, and the difference order d value is determined to be 1. If the p-value is greater than the significance level value, it is determined that the time series data is unstable, and a second-order difference is performed to calculate the difference series value Δ 2 D t ;

[0033] And construct a second-order time series data based on the second-order difference series value. Reintroduce ADF to perform a second-order time series data test. If the calculated second-order difference p-value is less than or equal to the significance level value, it is determined that the time series data is stable, and the difference order d value is determined to be 2;

[0034] Construct an ARIMA model according to the difference order d, moving average order q, and autoregressive order p, and determine the model parameters through maximum likelihood estimation MLE;

[0035] Use the Akaike information criterion AIC and Bayesian information criterion BIC to evaluate the model. Calculate the AIC value and BIC value for different combinations of difference order d, moving average order q, and autoregressive order p, and select the parameter combination that minimizes AIC and BIC as the difference order d, moving average order q, and autoregressive order p of the final ARIMA model;

[0036] Input the time series data of the order data to predict the order data sequence at future times.

[0037] As a preferred solution of the method for collaborative management of sheet stock based on home orders according to the present invention, wherein: construct a random forest model to predict non-linear order data, and perform dynamic time alignment paths based on the order data prediction sequence and historical data. Divide the time range based on the time range of the minimum alignment path, and screen the time range according to the correlation of the divided time intervals;

[0038] Construct a random forest model based on the random forest method and pre-train it with historical data;

[0039] Combine the predicted order data, historical demand order data, time features including weekends and holidays, and regional features including customer locations to form the input features of the random forest model, and output a non-linear order data prediction sequence including time features and regional features through the random forest model;

[0040] Based on historical time series data and the current order data prediction sequence, use dynamic time warping DTW to calculate the total distance of the minimum alignment path, expressed as:

[0041]

[0042] Where DDTW represents the total minimum distance of the dynamic alignment path, DTW(f a , f b ) represents the dynamic alignment distance between the f a -th time point and the f b -th time point, and G represents the set of optimal alignment paths of dynamic time warping;

[0043] Record the time indices of all data points in the alignment path G, and extract the time range of the path. Determine the corresponding time range according to the value of the total minimum distance D DTW ;

[0044] Based on the corresponding time range, use the sum of the mean and standard deviation of the historical segmentation interval as the segmentation number for time segmentation to form multiple sub-time intervals, and calculate the DTW distance value of the local alignment path for the historical time series values and the predicted sequence values of the order data in each sub-time interval;

[0045] Output the local DTW distance sequence, calculate the mean and standard deviation of the local DTW distance, and perform Pearson correlation coefficient calculation. Based on the sum of the historical mean and standard deviation as the correlation threshold, if the Pearson correlation coefficient of the calculated local DTW distance is greater than or equal to the correlation threshold, it is considered that the time range corresponding to the Pearson correlation coefficient of the local DTW distance has a good correlation with the historical pattern. Use the sum of the mean and standard deviation of the Pearson correlation coefficients with good correlations as the reference data, and select the time range corresponding to the good correlation coefficient closest to the reference data as the screening time range.

[0046] As a preferred solution of the method for collaborative management of sheet stock based on home orders according to the present invention, wherein: inventory collaboration is performed based on the screening time range to determine the optimal order quantity and inventory value, and the predicted value of the order data within the range is determined based on the screening time range;

[0047] Optimize the inventory collaboration between suppliers and home manufacturers in the supply chain, and set the objective function of maximizing profit and minimizing inventory cost, expressed as:

[0048]

[0049] where I t and I t-1 represent the inventory quantities at time t and t - 1 respectively, Q t represents the order quantity at time t, represents the predicted order quantity at time t, C represents the inventory cost, t end represents the end value of the time range of the predicted value sequence data, t strepresents the starting value of the time range of the predicted value sequence data, h represents the inventory holding cost, c represents the ordering cost, and w represents the shortage cost;

[0050] Based on the constraint that the manufacturer's order quantity cannot exceed the supply quantity, and based on the minimum inventory cost, the optimal order quantity and inventory value are determined.

[0051] As a preferred solution of the method for collaborative management of sheet inventory based on home orders according to the present invention, wherein: the data storage and backup are performed, a security log is generated, the data is stored based on the optimal order quantity and inventory value and the corresponding time range, and the data is transmitted through the RESTful API transmission protocol, and the sensitive fields are encrypted using RSA, and the identity authentication is performed using the API key;

[0052] The stored data backup is stored in the cloud platform, and at the same time, the backup storage is performed through the remote storage terminal, and the result of each identity authentication is recorded through the log generation tool to generate a security log.

[0053] In a second aspect, the present invention provides a system for a method for collaborative management of sheet inventory based on home orders, including,

[0054] A data collection and preprocessing module that collects data from the enterprise management system for preprocessing;

[0055] A time series analysis module that constructs an ARIMA model and optimizes the parameters to predict future order data;

[0056] A dynamic time alignment module that uses the dynamic time warping (DTW) method to calculate the minimum alignment path between historical data and the prediction sequence, and performs time interval segmentation and correlation analysis;

[0057] An inventory collaboration optimization module that determines the optimal inventory level and order quantity through the objective function, and optimizes the supply chain collaboration using the supply quantity constraint condition;

[0058] A data storage and security module that performs data transmission by encrypting and protecting sensitive fields, and automatically generates a security log for identity authentication.

[0059] In a third aspect, the present invention provides a computer device, including a memory and a processor, wherein: when the computer program stored in the memory is executed by the processor, any step of the method for collaborative management of sheet inventory based on home orders as described in the first aspect of the present invention is implemented.

[0060] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, wherein: when the computer program is executed by the processor, any step of the method for collaborative management of sheet inventory based on home orders as described in the first aspect of the present invention is implemented.

[0061] The beneficial effects of the present invention are as follows: By combining the ACF and PACF graphs to determine the moving average order q and the autoregressive order p, and automatically optimizing the parameter combination through the AIC / BIC criterion, the accuracy of model parameter selection is achieved. The multi-criteria collaboration avoids the limitations of a single evaluation method. By using the ARIMA model in combination with the random forest model, a comprehensive capture of the linear and non-linear characteristics of order data is realized, enhancing the accuracy and robustness of prediction. By dynamically combining the random forest prediction results and the DTW matching degree analysis, an efficient association between the historical pattern and the current prediction pattern can be achieved. Through the multi-level time range screening mechanism, the deviation of a single evaluation criterion is effectively avoided. Through the comprehensive analysis of local and global matching, the reliability and stability of time range selection are improved. By constructing the objective function and designing the constraint conditions based on the screened time range and the predicted value sequence data, the scientific nature and data-driven nature of inventory collaborative optimization are ensured, avoiding the subjectivity of traditional empirical decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0063] Figure 1 It is a schematic flow chart of the method for collaborative management of sheet stock based on home furnishing orders in Embodiment 1.

[0064] Figure 2 It is a schematic structural diagram of the system for collaborative management of sheet stock based on home furnishing orders in Embodiment 1. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0065] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the following detailed description of the specific embodiments of the present invention will be made in conjunction with the drawings in the specification.

[0066] Many specific details are set forth in the following description in order to provide a thorough understanding of the present invention. However, the present invention may be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0067] Secondly, the "one embodiment" or "embodiment" referred to herein means a specific feature, structure, or characteristic that may be included in at least one implementation manner of the present invention. The "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that excludes other embodiments from each other.

[0068] Example 1, referring to Figure 1 and Figure 2 , is the first embodiment of the present invention. This embodiment provides a method for collaborative management of sheet stock based on home furnishing orders, including the following steps:

[0069] S1, collect order data and inventory data for data preprocessing;

[0070] Preferably, collecting order data and inventory data for data preprocessing means obtaining order data, order time, and customer information based on the enterprise management system for home furnishing orders, and obtaining the inventory data of the warehouse, including inventory quantity and inventory cost data;

[0071] Perform data preprocessing on the collected data.

[0072] The multi-dimensional data collected through the enterprise management system can understand the order situation and inventory status in real time, reduce management problems caused by information lag, accurately grasp the order distribution and inventory status, support warehouse resource scheduling and logistics plan optimization, improve the consistency of inventory data through data preprocessing, make management decisions based on real information, and ensure that the input data is accurate and efficient.

[0073] S2, calculate the moving average order and autoregressive order based on the order data through the autocorrelation value and correction coefficient, determine the differencing order through time series data testing, and construct an ARIMA model for order data prediction;

[0074] Preferably, calculate the moving average order and autoregressive order based on the order data through the autocorrelation value and correction coefficient. Calculate the autocorrelation value based on the order data, which is expressed as:

[0075]

[0076] where ACF(k) represents the autocorrelation value at the lag order k, N represents the total number of data points, represents the average order quantity, k represents the lag order, D t represents the order quantity at time t in the time series, D t-k represents the order quantity at time t - k;

[0077] Draw an ACF graph based on the autocorrelation value, where the horizontal axis represents the lag order k and the vertical axis represents the autocorrelation value. Determine the value as the moving average order q according to the truncation position of the lag order on the horizontal axis of the ACF graph.

[0078] Define and calculate the recursive coefficient based on the autocorrelation value, expressed as:

[0079]

[0080] where a i,k represents the correction coefficient of the influence of the i-th order lag on the current lag, and p k represents the k-th autocorrelation value, and p k-j represents the (k - j)-th autocorrelation value. i represents the iteration variable of the current correction coefficient, starting from 1 and increasing step by step. j represents the intermediate variable for recursive calculation, and a j represents the correction coefficient of the j-th lag term for the current order i under the first i - 1 order lags;

[0081] Calculate the partial autocorrelation function based on the correction coefficient and the autocorrelation value, expressed as:

[0082]

[0083] where PACF(k) represents the partial autocorrelation value of the sequence at the lag order k, and p k-i represents the (k - i)-th autocorrelation value;

[0084] Draw the PACF graph based on the partial autocorrelation values. The horizontal axis is the lag order k, and the vertical axis is the partial autocorrelation value. Determine the autoregressive order p according to the truncation position of the lag order on the horizontal axis of the PACF graph.

[0085] By calculating the autocorrelation values and plotting the ACF graph, the short-term dependencies in the order data can be identified, ensuring that the selection of the moving average order q is based on the true characteristics of the data rather than subjective assumptions, thereby improving the accuracy of the model. The truncation position in the ACF graph intuitively reveals the randomness and periodicity characteristics in the time series, reducing the fitting of unnecessary parameters and enhancing the model's efficiency. The recursive calculation of the correction coefficient is based on the gradual adjustment of the autocorrelation values, which can accurately capture the true contributions of the lag terms in the time series, avoiding the mutual interference between lag terms and making the model more capable of interpreting historical data. The partial autocorrelation function, by correcting the autocorrelation values, clearly shows the independent influence of the lag terms on the current value in the time series. The plotting of the PACF graph further provides an intuitive basis for the truncation of the lag order, facilitating the selection of the appropriate autoregressive order p. By using the ACF graph and PACF graph simultaneously, the dynamic co-determination of the moving average order q and the autoregressive order p can be achieved, ensuring that the model can balance the ability to capture short-term fluctuations and long-term trends. The method avoids the subjectivity in the selection of the lag order in traditional models, making the determination process of the order data-driven and improving the generalization ability of the model. The overall method combines precise mathematical formulas with graphical analysis, providing an easy-to-understand and accurate basis for model parameter selection, laying a solid foundation for subsequent time series analysis and prediction.

[0086] Furthermore, the differencing order is determined through time series data tests, and an ARIMA model is constructed for order data prediction. Based on the order data, the data is differenced as follows:

[0087] ΔD t =D t -D t-1 ;

[0088] where ΔD t represents the differenced series value at time t, and D t and D t-1 represent the original order quantities at time t and time t - 1 respectively;

[0089] Based on ΔD t to form time series data, the ADF (Augmented Dickey-Fuller) is introduced for time series data tests, expressed as:

[0090]

[0091] where φ represents the unit root coefficient, ∈ t represents the random error term, β i represents the linear regression coefficient of the current series ΔD t with respect to the lag differenced term ΔD t-i , and z represents the lag order of the ADF test;

[0092] Based on ΔD t The time series data will be input and the unit root coefficient value φ will be determined by the least squares method, and the t-statistic will be calculated based on the unit root coefficient value φ, expressed as:

[0093]

[0094] where t' represents the t-statistic and SE(φ) represents the standard error of the unit root coefficient;

[0095] Use the t-distribution table and determine the corresponding p-value. Determine the significance level value based on historical data. If the p-value is less than or equal to the significance level value, it is judged that the time series data is stable, and the difference order d value is determined to be 1. If the p-value is greater than the significance level value, it is judged that the time series data is unstable, and the second-order difference is performed to calculate the difference series value Δ 2 D t , expressed as:

[0096] Δ 2 D t = ΔD t - ΔD t-1 ;

[0097] where Δ 2 D t represents the second-order difference series value at time t;

[0098] And construct the second-order time series data based on the second-order difference series value, reintroduce ADF to test the second-order time series data. If the calculated second-order difference p-value is less than or equal to the significance level value, it is judged that the time series data is stable, and the difference order d value is determined to be 2;

[0099] Construct an ARIMA model according to the difference order d, moving average order q, and autoregressive order p, and determine the model parameters through maximum likelihood estimation MLE;

[0100] Use the Akaike information criterion AIC and Bayesian information criterion BIC to evaluate the model and calculate the AIC value and BIC value for different combinations of difference order d, moving average order q, and autoregressive order p, and select the parameter combination that minimizes AIC and BIC as the difference order d, moving average order q, and autoregressive order p of the final ARIMA model;

[0101] Input the time series data of the order data and predict the order data sequence at future times.

[0102] By performing differential operations to eliminate non-stationarity, differential operations on the original order data can effectively remove trends and seasonal fluctuations, making the time series data more stationary, providing a data stability basis for the construction of subsequent models. Using the ADF test can quantitatively judge the stationarity of the time series. Through the t-statistic and p-value of the unit root test, the determination of the differential order becomes more scientific and data-driven, rather than relying on empirical judgment. Dynamically judge the demand for the differential order d according to the p-value to ensure that the time series reaches stationarity, while avoiding unnecessary over-differencing and retaining the valid information in the data. Combine the moving average order q, autoregressive order p, and differential order d to dynamically adjust the model structure to ensure that the ARIMA model can capture the linear relationship in the time series and effectively handle random disturbances. Optimize the model parameters through the MLE method to ensure the best fitting performance of the model, reduce the interference of the error term on the prediction result, and improve the accuracy of order data prediction. Through the calculation of AIC and BIC, the advantages and disadvantages of different parameter combinations can be scientifically evaluated, avoiding the situation of overly complex models or insufficient fitting, providing a reliable basis for the final parameter selection of the ARIMA model. Through the structure of the ARIMA model, the prediction result can intuitively reflect the linear trend, seasonal fluctuation, and random disturbance of the order data, improving the interpretability of the prediction result;

[0103] By combining the ACF and PACF graphs to determine the moving average order q and autoregressive order p, and automatically optimizing the parameter combination through the AIC / BIC criterion, the accuracy of model parameter selection is achieved. Multiple criteria cooperate to avoid the limitations of a single evaluation method. Combining the parameter selection mechanism of ACF and PACF, the linear prediction ability of ARIMA, and the stationarity control of the differential method, the effect of collaborative improvement of multiple algorithms in the prediction framework is achieved.

[0104] S3. Construct a random forest model for non-linear order data prediction, and perform a dynamic time alignment path based on the order data prediction sequence and historical data. Segment the time interval based on the time range of the minimum alignment path, and screen the time range according to the correlation of the segmented time intervals;

[0105] Preferably, construct a random forest model for non-linear order data prediction, and perform a dynamic time alignment path based on the order data prediction sequence and historical data. Segment the time interval based on the time range of the minimum alignment path, and screen the time range according to the correlation of the segmented time intervals;

[0106] Construct a random forest model based on the random forest method and pre-train it with historical data;

[0107] The predicted order data, historical demand order data, time features including weekends and holidays, and regional features including customer locations are jointly used as the input features of the random forest model. Through the random forest model, a non-linear order data prediction sequence containing time features and regional features is output;

[0108] Based on the historical time series data and the current order data prediction sequence, the dynamic time warping (DTW) is used to calculate the total distance of the minimized alignment path, expressed as:

[0109]

[0110] where D DTW represents the total minimum distance of the dynamic alignment path, and DTW(f a , f b ) represents the dynamic alignment distance between the f a -th time point and the f b -th time point. G represents the set of optimal alignment paths of the dynamic time warping;

[0111] Record the time indices of all data points in the alignment path G, and extract the time range of the path. Determine the corresponding time range according to the value of the total minimum distance D DTW ;

[0112] Based on the corresponding time range, use the sum of the mean and standard deviation of the historical segmentation interval as the number of segments for time segmentation to form multiple sub-time intervals, and calculate the DTW distance values of the local alignment paths for the historical time series values and the order data prediction sequence values of each sub-time interval;

[0113] Output the local DTW distance sequence, calculate the mean and standard deviation of the local DTW distance, and perform Pearson correlation coefficient calculation. Based on the sum of the historical mean and standard deviation as the correlation threshold, if the Pearson correlation coefficient of the calculated local DTW distance is greater than or equal to the correlation threshold, it is considered that the time range corresponding to the Pearson correlation coefficient of the local DTW distance has a good correlation with the historical pattern. Use the sum of the mean and standard deviation of the Pearson correlation coefficients with good correlations as the reference data, and select the time range corresponding to the good correlation coefficient closest to the reference data as the screening time range.

[0114] By inputting multi-dimensional features (such as time characteristics, regional characteristics, and historical order data) into a random forest model, it is possible to capture the complex non-linear characteristics in the order data, making the prediction sequence closer to the actual order change pattern, enhancing the accuracy of the prediction. By introducing dynamic time warping (DTW), the alignment path between the order prediction sequence and the historical sequence can dynamically adapt to time misalignments and non-linear relationships, overcoming the limitations of traditional sequence analysis methods in time point alignment. By providing a globally optimized matching metric through the total distance of the minimum alignment path, the time range selection is based on the historical pattern, making it more data-driven and practically valuable. By segmenting the time interval based on the mean and standard deviation of the historical segmentation interval, it can better reflect the fluctuation range of historical data, providing a scientific basis for subsequent local analysis. By calculating the mean and standard deviation of the local DTW distance sequence, the local alignment degree of different sub-time intervals can be quantified, thereby refining the correlation analysis of the time range and making the results more representative;

[0115] By calculating the Pearson correlation coefficient and combining the historical mean and standard deviation as thresholds, it ensures that the time range screening process is highly sensitive to the historical pattern, while avoiding the interference of outliers on the results. By using the sum of the mean and standard deviation as the reference data when screening the time range and selecting the time range corresponding to the closest good correlation coefficient, it balances the comprehensiveness of local correlation and the adaptability to the global pattern. By dynamically combining the random forest prediction results and DTW matching degree analysis throughout the method, it can achieve an efficient association between the historical pattern and the current prediction pattern, ensuring that the selected time range has practical guiding significance in the prediction. By the multi-level time range screening mechanism, it effectively avoids the deviation of a single evaluation criterion. Through the comprehensive analysis of local and global matching, it improves the reliability and stability of the time range selection. The selected time range can accurately reflect the best correlation between the prediction sequence and the historical pattern, providing data support for subsequent order management and inventory planning, and enhancing the accuracy and scientific nature of decision-making;

[0116] By combining the ARIMA model and the random forest model, it achieves a comprehensive capture of the linear and non-linear characteristics of the order data, enhancing the accuracy and robustness of the prediction. The ARIMA model provides long-term trend prediction, providing a basic guidance for overall inventory management and order planning. The random forest model combined with DTW provides a precise analysis of short-term fluctuations and local correlations, providing refined support for dynamic inventory adjustment and regional order management. The entire prediction and time range selection process realizes dynamic optimization from global to local, not only improving the prediction accuracy, but also enhancing the flexibility and reliability of decision-making, providing strong support for the enterprise's inventory management and order planning.

[0117] S4. Based on the filtered time range, conduct inventory collaboration, determine the optimal order quantity and inventory value, perform data storage and backup, and generate security logs;

[0118] Preferably, based on the filtered time range, conduct inventory collaboration, determine the optimal order quantity and inventory value, and determine the predicted value of order data within this range based on the filtered time range;

[0119] Optimize the inventory collaboration between suppliers and home furnishing manufacturers in the supply chain, and set up an objective function for maximizing profit and minimizing inventory cost, expressed as:

[0120]

[0121] where I t and I t-1 represent the inventory quantities at time t and t - 1 respectively, Q t represents the order quantity at time t, represents the predicted order quantity at time t, C represents the inventory cost, t end represents the end value of the time range of the predicted value sequence data, t st represents the start value of the time range of the predicted value sequence data, h represents the inventory holding cost, c represents the order cost, and w represents the shortage cost;

[0122] Based on the constraint that the order quantity of the manufacturer cannot exceed the supply quantity, and determine the optimal order quantity and inventory value based on the minimum inventory cost.

[0123] Improve the prediction accuracy through time range filtering: Based on the predicted value of order data within the filtered time range, it can more accurately reflect future order demands, enabling inventory collaboration optimization to be based on reliable prediction data, reducing inventory risks caused by prediction errors. By dynamically adjusting the inventory quantity and order quantity within the time range, the inventory configuration in the supply chain can quickly respond to changes in order demands, avoiding problems of excessive or insufficient inventory. With the collaboration between suppliers and home furnishing manufacturers as the goal, clarify the constraint condition that the order quantity cannot exceed the supply quantity, optimize the upstream and downstream resource allocation, improve the coordination of the entire supply chain. Through the objective function, while ensuring the response ability of the supply chain, significantly reduce the inventory holding cost, order cost, and shortage cost, thereby achieving the global optimization of the supply chain operation cost. Through the construction of the objective function and the design of the constraint conditions based on the filtered time range and predicted value sequence data, ensure the scientific nature and data-driven nature of inventory collaboration optimization, avoid the subjectivity of traditional experience-based decision-making. Under the goal of maximizing profit, by reasonably planning the order quantity and inventory level, reduce the sales losses caused by shortages and the capital occupation caused by overstock, and improve the overall profitability of the supply chain.

[0124] Furthermore, data storage and backup are performed to generate security logs. Data is stored based on the optimal order quantity, inventory value, and corresponding time range, and data transmission is carried out through the RESTful API transmission protocol. Sensitive fields are encrypted using RSA, and API keys are used for identity authentication.

[0125] The stored data backup is stored in the cloud platform, and at the same time, backup storage is performed through off-site storage terminals. Each identity authentication result is recorded through a log generation tool to generate security logs.

[0126] Sensitive fields (such as order data and inventory information) are encrypted using RSA to ensure protection against unauthorized access or leakage risks during data transmission and storage. Each data transmission is authenticated using API keys, effectively restricting access by illegal users, enhancing the security and reliability of the system. Through the combination of RESTful API and the encryption protocol RSA encryption, the security of the data transmission channel is ensured, while maintaining an efficient transmission speed, suitable for the synchronization requirements of real-time data. Through security logs, access behaviors are recorded in real-time, enabling the system to quickly detect potential abnormal access or attack behaviors, providing timely early warning support for defense and response.

[0127] This embodiment also provides a system for a sheet inventory collaborative management method based on home orders, including

[0128] A data collection and preprocessing module that collects data from the enterprise management system for preprocessing;

[0129] A time series analysis module that constructs an ARIMA model and optimizes parameters to predict future order data;

[0130] A dynamic time alignment module that uses the dynamic time warping (DTW) method to calculate the minimum alignment path between historical data and the prediction sequence, and performs time interval segmentation and correlation analysis;

[0131] An inventory collaboration optimization module that determines the optimal inventory level and order quantity through an objective function, and optimizes supply chain collaboration using supply quantity constraint conditions;

[0132] A data storage and security module that protects sensitive fields through encryption for data transmission and automatically generates security logs for identity authentication.

[0133] This embodiment also provides a computer device applicable to the case of a sheet inventory collaborative management method based on home orders, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the sheet inventory collaborative management method based on home orders as proposed in the above embodiment.

[0134] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer 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 operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a carrier network, NFC (Near Field Communication), or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the housing of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0135] This embodiment also provides a storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the method for collaborative management of sheet stock based on home orders proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, abbreviated as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, abbreviated as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, abbreviated as EPROM), programmable read-only memory (Programmable Red-Only Memory, abbreviated as PROM), read-only memory (Read-Only Memory, abbreviated as ROM), magnetic memory, flash memory, a magnetic disk, or an optical disc.

[0136] In summary, the present invention determines the moving average order q and the autoregressive order p by combining the ACF and PACF diagrams, and automatically optimizes the parameter combination through the AIC / BIC criterion, achieving the accuracy of model parameter selection and multi-criterion collaboration, avoiding the limitations of a single evaluation method. By using the combination of the ARIMA model and the random forest model, it comprehensively captures the linear and non-linear characteristics of order data, enhancing the accuracy and robustness of prediction. By dynamically combining the random forest prediction results and the DTW matching degree analysis, it can efficiently associate the historical pattern with the current prediction pattern. Through the multi-level time range screening mechanism, it effectively avoids the deviation of a single evaluation criterion. Through the comprehensive analysis of local and global matching, it improves the reliability and stability of time range selection. By constructing the objective function and designing the constraint conditions based on the screened time range and predicted value sequence data, it ensures the scientific nature and data-driven nature of inventory collaborative optimization, avoiding the subjectivity of traditional empirical decision-making.

[0137] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A method for collaborative management of sheet stock based on home furnishing orders, characterized in that, Including: Collecting order data and inventory data for data preprocessing; Calculating the moving average order and autoregressive order based on the order data through autocorrelation values and correction coefficients, determining the differencing order through time series data tests, and constructing an ARIMA model for order data prediction; Constructing a random forest model for non-linear order data prediction, and performing dynamic time alignment paths based on the order data prediction sequence and historical data, segmenting the time interval based on the time range of the minimum alignment path, and screening the time range according to the correlation of the segmented time intervals; Performing inventory collaboration based on the screened time range, determining the optimal order quantity and inventory value, storing and backing up data, and generating security logs; Drawing an ACF graph based on the autocorrelation value, where the horizontal axis represents the lag order k and the vertical axis represents the autocorrelation value, and determining the value as the moving average order q according to the truncation position of the lag order on the horizontal axis of the ACF graph; Defining and calculating the recursion coefficient based on the autocorrelation value, expressed as: where a i,k The correction coefficient of the effect of the i-th order lag on the current lag, p k represents the kth autocorrelation value, p k-j represents the kjth autocorrelation value, i represents the iterative variable of the current correction coefficient, and j represents the intermediate variable of the recursive calculation; Calculating the partial autocorrelation function based on the correction coefficient and autocorrelation value, drawing a PACF graph according to the partial autocorrelation value, where the horizontal axis is the lag order k and the vertical axis is the partial autocorrelation value, and determining the value as the autoregressive order p according to the truncation position of the lag order on the horizontal axis of the PACF graph; According to ΔD t Composed of time series data, according to the lag order z of ADF test, ADF is introduced to perform time series data test; Based on ΔD t Input the time series data composed thereof and determine the unit root coefficient value φ by the least squares method, and calculate the t-statistic based on the unit root coefficient value φ; Using the Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) to evaluate the model, calculating the AIC value and BIC value for different combinations of differencing order d, moving average order q, and autoregressive order p, and selecting the parameter combination that minimizes AIC and BIC as the differencing order d, moving average order q, and autoregressive order p of the final ARIMA model; Inputting the time series data of the order data to predict the order data sequence at future times; Record the time index of all data points in the alignment path G and extract the time range of the path according to the total minimum distance D DTW The value of determines the corresponding time range; Performing time segmentation based on the sum of the mean and standard deviation of the historical segmentation intervals as the segmentation number according to the corresponding time range, forming multiple sub-time intervals, and calculating the DTW distance value of the local alignment path for the historical time series values and order data prediction sequence values of each sub-time interval; The inventory collaboration based on the screened time range determines the optimal order quantity and inventory value, and determines the order data prediction value within this range based on the screened time range; Optimizing the inventory collaboration between suppliers and home furnishing manufacturers in the supply chain, setting the objective function of maximizing profit and minimizing inventory cost, expressed as: Among them I t and I t-1 represent the inventory quantity at time t and t-1, Q t represents the order quantity at time t, represents the predicted order quantity at time t, C represents the inventory cost, t end Indicates the end value of the time range of the forecast value series data, t st represents the starting value of the time range of the forecast value series data, h represents the inventory holding cost, c represents the ordering cost, and w represents the out-of-stock cost; Based on the constraint that the order quantity of the manufacturer cannot exceed the supply quantity, and determining the optimal order quantity and inventory value based on the minimum inventory cost.

2. The method for collaborative management of sheet stock based on home furnishing orders according to claim 1, wherein: The collection of order data and inventory data for data preprocessing refers to obtaining order data, order time, and customer information based on the enterprise management system of home furnishing orders, and obtaining the inventory data of the warehouse, including inventory quantity and inventory cost data; Performing data preprocessing on the collected data.

3. The method for collaborative management of sheet stock based on home furnishing orders according to claim 2, wherein: Calculating the moving average order and autoregressive order based on the order data through autocorrelation values and correction coefficients, and calculating the autocorrelation value based on the order data, expressed as: where ACF(k) represents the autocorrelation value at lag order k, and N represents the total number of data points. represents the average order quantity, k represents the lag order, D t represents the order quantity at time t in the time series, D t-k represents the order quantity at time t - k; Calculating the partial autocorrelation function based on the correction coefficient and autocorrelation value, expressed as: Where PACF(k) represents the partial autocorrelation value of the sequence at lag order k, p k-i represents the ki-th autocorrelation value.

4. The method for collaborative management of sheet stock based on home furnishing orders according to claim 3, wherein: Determining the differencing order through time series data tests, constructing an ARIMA model for order data prediction, and performing differencing operations on the data based on the order data, expressed as: ΔD t = D t - D t-1 ; where ΔD t represents the difference sequence value at time t, D t and D t-1 represent the original order quantities at time t and time t respectively; Calculate the t-statistic based on the unit root coefficient value φ, expressed as: where t' represents the t-statistic and SE(φ) represents the standard error of the unit root coefficient; Use the t-distribution table and determine the corresponding p-value. Determine the significance level value based on historical data. If the p-value is less than or equal to the significance level value, it is determined that the time series data is stable, and the difference order d value is determined to be 1. If the p-value is greater than the significance level value, it is determined that the time series data is unstable, and a second-order difference is performed to calculate the difference sequence value Δ 2 D t ; Construct second-order time series data based on the second-order difference sequence values, reintroduce ADF for second-order time series data testing. If the calculated second-order difference p-value is less than or equal to the significance level value, it is determined that the time series data is stable, and the difference order d value is determined to be 2; Construct an ARIMA model based on the difference order d, the moving average order q, and the autoregressive order p, and determine the model parameters through maximum likelihood estimation MLE.

5. The method for collaborative management of sheet stock based on home furnishing orders according to claim 4, wherein: Construct a random forest model for non-linear order data prediction, perform dynamic time alignment path based on the order data prediction sequence and historical data, perform time interval segmentation based on the time range of the minimum alignment path, and screen the time range according to the correlation of the segmented time intervals; Construct a random forest model based on the random forest method and perform pre-training with historical data; The predicted order data, historical demand order data, time features including weekends and holidays, and regional features including customer locations together form the input features of the random forest model, and a non-linear order data prediction sequence including time features and regional features is output through the random forest model; Based on historical time series data and the current order data prediction sequence, use dynamic time warping DTW to calculate the total distance of the minimum alignment path, expressed as: Among which D DTW represents the total minimum distance of the dynamic alignment path, DTW(f a , f b ) represents the dynamic alignment distance between the f a -th time point and the f b -th time point, and G represents the set of optimal alignment paths of dynamic time warping; Output the local DTW distance sequence, calculate the mean and standard deviation of the local DTW distance, and perform Pearson correlation coefficient calculation. Based on the sum of the historical mean and standard deviation as the correlation threshold, if the Pearson correlation coefficient of the calculated local DTW distance is greater than or equal to the correlation threshold, it is considered that the time range corresponding to the Pearson correlation coefficient of the local DTW distance has a good correlation with the historical pattern. Use the sum of the mean and standard deviation of the Pearson correlation coefficients with good correlations as the reference data, and select the time range corresponding to the good correlation coefficient closest to the reference data as the screened time range.

6. The collaborative management method for panel inventory based on home furnishing orders according to claim 5, characterized in that: Perform data storage and backup, generate security logs, store data based on the optimal order quantity and inventory value and the corresponding time range, perform data transmission through the RESTful API transmission protocol, encrypt sensitive fields using RSA, and perform identity authentication using API keys; Back up the stored data to the cloud platform, and at the same time perform backup storage through a remote storage terminal, record the results of each identity authentication through a log generation tool, and generate security logs.

7. A system for the collaborative management method of sheet stock based on home orders, based on the collaborative management method of sheet stock based on home orders according to any one of claims 1 to 6, characterized in that: Including, A data collection and preprocessing module that collects and preprocesses data from the enterprise management system; A time series analysis module that constructs an ARIMA model and performs parameter optimization to predict future order data; A dynamic time alignment module that uses the dynamic time warping DTW method to calculate the minimum alignment path between historical data and the prediction sequence, and performs time interval segmentation and correlation analysis; An inventory collaborative optimization module that determines the optimal inventory level and order quantity through the objective function, and optimizes supply chain collaboration using supply quantity constraint conditions; The data storage and security module encrypts and protects sensitive fields for data transmission, performs identity authentication, and automatically generates security logs.

8. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that: When the processor executes the computer program, the steps of the collaborative management method of panel inventory based on home orders according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the collaborative management method of panel inventory based on home orders according to any one of claims 1 to 6 are implemented.

Images