Inventory distribution decision-making method and system based on dynamic multi-objective optimization

Abstract

The application provides an inventory distribution decision method and system based on dynamic multi-objective optimization, wherein the method comprises the following steps: obtaining real-time order data, cargo weight data, historical demand data and environment state data of target goods; generating a distribution adjustment strategy of the target goods on the shelves based on the cargo weight data and in combination with a preset shelf load bearing constraint condition; combining the distribution adjustment strategy, the real-time order data, the historical demand data and the environment state data to construct a multi-objective decision model with the optimization objectives of minimizing the replenishment cycle, minimizing the warehouse cost and optimizing the risk balance; and outputting an optimal inventory distribution decision scheme in an interactive iterative manner based on the multi-objective decision model. The application improves the inventory turnover efficiency and reduces the operation risk.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent warehouse management, and particularly relates to an inventory distribution decision method and system based on dynamic multi-objective optimization. BACKGROUND

[0002] In e-commerce warehouse management, inventory distribution decision needs to consider multiple objectives such as replenishment efficiency, warehouse cost and shelf stability. Especially during the peak of promotion, the demand for commodities fluctuates greatly and the warehouse environment changes rapidly. Traditional static optimization methods are difficult to adapt to dynamic scenarios, and an intelligent decision-making scheme that can respond to order changes, environmental state and shelf load constraints in real time is urgently needed.

[0003] Currently, there is an inventory scheduling method based on single-objective optimization. This method predicts replenishment demand through historical sales, generates a fixed-period replenishment plan in combination with shelf load restrictions, and uses linear programming to solve the distribution scheme with the lowest warehouse cost. The system adjusts the storage location of goods according to preset rules to ensure that the single-layer load does not exceed the limit.

[0004] This scheme only takes warehouse cost as the optimization target, and replenishment cycle and risk control as fixed constraints, resulting in a scheme generated by the system that often sacrifices replenishment timeliness to meet cost requirements when orders surge or the environment is abnormal, and lacks dynamic evaluation of long-term shelf stability, resulting in a high risk of replenishment delay and shelf safety hazards in actual operation. SUMMARY

[0005] The present application provides an inventory distribution decision method and system based on dynamic multi-objective optimization to solve the problems of low inventory turnover efficiency and high operational risk in the prior art.

[0006] In a first aspect, the present application provides an inventory distribution decision method based on dynamic multi-objective optimization, comprising:

[0007] obtaining real-time order data, product weight data, historical demand data and environmental state data of target goods;

[0008] Based on the product weight data, in combination with the preset shelf load constraint condition, a distribution adjustment strategy of the target goods on the shelf is generated;

[0009] The distribution adjustment strategy, the real-time order data, the historical demand data and the environmental state data are combined to construct a multi-objective decision model with the optimization objectives of minimizing the replenishment cycle, minimizing the warehouse cost and optimizing the risk balance;

[0010] Based on the multi-objective decision model, an optimal inventory distribution decision scheme is output in an interactive iterative manner.

[0011] Optionally, the distribution adjustment strategy, the real-time order data, the historical demand data, and the environment state data are combined to construct a multi-objective decision model with optimization objectives of minimizing replenishment cycle, minimizing warehouse cost, and optimizing risk balance, including:

[0012] According to the distribution adjustment strategy, migration instructions are generated, and the migration instructions are converted into warehouse space occupation parameters;

[0013] According to the real-time order data and the historical demand data, replenishment urgency parameters are calculated and generated;

[0014] Based on the environment state data, in combination with the shelf stability change trend, a warehouse risk coefficient is generated;

[0015] The replenishment urgency parameters, the warehouse space occupation parameters, and the warehouse risk coefficient are used as variables to establish a multi-objective decision model.

[0016] Optionally, the replenishment urgency parameters, the warehouse space occupation parameters, and the warehouse risk coefficient are used as variables to establish a multi-objective decision model, including:

[0017] Based on the replenishment urgency parameters, a first function representing the minimization of the replenishment cycle is constructed, based on the warehouse space occupation parameters, a second function representing the minimization of the warehouse cost is constructed, and based on the warehouse risk coefficient, a third function representing the optimization of the risk balance is constructed;

[0018] According to the replenishment urgency parameters and the benchmark replenishment cycle, a fluctuation range constraint condition of the replenishment cycle is constructed;

[0019] Based on the warehouse space occupation parameters, a boundary condition of the warehouse cost is generated;

[0020] According to the dynamic change rate of the warehouse risk coefficient, a stability threshold of the risk balance is set;

[0021] The fluctuation range constraint condition, the boundary condition, and the stability threshold are integrated into an optimization framework containing the first function, the second function, and the third function;

[0022] In the optimization framework, a bidirectional conflict resolution mechanism for coordinating target conflicts is established to form a multi-objective decision model.

[0023] Optionally, in the optimization framework, a bidirectional conflict resolution mechanism for coordinating target conflicts is established to form a multi-objective decision model, including:

[0024] A feedback association relationship between the replenishment cycle and the warehouse cost is established;

[0025] establish a constraint intervention relationship of the feedback correlation relationship by balancing the risk;

[0026] establish a dynamic coupling relationship between the warehouse cost and the risk balance target;

[0027] Based on the feedback correlation relationship, the constraint intervention relationship and the dynamic coupling relationship, a bidirectional conflict resolution mechanism is constructed;

[0028] In the optimization framework, the bidirectional conflict resolution mechanism is used to real-time coordinate the conflict between the replenishment cycle, the warehouse cost and the risk balance, and generate a multi-objective decision model.

[0029] Optionally, based on the goods weight data, a distribution adjustment strategy of target goods on the shelf is generated in combination with a preset shelf load bearing constraint condition, comprising:

[0030] Identify the current load bearing distribution state of the goods weight data in the multi-layer space of the shelf;

[0031] Compare the current load bearing distribution state with the preset shelf load bearing constraint condition layer by layer, and identify the over-limit area that exceeds the preset load bearing threshold;

[0032] According to the position distribution and over-limit degree of the over-limit area, a goods migration scheme from the over-limit area to the non-over-limit area is generated;

[0033] In combination with the environment state data, adjust the migration path and migration order in the goods migration scheme, and take the adjusted goods migration scheme as the distribution adjustment strategy.

[0034] Optionally, the goods migration scheme from the over-limit area to the non-over-limit area is generated according to the position distribution and over-limit degree of the over-limit area, comprising:

[0035] Based on the position distribution of the over-limit area, determine the spatial distribution topological relationship of the over-limit goods;

[0036] According to the over-limit degree of the over-limit area, calculate the migration amount;

[0037] From all non-over-limit areas, select a target area with a load bearing surplus greater than the migration amount;

[0038] Based on the spatial distribution topological relationship and the target area, a goods migration scheme is generated.

[0039] Optionally, based on the multi-objective decision model, an optimal inventory distribution decision scheme is output in an interactive iterative manner, comprising:

[0040] Based on the multi-objective decision model, an initial decision solution set is generated by an initialization algorithm as the input of the first interactive iteration;

[0041] In each interactive iteration, a real-time change amount of the environment state data and the distribution adjustment strategy are taken as dynamic constraint conditions;

[0042] A non-inferior solution subset satisfying the dynamic constraint conditions is screened from the decision solution set;

[0043] It is judged whether a solution space change amount of the non-inferior solution subset is less than a preset convergence threshold;

[0044] If not, the non-inferior solution subset is expanded to obtain a non-inferior solution set, and the non-inferior solution set is taken as the decision solution set to repeat the iteration process until the solution space change amount of the non-inferior solution subset is less than the preset convergence threshold or a maximum iteration number is reached, and an optimal inventory distribution decision scheme is generated.

[0045] In a second aspect, the application provides an inventory distribution decision system based on dynamic multi-objective optimization, comprising:

[0046] An acquisition module is configured to acquire real-time order data, goods weight data, historical demand data and environment state data of a target commodity in a scenario where a real-time demand fluctuation amount of an e-commerce warehouse is greater than a preset fluctuation amount.

[0047] A generation module is configured to generate a distribution adjustment strategy of the target commodity on a shelf based on the goods weight data and in combination with a preset shelf load bearing constraint condition.

[0048] A construction module is configured to combine the distribution adjustment strategy, the real-time order data, the historical demand data and the environment state data to construct a multi-objective decision model with the optimization objectives of minimizing a replenishment cycle, minimizing a warehouse cost and optimizing a risk balance.

[0049] An output module is configured to output an optimal inventory distribution decision scheme in an interactive iteration manner based on the multi-objective decision model.

[0050] In a third aspect, the application provides a computing device comprising a processor and a memory, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the inventory distribution decision method based on dynamic multi-objective optimization according to any one of the first aspect.

[0051] In a fourth aspect, the application provides a computer storage medium storing computer program instructions, wherein the computer program instructions are executed by a processor to implement the inventory distribution decision method based on dynamic multi-objective optimization according to any one of the first aspect.

[0052] In the present application, an inventory distribution decision-making method based on dynamic multi-objective optimization is provided, which comprises the following steps: acquiring real-time order data, goods weight data, historical demand data and environment state data of target goods; generating a distribution adjustment strategy of the target goods on the shelf based on the goods weight data and in combination with a preset shelf load bearing constraint condition; combining the distribution adjustment strategy, the real-time order data, the historical demand data and the environment state data to construct a multi-objective decision-making model with the optimization objectives of minimizing the replenishment cycle, minimizing the warehouse cost and optimizing the risk balance; and outputting an optimal inventory distribution decision-making scheme in an interactive iterative manner based on the multi-objective decision-making model.

[0053] The technical scheme provided by the present application has the following beneficial effects:

[0054] The present application provides comprehensive and dynamic data support for decision-making by collecting orders, weights, historical demand and environment data in real time, ensuring that the system can respond to various changes in a timely manner. The goods position distribution is dynamically optimized based on the shelf load bearing constraint, which improves the space utilization rate while effectively avoiding the risk of shelf overloading. The replenishment cycle, cost and risk objectives are integrated to achieve collaborative optimization, improve inventory turnover efficiency and reasonably control operating costs. Through dynamic constraint adjustment and non-inferior solution screening, an optimal scheme that adapts to real-time changes is generated, greatly improving the decision-making accuracy.

[0055] Further, the present application also converts the distribution adjustment strategy into space occupation parameters, generates a replenishment urgency parameter in combination with order data, and generates a risk coefficient in association with environment data, to establish a multi-objective decision-making model with the three as variables, realizing the closed-loop conversion from data to model.

[0056] Moreover, through parameterized mapping and multi-objective coupling, the replenishment efficiency and warehouse cost are dynamically balanced on the basis of ensuring shelf safety, improving order processing capacity and reducing equipment maintenance requirements.

[0057] These aspects or other aspects of the present application will be more apparent in the following description of the embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0058] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0059] Figure 1 A flowchart of an inventory distribution decision-making method based on dynamic multi-objective optimization provided by an embodiment of the present application;

[0060] Figure 2 A structural schematic diagram of an inventory distribution decision system based on dynamic multi-objective optimization is provided for an embodiment of the present application.

[0061] Figure 3 A structural schematic diagram of a computing device is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0062] In order for those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application.

[0063] In some of the processes described in the specification and claims of the present application and in the above-described drawings, a plurality of operations appear in a specific order, but it should be clearly understood that these operations can be executed or in parallel without the order in which they appear in this text, and the serial numbers of the operations, such as 101, 102, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes can include more or fewer operations, and the operations can be executed in sequence or in parallel. It should be noted that the "first", "second", etc. in this text are used to distinguish different messages, devices, modules, etc., and do not represent the order, nor do "first" and "second" represent different types.

[0064] In the field of intelligent scheduling of e-commerce warehouses, the existing single-objective optimization scheme has limitations: it takes warehouse cost as a single optimization target, and treats replenishment cycle and risk control as fixed constraints, resulting in that the scheme generated by the system is often forced to extend the replenishment cycle to meet the cost requirements when facing order surges or environmental abnormalities, and at the same time, it lacks a dynamic evaluation mechanism for shelf stability, causing a double compromise of operational efficiency and safety. This defect is due to the fragmented handling of the multi-objective coordination relationship by the static optimization model, which is difficult to adapt to the dynamic complexity of modern warehouses.

[0065] To solve the above problems, the application provides an inventory distribution decision method based on dynamic multi-objective optimization. The method fuses real-time order data, shelf load status and environmental monitoring information to build a multi-objective decision model balancing replenishment cycle, storage cost and risk. The method first quantizes the product distribution strategy as a space occupation parameter, generates a dynamic urgency index combined with order fluctuation characteristics, and establishes a risk prediction mechanism associated with environmental data. Then, an interactive iterative algorithm is used to automatically balance the conflict between the three objectives under the premise of meeting the shelf load hard constraint, and output an optimal inventory distribution decision scheme that adapts to real-time working conditions. The scheme breaks through the limitations of traditional single-objective optimization, and through multi-source data fusion and dynamic weight adjustment mechanism, it not only guarantees the timely response capability of peak period orders, but also realizes the collaborative control of storage cost and safety risk, fundamentally solving the efficiency loss and safety hazard problems caused by the fragmentation of existing technology targets.

[0066] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the application.

[0067] Figure 1 A flowchart of an inventory distribution decision method based on dynamic multi-objective optimization provided by the embodiments of the application is shown in Figure 1 The method comprises the following steps.

[0068] Step 101: Obtain real-time order data, product weight data, historical demand data and environmental state data of the target goods.

[0069] In step 101, the real-time order data refers to the information of the goods orders to be processed at the current time, including the type, quantity and delivery time requirement of the goods, which is used to reflect the immediate market demand. The product weight data represents the actual weight of the goods and the distribution state on the shelf, which is used to evaluate the shelf load. The historical demand data represents the sales law and fluctuation characteristics of the goods in the past period, which is used to predict the replenishment demand. The environmental state data represents the physical parameters such as temperature, humidity and shelf vibration amplitude in the warehouse, which is used to evaluate the storage safety risk.

[0070] In the embodiments of the present application, the system collects order information of commodities, shelf weight sensor data and environmental monitoring data in real time through Internet of Things devices, and simultaneously retrieves historical sales records from a database. After cleaning, the order data extracts effective demand features, the product weight data is stored by shelf layer and zone, the historical data extracts periodic rules through time series analysis, and the environmental data is uploaded to the central processor in real time by a distributed sensor network. After standardization, the four types of data form structured input, providing multi-dimensional basis for subsequent decision-making.

[0071] For example, a certain e-commerce North China warehouse center monitors a sudden increase in real-time order quantity of air conditioner commodities during a promotion period, the shelf pressure sensor shows that the total weight of the refrigerators stored on the top layer exceeds the standard value, and the temperature and humidity sensor detects local high temperature. The system synchronously retrieves the sales data of this commodity in the past three years in June, analyzes its historical fluctuation rules, and generates a comprehensive data set combining the current environmental state. Among them, the real-time order quantity is directly obtained through the order management system, the shelf weight data is collected by the weighing sensor installed on each layer, the historical data is exported from the enterprise ERP system, and the environmental data is uploaded by the Internet of Things terminal deployed in the warehouse every five minutes.

[0072] Step 102: Based on the product weight data, a distribution adjustment strategy of the target commodity on the shelf is generated in combination with a preset shelf load bearing constraint condition.

[0073] In step 102, the shelf load bearing constraint condition represents the maximum allowable load weight and distribution density limit of each layer of the shelf, which is used to ensure warehouse safety. The distribution adjustment strategy includes the operation scheme of the migration path, sequence and quantity of goods, which is used to optimize the utilization rate of shelf space.

[0074] In the embodiments of the present application, first, the real-time weight distribution of each layer of the shelf is analyzed, which is compared with the preset load threshold layer by layer, and the over-limit area and its over-limit weight are marked. According to the over-limit degree, the number of goods to be migrated is calculated, and in combination with the shelf space topological relationship, the over-limit area with sufficient load remaining is selected as the target position. When generating the migration instruction, the area with serious over-limit is processed first, and the path planning is adjusted according to the environmental data to avoid high temperature or vibration abnormal area. Finally, the distribution adjustment strategy containing specific migration steps is output.

[0075] For example, the system detects that the current load of the top layer of the shelf exceeds the threshold value, and the over-limit weight of the refrigerators stored in a certain area in the region is 80 kg. According to the three-dimensional model of the shelf, two idle areas in the middle layer are selected as the migration target, and the remaining load is 50 kg and 30 kg, respectively. The generated instruction requires two-time migration: first, move the 50 kg refrigerator to the left area in the middle layer, and then move the remaining 30 kg to the right area in the middle layer. Since the environmental monitoring shows that the temperature of part of the path is high, the system automatically adjusts the transportation route to bypass the central channel.

[0076] Step 103: Combine the distribution adjustment strategy, real-time order data, historical demand data, and environmental state data to build a multi-objective decision-making model with the optimization objectives of minimizing replenishment cycle, minimizing warehouse cost, and optimizing risk balance.

[0077] In step 103, the generation process of the replenishment cycle target: based on the fluctuation characteristics of real-time order data and historical demand data, generate a replenishment urgency parameter, calculate the target value through the formula base replenishment cycle / urgency parameter, the influencing factors include order surge amplitude and historical sales periodicity, the correlation is that the larger the urgency parameter, the shorter the target replenishment cycle. The generation process of the warehouse cost target: calculate according to the space occupation parameter transformed from the weight data of goods and the load state of transportation nodes, quantify through the formula space density x transportation cost coefficient + load penalty term, the influencing factors include shelf space utilization and distribution center processing capacity, the correlation is that the higher the space density or the larger the load, the higher the cost target value. The generation process of the risk balance target: generate a warehouse risk coefficient through environmental state data and shelf physical state change, set a stability target interval combined with a safety threshold, the influencing factors include temperature and humidity change rate and shelf vibration amplitude, the correlation is that when the risk coefficient exceeds the threshold, the target interval is automatically tightened. The multi-objective decision-making model is a decision-making framework that coordinates the three targets of replenishment cycle, warehouse cost, and risk balance through mathematical optimization methods, among which the replenishment cycle target requires the fastest response speed, the warehouse cost target pursues the lowest operating cost, and the risk balance target ensures shelf safety and environmental stability. The model takes the space parameter converted from the distribution adjustment strategy, the urgency parameter generated from the order data, and the risk coefficient derived from the environmental data as input variables, establishes a multi-objective function system, and outputs a Pareto optimal solution set that balances the three conflicting targets through dynamic weight allocation and constraint coupling mechanism.

[0078] In the embodiments of the present application, the migration instruction in the distribution adjustment strategy is converted into a space occupation parameter, reflecting the change of shelf density; the replenishment urgency parameter is calculated according to the deviation of real-time order and historical data; the warehouse risk coefficient is derived based on the trend of environmental data. Taking the three as input variables, an optimization model containing three objective functions is established: Where P is the replenishment urgency parameter, S is the space density, L is the transportation load rate, AV is the risk coefficient change, and w, k are weight coefficients. The three targets are associated through a dynamic constraint network to form a collaborative optimization framework.

[0079] For example, the replenishment urgency parameter of the current air conditioner commodity is calculated as 2.0, the shelf space density parameter is 90 kg / m 2, the transport load rate is 75%, and the risk coefficient change amount is 0.15. When the model is constructed, the weight coefficients w1=0.6, w2=0.4 are set, the sensitive coefficient k=2, and the following objective functions are generated: the replenishment cycle target f1=0.5, the warehouse cost target f2=0.6*90+0.4*75=84, and the risk balance target f3=0.77. The model finds a balanced solution through iterative calculation.

[0080] Step 104: Based on the multi-objective decision-making model, an optimal inventory distribution decision-making scheme is output in an interactive iterative manner.

[0081] In step 104, the optimal inventory distribution decision-making scheme is a final execution strategy selected from the Pareto solution set generated by the multi-objective model. The scheme comprehensively considers the balance point of replenishment efficiency, cost control, and risk management under the premise of meeting the shelf load and environmental safety hard constraints, and specifically includes a product storage location adjustment scheme, a replenishment time arrangement, and environmental control requirements. It is a set of operational instructions for realizing dynamic optimization of warehouse.

[0082] In the embodiments of the present application, a first solution set containing random feasible solutions is initialized, and in each iteration, the constraint conditions are updated according to the latest environmental data and order changes, and non-inferior solutions that meet the current constraints are selected. By calculating the spatial distribution density of the solution set, solutions with high diversity are retained. When the solution space change amount of continuous multiple iterations is less than a threshold value, the solution with the largest crowding distance is selected as the final scheme.

[0083] For example, after five iterations of the model, the non-inferior solution set converges from the initial 50 groups to 15 groups. The current optimal solution is: replenishment cycle 1.2 days, warehouse cost 78 units, and risk coefficient 1.1. The system automatically generates execution instructions, including replenishment batch arrangement, shelf distribution adjustment diagram, and environmental control requirements.

[0084] This method realizes intelligent decision-making of inventory distribution through multi-source data fusion and dynamic multi-objective optimization. Under the premise of ensuring shelf safety, the replenishment response speed and warehouse space utilization rate are improved, and the operating risk is effectively controlled. The system can adapt to order fluctuations and environmental changes, providing efficient and reliable decision support for modern e-commerce warehouses.

[0085] To solve the technical problem of multi-objective collaborative optimization in e-commerce warehouses, in some embodiments, step 103: the distribution adjustment strategy, the real-time order data, the historical demand data, and the environmental state data are combined to construct a multi-objective decision-making model with the optimization objectives of minimizing the replenishment cycle, minimizing the warehouse cost, and optimizing the risk balance, including:

[0086] Step 201: According to the distribution adjustment strategy, generate migration instructions, and convert the migration instructions into warehouse space occupation parameters.

[0087] In step 201, the migration instruction is a standardized operation command generated according to the shelf out-of-limit area analysis, which contains the specific location of the goods to be migrated, the target storage area, the migration quantity and the priority ranking. The instruction converts the distribution adjustment strategy into a set of executable goods displacement operations by quantifying the out-of-limit degree and the space topology relationship, and is the basic action unit for realizing the dynamic optimization of the shelf. The warehouse space occupation parameter is a warehouse space utilization efficiency index quantified by the goods migration instruction, which reflects the actual load state of each area of the shelf. The parameter contains two dimensions of unit area bearing value and distribution uniformity, which is used to evaluate the utilization efficiency of the warehouse space.

[0088] In the embodiment of the present application, the system parses the migration instruction in the distribution adjustment strategy, extracts the weight and location information of the goods to be migrated, and calculates the space density change of each area before and after migration according to the shelf grid coordinates. By mapping the migration path to the three-dimensional model of the shelf, the bearing value of each layer and each area is dynamically updated, and finally the space occupation parameter containing the density distribution matrix is generated, providing a quantitative basis for cost optimization.

[0089] Step 202: Calculate and generate a replenishment urgency parameter according to the real-time order data and the historical demand data.

[0090] In step 202, the replenishment urgency parameter is a dynamic index generated by integrating real-time order fluctuations and historical rules, which represents the emergency degree of current replenishment demand. This parameter overcomes the lag defect of static prediction by integrating immediate sales trends and periodic characteristics.

[0091] In the embodiment of the present application, first, the real-time order data is analyzed by sliding window, and the deviation amplitude of the current sales from the historical same period average is calculated; at the same time, the seasonal and periodic characteristics of the historical demand data are extracted as a reference. The real-time fluctuation amplitude and the historical rule coefficient are weighted and integrated to generate a replenishment urgency parameter with time sequence sensitivity, which is used to dynamically adjust the replenishment cycle target. The specific process is as follows: extract the periodic characteristics of the historical demand data, calculate the baseline daily sales μ and the standard deviation σ; obtain the real-time order data, calculate the current sales x; calculate the fluctuation coefficient α=(x-μ) / σ; the replenishment urgency parameter P=β×(1+α), where β is the historical adjustment coefficient. Embodiment: The historical 6-month daily sales of air conditioner commodities μ=120 units, σ=30 units; the real-time sales x=200 units; then α=(200-120) / 30=2.67; take β=1.2, P=1.2×3.67=4.4.

[0092] Step 203: Generate a warehouse risk coefficient based on the environmental state data and in combination with the shelf stability change trend.

[0093] In step 203, the shelf stability change trend refers to the dynamic characteristics of the evolution of the shelf structure safety state over time, which is obtained by continuously monitoring the associated changes of environmental parameters (temperature and humidity, vibration) and physical state (deformation, load distribution). Specifically, it is derived from the fluctuation law obtained by time series analysis of the data collected by the environmental sensor in real time, and the cumulative trend of micro-deformation recorded by the shelf strain gauge. After weighted fusion, the stability evaluation curve is formed. The warehouse risk coefficient is a composite index for quantifying the correlation between environmental state and shelf stability. This coefficient predicts potential operational risks by monitoring dynamic parameters such as temperature and humidity gradient, vibration amplitude, and combining with shelf physical deformation data.

[0094] In the embodiments of the present application, the temperature and humidity data collected by the environmental sensor in real time is filtered and processed, and the rate of change over time is calculated; at the same time, the micro-deformation after load adjustment is monitored by the shelf strain gauge. The environmental change rate and physical deformation are normalized, and the comprehensive risk coefficient is generated according to the preset weight formula as the safety constraint boundary of the model.

[0095] Step 204: Establish a multi-objective decision-making model with the replenishment urgency parameter, the warehouse space occupancy parameter, and the warehouse risk coefficient as variables.

[0096] In the embodiments of the present application, the space occupancy parameter is mapped as the input variable of the warehouse cost function, the replenishment urgency parameter is converted into the adjustment coefficient of the periodic target, and the risk coefficient is used as the boundary value of the constraint condition. By using the nonlinear programming method to construct the correlation matrix of the three objectives, the interactive weight adjustment mechanism is used to balance the target conflict, and finally a decision-making model that can dynamically respond is formed.

[0097] The following is a specific example:

[0098] A large e-commerce warehouse center during the air conditioning sales season, the system monitoring A commodity area real-time order volume increased by 150% than usual, which top shelf storage of B class goods total weight of 580 kg, more than the default 500 kg threshold. Through the analysis of the three-dimensional model of the shelf, determine the overrun area concentrated in the coordinates X10-Y15 to X15-Y20 range, overrun weight according to the formula 80 kg = measured weight - bearing threshold. According to the distribution adjustment strategy, the system generates migration instructions: 2 sets of total weight of 96 kg of B goods are migrated twice, the first 48 kg to the middle layer left area coordinates X5-Y8 to X8-Y11, the remaining 48 kg to the bottom right area coordinates X3-Y5 to X6-Y8, the migration path is automatically adjusted to bypass the central channel due to the monitoring of the X12-Y14 area temperature of 38℃. At the same time, the system calls the A commodity sales data in the same period for three years, the calculation of the baseline daily sales of 120 units, the standard deviation of 30 units, the current sales of 200 units corresponding to the replenishment urgency parameter is calculated according to the formula P = β × (1 + (x-μ) / σ) as 4.4, where β takes the historical adjustment coefficient 1.2. Environmental monitoring shows that the current temperature of 38℃ is 0.086 higher than the safety threshold of 35℃, and the risk contribution value of the shelf vibration amplitude of 0.12mm is 0.04. The comprehensive warehouse risk coefficient is 1.126. With the space density of 90 kg / m 2 , the transportation load rate of 75%, a multi-objective function is established: the replenishment cycle target f1 = 1 / P = 0.227, the warehouse cost target f2 = w1S + w2L = 0.6 × 90 + 0.4 × 75 = 84, and the risk balance target f3 = 1 / (1+kΔV) = 0.816, where k is the risk sensitivity coefficient 2, and ΔV is the risk coefficient variation 0.15.

[0099] In the embodiments of the present application, the method realizes intelligent decision of inventory distribution strategy through parameterized modeling and dynamic optimization, improves the replenishment response capability and warehouse resource utilization rate on the basis of ensuring the safety of the shelf, and provides reliable technical support for inventory management in complex environment.

[0100] In order to solve the problem of multi-objective collaborative optimization in e-commerce warehouse, in some embodiments, step 204: taking the replenishment urgency parameter, the warehouse space occupation parameter and the warehouse risk coefficient as variables, a multi-objective decision model is established, including:

[0101] Step 301: constructing a first function representing the minimum replenishment cycle based on the replenishment urgency parameter, constructing a second function representing the minimum warehouse cost based on the warehouse space occupation parameter, and constructing a third function representing the optimization of risk balance based on the warehouse risk coefficient.

[0102] In step 301, the first function is a mathematical expression reflecting the replenishment efficiency, and the smaller the value, the faster the replenishment response. The second function is a mathematical expression quantifying the warehouse operating cost, and the smaller the value, the better the cost control. The third function is a mathematical expression evaluating the safety risk, and the larger the value, the higher the system stability

[0103] In the embodiment of the present application, the system first converts the replenishment urgency parameter into an adjustment coefficient of a periodic function, so that the function value dynamically changes with demand fluctuations; then decomposes the space occupation parameter into storage density and transportation load, and constructs a linearly weighted cost function; finally, the risk coefficient is processed by taking its reciprocal, and is converted into a stability function for maximum optimization. The three functions are kept dimensionally consistent through normalization processing.

[0104] Step 302: According to the replenishment urgency parameter and the reference replenishment period, a fluctuation range constraint condition of the replenishment period is constructed.

[0105] In step 302, the reference replenishment period refers to the standard replenishment time interval of a commodity under normal sales conditions, and its value is derived from statistical analysis of historical sales data of the commodity, and is specifically determined by calculating the ratio of the average inventory consumption rate to the safety stock level in the past multiple sales periods. The system will modify it in combination with factors such as commodity characteristics and supplier delivery period, and use it as a reference value for the optimization model. The fluctuation range constraint condition refers to the upper and lower limit interval within which the replenishment period is allowed to vary, and its boundaries are determined by the reference period and the urgency parameter, and is used to limit the reasonable range of the optimization solution.

[0106] In the embodiment of the present application, the reference replenishment period is set according to the characteristics of the commodity, and is multiplied or divided by the replenishment urgency parameter to obtain the dynamic constraint interval. When the order surges, the urgency parameter increases, causing the constraint interval to move up, allowing a shorter replenishment period; otherwise, it moves down, ensuring that the replenishment frequency is not too high.

[0107] Step 303: Based on the warehouse space occupation parameter, a boundary condition of the warehouse cost is generated.

[0108] In step 303, the boundary condition refers to the floating range allowed for the warehouse cost target, and its upper and lower limits are dynamically adjusted according to the space occupation parameter and the current transportation resource status, and is used to balance the storage efficiency and logistics capacity.

[0109] In the embodiment of the present application, the basic value of the cost boundary is calculated by analyzing the space density distribution of each area of the shelf and combining the real-time load rate of the transportation node. When the density of a certain area is too high or the transportation load is too large, the corresponding boundary condition is automatically relaxed to avoid excessive constraints leading to no feasible solution.

[0110] Step 304: According to the dynamic change rate of the warehouse risk coefficient, a stability threshold for risk balance is set.

[0111] In step 304, the dynamic change rate of the warehouse risk coefficient refers to the degree of change of the risk coefficient over time, which is obtained by calculating the relative change amount of the current risk coefficient and the previous value. Specifically, a sliding time window algorithm is used to perform linear regression analysis on the risk coefficients of the latest number of monitoring periods, and the slope is taken as the change rate index, which is used to judge the degree of change of the risk development trend. The stability threshold refers to the minimum safety standard that the risk balance target must meet, and its value is dynamically updated according to the change trend of the risk coefficient, which is used to ensure the safety of the warehouse environment.

[0112] In the embodiments of the present application, the recent change slope of the monitoring risk coefficient is monitored, and when the change rate exceeds the preset warning value, the stability threshold is tightened in proportion; when the change is stable, the threshold is appropriately relaxed to improve the optimization space. This dynamic adjustment mechanism ensures safety and flexibility.

[0113] Step 305: integrating the fluctuation range constraint condition, the boundary condition and the stability threshold into an optimization framework containing the first function, the second function and the third function.

[0114] In step 305, the optimization framework refers to the mathematical solution space integrating the three objective functions and three types of constraint conditions, and its structure design directly affects the efficiency and effect of multi-objective optimization.

[0115] In the embodiments of the present application, a hierarchical nested method is used to construct the framework: the outermost layer is the parallel optimization space of the three objective functions, the middle layer is embedded with a dynamic constraint network, and the inner layer is provided with a variable interaction channel. Through this structure, the objectives and constraints are cooperatively processed.

[0116] Step 306: establishing a bidirectional conflict resolution mechanism in the optimization framework for coordinating the conflict between objectives to form a multi-objective decision model.

[0117] In step 306, the bidirectional conflict resolution mechanism refers to the regulation system for coordinating the competitive relationship between objectives, which dynamically balances the optimization needs of different objectives through two paths of forward conduction and reverse feedback.

[0118] In the embodiments of the present application, when the replenishment period and the warehouse cost conflict, the mechanism first attempts to adjust the cost boundary condition; if it still cannot be coordinated, the weight redistribution program is started. At the same time, a risk warning trigger mechanism is set to prioritize the safety target when the stability is threatened. This bidirectional adjustment ensures that the system can still output reasonable solutions in complex environments.

[0119] The following is a specific example:

[0120] In a large e-commerce warehouse center during the air conditioning sales season, the system constructs a multi-objective decision model based on the real-time data of the weight 2 embodiment. First, according to the replenishment urgency parameter 4.4 and the benchmark replenishment cycle 1.5 days, the replenishment cycle constraint interval is calculated to be 0.34-6.6 days through the formula fluctuation lower limit = benchmark cycle / urgency parameter, fluctuation upper limit = benchmark cycle x urgency parameter, wherein the benchmark cycle 1.5 days is derived from the historical average turnover days of the commodity. For the warehouse space occupancy parameter 90 kg / m 2 And the transportation load rate 75%, the cost boundary condition 56.25-75 yuan is generated through the formula cost lower limit = basic cost 50 x space density coefficient 90 / 80 = 56.25, cost upper limit = basic cost 50 x load coefficient 75 / 50 = 75. According to the warehouse risk coefficient 1.126 and its change amount 0.15 in the last 3 hours, the risk threshold is set through the formula stability threshold = benchmark value 1.0 x risk coefficient 1.126 + change amount correction value 0.15 x sensitivity coefficient 2 = 1.426, wherein the sensitivity coefficient 2 is determined by historical data analysis. The three objective functions replenishment cycle f1 = 1 / 4.4 ≈ 0.227, warehouse cost f2 = 0.6 x 90 + 0.4 x 75 = 84, and risk balance f3 = 1 / (1+2x0.15) ≈ 0.769, wherein the weight coefficients 0.6 and 0.4 come from expert evaluation, and the sensitivity coefficient 2 reflects the risk tolerance. After integrating into the optimization framework, when the conflict resolution mechanism detects a conflict between the replenishment cycle target and the risk target, the risk threshold 1.426 is prioritized to be not broken through.

[0121] In the embodiments of the present application, the method realizes the dynamic balance of replenishment efficiency, cost control and risk management by establishing a scientific multi-objective decision model, so that the warehouse system can intelligently adapt to market demand changes and environmental fluctuations, and improves the quality and reliability of operation decision.

[0122] In order to further improve the coordination ability of multi-objective optimization, in some embodiments, step 306: the two-way conflict resolution mechanism for coordinating target conflicts is established in the optimization framework to form a multi-objective decision model, comprising:

[0123] Step 401: establish the feedback association relationship between the replenishment cycle and the warehouse cost.

[0124] In step 401, the feedback association relationship refers to the dynamic adjustment link between the replenishment cycle and the warehouse cost. When the replenishment cycle is shortened, the cost constraint is automatically relaxed, and when the warehouse cost is over the limit, the replenishment frequency is adjusted accordingly, forming a two-way adjustment channel. The relationship is quantified by the target sensitivity matrix.

[0125] In the embodiment of the present application, the system first establishes the elastic coefficient of the change of the replenishment cycle on the warehouse cost and the feedback coefficient of the cost fluctuation on the replenishment cycle, and constructs a dynamic response model between the two. When it is detected that the target optimization of the replenishment cycle leads to a sharp increase in the cost, the cost boundary condition adjustment is automatically triggered; otherwise, when the cost constraint is tightened, the replenishment cycle target weight is corrected accordingly, forming a closed-loop regulation.

[0126] Step 402: Establish the constraint intervention relationship of the risk balance on the feedback correlation.

[0127] In step 402, the constraint intervention relationship refers to the forced adjustment of the risk balance target on the previous feedback link. When the risk coefficient exceeds the warning value, the safety target is prioritized, and part of the optimization space of the cycle and the cost is temporarily frozen. This relationship is realized through a risk warning trigger mechanism.

[0128] In the embodiment of the present application, multiple thresholds of the risk coefficient are set. When the primary threshold is reached, a warning signal is sent, when the intermediate threshold is reached, the adjustment amplitude of the replenishment cycle is limited, and when the high-level threshold is reached, the safety priority mode is forced to switch. This hierarchical intervention not only guarantees safety, but also minimizes the disturbance to normal optimization.

[0129] Step 403: Establish a dynamic coupling relationship between the warehouse cost and the risk balance target.

[0130] In step 403, the dynamic coupling relationship refers to the indirect correlation mechanism between the warehouse cost and the risk balance target. Through the increase and decrease of the environmental control cost and the safety maintenance cost, the economic balance between the two is reflected. This relationship is quantitatively related through a cost-risk conversion coefficient matrix.

[0131] In the embodiment of the present application, the corresponding relationship between the warehouse cost input and the risk coefficient change in the historical data is analyzed, and the risk improvement coefficient brought by the unit cost input is established. In the optimization process, when the safety level needs to be improved, the corresponding minimum cost increment is automatically calculated as a constraint condition embedded in the model.

[0132] Step 404: Based on the feedback correlation, the constraint intervention relationship, and the dynamic coupling relationship, a bidirectional conflict resolution mechanism is constructed.

[0133] In the embodiment of the present application, the feedback correlation is taken as the basic adjustment layer, the constraint intervention relationship is taken as the safety protection layer, and the dynamic coupling relationship is taken as the economic balance layer. A complete resolution mechanism is constructed through hierarchical fusion. The system evaluates the state of each target in real time and automatically selects the optimal coordination strategy.

[0134] Step 405: In the optimization framework, the bidirectional conflict resolution mechanism is used to real-time coordinate the conflict between the replenishment cycle, the warehouse cost, and the risk balance, and a multi-objective decision model is generated.

[0135] In the embodiments of the present application, the model runtime continuously monitors the achievement degrees of the three targets, and when a conflict is detected, a corresponding resolution rule library is called to generate an adjustment scheme. Through iterative optimization, each target gradually approaches a balanced state, and finally an optimal decision that takes into account efficiency, cost and safety is output.

[0136] The following is a specific example:

[0137] During the air conditioner sales peak season of a large e-commerce warehouse center, the system implements a bidirectional conflict resolution mechanism based on the optimization framework of the third embodiment. First, the feedback relationship between the replenishment cycle and the warehouse cost is established. When it is detected that the current scheme of a replenishment cycle of 0.5 days causes the cost to reach 74 yuan close to the upper limit of 75 yuan, the new cost upper limit is calculated to be 82.5 yuan by the formula new cost upper limit = original upper limit x adjustment coefficient 1.1, wherein the adjustment coefficient 1.1 comes from the regression analysis of historical similar scenarios. At the same time, the risk monitoring shows that the coefficient rises to 1.38 close to the threshold value 1.426, triggering the constraint intervention relationship, and the allowed maximum cycle is calculated to be 0.6 days by the formula allowed maximum cycle = original cycle x risk buffer coefficient 1.2, wherein the risk buffer coefficient 1.2 is determined by safety specifications. At this time, the dynamic coupling relationship calculates that the cost input needs to be increased to reduce the risk, and the cost increment is calculated to be 5 yuan by the formula cost increment = risk difference x conversion coefficient, wherein the risk difference is 1.426-1.38 = 0.046, and the conversion coefficient takes the historical average value 108.7. After the system integrates the three relationships, the conflict resolution mechanism outputs a balanced scheme: the replenishment cycle is adjusted to 0.55 days, the cost is controlled at 79 yuan, and the risk coefficient is stabilized at 1.4, and execution instructions are generated including moderately reducing the replenishment frequency, increasing the shelf ventilation equipment, and adjusting some goods to the standby warehouse area. The cycle adjustment value is a compromise value between the two conflicting parties, the cost control value 79 yuan is derived from the new upper limit 82.5 yuan minus the risk adjustment reserved space 3.5 yuan, and the final risk coefficient 1.4 is confirmed to meet the safety requirements through real-time monitoring. The entire decision-making process is optimized through multiple iterations to ensure that the three targets all reach the optimal state within an acceptable range.

[0138] In the embodiments of the present application, the method establishes an intelligent conflict resolution mechanism, enabling the multi-objective optimization system to have self-adaptive coordination capability, and still maintaining the rationality and reliability of the decision-making in complex and variable operating environments, thereby improving the comprehensive efficiency of warehouse management.

[0139] In order to further improve the intelligent level of shelf load safety management, in some embodiments, step 102: based on the goods weight data, a distribution adjustment strategy of the target goods on the shelf is generated in combination with a preset shelf load constraint condition, including:

[0140] Step 501: identify the current load distribution state of the goods weight data in the multi-layer space of the shelf.

[0141] In step 501, the current load distribution state refers to the actual load condition of each area of each layer of the shelf. A three-dimensional load heat map is constructed by real-time data collected by the weight sensor to reflect the weight distribution characteristics of goods in space. The state includes the position coordinates of each storage unit, the load value, and the correlation relationship of adjacent areas.

[0142] In the embodiment of the present application, the system obtains real-time weight data of each layer and each partition through the shelf embedded sensor network, and generates a load distribution matrix with position coding in combination with the three-dimensional digital model of the shelf. High-density aggregation areas are identified through a spatial clustering algorithm, and the weight gradient change trend between each area is analyzed to form a complete load state evaluation report.

[0143] Step 502: Compare the current load distribution state with the preset shelf load constraint condition layer by layer to identify over-limit areas that exceed the preset load threshold.

[0144] In step 502, the preset load threshold is a quantitative embodiment of the shelf load constraint condition, which is directly derived from the maximum allowed load value of each layer of the shelf specified in the constraint condition. For example, the constraint condition specifies "≤ 500 kg per layer", and the load threshold is 500 kg, which is used to specifically compare and judge the over-limit state. The over-limit area refers to the shelf space range whose actual load exceeds the preset safety threshold, and its identification result includes over-limit position coordinates, over-limit weight value, and over-limit proportion and other core parameters. The area is determined through the cross analysis of spatial position and load data.

[0145] In the embodiment of the present application, the system compares the real-time load distribution matrix with the preset shelf load standard matrix element by element to mark all over-limit units. The adjacent over-limit units are merged into continuous over-limit areas through a region growing algorithm, the over-limit total amount and the average over-limit proportion of each area are calculated, and an over-limit area list is generated.

[0146] Step 503: According to the position distribution and over-limit degree of the over-limit area, a goods migration scheme from the over-limit area to the non-over-limit area is generated.

[0147] In step 503, the location distribution of the over-limit area and the over-limit degree: by comparing the current load distribution state of each layer of the shelf with the preset load threshold layer by layer, the specific shelf layer and area exceeding the threshold (location distribution) are located, and the over-limit degree is quantified according to the percentage difference between the over-limit weight and the threshold; the specific implementation is: the location distribution is determined by the shelf pressure sensor data and three-dimensional coordinate mapping, and the over-limit degree is calculated by (measured weight-threshold) / threshold x 100%. The non-over-limit area refers to the area in the shelf whose measured load is less than the preset load threshold, and the physical adjacency of the over-limit area is not a necessary condition. The system locates all areas that meet the condition "measured load < threshold" as candidate target areas through shelf coordinate positioning, regardless of whether they are adjacent to the over-limit area, and preferentially selects areas with sufficient load and optimal path. The goods migration scheme refers to the specific operation plan to eliminate the over-limit state, including the list of goods to be migrated, the target storage location, the migration path planning and the execution priority sorting. The scheme is generated by a space optimization algorithm to ensure that the over-limit problem is solved with the minimum handling amount.

[0148] In the embodiments of the present application, the system first calculates the theoretical migration amount of each area according to the over-limit area list, and then searches for candidate target areas that meet the load requirement in the shelf model. The optimal migration combination is selected through a path cost evaluation function to generate a preliminary migration scheme containing the source location, target location, goods quantity and recommended path. The execution order of the scheme is arranged in descending order of over-limit severity.

[0149] Step 504: Adjust the migration path and migration order in the goods migration scheme in combination with the environmental state data, and take the adjusted goods migration scheme as the distribution adjustment strategy.

[0150] In step 504, the migration path generates the shortest handling route according to the spatial topological relationship between the over-limit area and the low-load area, and the migration order is sorted from high to low according to the over-limit degree; the specific implementation is: the path planning uses the shelf navigation grid data to calculate the optimal path, and the order strategy preferentially processes goods with an over-limit degree > 20%, and then processes goods with an over-limit degree of 10-20%. The adjusted goods migration scheme refers to the final execution strategy that integrates environmental constraints, which adds environmental adaptability correction to the basic migration scheme, including path obstacle avoidance, operation time adjustment and other optimization measures. The scheme ensures the safety and feasibility of the migration operation.

[0151] In the embodiments of the present application, the system superimposes environmental monitoring data onto the three-dimensional model of the shelf to identify high-temperature, high-humidity or vibration abnormal areas. The path in the preliminary migration scheme is optimized to avoid obstacles, the operation time is adjusted to avoid peak periods of the environment, and if necessary, large-weight migration is split into multiple batches for execution. Finally, a distribution adjustment strategy document with environmental adaptation markers is generated.

[0152] The following is a specific example:

[0153] At the peak of air conditioner sales in a large e-commerce warehouse, the system monitors that the top shelf in the A commodity area bears 580 kg, exceeding the preset threshold of 500 kg. Among them, the over-limit weight of B class goods stored in the X10-Y15 to X15-Y20 area is calculated to be 80 kg according to the formula over-limit weight = measured weight - bearing threshold. Through analysis of the three-dimensional model of the shelf, the remaining bearing of 50 kg in the middle left area and the remaining bearing of 30 kg in the bottom right area are selected as the migration target, and a preliminary migration scheme is generated: 2 B goods with a total weight of 96 kg are migrated twice according to the over-limit degree, and the first migration of 48 kg is moved to the middle left area coordinates X5-Y8 to X8-Y11, and the remaining 48 kg is moved to the bottom right area coordinates X3-Y5 to X6-Y8. Environmental monitoring shows that the temperature of X12-Y14 channel reaches 38℃, exceeding the safety threshold, and the system automatically adjusts the path to bypass the central channel, and arranges the migration operation period in the night when the temperature is lower. Combined with the urgency parameter of 2.0 calculated from the real-time order data, the space density of the shelf is 90 kg / m 2 and the transportation load rate is 75%, a multi-objective function is constructed: the replenishment cycle target f1 = 1 / 2.0 = 0.5, the warehouse cost target f2 = 0.6*90 + 0.4*75 = 84, where the weight coefficients 0.6 and 0.4 come from expert evaluation, and the risk balance target f3 = 1 / (1+2*0.15)≈0.77, where k = 2 is the risk sensitivity coefficient. After multiple rounds of iterative optimization, the final output execution scheme is: under the premise that the risk coefficient does not exceed 1.1, a distribution strategy with a replenishment cycle of 1.2 days and a warehouse cost of 78 units is adopted, and the adjusted goods migration scheme is executed simultaneously, and the migration operation is completed in two nights and auxiliary cooling measures are started.

[0154] In the embodiments of the present application, the method realizes precise control of shelf load through intelligent bearing state recognition and migration scheme generation, optimizes space resource utilization while ensuring warehouse safety, and improves the operation reliability and management efficiency of large-scale warehouse systems.

[0155] In order to further improve the accuracy and feasibility of the goods migration scheme, in some embodiments, step 503: generating a goods migration scheme from the over-limit area to the non-over-limit area according to the position distribution and over-limit degree of the over-limit area, comprising:

[0156] Step 601: determining the spatial distribution topological relationship of the over-limit goods based on the position distribution of the over-limit area.

[0157] In step 601, the spatial distribution topological relationship refers to the position correlation characteristics of the over-limit goods in the three-dimensional space of the shelf, including the geometric shape of the over-limit area, the connection mode of the adjacent area, and the spatial distance from other functional areas. The relationship is represented by the graph structure of the shelf digital model, where the nodes represent storage units and the edges represent the reachable paths of the transfer.

[0158] In the embodiment of the present application, the system parses the coordinate data of the over-limit area, constructs a weighted undirected graph with the storage units as vertices and the transfer channels as edges. The aggregation form of the over-limit goods is identified through the graph traversal algorithm, the spatial adjacency relationship between each over-limit unit is analyzed, and a topological network describing the distribution density of the goods is formed.

[0159] Step 602: Calculate the migration amount according to the over-limit degree of the over-limit area.

[0160] In step 602, the migration amount refers to the total value of the goods that need to be transferred to eliminate the over-limit state, which takes into account the absolute value of over-limit and safety margin to ensure a reasonable buffer space after migration. This value is the core basis for formulating the migration plan. Migration amount = over-limit weight x safety margin coefficient, where over-limit weight = measured load - load threshold.

[0161] In the embodiment of the present application, the over-limit weight data in the over-limit area list is multiplied by a preset safety adjustment coefficient to obtain the theoretical migration amount. For continuous over-limit areas, the region merging calculation method is used, and for discrete over-limit points, separate calculation is performed, and finally the migration amount demand table of each region is generated.

[0162] Step 603: Select the target area with a load remaining amount greater than the migration amount from all non-over-limit areas.

[0163] In step 603, the load remaining amount = load threshold - measured load. Target area selection refers to selecting the most suitable storage location to receive the migrated goods from all candidate areas that meet the load requirements. The selection criteria include load remaining amount matching degree, space utilization optimization potential, and path convenience with the over-limit area.

[0164] In the embodiment of the present application, the system traverses all non-over-limit areas in the shelf model, calculates the matching degree of the current remaining amount of each area with the migration demand. Through a multi-level filtering mechanism, areas with insufficient remaining amount are excluded first, and then the transfer path cost of the remaining candidate areas is evaluated, and finally a target area list sorted by adaptation degree is generated.

[0165] Step 604: Generate a goods migration plan based on the spatial distribution topological relationship and the target area.

[0166] In the embodiment of the present application, based on the topological relationship network and the target area list, a heuristic search algorithm is used to generate an initial migration path. The path is checked for feasibility in combination with the shelf operation specifications, and the loading order is optimized according to the goods attributes, and finally a detailed migration plan document including time scheduling, equipment scheduling and personnel configuration is formed.

[0167] The following is a specific example:

[0168] In a large e-commerce warehouse center during the air conditioner sales peak, based on the over-limit analysis result of the system according to the fifth embodiment, for the B-class goods with over-limit of 80 kg in the top X10-Y15 to X15-Y20 area, first, through the graph structure analysis of the three-dimensional model of the shelf, the spatial topological relationship of 5 refrigerators in a 2*3 matrix distribution is determined, and the core over-limit nodes are concentrated in the X12-Y16 to X14-Y18 area. According to the over-limit degree calculation migration amount, the formula migration amount = over-limit weight * safety factor is used, wherein the over-limit weight is 80 kg, the safety factor is 1.2, and the theoretical migration amount is 96 kg. The system traverses the shelf database to filter the non-over-limit area, and calculates the score of each candidate area according to the formula area adaptation degree = residual matching degree * path coefficient, wherein the residual of the middle left area is 100 kg, and the score is 0.85, the residual of the bottom right area is 80 kg, and the score is 0.72, and the middle left area is selected as the primary target. Based on the topological relationship, the migration scheme is generated: the over-limit area is sorted by distance, and the total weight of 3 refrigerators in the central position is preferentially migrated to the middle left area X5-Y8 to X8-Y11, and the remaining 24 kg is supplemented by 1 refrigerator at the edge position.

[0169] In the embodiments of the present application, through systematic spatial analysis and intelligent matching mechanism, the migration scheme generated not only effectively solves the shelf over-limit problem, but also fully considers various constraint conditions in actual operation, and improves the safety and execution efficiency of warehouse scheduling.

[0170] In order to further improve the optimization efficiency and quality of the inventory decision scheme, in some embodiments, step 104: based on the multi-objective decision model, an interactive iterative method is used to output an optimal inventory distribution decision scheme, comprising:

[0171] Step 701: based on the multi-objective decision model, an initial decision solution set is generated by an initialization algorithm as the input of the first interactive iteration.

[0172] In step 701, the initial decision solution set refers to the feasible solution population of the first iteration in the multi-objective optimization process, which includes several groups of candidate schemes that meet the basic constraint conditions, and each scheme corresponds to a specific combination of replenishment period, warehouse cost and risk balance target. The set is generated by a spatial sampling algorithm to ensure the diversity and universality of the solution distribution.

[0173] In the embodiment of the present application, the system generates an initial solution set using the Latin hypercube sampling method according to the variable range of the multi-objective decision model. Each solution contains parameter values of three dimensions, and the solutions that meet the hard constraints such as shelf load-bearing and environmental safety are retained through feasibility check to form a representative initial decision solution set. The specific process is: based on the multi-objective decision model, an initial decision solution set is generated through interactive iterative calculation, specifically including: generating a feasible solution that meets the shelf load-bearing constraint and the environmental state constraint as an initial population, and screening non-dominated solutions to form a decision solution set through non-dominated sorting and crowded distance calculation. For example, during the e-commerce promotion period, a warehouse initializes 100 groups of feasible solutions for home appliance goods, each group of solutions contains different shelf distribution schemes (such as the combination of refrigerator centralized storage area and washing machine scattered storage), and after evaluating the replenishment cycle, warehouse cost and risk coefficient of each scheme through the model, 30 groups of mutually non-dominated Pareto optimal solutions are retained to form the decision solution set.

[0174] Step 702: In each interactive iteration, the real-time change of the environmental state data and the distribution adjustment strategy are taken as dynamic constraint conditions.

[0175] In step 702, the dynamic constraint condition refers to the optimization limit that is updated in real time with the change of the environment and the adjustment of the strategy, which converts external changes into model parameter boundary adjustment to ensure that the optimization direction is consistent with the actual demand. This condition acts on the objective function space through a constraint propagation mechanism.

[0176] In the embodiment of the present application, the system continuously monitors the environmental sensor data and the execution feedback of the distribution adjustment strategy, and when detecting changes in key indicators such as temperature, humidity or shelf vibration, automatically updates the risk constraint boundary; according to the implementation progress of the migration scheme, dynamically adjusts the value range of the space occupation parameter to form a constraint condition set that adapts to the latest state.

[0177] Step 703: Screen a non-dominated solution subset from the decision solution set that meets the dynamic constraint condition.

[0178] In step 703, the non-dominated solution subset refers to a candidate scheme set that is not dominated by other solutions in the current iteration round, which meets the dynamic constraint condition and the Pareto optimality requirement. This subset is extracted through a multi-dimensional screening mechanism and represents the candidate group of the current optimal solution.

[0179] In the embodiment of the present application, the system matches each candidate scheme in the decision solution set with the dynamic constraint condition and eliminates the solutions that violate the constraint. The remaining solutions are non-dominated sorted, and the solutions with high sorting and uniform distribution are selected to form the non-dominated solution subset, ensuring the quality and diversity of the solutions.

[0180] Step 704: Determine whether the solution space change of the non-dominated solution subset is less than a preset convergence threshold.

[0181] In step 704, the solution space variation refers to the difference in the distribution of the non-inferior solution subset in the objective function space between adjacent iteration rounds, which reflects the convergence state of the optimization process. This index is calculated by a solution set distance measurement function and is used to determine the termination condition.

[0182] In the embodiments of the present application, the average distance of the current non-inferior solution subset and the solution set of the last round in each objective dimension is calculated to comprehensively evaluate the evolution degree of the solution space. When the variation continues to be lower than the threshold value, it is considered that the optimization process has reached a stable state.

[0183] Step 705: If not, the non-inferior solution subset is expanded to obtain a non-inferior solution set, and the non-inferior solution set is used as a decision solution set to repeat the iteration process until the solution space variation of the non-inferior solution subset is less than a preset convergence threshold value or the maximum number of iterations is reached, and an optimal inventory distribution decision scheme is generated.

[0184] In step 705, the expansion processing refers to an intelligent expansion operation performed on the non-inferior solution subset to maintain the optimization momentum, which generates new solutions through mutation and recombination strategies to avoid premature convergence of the algorithm. This processing balances the breadth and depth of the search.

[0185] In the embodiments of the present application, the non-inferior solution subset is subjected to cluster analysis, and new solutions are generated in sparse areas through Gaussian mutation; high-quality solutions in dense areas are subjected to cross-recombination to produce improved offspring. After the new solutions are combined with the original solution set, feasibility checking and non-dominant screening are performed to form a new decision solution set with better quality for continued iteration.

[0186] The following is a specific example:

[0187] In a large e-commerce warehouse center during the air conditioning sales peak season, the system starts interactive iterative optimization based on the multi-objective decision model of the first embodiment of the weight. First, Latin hypercube sampling is used to generate 50 sets of initial solutions, each set of solution containing three-dimensional parameters of replenishment cycle, warehouse cost and risk coefficient, and the replenishment cycle value range is determined by the formula lower limit = benchmark cycle 1.5 days / replenishment urgency parameter 2.0 = 0.75 days, upper limit = benchmark cycle 1.5 days x replenishment urgency parameter 2.0 = 3.0 days. In the first iteration, the environmental monitoring shows that the local temperature rises to 38℃, and the risk threshold is reset by adjusting the formula risk threshold = original threshold 1.1 x temperature influence coefficient 1.15 ≈ 1.27, 12 non-inferior solutions are selected after screening, and the solution space changes by 0.28. The system performs Gaussian variation and simulated binary crossover on the non-inferior solutions, expands to 30 solutions for further iteration. In the second iteration, the new storage space is released by the shelf migration strategy update, and the upper limit of the warehouse cost is relaxed to 1.1 times of the original value of 75 yuan, that is, 82.5 yuan, and 18 non-inferior solutions are selected, with a change of 0.18. In the third iteration, the risk coefficient change rate monitoring value falls to 0.1, and the judgment standard is adjusted by the formula convergence threshold = basic value 0.15 x change rate attenuation coefficient 0.9 = 0.135, and finally 15 solutions change by 0.12 to reach convergence, and the solution with the largest crowding distance is selected as the output scheme: the replenishment cycle is 1.1 days calculated by the formula cycle = benchmark cycle 1.5 days / optimization coefficient 1.36, the warehouse cost is 79 yuan within the relaxed constraint range, and the risk coefficient is 1.08, which meets the latest threshold requirement. The system generates the final execution scheme accordingly, including the specific time arrangement of three daily replenishments, the load distribution diagram after shelf adjustment, and the temperature control measures, to ensure the efficient and safe operation of the inventory during the peak sales period.

[0188] In the embodiments of the present application, the method dynamically adapts the inventory decision scheme to environmental changes and business needs through an intelligent interactive iterative mechanism, improves the optimization efficiency while ensuring the quality of the solution, and provides reliable multi-objective decision support for complex warehouse scenarios.

[0189] Figure 2 The structure diagram of an inventory distribution decision system based on dynamic multi-objective optimization provided in the embodiments of the present application is shown in Figure 2 The system comprises:

[0190] The acquisition module 21 is configured to acquire real-time order data, product weight data, historical demand data and environmental state data of the target product in a scenario where the real-time demand fluctuation of the e-commerce warehouse is greater than the preset fluctuation.

[0191] The generation module 22 is configured to generate a distribution adjustment strategy of the target product on the shelf based on the product weight data and in combination with a preset shelf load constraint condition.

[0192] The constructing module 23 is configured to combine the distribution adjustment strategy, the real-time order data, the historical demand data, and the environment state data to construct a multi-objective decision model with an optimization objective of minimizing a replenishment cycle, minimizing a warehouse cost, and optimizing a risk balance.

[0193] The output module 24 is configured to output an optimal inventory distribution decision scheme in an interactive iterative manner based on the multi-objective decision model.

[0194] Figure 2 The inventory distribution decision system based on dynamic multi-objective optimization can perform Figure 1 The implementation principle and technical effects of the inventory distribution decision method based on dynamic multi-objective optimization are not repeated here. The specific manner in which each module and unit of the inventory distribution decision system based on dynamic multi-objective optimization performs operations has been described in detail in the embodiments of the method, and will not be described in detail here.

[0195] In one possible design, Figure 2 The inventory distribution decision system based on dynamic multi-objective optimization can be implemented as a computing device, such as a computer. Figure 3 As shown in the figure, the computing device can include a storage component 31 and a processing component 32.

[0196] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32.

[0197] The processing component 32 executes the above Figure 1 The inventory distribution decision method based on dynamic multi-objective optimization of the embodiments.

[0198] The processing component 32 can include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component can also be one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components, for executing the above method.

[0199] The storage component 31 is configured to store various types of data to support the operation of the terminal. The storage component can be implemented by any type of volatile or nonvolatile storage devices or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read only memory (EEPROM), erasable programmable read only memory (EPROM), programmable read only memory (PROM), read only memory (ROM), magnetic storage, flash memory, magnetic or optical disk.

[0200] Of course, the computing device can also include other components, such as an input / output interface, a display component, a communication component, etc.

[0201] The input / output interface provides an interface between the processing component and peripheral interface modules, which can be output devices, input devices, etc.

[0202] The communication component is configured to facilitate wired or wireless communication between the computing device and other devices, etc.

[0203] The computing device can be a physical device or an elastic computing host provided by a cloud computing platform, and the computing device can be a cloud server, and the processing component, the storage component, etc. can be basic server resources rented or purchased from the cloud computing platform.

[0204] The embodiments of the present application also provide a computer storage medium storing a computer program, and the computer program can implement the above-mentioned Figure 1 The embodiment shown in the figure provides an inventory distribution decision method based on dynamic multi-objective optimization.

[0205] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-mentioned system, device and unit can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

[0206] The device embodiments described above are merely illustrative, wherein the units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units, i.e., can be located in one place, or can be distributed to multiple network units. Part or all of the modules can be selected to achieve the purposes of the embodiments according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0207] Through the description of the above embodiments, those skilled in the art can clearly understand that the embodiments can be realized by means of software and necessary universal hardware platforms, and of course can also be realized by hardware. Based on such understanding, the above technical solutions can be embodied in the form of software products, and the computer software products can be stored in a computer readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and include a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0208] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for inventory distribution decision based on dynamic multi-objective optimization, characterized in that, The method comprises the following steps: acquiring real-time order data, product weight data, historical demand data and environmental state data of target commodities; generating a distribution adjustment strategy of the target commodities on the shelves based on the product weight data and in combination with a preset shelf load bearing constraint condition; combining the distribution adjustment strategy, the real-time order data, the historical demand data and the environmental state data to construct a multi-objective decision model with the optimization objectives of minimizing the replenishment cycle, minimizing the warehouse cost and optimizing the risk balance; outputting an optimal inventory distribution decision scheme in an interactive iterative manner based on the multi-objective decision model; the combining the distribution adjustment strategy, the real-time order data, the historical demand data and the environmental state data to construct a multi-objective decision model with the optimization objectives of minimizing the replenishment cycle, minimizing the warehouse cost and optimizing the risk balance comprises: generating a migration instruction according to the distribution adjustment strategy and converting the migration instruction into a warehouse space occupation parameter; calculating a replenishment urgency parameter based on the real-time order data and the historical demand data; generating a warehouse risk coefficient based on the environmental state data and in combination with a shelf stability change trend; establishing a multi-objective decision model with the replenishment urgency parameter, the warehouse space occupation parameter and the warehouse risk coefficient as variables; the generating a distribution adjustment strategy of the target commodities on the shelves based on the product weight data and in combination with a preset shelf load bearing constraint condition comprises: identifying a current load bearing distribution state of the product weight data in a multi-layer space of the shelves; comparing the current load bearing distribution state with a preset shelf load bearing constraint condition layer by layer to identify an over-limit area that exceeds a preset load bearing threshold; generating a product migration scheme from the over-limit area to a non-over-limit area according to the position distribution and over-limit degree of the over-limit area; adjusting a migration path and a migration sequence in the product migration scheme in combination with the environmental state data and taking the adjusted product migration scheme as the distribution adjustment strategy.

2. The method of claim 1, wherein, the establishing a multi-objective decision model with the replenishment urgency parameter, the warehouse space occupation parameter and the warehouse risk coefficient as variables comprises: constructing a first function representing the minimization of the replenishment cycle based on the replenishment urgency parameter, a second function representing the minimization of the warehouse cost based on the warehouse space occupation parameter and a third function representing the optimization of the risk balance based on the warehouse risk coefficient; constructing a fluctuation range constraint condition of the replenishment cycle according to the replenishment urgency parameter and a benchmark replenishment cycle; generating a boundary condition of the warehouse cost based on the warehouse space occupation parameter; setting a stability threshold of the risk balance according to a dynamic change rate of the warehouse risk coefficient; integrating the fluctuation range constraint condition, the boundary condition and the stability threshold into an optimization framework containing the first function, the second function and the third function; establishing a bidirectional conflict resolution mechanism for coordinating the objective conflicts in the optimization framework to form the multi-objective decision model.

3. The method of claim 2, wherein, The two-way conflict resolution mechanism for coordinating target conflicts is established in the optimization framework to form a multi-objective decision model, including: establishing a feedback correlation between the replenishment cycle and the warehouse cost; establishing a constraint intervention relationship of the risk balance to the feedback correlation; establishing a dynamic coupling relationship between the warehouse cost and the risk balance target; based on the feedback correlation, the constraint intervention relationship and the dynamic coupling relationship, a two-way conflict resolution mechanism is constructed; In the optimization framework, the conflicts between the replenishment cycle, the warehouse cost and the risk balance are coordinated in real time through the two-way conflict resolution mechanism, and a multi-objective decision model is generated.

4. The method of claim 1, wherein, The position distribution and the over-limit degree of the over-limit area are used to generate a goods migration scheme from the over-limit area to the non-over-limit area, including: Based on the position distribution of the over-limit area, the spatial distribution topological relationship of the over-limit goods is determined; According to the over-limit degree of the over-limit area, the migration amount is calculated; From all the non-over-limit areas, the target area with a bearing residual greater than the migration amount is selected; Based on the spatial distribution topological relationship and the target area, a goods migration scheme is generated.

5. The method of claim 1, wherein, Based on the multi-objective decision model, an optimal inventory distribution decision scheme is output in an interactive iterative manner, including: Based on the multi-objective decision model, an initial decision solution set is generated by an initialization algorithm as the input of the first interactive iteration; In each interactive iteration, the real-time change of the environment state data and the distribution adjustment strategy are used as dynamic constraint conditions; From the decision solution set, a non-inferior solution subset that meets the dynamic constraint conditions is selected; Judge the solution space change of the non-inferior solution subset is less than the preset convergence threshold; If not, the non-inferior solution subset is expanded to obtain a non-inferior solution set, and the non-inferior solution set is used as the decision solution set to repeat the iteration process until the solution space change of the non-inferior solution subset is less than the preset convergence threshold or the maximum iteration number is reached, and an optimal inventory distribution decision scheme is generated.

6. An inventory distribution decision system based on dynamic multi-objective optimization, characterized by, Including: The acquisition module is used to acquire real-time order data, goods weight data, historical demand data and environment state data of the target goods in the scenario where the real-time demand fluctuation of the e-commerce warehouse is greater than the preset fluctuation; The generation module is used to generate a distribution adjustment strategy of the target goods on the shelf based on the goods weight data and in combination with a preset shelf bearing constraint condition; The construction module is used to combine the distribution adjustment strategy, the real-time order data, the historical demand data and the environment state data to construct a multi-objective decision model with the optimization objectives of minimizing the replenishment cycle, minimizing the warehouse cost and optimizing the risk balance; The output module is used to output an optimal inventory distribution decision scheme in an interactive iterative manner based on the multi-objective decision model; The distribution adjustment strategy, the real-time order data, the historical demand data and the environment state data are combined to construct a multi-objective decision model with the optimization objectives of minimizing the replenishment cycle, minimizing the warehouse cost and optimizing the risk balance, including: According to the distribution adjustment strategy, migration instructions are generated, and the migration instructions are converted into warehouse space occupation parameters; According to the real-time order data and the historical demand data, a replenishment urgency parameter is calculated and generated; Based on the environmental state data, in combination with the shelf stability change trend, a warehouse risk coefficient is generated; A multi-objective decision-making model is established with the replenishment urgency parameter, the warehouse space occupation parameter, and the warehouse risk coefficient as variables; The distribution adjustment strategy of the target goods on the shelf is generated based on the goods weight data and in combination with a preset shelf load-bearing constraint condition, including: The current load-bearing distribution state of the goods weight data in the multi-layer space of the shelf is identified; The current load-bearing distribution state is compared layer by layer with the preset shelf load-bearing constraint condition, and an over-limit area that exceeds the preset load-bearing threshold is identified; According to the position distribution and over-limit degree of the over-limit area, a goods migration scheme from the over-limit area to a non-over-limit area is generated; In combination with the environmental state data, the migration path and migration sequence in the goods migration scheme are adjusted, and the adjusted goods migration scheme is taken as the distribution adjustment strategy.

7. A computing device, comprising: The storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement the inventory distribution decision-making method based on dynamic multi-objective optimization according to any one of claims 1-5.

8. A computer storage medium, characterized in that, The computer program is stored in the computer and is executed by the computer to implement the inventory distribution decision-making method based on dynamic multi-objective optimization according to any one of claims 1-5.

Images