Cross-platform data terminal intelligent warehouse management system and method

Abstract

The invention relates to the field of intelligent warehouse management, in particular to a cross-platform data terminal intelligent warehouse management system and method. The cross-platform data terminal intelligent warehouse management system comprises a commodity classification module, a long-term commodity prediction module and a short-term commodity prediction module. By analyzing the historical order data of the commodities and considering fluctuation characteristic values such as the variable coefficient and the change rate standard deviation, the system can automatically divide the commodities into long-term supply commodities and short-term supply commodities, so that the purchase strategy can be more flexible and adapt to market changes, and the purchase efficiency is improved. A more stable purchase plan can be adopted for long-term commodity supply, and market fluctuation can be quickly responded for short-term commodity supply, so that more efficient inventory turnover is realized, and warehouse management is more refined and personalized.

Description

Technical Field

[0001] The present invention relates to the field of intelligent warehouse management, and in particular to a cross-platform data terminal intelligent warehouse management system and method. Background Art

[0002] In the context of globalization, Chinese-made products are favored by overseas consumers for their high cost-effectiveness. With the rise of overseas shopping platforms, more and more foreign customers choose to buy Chinese goods through these platforms. However, overseas shopping platforms still face a series of challenges in warehouse management, especially in the process of sending goods from China to overseas. Since the goods need to cross national borders, the logistics time is long, which requires the warehouse management system to accurately predict demand in order to reduce inventory backlogs and out-of-stock risks. At the same time, the demand of overseas markets is changeable and is affected by various factors such as culture, seasons, and economy, which puts higher requirements on the flexibility and response speed of warehouse management systems. In addition, due to factors such as long-distance transportation and tariffs, inventory costs are high, so more precise inventory management is needed to reduce costs. On the other hand, due to the lack of in-depth understanding of overseas markets and data analysis, it is difficult for traditional warehouse management systems to accurately predict the order volume of overseas customers. Overseas shopping platforms often need to process data from different countries and platforms, and these data are scattered in different systems, making it difficult to achieve effective data integration and analysis.

[0003] In order to solve the above problems, the present invention proposes a cross-platform data terminal intelligent warehousing management method, which is specifically aimed at the overseas online shopping platform sent by China. This method can achieve accurate prediction of order quantity and dynamic adjustment of purchase quantity through intelligent classification of commodities; through this intelligent warehousing management method, the online shopping platform can better adapt to the rapid changes in the overseas market and realize refined management of inventory, so as to maintain a leading position in the fierce international market competition; at the same time, this method can also help the platform better integrate cross-platform data, improve the accuracy and efficiency of data analysis, and provide faster and more reliable services for overseas customers. Summary of the invention

[0004] By analyzing the historical order data of commodities and considering fluctuation characteristic values ​​such as coefficient of variation and standard deviation of rate of change, the system can automatically classify commodities into long-term supply commodities and short-term supply commodities. Such classification enables procurement strategies to be more flexible and adaptable to market changes. Long-term supply commodities can adopt more stable procurement plans, while short-term supply commodities can respond quickly to market fluctuations and achieve more efficient inventory turnover, which makes warehouse management more refined and personalized.

[0005] The cross-platform data terminal intelligent warehouse management method includes:

[0006] Set a long-term procurement cycle, where a long-term procurement cycle includes n short-term procurement cycles, where the value of n is selected by the simulated annealing algorithm; at the beginning of each long-term procurement cycle, for any commodity to be sold, analyze the historical order data of the commodity to determine whether the analysis result meets the preset conditions. If so, the current commodity is classified as a long-term supply commodity; if not, the current commodity is classified as a short-term supply commodity;

[0007] For any long-term supply product, at the beginning of each long-term procurement cycle, obtain the historical order data of the long-term supply product in the most recent N long-term procurement cycles, use the obtained historical order data of the N long-term procurement cycles as the order demand prediction set, use the order demand prediction set as the input of the order prediction model, and output the predicted order quantity of the long-term supply product in the current long-term procurement cycle; based on the predicted order quantity and actual order quantity of the most recent long-term procurement cycle, calculate the prediction deviation rate of the most recent long-term procurement cycle; according to the current inventory of the long-term supply product and the calculated prediction deviation rate, obtain the purchase quantity of the long-term supply product;

[0008] For any short-term supply product, at the beginning of each short-term procurement cycle, the historical order data of the short-term supply product in the most recent N short-term procurement cycles is obtained, and the historical order data of the N short-term procurement cycles obtained are analyzed by using the exponential smoothing method to obtain the predicted order quantity of the short-term supply product in the current short-term procurement cycle; based on the current inventory of the short-term supply product, the purchase quantity of the short-term supply product is calculated.

[0009] Preferably, the historical order data of the goods is analyzed to divide the goods into long-term supply goods and short-term supply goods. The specific operations are as follows:

[0010] At the beginning of the current long-term procurement cycle, for any product, obtain the order volume data x of the product in the last N long-term procurement cycles i , i = 1, 2, ..., N; calculate the order quantity data x i The average value of the order quantity Then use the formula Calculate and obtain order quantity data x i The standard deviation σ 1 ; Based on the average order volume obtained and standard deviation σ 1 , using the formula Calculate the coefficient of variation and use it as the first fluctuation characteristic value; at the same time, for the order volume data x i The last N-1 order quantity data in the formula Calculate the change rate B of the N-1 order quantity data after acquisition i; Then use the formula Calculate the standard deviation of the rate of change σ 2 , the standard deviation of the rate of change σ 2 as the second fluctuation characteristic value; setting the first discrimination threshold and the second discrimination threshold to determine whether the first fluctuation characteristic value and the second fluctuation characteristic value of the current commodity are respectively greater than the first discrimination threshold and the second discrimination threshold; if so, the current commodity is classified as a short-term supply commodity; if not, the current commodity is classified as a long-term supply commodity.

[0011] Preferably, the historical order data of N short-term procurement cycles are analyzed using an exponential smoothing method to obtain the predicted order volume of the short-term supply product in the current short-term procurement cycle. The specific operations are as follows:

[0012] For any short-term supply product, at the beginning of each short-term procurement cycle, obtain the historical order data y of the most recent N short-term procurement cycles i , using the formula Y i+1 =αy i +(1-α)Y i Calculate and obtain historical order data y i The predicted order quantity Y of the i+1th short-term procurement cycle i+1 , where α is the adjustment coefficient, which is used to represent the weight of the historical order data of the most recent short-term procurement cycle; when i = 1, Y 1 =y 1 ; When i=N, the calculation result is the predicted order quantity of the short-term supply product in the current short-term procurement cycle.

[0013] Preferably, the purchase quantity of long-term supply goods and the purchase quantity of short-term supply goods are calculated and obtained respectively, and the specific operations are as follows:

[0014] For any long-term supply product, the predicted order volume x based on the most recent long-term procurement cycle of the long-term supply product forecast and the actual order quantity x actual , using the formula Calculate the prediction deviation rate S; then use the formula x buy =X forecast +SX forecast -x stock Calculate the purchase quantity x of the long-term supply product buy ; Among them, X forecast represents the predicted order quantity of the long-term supply product in the current long-term procurement cycle, x stock Indicates the current inventory of the long-term supply product;

[0015] For any short-term supply product, based on the current inventory y of the short-term supply product stockand the forecast order quantity Y of the current short-term supply goods forecast , using the formula y buy =Y forecast -y stock Calculate the purchase quantity y of the short-term supply product buv .

[0016] Preferably, the value of n is selected by a simulated annealing algorithm, and the specific operation is as follows:

[0017] Step 1: Randomly select an initial n value as the current solution, set the initial temperature, maximum number of iterations and temperature attenuation coefficient; use the objective function Calculate the fitness value of the current solution, where n is the value of n; H represents the accuracy of the predicted order quantity of short-term supply goods when the current value of n is applied; ω 1 and ω 2 are weight coefficients, ω 1 <ω 2 ;

[0018] Step 2: Add or subtract a random value to the current solution to generate a neighborhood solution; use the objective function to calculate the fitness value of the neighborhood solution and compare it with the fitness value of the current solution; if the fitness value of the current solution is less than the fitness value of the neighborhood solution, update the current solution to the selected neighborhood solution; if the fitness value of the current solution is greater than the fitness value of the neighborhood solution, update the neighborhood solution according to the probability Decide whether to accept the neighborhood solution as the current solution; where F new represents the fitness value of the neighborhood solution, F current Represents the fitness value of the current solution, T current Indicates the current temperature; using the formula T new =kT current Update the temperature, where k is the temperature attenuation coefficient;

[0019] Step 3: Repeat step 2 until the maximum number of iterations is reached, and the current solution finally obtained is the optimal n value.

[0020] Preferably, the order prediction model is established based on the LSTM model, including N demand prediction LSTM units, 1 fully connected layer and 1 output layer, wherein the demand prediction LSTM unit is used to extract the temporal characteristics of the order demand prediction set; the fully connected layer is used to extract features from the outputs of all demand prediction LSTM units; and the output layer is used to output the predicted order quantity.

[0021] Preferably, the specific operations for training the order prediction model are as follows:

[0022] Obtain several training samples, each of which stores N+1 historical order data sorted by time. The first N historical order data in the training sample are used as the input of the order prediction model, and the last historical order data in the training sample is used as the output of the order prediction model. Divide all training samples into an order prediction training set and an order prediction verification set, send the order prediction training set to the order prediction model with initialized parameters for training, and then input the order prediction verification set to the order prediction model to obtain the verification result. Set the training conditions and determine whether the verification result meets the training conditions. If the training conditions are met, output the trained order prediction model; otherwise, continue to train the order prediction model through the order prediction training set.

[0023] Preferably, the cross-platform data terminal intelligent warehouse management system includes:

[0024] The commodity classification module is used to analyze the historical order data of any commodity to be sold at the beginning of each long-term procurement cycle, and determine whether the analysis result meets the preset conditions. If so, the current commodity is classified as a long-term supply commodity; if not, the current commodity is classified as a short-term supply commodity;

[0025] The long-term commodity forecasting module includes a long-term order forecasting unit, a deviation rate calculation unit and a long-term purchase quantity calculation unit; the long-term order forecasting unit is used to obtain the historical order data of the long-term supply commodity in the latest N long-term purchase cycles at the beginning of each long-term purchase cycle, use the obtained historical order data of the N long-term purchase cycles as the order demand forecast set, use the order demand forecast set as the input of the order forecasting model, and output the predicted order quantity of the long-term supply commodity in the current long-term purchase cycle; the deviation rate calculation unit is used to calculate the predicted deviation rate of the latest long-term purchase cycle using the predicted order quantity and actual order quantity of the latest long-term purchase cycle; the purchase quantity calculation unit is used to obtain the purchase quantity of the long-term supply commodity according to the current inventory of the long-term supply commodity and the calculated predicted deviation rate;

[0026] The short-term commodity forecasting module includes a short-term order forecasting unit and a short-term purchase quantity calculation unit; the short-term order forecasting unit is used to obtain the historical order data of the short-term supply commodities in the most recent N short-term purchase cycles at the beginning of each short-term purchase cycle, and to analyze the historical order data of the N short-term purchase cycles obtained by using the exponential smoothing method to obtain the predicted order quantity of the short-term supply commodities in the current short-term purchase cycle; the short-term purchase quantity calculation unit is used to calculate the purchase quantity of the short-term supply commodities according to the current inventory quantity and predicted order quantity of the short-term supply commodities.

[0027] The present invention has the following advantages:

[0028] 1. By analyzing the historical order data of commodities and considering the fluctuation characteristic values ​​such as the coefficient of variation and the standard deviation of the rate of change, the system can automatically divide commodities into long-term supply commodities and short-term supply commodities. Such classification makes the procurement strategy more flexible and adaptable to market changes. Long-term supply commodities can adopt a more stable procurement plan, while short-term supply commodities can respond quickly to market fluctuations and achieve more efficient inventory turnover, which makes warehouse management more refined and personalized.

[0029] 2. The present invention can better predict and respond to possible delays and changes in international logistics through the intelligent warehouse management system. The system can predict the order volume in different time periods based on historical order data and market trends, so as to plan logistics resources in advance and optimize inventory allocation, which helps to reduce transportation costs, shorten delivery time, and improve customer satisfaction; at the same time, the intelligent warehouse management system can also quickly adjust inventory and delivery plans in the face of various possible situations to ensure that goods can reach overseas customers in a timely manner, thereby enhancing the resilience and reliability of the entire supply chain. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a schematic diagram of the structure of the cross-platform data terminal intelligent warehouse management system adopted in an embodiment of the present invention. DETAILED DESCRIPTION

[0031] In order to enable persons skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0032] Embodiment 1, a cross-platform data terminal intelligent warehouse management method, comprising:

[0033] Set a long-term procurement cycle. A long-term procurement cycle includes n short-term procurement cycles. By setting a long-term procurement cycle, long-term inventory demand can be better planned and predicted, thereby reducing inventory costs and avoiding excessive inventory or out-of-stock. This number "n" is dynamically determined and will be adjusted according to the actual situation of the market and supply chain. The value of n is selected by the simulated annealing algorithm. The simulated annealing algorithm is a probabilistic optimization algorithm that simulates the annealing process in physics to find the optimal solution to the problem. Here, it is used to determine the optimal value of n. At the beginning of each long-term procurement cycle, for any commodity that needs to be sold, the historical order data of the commodity is analyzed. By analyzing the historical order data, the enterprise can better understand the demand pattern of the commodity, provide data support for subsequent inventory management and procurement decisions, and judge whether the analysis results meet the preset conditions. If so, the current commodity is classified as a long-term supply commodity; if not, the current commodity is classified as a short-term supply commodity. This classification method allows enterprises to adopt different inventory and procurement strategies according to the demand characteristics of the commodity, thereby improving the overall supply chain efficiency and the ability to respond to market changes. Long-term supply commodities may require more stable inventory management, while short-term supply commodities require more flexible procurement strategies.

[0034] For any long-term supply product, at the beginning of each long-term procurement cycle, the historical order data of the long-term supply product in the most recent N long-term procurement cycles is obtained, and the obtained historical order data of the N long-term procurement cycles is used as the order demand forecast set to help the system understand the demand history and trend of the product, so as to make more accurate forecasts; the order demand forecast set is used as the input of the order forecast model to output the predicted order quantity of the long-term supply product in the current long-term procurement cycle; based on the predicted order quantity and actual order quantity of the most recent long-term procurement cycle, the forecast deviation rate of the most recent long-term procurement cycle is calculated; based on the current inventory of the long-term supply product and the calculated forecast deviation rate, the required purchase quantity of the long-term supply product is obtained; the determination of the required purchase quantity is the core of inventory management, which ensures that the enterprise can adjust the purchase quantity according to the actual demand and forecast uncertainty, so as to optimize the inventory level, reduce costs and improve customer satisfaction;

[0035] For any short-term supply product, at the beginning of each short-term procurement cycle, the historical order data of the short-term supply product in the most recent N short-term procurement cycles is obtained, and the historical order data of the N short-term procurement cycles obtained are analyzed by using the exponential smoothing method. The exponential smoothing method is a time series prediction method that gives more weight to the most recent data, while the weight of earlier data gradually decreases, so as to obtain the predicted order quantity of the short-term supply product in the current short-term procurement cycle; based on the current inventory of the short-term supply product, the purchase quantity of the short-term supply product is calculated.

[0036] Analyze the historical order data of the goods and divide them into long-term supply goods and short-term supply goods. The specific operations are as follows:

[0037] At the beginning of the current long-term procurement cycle, for any product, obtain the order volume data x of the product in the last N long-term procurement cycles i , i = 1, 2, ..., N; calculate the order quantity data x i The average value of the order quantity Then use the formula Calculate and obtain order quantity data x i The standard deviation σ 1 ; Based on the average order quantity k and standard deviation σ 1 , using the formula Calculate the coefficient of variation and use it as the first fluctuation characteristic value; at the same time, for the order volume data x i The last N-1 order quantity data in the formula Calculate the change rate B of the N-1 order quantity data after acquisition i ; Then use the formula Calculate the standard deviation of the rate of change σ 2 , the standard deviation of the rate of change σ 2 as the second fluctuation characteristic value; setting the first discrimination threshold and the second discrimination threshold to determine whether the first fluctuation characteristic value and the second fluctuation characteristic value of the current commodity are respectively greater than the first discrimination threshold and the second discrimination threshold; if so, the current commodity is classified as a short-term supply commodity; if not, the current commodity is classified as a long-term supply commodity.

[0038] Use the exponential smoothing method to analyze the historical order data of N short-term procurement cycles to obtain the predicted order volume of the short-term supply goods in the current short-term procurement cycle. The specific operations are as follows:

[0039] For any short-term supply product, at the beginning of each short-term procurement cycle, obtain the historical order data y of the most recent N short-term procurement cycles i , using the formula Y i+1 =αy i +(1-α)Y i Calculate and obtain historical order data y i The predicted order quantity Y of the i+1th short-term procurement cycle i+1 , where α is the adjustment coefficient, which is used to represent the weight of the historical order data of the most recent short-term procurement cycle; when i = 1, Y 1 =y 1 ; When i=N, the calculation result is the predicted order quantity of the short-term supply product in the current short-term procurement cycle.

[0040] Calculate the purchase volume of long-term supply goods and short-term supply goods respectively. The specific operations are as follows:

[0041] For any long-term supply product, the predicted order volume x based on the most recent long-term procurement cycle of the long-term supply product forecast and the actual order quantity x actual , using the formula Calculate the prediction deviation rate S; then use the formula x buy =X forecast +SX forecast -x stock Calculate the purchase quantity x of the long-term supply product buy ; Among them, X forecast represents the predicted order quantity of the long-term supply product in the current long-term procurement cycle, x stock Indicates the current inventory of the long-term supply product;

[0042] For any short-term supply product, based on the current inventory of the short-term supply product ysto ck and the forecast order quantity Y of the current short-term supply goods forecast , using the formula y buy =Y forecast -y stock Calculate the purchase quantity y of the short-term supply product buy .

[0043] The value of n is selected by the simulated annealing algorithm. The specific operations are as follows:

[0044] Step 1: Randomly select an initial n value as the current solution, set the initial temperature, maximum number of iterations and temperature attenuation coefficient; use the objective function Calculate the fitness value of the current solution, where n is the value of n; H represents the accuracy of the predicted order quantity of short-term supply goods when the current value of n is applied; ω 1 and ω 2 are weight coefficients, ω 1 <ω 2 ;

[0045] Step 2: Add or subtract a random value to the current solution to generate a neighborhood solution; use the objective function to calculate the fitness value of the neighborhood solution and compare it with the fitness value of the current solution; if the fitness value of the current solution is less than the fitness value of the neighborhood solution, update the current solution to the selected neighborhood solution; if the fitness value of the current solution is greater than the fitness value of the neighborhood solution, update the neighborhood solution according to the probability Decide whether to accept the neighborhood solution as the current solution; where F new represents the fitness value of the neighborhood solution, F current Represents the fitness value of the current solution, Tcurrent Indicates the current temperature; using the formula T new =kT current Update the temperature, where k is the temperature attenuation coefficient;

[0046] Step 3: Repeat step 2 until the maximum number of iterations is reached, and the current solution finally obtained is the optimal n value.

[0047] Preferably, the order prediction model is established based on the LSTM model. The long short-term memory network, i.e. LSTM, is a special type of recurrent neural network that can learn and memorize long-term dependencies and is particularly suitable for processing and predicting complex patterns in time series data. It includes N demand prediction LSTM units, 1 fully connected layer and 1 output layer, wherein the demand prediction LSTM unit is used to extract the temporal features in the order demand prediction set; the fully connected layer is used to extract features from the outputs of all demand prediction LSTM units; and the output layer is used to output the predicted order quantity.

[0048] The specific operations for training the order prediction model are as follows:

[0049] Obtain several training samples, each of which stores N+1 historical order data sorted by time. The first N historical order data in the training sample are used as the input of the order prediction model, and the last historical order data in the training sample is used as the output of the order prediction model. Divide all training samples into an order prediction training set and an order prediction verification set, send the order prediction training set to the order prediction model with initialized parameters for training, and then input the order prediction verification set to the order prediction model to obtain the verification result. Set the training conditions and determine whether the verification result meets the training conditions. If the training conditions are met, output the trained order prediction model; otherwise, continue to train the order prediction model through the order prediction training set.

[0050] Example 2, cross-platform data terminal intelligent warehouse management system, such as Figure 1 As shown, including:

[0051] The commodity classification module is used to analyze the historical order data of any commodity to be sold at the beginning of each long-term procurement cycle, and determine whether the analysis result meets the preset conditions. If so, the current commodity is classified as a long-term supply commodity; if not, the current commodity is classified as a short-term supply commodity;

[0052] The long-term commodity forecasting module includes a long-term order forecasting unit, a deviation rate calculation unit and a long-term purchase quantity calculation unit; the long-term order forecasting unit is used to obtain the historical order data of the long-term supply commodity in the latest N long-term purchase cycles at the beginning of each long-term purchase cycle, use the obtained historical order data of the N long-term purchase cycles as the order demand forecast set, use the order demand forecast set as the input of the order forecasting model, and output the predicted order quantity of the long-term supply commodity in the current long-term purchase cycle; the deviation rate calculation unit is used to calculate the predicted deviation rate of the latest long-term purchase cycle using the predicted order quantity and actual order quantity of the latest long-term purchase cycle; the purchase quantity calculation unit is used to obtain the purchase quantity of the long-term supply commodity according to the current inventory of the long-term supply commodity and the calculated predicted deviation rate;

[0053] The short-term commodity forecasting module includes a short-term order forecasting unit and a short-term purchase quantity calculation unit; the short-term order forecasting unit is used to obtain the historical order data of the short-term supply commodities in the most recent N short-term purchase cycles at the beginning of each short-term purchase cycle, and to analyze the historical order data of the N short-term purchase cycles obtained by using the exponential smoothing method to obtain the predicted order quantity of the short-term supply commodities in the current short-term purchase cycle; the short-term purchase quantity calculation unit is used to calculate the purchase quantity of the short-term supply commodities according to the current inventory quantity and predicted order quantity of the short-term supply commodities.

[0054] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all such improvements and changes should fall within the scope of protection of the appended claims of the present invention. Parts not described in detail in this specification belong to the prior art known to those skilled in the art.

Claims

1. A cross-platform data terminal intelligent warehouse management method, characterized in that: include: Set a long-term procurement cycle, where a long-term procurement cycle includes n short-term procurement cycles, where the value of n is selected by the simulated annealing algorithm; at the beginning of each long-term procurement cycle, for any commodity to be sold, analyze the historical order data of the commodity to determine whether the analysis result meets the preset conditions. If so, the current commodity is classified as a long-term supply commodity; if not, the current commodity is classified as a short-term supply commodity; For any long-term supply product, at the beginning of each long-term procurement cycle, obtain the historical order data of the long-term supply product in the most recent N long-term procurement cycles, use the obtained historical order data of the N long-term procurement cycles as the order demand prediction set, use the order demand prediction set as the input of the order prediction model, and output the predicted order quantity of the long-term supply product in the current long-term procurement cycle; based on the predicted order quantity and actual order quantity of the most recent long-term procurement cycle, calculate the prediction deviation rate of the most recent long-term procurement cycle; according to the current inventory of the long-term supply product and the calculated prediction deviation rate, obtain the purchase quantity of the long-term supply product; For any short-term supply product, at the beginning of each short-term procurement cycle, the historical order data of the short-term supply product in the most recent N short-term procurement cycles is obtained, and the historical order data of the N short-term procurement cycles obtained are analyzed by using the exponential smoothing method to obtain the predicted order volume of the short-term supply product in the current short-term procurement cycle; Based on the current inventory of the short-term supply product, calculate the required purchase quantity of the short-term supply product.

2. The cross-platform data terminal intelligent warehouse management method according to claim 1 is characterized in that: Analyze the historical order data of the goods and divide them into long-term supply goods and short-term supply goods. The specific operations are as follows: At the beginning of the current long-term procurement cycle, for any product, obtain the order volume data x of the product in the last N long-term procurement cycles i , i=1,2,…,N; Calculate the order quantity data x i The average value of the order quantity Then use the formula Calculate and obtain order quantity data x i The standard deviation σ1 is based on the average order quantity obtained and standard deviation σ1, using the formula Calculate and obtain the coefficient of variation, and use the coefficient of variation as the first fluctuation characteristic value; At the same time, for the order volume data x i The last N-1 order quantity data in the formula Calculate the change rate B of the N-1 order quantity data after acquisition i ; Then use the formula Calculate and obtain the standard deviation σ2 of the rate of change, and use the standard deviation σ2 of the rate of change as the second fluctuation characteristic value; A first discrimination threshold and a second discrimination threshold are set to determine whether the first fluctuation characteristic value and the second fluctuation characteristic value of the current commodity are respectively greater than the first discrimination threshold and the second discrimination threshold. If so, the current commodity is classified as a short-term supply commodity; if not, the current commodity is classified as a long-term supply commodity.

3. The cross-platform data terminal intelligent warehouse management method according to claim 2 is characterized in that: Use the exponential smoothing method to analyze the historical order data of N short-term procurement cycles to obtain the predicted order volume of the short-term supply goods in the current short-term procurement cycle. The specific operations are as follows: For any short-term supply product, at the beginning of each short-term procurement cycle, obtain the historical order data y of the most recent N short-term procurement cycles i , using the formula Y i+1 =αy i +(1-α)Y i Calculate and obtain historical order data y i The predicted order quantity Y of the i+1th short-term procurement cycle i+1 , where α is the adjustment coefficient, which is used to represent the weight of the historical order data of the most recent short-term procurement cycle; when i=1, Y1=y1; when i=N, the calculation result is the predicted order quantity of the short-term supply product in the current short-term procurement cycle.

4. The cross-platform data terminal intelligent warehouse management method according to claim 3 is characterized in that: Calculate the purchase volume of long-term supply goods and short-term supply goods respectively. The specific operations are as follows: For any long-term supply product, the predicted order volume x based on the most recent long-term procurement cycle of the long-term supply product forecast and the actual order quantity x actual , using the formula Calculate and obtain the prediction deviation rate S; Then use the formula x buy =X forecast +SX forecast -x stock Calculate the purchase quantity x of the long-term supply product buy ; Among them, X forecast represents the predicted order quantity of the long-term supply product in the current long-term procurement cycle, x stock Indicates the current inventory of the long-term supply product; For any short-term supply product, based on the current inventory y of the short-term supply product stock and the forecast order quantity Y of the current short-term supply goods forecast , using the formula y buy =Y forecast -y stock Calculate the purchase quantity y of the short-term supply product buy .

5. The cross-platform data terminal intelligent warehouse management method according to claim 4 is characterized in that: The value of n is selected by the simulated annealing algorithm. The specific operations are as follows: Step 1: Randomly select an initial n value as the current solution, set the initial temperature, maximum number of iterations and temperature attenuation coefficient; use the objective function Calculate the fitness value of the current solution, where n is the value of n; H represents the accuracy of the predicted order quantity of short-term supply goods when the current value of n is applied; ω1 and ω2 are weight coefficients, ω1<ω2; Step 2: Add or subtract a random value to the current solution to generate a neighborhood solution; use the objective function to calculate the fitness value of the neighborhood solution and compare it with the fitness value of the current solution; if the fitness value of the current solution is less than the fitness value of the neighborhood solution, update the current solution to the selected neighborhood solution; if the fitness value of the current solution is greater than the fitness value of the neighborhood solution, update the neighborhood solution according to the probability Decide whether to accept the neighborhood solution as the current solution; where F new represents the fitness value of the neighborhood solution, F current Represents the fitness value of the current solution, T current Indicates the current temperature; using the formula T new =kT current Update the temperature, where k is the temperature attenuation coefficient; Step 3: Repeat step 2 until the maximum number of iterations is reached, and the current solution finally obtained is the optimal n value.

6. The cross-platform data terminal intelligent warehouse management method according to claim 5 is characterized in that: The order prediction model is built based on the LSTM model, which includes N demand prediction LSTM units, 1 fully connected layer and 1 output layer. The demand prediction LSTM unit is used to extract the temporal characteristics of the order demand prediction set. The fully connected layer is used to extract features from the outputs of all demand prediction LSTM units; the output layer is used to output the predicted order quantity.

7. The cross-platform data terminal intelligent warehouse management method according to claim 6 is characterized in that: The specific operations for training the order prediction model are as follows: Obtain several training samples, each of which stores N+1 pieces of historical order data sorted by time. The first N pieces of historical order data in the training sample are used as the input of the order prediction model, and the last piece of historical order data in the training sample is used as the output of the order prediction model. All training samples are divided into an order prediction training set and an order prediction verification set. The order prediction training set is fed into the order prediction model with initialized parameters for training. The order prediction verification set is then fed into the order prediction model to obtain verification results. Set the training conditions and determine whether the verification results meet the training conditions. If so, output the trained order prediction model; otherwise, continue to train the order prediction model using the order prediction training set.

8. Cross-platform data terminal intelligent warehouse management system, characterized by: The system is applied to the cross-platform data terminal intelligent warehouse management method according to any one of claims 1 to 7, including: The commodity classification module is used to analyze the historical order data of any commodity to be sold at the beginning of each long-term procurement cycle, and determine whether the analysis result meets the preset conditions. If so, the current commodity is classified as a long-term supply commodity; if not, the current commodity is classified as a short-term supply commodity; The long-term commodity forecasting module includes a long-term order forecasting unit, a deviation rate calculation unit and a long-term purchase quantity calculation unit; the long-term order forecasting unit is used to obtain the historical order data of the long-term supply commodity in the latest N long-term purchase cycles at the beginning of each long-term purchase cycle, use the obtained historical order data of the N long-term purchase cycles as the order demand forecast set, use the order demand forecast set as the input of the order forecasting model, and output the predicted order quantity of the long-term supply commodity in the current long-term purchase cycle; the deviation rate calculation unit is used to calculate the predicted deviation rate of the latest long-term purchase cycle using the predicted order quantity and actual order quantity of the latest long-term purchase cycle; the purchase quantity calculation unit is used to obtain the purchase quantity of the long-term supply commodity according to the current inventory of the long-term supply commodity and the calculated predicted deviation rate; The short-term commodity forecasting module includes a short-term order forecasting unit and a short-term purchase quantity calculation unit; the short-term order forecasting unit is used to obtain the historical order data of the short-term supply commodities in the most recent N short-term purchase cycles at the beginning of each short-term purchase cycle, and to analyze the historical order data of the N short-term purchase cycles obtained by using the exponential smoothing method to obtain the predicted order quantity of the short-term supply commodities in the current short-term purchase cycle; the short-term purchase quantity calculation unit is used to calculate the purchase quantity of the short-term supply commodities according to the current inventory quantity and predicted order quantity of the short-term supply commodities.

Images