Warehouse cargo supply chain management method and system based on AI big data

Abstract

The invention discloses a warehouse cargo supply chain management method and system based on AI big data, and relates to the technical field of warehouse management, and the method comprises the steps: collecting and preprocessing warehouse multi-source data, and predicting the cargo delivery frequency based on an LSTM neural network; and a multi-target scheduling model combining pickup path time consumption and storage gravity center control is constructed based on a prediction result, a cargo task and AGV path planning are jointly modeled into an open path multi-starting-point asymmetric TSP problem, an FWA-GA hybrid optimization algorithm is used for model solving, and an optimal cargo allocation matrix and an optimal AGV scheduling path table are obtained. Therefore, integrated joint optimization of task allocation, path planning and goods allocation layout is achieved, a supply chain platform is further linked to complete stockout early warning and supplier intelligent selection, and multiple limitations of a traditional method in the aspects of prediction precision, scheduling coupling, structure control, supply collaboration and the like are effectively broken through.

Description

Technical Field

[0001] The present invention relates to the technical field of warehouse management, and in particular to a warehouse goods supply chain management method and system based on AI (Artificial Intelligence) big data, a computer device, and a computer-readable storage medium. Background Art

[0002] With the development of industrial automation and intelligent logistics, modern warehousing systems are evolving continuously towards intelligence, flexibility, and collaboration. The traditional static manual management mode has gradually been replaced by automated equipment, information systems, and artificial intelligence models. Against this background, the warehouse management system, enterprise resource planning system, Internet of Things sensing network, and various mobile handling devices have achieved initial integration, providing a basic support for realizing goods inbound and outbound scheduling, path optimization, inventory warning, and supply chain linkage.

[0003] However, most current systems still process modules such as prediction, scheduling, execution, and supply response separately, and have not yet formed a prediction-driven full-chain optimization strategy; at the same time, there are still deficiencies in the global modeling ability and algorithm solving efficiency for complex path scheduling in multi-AGV (Automated Guided Vehicle) scenarios, making it difficult to balance real-time performance and system stability. Summary of the Invention

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

[0005] Therefore, the present invention provides a warehouse goods supply chain management method and system based on AI big data, which solves the problem that most current systems still process modules such as prediction, scheduling, execution, and supply response separately and have not yet formed a prediction-driven full-chain optimization strategy; at the same time, it solves the problem that there are still deficiencies in the global modeling ability and algorithm solving efficiency for complex path scheduling in multi-AGV scenarios, making it difficult to balance real-time performance and system stability.

[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 warehouse goods supply chain management method based on AI big data, which includes: collecting multi-source warehouse data and performing preprocessing, predicting the goods outbound frequency based on real-time goods data using an LSTM (Long Short-Term Memory) neural network, wherein the multi-source warehouse data includes goods basic information and goods location data; constructing a multi-objective scheduling model combining picking path time consumption and warehouse center of gravity control based on the prediction result, jointly modeling the goods task and the AGV path planning as an open path multi-start asymmetric TSP (Traveling Salesman Problem), and using a FWA-GA (Fireworks Algorithm-Genetic Algorithm) hybrid optimization algorithm to solve the model to obtain an optimal goods location allocation matrix and an optimal AGV scheduling path table; implementing the optimal scheduling plan, comparing the predicted goods outbound frequency with the current inventory quantity after the scheduling optimization, automatically generating a shortage warning and a procurement plan and pushing them to the supply chain control system, and selecting the optimal supplier to respond to the goods.

[0008] As a preferred solution of the warehouse goods supply chain management method based on AI big data according to the present invention, wherein: the constructing a multi-objective scheduling model combining picking path time consumption and warehouse center of gravity control based on the prediction result includes: using a Flying-V layout as the warehouse basic structure, combining the outbound frequency, goods quality and structural location to construct a two-objective function for scheduling optimization, including minimizing the path time f1 and minimizing the warehouse center of gravity f2.

[0009] As a preferred solution of the warehouse goods supply chain management method based on AI big data according to the present invention, wherein: the jointly modeling the goods task and the AGV path planning as an open path multi-start asymmetric TSP problem and using a FWA-GA hybrid optimization algorithm to solve the model to obtain an optimal goods location allocation matrix and an optimal AGV scheduling path table includes:

[0010] Extracting all the goods tasks to be scheduled to form a task set A, recording the current number of online AGVs, initializing the current position of each AGV, and calculating the corresponding path cost d as the task movement cost input for any two points in the task set;

[0011] Combining the overall efficiency and the single-AGV response time to construct a scheduling objective function F and adding scheduling constraints;

[0012] Setting the constraint condition that each task must be assigned and only assigned to one AGV, and the goods locations of all tasks must meet the warehouse physical capacity;

[0013] Construct the initial feasible solution population, set the total number of individuals and the maximum number of iterations, and each individual is represented as a scheduling plan;

[0014] Use the Logistic chaotic map to generate the initialization perturbation sequence;

[0015] Perform integer encoding for the first O variables, map them to the AGV number and execution priority to which the task belongs, and for the latter O variables, map them to the layer number where the goods are located. Decode the task scheduling result and the goods location allocation into the task path table and the goods location allocation matrix;

[0016] Use the scheduling objective function as the fitness function to calculate the fitness, and sort the individuals in descending order according to the fitness. Perform the spark explosion operation on the first U individuals to generate the perturbed candidate scheduling solutions;

[0017] Set the maximum perturbation radius of the fireworks explosion and calculate the explosion radius of the t-th generation;

[0018] For each main individual, calculate the number of sparks S according to the corresponding fitness value and the worst fitness in the population Ω ;

[0019] Traverse each scheduling individual, initialize the spark set, loop Q times for each individual, generate the spark individuals in turn, automatically select the current perturbation strategy according to the stage t, compare the stage t with the preset threshold. If t is less than the threshold, enter the early global search stage and perform the uniform explosion perturbation;

[0020] Otherwise, enter the late local convergence stage and use the Cauchy-Gaussian mixed perturbation;

[0021] Adopt the Cauchy-Gaussian mixed perturbation strategy to perform global structural perturbation on the individuals of the initial main population, providing new scheduling solution candidates for elite screening;

[0022] Immediately perform the legality check and out-of-bounds repair for each generated spark individual;

[0023] Store all legal spark individuals in the spark set of this subject to form the spark individual set;

[0024] For each spark individual, recalculate using the fitness function, bind the spark individual with the new fitness value to form the extended population;

[0025] Sort all individuals in the extended population in ascending order of fitness, intercept the first n individuals to form the (t + 1)-th generation main population, retain the current global optimal individual, randomly select M pairs of individuals, and perform partial mapping crossover on their task path encodings;

[0026] Update the path structures of the two individuals after crossover to form the new intermediate individual set Xpmx ;

[0027] For each individual in X pmx Set a perturbation probability, randomly select some tasks, and perform mutation operations;

[0028] During the genetic evolution stage, perform a Cauchy-Gaussian (a hybrid model combining the Cauchy distribution and the Gaussian distribution) perturbation strategy on the crossover individuals to further optimize the task level sorting and the fine layout of storage locations;

[0029] If the perturbation causes the variable to cross the boundary, immediately perform boundary repair to obtain the population X of individuals after perturbation optimization mut ;

[0030] For each individual in X mut Extract the task path of each AGV, based on the Flying-V storage structure, perform path planning inspection to obtain the population X1 of optimized scheduling solutions, sort X1 in ascending order according to the fitness value, extract the individual w with the smallest fitness, compare w with the optimal individual w' recorded last time. If the fitness value of w is less than the optimal individual w', then update w to the global optimal individual.

[0031] As a preferred solution of the warehouse goods supply chain management method based on AI big data according to the present invention, wherein: the implementation of the optimal scheduling plan means extracting the optimal task location allocation matrix, the AGV path scheduling table, the total task path time, and the maximum AGV task time from the optimal individual coding, converting the location matrix into a visible structure in a three-dimensional coordinate system for inbound / outbound area instruction matching, and converting the AGV path scheduling table into a scheduling queue to implement task push.

[0032] As a preferred solution of the warehouse goods supply chain management method based on AI big data according to the present invention, wherein: after the scheduling optimization is completed, compare the predicted goods outbound frequency with the current inventory, automatically generate a shortage warning and a procurement plan and push them to the supply chain control system. Selecting the optimal supplier for goods response means comparing based on the prediction results and real-time data, analyzing the goods shortage risk, for each out-of-stock good, automatically calling the list of available suppliers pre-registered in the supply chain platform, performing a weighted sum of the historical average supply cycle, the performance success rate, and the unit supply price of the suppliers, sorting the suppliers in descending order based on the sum score, selecting the supplier with the highest score, and immediately generating a corresponding replenishment task order.

[0033] As a preferred solution of the warehouse goods supply chain management method based on AI big data according to the present invention, wherein: the collection and preprocessing of multi-source data in the warehouse include:

[0034] Before all goods are warehoused, a unique RFID (Radio Frequency Identification) tag is bound to each piece of goods to obtain the triple of the unique number of the goods, the warehousing / outbound timestamp, and the current location coordinates (x i , y i , z i ) in real time, corresponding to the row number, layer number, and column number of the shelf;

[0035] Collect the current weight of the goods and the real-time inventory. Combining the warehousing / outbound time, the inbound / outbound frequency sequence and the current storage coordinates of the goods can be generated;

[0036] Through the data interface docking with the ERP (Enterprise Resource Planning) and WMS (Warehouse Management System) systems, each piece of goods can obtain the corresponding supplier attributes, including the supply cycle, the performance rate, and the unit procurement cost;

[0037] Process the missing values and normalize the collected data.

[0038] As a preferred solution of the warehouse goods supply chain management method based on AI big data according to the present invention, wherein: the prediction of the goods outbound frequency based on the LSTM neural network means inputting the real-time goods outbound data into the LSTM model to obtain the future goods outbound frequency.

[0039] In the second aspect, the present invention provides a warehouse goods supply chain management system based on AI big data, including

[0040] A data acquisition module for collecting real-time goods data and performing preprocessing;

[0041] A prediction module for predicting the goods outbound frequency;

[0042] A scheduling optimization module for performing multi-objective modeling;

[0043] A scheduling solution module for solving the optimal scheduling result through FWA-GA;

[0044] An implementation module for performing real-time task scheduling based on the optimal scheduling result;

[0045] A response module for performing out-of-stock warnings and supplier responses.

[0046] In a third aspect, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and: when the computer program is executed by the processor, any step of the warehouse goods supply chain management method based on AI big data as described in the first aspect of the present invention is implemented.

[0047] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and: when the computer program is executed by the processor, any step of the warehouse goods supply chain management method based on AI big data as described in the first aspect of the present invention is implemented.

[0048] In a second aspect, the present invention provides a warehouse goods supply chain management system based on AI big data, including a data collection module for collecting real-time goods data and performing preprocessing; a prediction module for predicting the goods outbound frequency based on the real-time goods data using an LSTM neural network; a scheduling optimization module for constructing a multi-objective scheduling model that combines the picking path time consumption and the storage center of gravity control based on the goods outbound frequency, and jointly modeling the goods task and the AGV path planning as an open path multi-start non-symmetric TSP problem; a scheduling solution module for using a FWA-GA hybrid optimization algorithm to solve the multi-objective scheduling model to obtain an optimal goods location allocation matrix and an optimal AGV scheduling path table; an implementation module for implementing the optimal scheduling plan according to the optimal task goods location allocation matrix and the optimal AGV path scheduling table; and a response module for comparing the predicted goods outbound frequency with the current inventory, automatically generating a shortage warning and a procurement plan, pushing them to the supply chain control system, and selecting the optimal supplier to respond to the goods.

[0049] In a third aspect, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and: when the computer program is executed by the processor, any step of the warehouse goods supply chain management method based on AI big data as described in the first aspect of the present invention is implemented.

[0050] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and: when the computer program is executed by the processor, any step of the warehouse goods supply chain management method based on AI big data as described in the first aspect of the present invention is implemented.

[0051] The beneficial effects of the present invention are as follows: Based on the AI-driven prediction of outbound frequency, the present invention constructs a two-objective scheduling model that takes into account both path time consumption and warehouse center of gravity control, and introduces the FWA-GA hybrid optimization algorithm for solution based on the open path multi-start asymmetric TSP problem. At the same time, the Cauchy-Gaussian perturbation mechanism is integrated to strengthen the local convergence ability. Finally, the integrated joint optimization of task allocation, path planning and storage location layout is realized, and further linked with the supply chain platform to complete out-of-stock warning and intelligent supplier selection, effectively breaking through the multiple limitations of traditional methods in aspects such as prediction accuracy, scheduling coupling, structure control and supply coordination. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 is a flowchart of a warehouse goods supply chain management method based on AI big data according to an embodiment of the present invention.

[0053] Figure 2 is a schematic diagram of Flying-V storage layout and AGV path according to an embodiment of the present invention

[0054] Figure 3 is a structural diagram of a warehouse goods supply chain management system based on AI big data according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0055] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0056] Figure 1 is a flowchart of a warehouse goods supply chain management method based on AI big data according to an embodiment of the present invention, as Figure 1 shown, the method includes the following steps:

[0057] S1, collect multi-source data of the warehouse and perform preprocessing, and predict the outbound frequency of goods based on the LSTM neural network according to real-time goods data. Among them, the multi-source data of the warehouse includes basic information of goods and goods location data.

[0058] In a specific embodiment of the present invention, collecting and preprocessing multi-source data in the warehouse includes: before all goods are warehoused, binding a unique RFID tag to each piece of goods, which can be batch-initialized through UHF (Ultra High Frequency, referring to radio waves with frequencies between 300 MHz and 3000 MHz) electronic tags, and deploying RFID reading and writing antennas at the warehouse entrance, exit, intersections of main channels, and key picking points to obtain in real time the triple of the unique number of the goods, the warehousing / outbound timestamp, and the current location coordinates (x i , y i , z i ), corresponding to the row number, layer number, and column number of the shelf, as well as operation events such as the operation type (warehousing / outbound / allocation); a weight sensor and a location coding device are deployed for each shelf unit to collect the following parameters: the current weight m of the goods i , the real-time inventory s i (indicating the number of pieces of the current goods at this location), and the inbound / outbound frequency sequence r i (t) can be generated by combining the warehousing / outbound time (where r i (t) represents the number of outbound times of goods i at time point t), and the current storage coordinates of the goods; through the data interface docking with the ERP and WMS systems, each piece of goods can obtain the corresponding supplier attributes, including the supply cycle τ q , the fulfillment rate ρ q (reflecting the success rate of supplier order fulfillment), and the unit procurement cost ε q , where q refers to the qth supplier; perform missing value processing on the collected data, perform logarithmic normalization on the outbound frequency, perform Min-Max (range normalization) normalization on the goods quality, inventory, and supply cycle, and perform Z-score (standard score) normalization on the procurement cost.

[0059] Specifically, by deploying RFID readers at key warehouse nodes (such as in / out ports, intersection channels, picking stations, etc.) and combining with the unique electronic tags attached to each piece of goods, the system can collect key data such as its number, location, transfer time, operation type, etc. in real time, and build a high-resolution trajectory chain of the goods flow. Compared with the traditional barcode scanning recording method, this solution has significant advantages such as non-contact, automation, and high reliability, and can achieve continuous data collection without human intervention. At the same time, the combination of the deployment of weight sensors and the goods location coding device can obtain information such as the change in the weight of goods, the real-time inventory number of each goods location, and the space utilization rate in real time, so as to realize dynamic warehouse status monitoring and refined inventory management. By counting the in / out warehouse behaviors of goods at each time point, an outbound frequency sequence in the form of a time series can also be constructed, which serves as an important input variable for the training of the prediction model. Through the above preprocessing, each piece of goods data is finally organized into a multi-dimensional vector in a unified format, providing a consistent, operable, and learnable input interface for the subsequent LSTM prediction model and optimization scheduling algorithm.

[0060] Furthermore, predicting the outbound frequency of goods based on the LSTM neural network means using the LSTM model to predict the outbound frequency of goods. The historical outbound data of goods is used as the training set to input into the LSTM model for model training. The loss function and the Adam (an optimization algorithm based on the gradient descent method) optimizer are defined to perform iterative optimization of the model parameters. When the loss of the LSTM model no longer decreases significantly during continuous iteration, the iteration stops, the model parameters are output, the LSTM model is updated, and the real-time outbound data of goods is input into the LSTM model to obtain the future outbound frequency of goods.

[0061] Embedding the LSTM prediction output into the scheduling and replenishment decision-making chain to form a "prediction - optimization - execution - response" closed loop is significantly better than the low-precision prediction strategies such as static rules or moving averages in the prior art. Through multi-stage fine modeling and optimization, the accuracy of the outbound trend prediction and the adaptability of the scheduling model are significantly improved.

[0062] S2. Based on the outbound frequency of goods, construct a multi-objective scheduling model that combines the picking path time consumption and the control of the warehouse center of gravity. Jointly model the goods tasks and the AGV path planning as an open path multi-start asymmetric TSP problem, and use the FWA-GA hybrid optimization algorithm to solve the multi-objective scheduling model to obtain the optimal goods location allocation matrix and the optimal AGV scheduling path table.

[0063] Among them, constructing a multi-objective scheduling model that combines the picking path time consumption and the control of the warehouse center of gravity based on the prediction results includes: taking the Flying-V layout as the warehouse infrastructure, and combining the outbound frequency, the quality of goods, and the structural location to construct a two-objective function for scheduling optimization, including minimizing the path time f1 and minimizing the warehouse center of gravity f2.

[0064] Specifically, as Figure 2 shown, the Flying-V warehouse structure is in a "V" shape. Under the Flying-V warehouse structure, the AGV needs to walk in the main channel and then enter each sub-channel, and then perform lifting operations to complete the handling of goods. The moving time of various goods on the plane path and the vertical lifting time are weighted and superimposed to construct the following path time-consuming objective function f1:

[0065]

[0066] In the formula, R i is the predicted outbound frequency of goods, α i is the scheduling priority weight fitted according to the historical order urgency and prediction trend, L x is the basic length of the main channel, y i is the row number of the goods in the channel, l is the row spacing of the shelves, z i is the layer where the goods are located, multiplied by h represents the height, v1 and v2 are the horizontal movement and vertical lifting speeds of the AGV, N is the total number of goods types or total tasks in the warehouse, and h is the shelf layer height.

[0067] Different from the traditional model, the present invention embeds the AI-driven priority coefficient α i into the path distance weight, so that the system not only considers the location of the goods, but also considers the urgency of its actual scheduling, thus being more in line with the "AI prediction-driven scheduling scenario".

[0068] To avoid the increase in handling load or the instability of the shelf center of gravity caused by the concentrated stacking of frequently outbound goods on the high floors, a center of gravity control mechanism is introduced at the same time. Considering that there may be a certain uneven weight distribution for some high-frequency goods, in order to enhance the comprehensiveness of the scheduling model, the frequency ranking influence factor β j is introduced into the center of gravity model, and the following objective function f2 is reconstructed:

[0069]

[0070] In the formula, m i is the weight of the goods, and β i is the quantile value of the goods in the predicted outbound frequency ranking.

[0071] Considering the limitations of the actual warehouse structure and the AGV operation ability, it is necessary to set a hard boundary for the spatial position where the goods can be allocated to ensure that all scheduling results can be implemented in the physical space. The constraint expression is as follows:

[0072] x i ≤x max , y i ≤y max , z i ≤z max

[0073] where x i , y i , z i represent the column, row, and layer positions to which the i-th item of goods is assigned respectively, and x max , y max , z max represent the maximum accommodation values of the three-dimensional structure of the warehouse in each direction respectively;

[0074] In summary, the structure of the multi-objective scheduling model has been built, and the following optimization objective structure is formed:

[0075] Minimize{f1,f2}Subject to:x i ≤x max ,y i ≤y max ,z i ≤z max ;

[0076] Minimize is the minimization parameter to satisfy the objective function.

[0077] This model is guided by the AI prediction results, constrained by the characteristics of the warehousing structure, and uses the execution scheduling time consumption and structural stability as the evaluation criteria to form an evaluation objective function for the operation of the scheduling algorithm.

[0078] By combining the predicted outbound frequency with the path cost, not only the physical distance in space is considered, but also the dynamic scheduling dimension of "time urgency" is introduced. Compared with the traditional scheduling model that only uses distance or operation time as the optimization objective, this method realizes the integrated modeling of "frequency-driven + location-oriented", which is closer to the management logic of prioritizing frequent tasks in real business, and significantly improves the intelligence level and actual combat response ability of scheduling. The traditional scheduling model rarely considers the load distribution in the vertical direction of the shelves. Especially in high-rise stereoscopic warehouses, frequent high-level operations will lead to an increase in the lifting energy consumption and a decrease in the stability of AGVs. The present invention establishes a quality-heat coupling model by adding the weight and frequency quantile values of the goods, restricts the concentration of heavy or high-frequency goods to the middle and lower layers, realizes the optimization of the warehousing center of gravity, and effectively prevents structural safety problems such as shelf inclination and AGV lifting conflicts. By explicitly setting the maximum column / row / layer positions where the goods can be assigned, the hard constraint processing is performed on the spatial range of all variables during the optimization solution process to prevent the problem that the scheduling task cannot be executed due to the out-of-bounds result of the optimization function. The coordinated optimization of the minimum path time consumption and the center of gravity control objective enables the scheduling result to achieve a balance between execution efficiency and system safety. Especially in the complex scheduling scenario of multiple AGVs and multiple tasks, the present invention provides a scheduling control architecture for unified global objective modeling and dynamic local conflict optimization, with higher operation stability and expansion elasticity.

[0079] In one embodiment of the present invention, jointly modeling the goods task and the AGV path planning as an open-path multi-start asymmetric TSP problem, and using the FWA-GA hybrid optimization algorithm to solve the model to obtain the optimal goods location allocation matrix and the optimal AGV scheduling path table includes the following steps S21 - S220:

[0080] S21, Extract all the goods tasks to be scheduled to form a task set A, record the current number of on-line AGVs, initialize the current position of each AGV, and for any two points in the task set, calculate the corresponding path cost d as the task movement cost input.

[0081]

[0082] Calculate the path cost from the starting point of each AGV to the task, accumulate the costs of all AGVs to complete their respective task paths to obtain the total path cost. In the path of each AGV, encode the task order as a set P = {p1, p2,..., p k}, where p k is the kth task;

[0083] S22, Construct a scheduling objective function F by combining the overall efficiency and the response time of a single AGV and add scheduling constraints.

[0084] F = ω1·SOC + ω2·MS + ω3·GC

[0085] In the formula, ω1, ω2, and ω3 are weighting coefficients, SOC is the total time consumed by all AGV task paths, MS is the maximum task execution time, and GC is the total impact of the center of gravity of all tasks;

[0086]

[0087]

[0088] In the formula, n j is the total number of tasks executed by the jth AGV, f1(p k ) is the path time consumed corresponding to the kth task, e is the total number of AGVs, g ∈ E is the gth goods task in the current scheduling task set E, f2(g) is the single impact amount of the gth goods on the storage center of gravity, j ∈ [1, e] is the jth AGV, and there are e AGVs in the system;

[0089] S23, Set the constraint condition that each task must be assigned and only assigned to one AGV, and the storage locations of all tasks must meet the physical capacity of the warehouse.

[0090] S24, Construct an initial feasible solution population, set the total number of individuals and the maximum number of iterations, and each individual represents a scheduling scheme;

[0091] S25, generate an initial perturbation sequence using the Logistic chaotic map.

[0092] z o+1 = μ · z o · (1 - z o )

[0093] In the formula, z o+1 is the next chaotic value, μ is the control parameter of the Logistic map, and z o is the chaotic variable value generated in the o-th iteration.

[0094] S26, perform integer encoding on the first O variables, map them to the AGV number and execution priority to which the task belongs. For the last O variables, map them to the layer number where the goods are located, and decode the task scheduling result and the goods location allocation into a task path table and a goods location allocation matrix.

[0095] S27, use the scheduling objective function as the fitness function to calculate the fitness, sort the individuals in descending order according to the fitness, and perform a spark explosion operation on the first U individuals to generate a perturbed candidate scheduling solution.

[0096] To expand the search space and improve the global optimal solution finding ability of the scheduling model, it is necessary to perform a spark explosion operation on the individuals with better fitness to generate several perturbed candidate scheduling solutions;

[0097] S28, set the maximum perturbation radius of the fireworks explosion and calculate the explosion radius of the t-th generation.

[0098] Set the maximum perturbation radius of the fireworks explosion and calculate the explosion radius of the t-th generation:

[0099]

[0100] In the formula, r t is the explosion radius of the current t-th generation iteration, which is used to control the perturbation amplitude, r initial is the initial maximum perturbation radius, which is set manually, r end is the final convergence perturbation radius, which limits the minimum perturbation granularity, and is set manually, usually an empirical value. t is the current iteration round, T is the maximum iteration round, and ξ is the power factor, which controls the convergence speed and is a constant setting.

[0101] S29, for each main individual, calculate the number of sparks S Ω .

[0102]

[0103] In the formula, S Ωis the number of sparks to be generated by the Ω-th scheduling individual, M is the total spark budget, f(X Ω ) is the fitness value of the Ω-th scheduling individual, f max is the worst fitness value in the current population, ε is a small constant to prevent the denominator from being zero, Y is the total number of scheduling individuals, is the -th fitness value of the scheduling individual.

[0104] Round down the number of sparks, set the minimum and maximum numbers to control the complexity of the single explosion, and pair each number of sparks with the explosion radius.

[0105] S210, traverse each scheduling individual, initialize the spark set, loop Q times for each individual, generate spark individuals in turn, automatically select the current perturbation strategy according to stage t, compare stage t with the preset threshold. If t is less than the threshold, enter the early global search stage and perform uniform explosion perturbation; otherwise, enter the late local convergence stage and use Cauchy-Gaussian mixed perturbation.

[0106]

[0107] In the formula, is the new value of the y-th variable obtained under the current perturbation, x Ωy is the y-th variable value of the Ω-th scheduling individual, rand(0, r t ) is a random number generated between 0 and r t .

[0108]

[0109] In the formula, is the new value of the a-th dimension of the Ω-th individual after the mixed perturbation, used to generate a new solution individual, x k is the k-th variable to be perturbed, which may be discrete values such as task number, storage location layer number, AGV number, etc. Gaussian(1, 1) is a standard normal distribution sample (mean 1, variance 1), used for small-scale and symmetric perturbation. Cauchy(1, 1) is a Cauchy distribution sample (location parameter 1, scale 1), with thick-tail characteristics, used for large-scale jump perturbation. λ is a stage control factor, determining the dominant degree of the perturbation type. Round() is a rounding function to ensure the result is an integer, meeting the requirement of the discrete attribute of the storage scheduling variable being a number type;

[0110]

[0111] In the formula, t is the current iteration number, and T is the total set maximum number of iteration rounds;

[0112] S211, the Cauchy-Gaussian mixed perturbation strategy is used to perform global structural perturbations on the individuals in the initial main population to provide new scheduling solution candidates for elite screening.

[0113] S212, every time a spark individual is generated, a legality check and out-of-bounds repair are immediately performed.

[0114] S213 stores all legal spark individuals into the spark set of the subject to form a spark individual set.

[0115] S214, recalculate each spark individual using the fitness function, bind the spark individual to the new fitness value, and form an extended population.

[0116] S215, sort all individuals in the extended population in ascending order of fitness, intercept the first n individuals to form the t+1th generation main population, retain the current global optimal individual, randomly select M pairs of individuals, and perform partial mapping crossover on their task path encoding to exchange some task segments assigned by AGV and their order, to ensure that each task is still uniquely assigned after the crossover, all tasks are fully covered, and the length of the crossover range is randomly set, with a maximum of no more than 30% of the number of tasks.

[0117] S216, update the path structure of the two individuals after the crossover to form a new intermediate individual set X pmx .

[0118] S217, to X pmx For each individual, set the perturbation probability, randomly select some tasks, and perform mutation operations.

[0119]

[0120] In the formula, It is the new variable value after disturbance, which is used to generate new solution individuals.

[0121] S218, in the genetic evolution stage, the Cauchy-Gaussian perturbation strategy is implemented on the crossover individuals to further optimize the task hierarchy sorting and the detailed layout of the cargo positions. This can enhance the local convergence stability and mutation depth control.

[0122] S219, if the disturbance causes the variable to cross the boundary, immediately perform boundary repair to obtain the individual group X after disturbance optimization mut As a result, there is a better distribution of cargo locations and local sorting of tasks.

[0123] S220, for X mutFor each individual in it, extract the task path of each AGV. Based on the Flying-V storage structure, perform path planning inspection: Use the improved A* algorithm to simulate the movement of AGVs, detect problems such as path intersections, reverse conflicts, and simultaneous lifting deadlocks. If conflicts are found: First, adjust the task order (swap local tasks). If there is still no solution, insert waiting time or a fallback path to obtain an optimized scheduling solution individual group X1. Sort X1 in ascending order according to the fitness value, extract the individual w with the smallest fitness, and compare w with the optimal individual w' recorded last time. If the fitness value of w is less than the optimal individual w', then update w to the global optimal individual.

[0124] Specifically, the "Open Path Multi-Start Asymmetric TSP Problem" (OP-MATSP) is a scheduling modeling method for multi-start, multi-task, and path-cost-asymmetric scenarios. Different from the "closed-loop path" and "symmetric path distance" assumed in the traditional TSP problem, OP-MATSP allows each AGV to start from different positions, not return to the starting point on the path, and the path cost can change due to the spatial structure or task order, especially suitable for multi-channel storage structures in the Flying-V layout. This modeling method can truly reflect the complex coupling characteristics of task allocation and path sorting in a multi-AGV system, significantly improving the expression ability and practicality of the scheduling model. The FWA-GA hybrid optimization algorithm combines the advantages of the "Fireworks Algorithm" and the "Genetic Algorithm", generates search solutions with extensive perturbations through the spark explosion mechanism, and at the same time improves the evolutionary efficiency of the population by means of genetic crossover and mutation. Among them, the "Logistic chaos mapping" is used to generate perturbation variables with chaotic distribution characteristics during the population initialization stage, effectively preventing the initial population from falling into local optima; the "spark explosion radius" uses a dynamic convergence function to control the perturbation amplitude, combined with the fitness-guided spark quantity calculation mechanism, to ensure the dynamic balance between early global exploration and late local convergence. In addition, the "Cauchy-Gaussian hybrid perturbation strategy" is introduced, which combines the long-tailed and high-jump characteristics of the Cauchy distribution with the local smoothing characteristics of Gaussian perturbation to achieve in-depth fine-tuning of the local space and jump exploration of the global space, greatly enhancing the diversity of the optimization algorithm and the solution space coverage ability.

[0125] The scheduling objective function not only considers the traditional total path time (SOC) and the maximum single-vehicle task time (MS), but also introduces a warehouse center of gravity control index (GC). By frequency and weight weighted modeling, the center of gravity stability of goods in the warehouse structure is considered, avoiding the problems of unstable shelves or increased handling costs caused by the concentration of high-frequency heavy goods on the upper floors. The setting of this three-objective scheduling function takes into account efficiency, safety and scheduling balance, making the final optimization result achieve a reasonable balance among executability, system stability and operation timeliness. In terms of scheduling execution, the system simultaneously includes two sub-structures of "task assignment + location mapping" in the individual coding, and controls the crossover and perturbation behaviors of the coding through evolutionary operations to achieve the joint optimization of task path sorting and location layer numbers. During the optimization process, an improved A* path checking mechanism is introduced to detect conflicts and perform feasibility verification on the generated AGV paths, automatically handling path intersections, reverse conflicts and lift resource deadlocks, further enhancing the stability and controllability of the scheduling scheme in actual deployment.

[0126] S3. Implement the optimal scheduling scheme. After the scheduling optimization is completed, compare the predicted goods outbound frequency with the current inventory, automatically generate out-of-stock warnings and procurement plans, and push them to the supply chain control system, and select the optimal supplier to respond to the goods.

[0127] In a specific implementation of the present invention, implementing the optimal scheduling scheme means extracting the optimal task location allocation matrix, AGV path scheduling table, total task path time and maximum AGV task time from the optimal individual coding, converting the location matrix into a visible structure in a three-dimensional coordinate system for inbound / outbound area instruction matching, and converting the AGV path scheduling table into a scheduling queue to implement task push.

[0128] Specifically, by directly parsing the scheduling structure from the individual coding, additional data conversion and secondary calculations are avoided, improving the scheduling efficiency; secondly, through the combination of three-dimensional coordinate mapping and path instruction generation, it is ensured that the scheduling result has a high degree of execution feasibility; thirdly, this structure supports subsequent standardized docking with digital twin systems and AGV scheduling control systems, providing a good interface foundation for multi-platform heterogeneous system integration; fourthly, by extracting the total path time and maximum task time, refined evaluation of the scheduling strategy in different target dimensions can be realized, providing support for dynamic strategy adjustment and system load prediction; finally, the AGV task scheduling queue directly generated based on the scheduling solution can achieve millisecond-level task distribution and path update, significantly improving the task response speed and path conflict avoidance ability of the warehousing system.

[0129] S4. After the scheduling optimization is completed, compare the predicted goods outbound frequency with the current inventory, automatically generate out-of-stock warnings and procurement plans, and push them to the supply chain control system, and select the optimal supplier to respond to the goods.

[0130] Specifically, after the scheduling optimization is completed, the predicted goods outbound frequency is compared with the current inventory level, and out-of-stock warnings and procurement plans are automatically generated and pushed to the supply chain control system. Selecting the optimal supplier for goods response means multiplying the predicted goods outbound frequency by the prediction period to obtain v, and multiplying the current inventory quantity by the goods safety inventory redundancy (set as a multiple of the standard deviation of the outbound frequency) to obtain l. If v is greater than l, it indicates that there is a risk of out-of-stock for the goods, realizing the early quantitative identification of potential out-of-stock risks, significantly improving the pre-warning ability of the warehousing system, reducing logistics delays and production line stagnation caused by inventory shortages. For each out-of-stock good, the pre-registered list of available suppliers in the supply chain platform is automatically called, and the historical average supply cycle, performance success rate, and unit supply price of the suppliers are weighted and summed. This mechanism can not only comprehensively consider supply stability and cost control, but also support dynamic adjustment of weights due to the urgency of materials to meet the intelligent adaptation of multi-level inventory response strategies. Based on the summation score, the suppliers are sorted in descending order, the supplier with the highest score is selected, and a corresponding replenishment task order is immediately generated, including the replenishment goods ID, quantity, latest arrival time, and executing supplier.

[0131] By generating a task order containing "goods ID, required quantity, latest arrival time, and designated executor" for the supplier with the highest score, the system can achieve a closed-loop control of "prediction - judgment - response - execution" from warehousing to the supplier, break down the data barrier between warehousing operations and the supply chain control system, and realize the intelligence and real-time of supply response.

[0132] This embodiment also provides a warehouse goods supply chain management system based on AI big data, as Figure 3 shown. The warehouse goods supply chain management system based on AI big data includes: a data collection module for collecting and preprocessing multi-source warehouse data, where the multi-source warehouse data includes goods basic information and goods location data; a prediction module for predicting the goods outbound frequency based on real-time goods data using an LSTM neural network; a scheduling optimization module for constructing a multi-objective scheduling model combining the picking path time consumption and warehousing center control based on the prediction results; a scheduling solution module for jointly modeling the goods tasks and AGV path planning as an open path multi-start asymmetric TSP problem and using a FWA-GA hybrid optimization algorithm to solve the model to obtain the optimal goods location allocation matrix and the optimal AGV scheduling path table; an implementation module for comparing the predicted goods outbound frequency with the current inventory level after the scheduling optimization is completed, automatically generating out-of-stock warnings and procurement plans and pushing them to the supply chain control system, and selecting the optimal supplier for goods response.

[0133] This embodiment also provides a computer device, which is applicable to the situation of the warehouse goods supply chain management method based on AI big data, 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 warehouse goods supply chain management method based on AI big data as proposed in the above embodiment.

[0134] The computer device can be a terminal. 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. The wireless manner can be achieved through WIFI, a carrier network, NFC (Near Field Communication), or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or a button, a trackball, or a touchpad provided on the housing of the computer device. It can also be an external keyboard, a touchpad, or a mouse, etc.

[0135] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the warehouse goods supply chain management method based on AI big data as 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 (SRAM for short), Electrically Erasable Programmable Read-Only Memory (EEPROM for short), Erasable Programmable Read-Only Memory (EPROM for short), Programmable Read-Only Memory (PROM for short), Read-Only Memory (ROM for short), magnetic memory, flash memory, a magnetic disk, or an optical disc.

[0136] In summary, based on the AI-driven prediction of outbound frequency, the present invention constructs a two-objective scheduling model that takes into account both path time consumption and warehouse center of gravity control. It introduces the FWA-GA hybrid optimization algorithm for solution based on the open path multi-start asymmetric TSP problem. At the same time, it integrates the Cauchy-Gaussian perturbation mechanism to strengthen the local convergence ability. Finally, it realizes the integrated joint optimization of task assignment, path planning, and storage location layout, and further links with the supply chain platform to complete out-of-stock warning and intelligent supplier selection, effectively breaking through the multiple limitations of traditional methods in aspects such as prediction accuracy, scheduling coupling, structural control, and supply coordination.

[0137] In the description of this specification, the descriptions referring to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0138] In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, "a plurality" means at least two, such as two, three, etc., unless otherwise specifically defined.

[0139] Any process or method description shown in the flowchart or described in other ways herein can be understood as representing a module, segment, or part of code including one or more executable instructions for implementing a customized logical function or process. The scope of the preferred embodiments of the present invention includes additional implementations, where the functions can be executed in a substantially simultaneous manner or in the reverse order according to the functions involved, rather than in the order shown or discussed, which should be understood by those skilled in the art to which the embodiments of the present invention belong.

[0140] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a definable sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or in combination with these instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in combination with an instruction execution system, apparatus, or device. More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection portion having one or more wirings (electronic device), a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or otherwise processing as appropriate, and then storing it in a computer memory.

[0141] It should be understood that various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system.

[0142] Those of ordinary skill in the art of this technology can understand that all or part of the steps carried by the method of implementing the above embodiments can be completed by instructing relevant hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiments.

[0143] Furthermore, in each of the various embodiments of the present invention, the functional units can be integrated into one processing module, or each unit can exist physically alone, or two or more units can be integrated into one module. The above integrated module can be implemented in the form of hardware or in the form of a software functional module. When the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0144] The storage medium mentioned above may be a read-only memory, a magnetic disk, an optical disk, etc. Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.

Claims

1. A warehouse goods supply chain management method based on AI big data, characterized in that: Including, Collecting multi-source data of the warehouse and preprocessing it, and predicting the outbound frequency of goods based on the LSTM neural network; The multi-source data of the warehouse includes basic information of goods and goods location data; Based on the prediction results, constructing a multi-objective scheduling model that combines the time-consuming of the picking path and the control of the warehouse center of gravity, jointly modeling the goods task and the AGV path planning as an open-path multi-start asymmetric TSP problem, and using the FWA-GA hybrid optimization algorithm to solve the model to obtain the optimal goods location allocation matrix and the optimal AGV scheduling path table; Implementing the optimal scheduling plan, comparing the predicted outbound frequency of goods with the current inventory after the scheduling optimization, automatically generating a shortage warning and a procurement plan and pushing them to the supply chain control system, and selecting the optimal supplier to respond to the goods.

2. The warehouse goods supply chain management method based on AI big data according to claim 1, wherein: The multi-objective scheduling model that combines the time-consuming of the picking path and the control of the warehouse center of gravity based on the prediction results includes: taking the Flying-V layout as the warehouse infrastructure, combining the outbound frequency, the quality of goods and the structural location, and constructing a two-objective function for scheduling optimization, including minimizing the path time f1 and minimizing the warehouse center of gravity f2.

3. The warehouse goods supply chain management method based on AI big data according to claim 2, characterized in that: The joint modeling of the goods task and the AGV path planning as an open-path multi-start asymmetric TSP problem and using the FWA-GA hybrid optimization algorithm to solve the model to obtain the optimal goods location allocation matrix and the optimal AGV scheduling path table includes: Extracting all the goods tasks to be scheduled to form a task set A, recording the current number of online AGVs, initializing the current position of each AGV, and calculating the corresponding path cost d as the task movement cost input for any two points in the task set; Constructing a scheduling objective function F by combining the overall efficiency and the response time of a single AGV and adding scheduling constraints; Setting the constraint conditions that each task must be assigned and only assigned to one AGV, and the goods locations of all tasks must meet the physical capacity of the warehouse; Constructing an initial feasible solution population, setting the total number of individuals and the maximum number of iterations, and each individual represents a scheduling plan; Using the Logistic chaotic map to generate an initialization perturbation sequence; Performing integer encoding on the first O variables, mapping them to the AGV number and execution priority to which the task belongs, and mapping the latter O variables to the layer number where the goods are located, and decoding the task scheduling result and the goods location allocation into a task path table and a goods location allocation matrix; Calculating the fitness using the scheduling objective function as the fitness function, sorting the individuals in descending order according to the fitness, performing a spark explosion operation on the first U individuals to generate a perturbed candidate scheduling solution; Setting the maximum perturbation radius of the spark explosion and calculating the explosion radius of the t-th generation; For each main individual, calculate the number of sparks S according to the corresponding fitness value and the worst fitness in the population Ω ; Traversing each scheduling individual, initializing the spark set, looping Q times for each individual, generating spark individuals in turn, automatically selecting the current perturbation strategy according to the stage t, comparing the stage t with the preset threshold, if t is less than the threshold, then entering the early global search stage and performing a uniform explosion perturbation; Otherwise, entering the late local convergence stage and using the Cauchy-Gaussian hybrid perturbation; Use the Cauchy-Gaussian mixed perturbation strategy to perform global structural perturbation on the individuals of the initial main population, providing new scheduling solution candidates for elite screening; Immediately conduct legality inspection and out-of-bounds repair for each generated spark individual; Store all legal spark individuals in the spark set of this subject to form a spark individual set; For each spark individual, recalculate it using the fitness function, bind the spark individual with the new fitness value to form an extended population; Sort all individuals in the extended population in ascending order of fitness, intercept the first n individuals to form the (t + 1)-th generation main population, retain the current global optimal individual, randomly select M pairs of individuals, and perform partial mapping crossover on their task path encodings; Update the path structures of the two individuals after crossover to form a new set X of intermediate individuals pmx ; For X pmx For each individual in pmx , set the perturbation probability, randomly select some tasks, and perform the mutation operation; Execute the Cauchy-Gaussian perturbation strategy on the crossed individuals in the genetic evolution stage to further optimize the task level sorting and the fine layout of storage locations; If the perturbation causes a variable to exceed the boundary, immediately perform boundary repair to obtain the population X of individuals after perturbation optimization mut ; For each individual in X mut Extract the task path of each AGV, perform path planning inspection based on the Flying-V storage structure to obtain the optimized scheduling solution individual group X1. Sort X1 in ascending order according to the fitness value, extract the individual w with the smallest fitness, and compare w with the previously recorded optimal individual w'. If the fitness value of w is less than the optimal individual w', then update w to the global optimal individual.

4. The method for warehouse goods supply chain management based on AI big data according to claim 3, characterized in that: The implementation of the optimal scheduling plan refers to extracting the optimal task storage location allocation matrix, AGV path scheduling table, total task path time, and maximum AGV task time from the optimal individual encoding, converting the storage location matrix into a visible graph structure in a three-dimensional coordinate system for inbound / outbound area instruction matching, and converting the AGV path scheduling table into a scheduling queue to implement task pushing.

5. The warehouse goods supply chain management method based on AI big data according to claim 4, characterized in that: After the scheduling optimization is completed, compare the predicted goods outbound frequency with the current inventory, automatically generate a shortage warning and a procurement plan, and push them to the supply chain control system. Selecting the optimal supplier for goods response means comparing based on the prediction results and real-time data, analyzing the goods shortage risk. For each out-of-stock good, automatically call the list of available suppliers pre-registered in the supply chain platform, perform a weighted sum of the historical average supply cycle, fulfillment success rate, and unit supply price of the suppliers, sort the suppliers in descending order based on the sum score, select the supplier with the highest score, and immediately generate the corresponding replenishment task order.

6. The method for managing the warehouse goods supply chain based on AI big data according to claim 5, characterized in that: The collection and preprocessing of multi-source data in the warehouse include: Before all goods are warehoused, a unique RFID tag is bound to each piece of goods, and the triple of the unique number of the goods, the timestamp of warehousing / outbound, and the current location coordinates (x i , y i , z i ) is obtained in real time, corresponding to the row number, layer number, and column number of the shelf; Collect the current goods weight, real-time inventory, and generate an inbound / outbound frequency sequence and the current storage coordinates of the goods in combination with the inbound / outbound time; Through the data interface docking with the ERP and WMS systems, each good can obtain the corresponding supplier attributes, including the supply cycle, fulfillment rate, and unit procurement cost; Perform missing value processing and normalization on the collected data.

7. The method for managing the warehouse goods supply chain based on AI big data according to claim 6, wherein: The prediction of the goods outbound frequency based on the LSTM neural network means inputting the real-time goods outbound data into the LSTM model to obtain the future goods outbound frequency.

8. A warehouse goods supply chain management system based on AI big data, based on the method for managing the warehouse goods supply chain based on AI big data according to any one of claims 1 to 7, characterized in that: Including, A data collection module for collecting real-time goods data and performing preprocessing; A prediction module for predicting the goods outbound frequency; A scheduling optimization module for performing multi-objective modeling; A scheduling solution module for solving the optimal scheduling result through FWA-GA; An implementation module for performing real-time task scheduling based on the optimal scheduling result; A response module for performing shortage warning and supplier response.

9. 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, it implements the steps of the AI big data-based warehouse goods supply chain management method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, it implements the steps of the AI big data-based warehouse goods supply chain management method according to any one of claims 1 to 7.

Images