Labor insurance appliance order dynamic scheduling and inventory management method based on industrial internet

Abstract

The invention provides a labor insurance article order dynamic scheduling and inventory management method based on the industrial internet, and relates to the technical field of logistics scheduling, and the method comprises the steps: obtaining the order demand information of each enterprise in an industrial park, and the real-time inventory data and historical delivery records of suppliers, and dynamically constructing a probability density function of the delivery capability of the suppliers, and a gravitation model is combined to calculate the adaptation degree between the supplier and the order demand, so that an accurate order distribution scheme is generated. In this way, order allocation rationality and efficiency can be improved, and order delay and resource waste are reduced. Besides, through an improved isochronous circle division method and a dynamic space-time combination optimization algorithm, clustering analysis and scheduling optimization are performed on the distribution path, so that the timeliness of distribution and the rationality of the path can be remarkably improved, the logistics cost is reduced, and the overall logistics management efficiency of the industrial park is improved.

Description

Technical Field

[0001] The present invention relates to the field of logistics scheduling technology, and more specifically, to a method for dynamic scheduling and inventory management of labor protection product orders based on the Industrial Internet. Background Art

[0002] In recent years, with the rapid development of industrial internet technologies, traditional supply chain management models are gradually transforming towards digitalization and intelligentization. In industrial park environments, labor protection supplies are essential for production operations, and the efficiency of their order management and inventory scheduling directly impacts a company's production continuity and safety. Currently, most companies rely on manual experience or simple spreadsheet tools to manage labor protection supply order scheduling and inventory management. While this traditional management model is somewhat operational in low-complexity supply chain environments, it struggles to meet real-time demands in scenarios with complex order requirements, numerous suppliers, and strict delivery deadlines. In recent years, some research has begun exploring labor protection supply management methods based on the Industrial Internet, leveraging IoT sensing technologies to obtain real-time inventory data and combining them with supply chain analysis models to optimize order scheduling. However, these studies are largely limited to fixed-rule order allocation models and lack in-depth modeling of suppliers' dynamic delivery capabilities and geospatial relationships, resulting in low order allocation accuracy and efficiency.

[0003] Existing technologies for labor protection product order scheduling and inventory management still have many shortcomings. On the one hand, suppliers' delivery capabilities are often affected by multiple factors, including inventory status, historical fulfillment capabilities, and geographic location. Existing technologies fail to fully consider the combined effects of these dynamic factors, resulting in high delay rates or increased costs in actual order allocation plans. On the other hand, existing distribution route planning mostly uses static algorithms, which fail to effectively combine the order time window with the dynamic changes in the distribution area, making it difficult to achieve the dual goals of spatial aggregation and time optimization. In addition, traditional inventory management methods are often based on single-dimensional data analysis, such as inventory volume or historical demand, while ignoring the overall collaborative optimization of the supply chain. This makes it impossible to achieve efficient resource allocation in industrial park environments with multiple enterprises and suppliers. To address these shortcomings, how to build a dynamic scheduling and inventory management method based on the Industrial Internet that fully utilizes real-time data and historical delivery records for comprehensive analysis has become an important research direction in the current technology field. Summary of the Invention

[0004] To address the aforementioned technical issues, the present invention provides a method for dynamically scheduling orders and managing inventory for labor protection products based on the Industrial Internet. This method can, to a certain extent, address the supply-demand imbalance and waste of logistics resources in centralized procurement of labor protection products within industrial parks.

[0005] According to one aspect of the present invention, a method for dynamic scheduling and inventory management of labor protection supplies orders based on the Industrial Internet is provided, which includes:

[0006] Obtain labor protection supplies order demand information reported by enterprises within the industrial park, and collect real-time inventory data and historical delivery records of suppliers within the park;

[0007] Based on the historical delivery records, calculate the order fulfillment time distribution curve of each supplier to obtain the delivery capacity probability density function of each supplier, and use the probability density function as the basic parameter for the supplier dynamic scoring;

[0008] Based on the delivery capability probability density function and the supplier's real-time inventory data, an order allocation weight matrix is ​​constructed. Based on the geographic spatial relationship between suppliers and enterprises, a gravity model is used to calculate each supplier's suitability for each order demand, and an initial order allocation plan is generated.

[0009] Based on the improved isochronous circle division method, orders with similar delivery paths are automatically aggregated, and the final order scheduling plan is generated through a dynamic spatiotemporal combination optimization algorithm.

[0010] Furthermore, the order fulfillment time distribution curve is obtained by extracting the fulfillment time data set from the historical delivery records, constructing a Gaussian kernel function centered on the fulfillment time records of the fulfillment time data set, superimposing the Gaussian kernel function and normalizing it.

[0011] Furthermore, a Gaussian kernel function centered on the performance time record is constructed to construct a symmetrical bell-shaped distribution curve with weight adjustment according to the number of repetitions of the performance time record. The function value is then calculated through the time interval, and after normalization, a standardized Gaussian kernel function with an integral value of one is obtained.

[0012] Furthermore, the dynamic score of the supplier is obtained by constructing an expected delivery time window, calculating the cumulative probability value within it, and dynamically adjusting the judgment interval in combination with the order type. Finally, the cumulative probability value is weightedly combined with the historical order size to obtain a scoring indicator for measuring the supplier's comprehensive delivery capability.

[0013] Furthermore, the order allocation weight matrix is ​​constructed by combining the basic rating value and inventory coverage of the supplier, calculating the initial weight value of the supplier and order demand combination, and introducing a penalty factor to correct the initial weight value, and finally generating the order allocation weight matrix.

[0014] Furthermore, the penalty factor is set based on whether the supplier has breached the contract;

[0015] When a supplier defaults for the first time, a basic penalty coefficient is determined based on the inventory coverage ratio at the time of default, where the basic penalty coefficient is set by a piecewise function:

[0016] When the inventory coverage ratio is lower than the first warning threshold, the basic penalty coefficient is taken from the severe penalty range, and the upper limit of the severe penalty range does not exceed the first critical value;

[0017] When the inventory coverage ratio is between the first warning threshold and the second warning threshold, the basic penalty coefficient is taken from the light penalty range, and the lower limit of the light penalty range is not lower than the second critical value;

[0018] When a supplier defaults multiple times, the penalty factor is calculated using a cumulative multiplication method.

[0019] Furthermore, the gravity model uses the probability density function value of the supplier's delivery capability as the subject quality item in the gravity formula, uses the enterprise's order demand as the object quality item in the gravity formula, and uses the geographical distance between two points as the distance item in the gravity formula.

[0020] Furthermore, the orders with similar delivery paths are automatically aggregated based on the supplier warehouse as the origin, and delivery clusters are constructed through grid division, adaptive adjustment of delivery units and dynamically adjusted time circles, and the delivery clusters are gradually optimized in combination with restrictions and the geometric center of the order address.

[0021] Furthermore, the order scheduling plan is generated by constructing an order spatiotemporal feature matrix and rolling time domain division, allocating orders to scheduling cycles according to priority, time window and delivery cost, and combining the optimization of delivery clusters to generate a detailed delivery schedule including vehicle arrangement and time arrangement.

[0022] Furthermore, the optimization of the delivery cluster optimizes the delivery sequence through neighborhood operations, uses a combination of dynamic step size, perturbation mechanism and feature guidance to jump out of the local optimum, and finally outputs the optimal solution that meets the termination conditions.

[0023] Compared with the existing technology, the method for dynamic scheduling and inventory management of labor protection supplies orders based on the industrial Internet provided by the present invention obtains the order demand information of each enterprise in the industrial park and the real-time inventory data and historical delivery records of the supplier, dynamically constructs the probability density function of the supplier's delivery capacity, and combines the gravity model to calculate the compatibility between the supplier and the order demand, thereby generating an accurate order allocation plan. In this way, the rationality and efficiency of order allocation can be improved, and order delays and resource waste can be reduced. In addition, through the improved isochronous circle division method and dynamic spatiotemporal combination optimization algorithm, cluster analysis and scheduling optimization of the distribution path can be performed, which can significantly improve the timeliness of distribution and the rationality of the path, thereby reducing logistics costs and improving the overall logistics management efficiency of the industrial park. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0025] Figure 1 The present invention is a flowchart of a method for dynamic scheduling and inventory management of labor protection product orders based on the Industrial Internet according to an embodiment of the present invention. DETAILED DESCRIPTION

[0026] Below, the exemplary embodiments according to the present invention will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments of the present invention, and it should be understood that the present invention is not limited to the exemplary embodiments described herein.

[0027] Figure 1 Flowchart of the method for dynamic scheduling and inventory management of labor protection products orders based on the industrial Internet according to an embodiment of the present invention. Figure 1 As shown in the figure, the dynamic scheduling and inventory management method for labor protection supplies orders based on the Industrial Internet includes:

[0028] S1: Obtain labor protection supplies order demand information reported by enterprises in the industrial park. The labor protection supplies order demand information includes item type, quantity, delivery time, and enterprise geographic coordinates. The real-time inventory data and historical delivery records of each supplier in the park are also collected.

[0029] Obtaining labor protection supply order demand information reported by various enterprises in the industrial park, wherein the process of obtaining the labor protection supply order demand information includes: receiving an electronic order form filled out by the enterprise through the enterprise-side interface of the industrial Internet platform, wherein the electronic order form includes item type information, order quantity information, expected delivery time information, and the enterprise's geographic location coordinate information. The item type information includes specific category specifications of labor protection supplies such as hard hats, protective gloves, protective clothing, and safety shoes. The order quantity information indicates the specific purchase quantity of each type of labor protection supply. The expected delivery time information indicates the latest delivery date required by the enterprise. The geographic location coordinate information is automatically converted from the delivery address pre-entered by the enterprise. At the same time, collecting real-time inventory data and historical delivery records of various suppliers in the park. The real-time inventory data is synchronized in real time through the data interface of the supplier's warehouse management system and includes dynamic information such as the current inventory, incoming inventory, and outgoing inventory of various types of labor protection supplies. The historical delivery records include order information completed by the supplier in the past six months, recording detailed data such as the order receipt time, actual delivery time, delivery quantity, and delivery location. The order demand information and supplier data are both standardized and stored using the unified data format of the industrial Internet platform.

[0030] S2: Based on the historical delivery records, calculate the order fulfillment time distribution curve of each supplier, use the kernel density estimation method to obtain the delivery capability probability density function of each supplier, and use the probability density function as the basic parameter for the supplier dynamic scoring (Innovation 1: Introducing the probability density function to evaluate the supplier's delivery capability);

[0031] Based on the historical delivery records, the order fulfillment time data of each supplier is first extracted, where the order fulfillment time refers to the time interval from the time the order is received to the actual delivery time, and the time interval is independently aggregated according to each supplier; for each supplier's order fulfillment time data set, a Gaussian kernel function is used for kernel density estimation, where the bandwidth parameter of the Gaussian kernel function is adaptively determined by the Silverman criterion, and a continuous probability density curve is obtained by performing kernel density estimation on the fulfillment time data, which reflects the probability distribution characteristics of the supplier completing the order within different fulfillment time intervals; on this basis, the probability density curve is numerically integrated to calculate the cumulative probability value of the supplier within the expected delivery time window, and the cumulative probability value is weightedly combined with the supplier's historical order completion volume to obtain a probability density function reflecting the supplier's delivery capability.

[0032] The process of performing kernel density estimation on fulfillment time data to obtain a continuous probability density curve is as follows: first, a fulfillment time dataset is extracted from the supplier's historical delivery records. When the sample size of the dataset exceeds one thousand records, the most recent one thousand fulfillment time records are preferentially selected as calculation samples. When the sample size is less than one thousand records, all fulfillment time records are used. For each fulfillment time record, a Gaussian kernel function is constructed with the time point as the center. The bandwidth parameter of the Gaussian kernel function adopts an adaptive adjustment mechanism. When the local data points are densely distributed, the bandwidth parameter is automatically reduced to improve the local detail performance of the curve. When the local data points are sparsely distributed, the bandwidth parameter is automatically increased to ensure the smoothness of the curve. The Gaussian kernel functions corresponding to all fulfillment time records are superimposed, and the superposition results are normalized to obtain a probability density estimation function reflecting the overall delivery time characteristics of the supplier. Finally, the probability density estimation function is standardized within the time interval of actual business concern so that its integral value within the interval is one, thereby obtaining a standardized probability density curve. The probability density curve intuitively reflects the probability distribution characteristics of the supplier completing order delivery within different fulfillment time periods.

[0033] The process of constructing a Gaussian kernel function with the time point as the center is as follows: first, mark each fulfillment time record as t and set it as the center point of the kernel function. When the center point corresponds to multiple repeated fulfillment records, record the number of occurrences of the center point n; extend the center point to both sides to construct a symmetrical bell-shaped distribution curve, so that the bell-shaped distribution curve reaches a peak at the center point t. When moving from the center point t to the time axis in any direction, the function value decreases according to the law of exponential decay; for any time point x, calculate the function value based on the time interval between it and the center point t. The larger the time interval, the smaller the corresponding function value, reflecting the difference in the degree of influence of the center point on different time points; if the center point t corresponds to multiple repeated records, then multiply the kernel function constructed by the center point by the number of repetitions n as a whole, so that the repeated fulfillment time has a higher weight in the overall distribution; after the calculation is completed, normalize the kernel function so that the integral value of the kernel function in the entire time domain is one, thereby obtaining a standardized Gaussian kernel function. More specifically, the Gaussian kernel function constructed based on the center point It can be expressed as follows:

[0034]

[0035] in, is the time point to be evaluated, is the center point of the kernel function, i.e., the fulfillment time record. is the number of repetitions at the center point t, The bandwidth parameter is used to control the broadening degree of the curve. is the time window parameter used to determine the integration interval, is the integration variable.

[0036] Furthermore, the cumulative probability value of the supplier within the expected delivery time window is calculated. First, the expected delivery time window of each order is determined. The expected delivery time window takes the latest delivery date required by the enterprise as the end point and extends forward three working days as the starting point to form a delivery time determination interval. Within the delivery time determination interval, the time step is selected as one hour, and the determination interval is evenly divided into several discrete time points. For each discrete time point, the probability density function obtained in the previous step is substituted into the probability density function to calculate the probability density value corresponding to the time point. After traversing all discrete time points in the determination interval, the trapezoidal integral method is used to calculate the probability density value of adjacent time points. The probability values ​​between the points in the delivery time window are calculated, and the probability values ​​of all intervals are accumulated to obtain the cumulative probability value within the delivery time determination interval; if the cumulative probability value of a supplier is significantly higher than that of other suppliers, it means that the supplier is more likely to complete the order delivery within the expected delivery time window; for different types of labor protection product orders, the determination interval is dynamically adjusted according to their urgency. When the order is an expedited type, the time span of the determination interval is shortened, and when the order is a regular type, the standard determination interval is maintained; finally, the cumulative probability value is weighted and combined with the supplier's historical order scale to obtain a scoring index that reflects the supplier's comprehensive delivery capability. More specifically, the scoring index can be expressed as follows:

[0037]

[0038] in, Provide comprehensive delivery capability scoring indicators for suppliers. is the order type adjustment factor (1.5 for expedited orders and 1.0 for regular orders), For time point The probability density value at is the interval between adjacent time points (fixed at 1 hour), is the supplier's historical order completion volume, is the benchmark order volume (the average order volume of all suppliers in the current evaluation period), is the total number of time points within the judgment interval. It should be noted that the comprehensive delivery capability scoring indicator is an application scenario of the probability density function reflecting the supplier's delivery capability, which has the same characteristics.

[0039] The process of calculating the probability value between adjacent time points using the trapezoidal integration method is as follows: first, obtain two adjacent time points t1 and t2 within the delivery time determination interval, substitute these two time points into the probability density function to obtain the corresponding probability density values ​​p1 and p2 respectively; when the probability density values ​​of two adjacent time points are both positive, use these two time points as endpoints to construct a trapezoidal area, where the upper base of the trapezoid is the larger probability density value, the lower base is the smaller probability density value, and the height is the time interval between the two time points; if the probability density values ​​corresponding to the two time points are equal, the constructed trapezoid degenerates into a rectangle; when the probability density values ​​are calculated, the trapezoid is converted into a rectangle. When calculating the trapezoidal area, the probability density values ​​of the two endpoints are added and divided by two to obtain the average probability density value, and then the average probability density value is multiplied by the time interval to obtain the probability value within this time period; if the probability density value within a certain time period fluctuates violently, the time period is further subdivided into smaller time intervals, and the trapezoidal area calculation process is repeated until the probability density value within the subdivided interval changes smoothly; after completing the probability value calculation between all adjacent time points, these probability values ​​are accumulated to obtain the cumulative probability value within the entire delivery time determination interval; finally, the cumulative probability value is normalized.

[0040] S3: Based on the delivery capability probability density function and the supplier's real-time inventory data, an order allocation weight matrix is ​​constructed. Based on the geographic spatial relationship between suppliers and enterprises, a gravity model is used to calculate each supplier's suitability for each order demand, and an initial order allocation plan is generated.

[0041] First, the probability density function value of each supplier's delivery capacity is obtained and used as the supplier's basic scoring indicator; when the supplier's real-time inventory data changes, the supplier's available inventory is updated in real time, and the ratio of available inventory to order demand is used as the inventory adequacy indicator; based on the supplier's basic scoring indicator and inventory adequacy indicator, an initial weight matrix is ​​constructed. When the supplier's inventory is insufficient, its score in the weight matrix is ​​reduced through an exponential decay function; the geographic coordinate information of the supplier's warehouse and the enterprise's delivery location is obtained, the straight-line distance between the supplier and the enterprise is calculated, and the said straight-line distance is substituted into the gravity model to calculate the spatial attenuation coefficient; when the distance between the supplier and the enterprise exceeds the set threshold, the corresponding spatial attenuation coefficient is increased. The number is set to zero, indicating that the supplier does not participate in the allocation of the enterprise's orders; the initial weight matrix is ​​multiplied by the spatial attenuation coefficient to obtain a comprehensive weight matrix that takes into account the geographical location factor; each cell in the comprehensive weight matrix is ​​normalized so that the sum of the weights of all suppliers corresponding to each order is one; based on the normalized comprehensive weight matrix, a proportional allocation method is used to allocate suppliers to each order. When the remaining inventory of a supplier is insufficient to meet the allocation share, the excess will be reallocated to other suppliers with supply capabilities according to the weight ratio; finally, an initial order allocation plan is generated based on the allocation results. The initial order allocation plan contains information such as the supplier allocation ratio, supply quantity and delivery distance for each order.

[0042] The specific construction of the gravity model: first, obtain the latitude and longitude coordinates of the supplier's warehouse location and the latitude and longitude coordinates of the company's delivery location, and use the spherical distance formula to calculate the actual geographical distance between the two points; when the geographical distance is less than one kilometer, set the distance value to one kilometer to avoid abnormal gravity values ​​at extremely close distances; construct a gravity calculation formula, use the supplier's delivery capacity probability density function value as the subject quality item in the gravity formula, use the company's order demand as the object quality item in the gravity formula, and use the geographical distance between the two points as the distance item in the gravity formula.

[0043] The process of constructing the initial weight matrix based on the supplier's basic scoring index and inventory adequacy index is as follows: first, read the probability density function value of each supplier within the expected delivery time window as the basic scoring value. When the basic scoring value of a supplier is significantly lower than the average level, it is marked as an inefficient supplier; obtain the current real-time inventory data of each supplier, and calculate its inventory coverage ratio for various labor protection product order demands. When the inventory is greater than the order demand, the inventory coverage ratio is recorded as one; when the inventory is less than the order demand, the inventory coverage ratio is the ratio of the actual inventory to the order demand; for each combination of supplier and order demand, the basic scoring value and inventory coverage ratio are compared. The initial weight value is obtained by row-weighted calculation, where the weight coefficient of the basic score value is 0.6, and the weight coefficient of the inventory coverage ratio is 0.4; when the supplier's inventory coverage ratio is lower than 0.2, the exponential decay function is used to quickly reduce its initial weight value to prevent suppliers with serious inventory shortages from participating in order allocation; if the supplier has recently defaulted on an order due to inventory shortages, an additional penalty factor is introduced when calculating the initial weight value to further reduce its weight value; the initial weight values ​​of all supplier and order demand combinations are organized into a matrix form, where the rows of the matrix represent suppliers and the columns represent order demands, and each element in the matrix represents the initial weight value of the corresponding supplier and order combination.

[0044] The calculation and setting process of the penalty factor is as follows: first, the number of order defaults caused by inventory shortages by suppliers in the most recent quarter is counted. When there is no record of default, the penalty factor takes the full value; when the first default occurs, the basic penalty coefficient is determined according to the inventory coverage ratio at the time of default, wherein the basic penalty coefficient is set using a piecewise function. When the inventory coverage ratio is lower than the first warning threshold, the basic penalty coefficient takes a value within the severe punishment interval, and the upper limit of the severe punishment interval does not exceed the first critical value; when the inventory coverage ratio is between the first warning threshold and the second warning threshold, the basic penalty coefficient takes a value within the light punishment interval, and the lower limit of the light punishment interval is not lower than the second critical value; when a supplier defaults multiple times within a month, the penalty factor is calculated using a cumulative multiplication method, that is, for each additional default, the current penalty factor is multiplied by the corresponding basic penalty coefficient. A lower limit is set for the cumulative multiplication result to prevent excessive punishment; if the supplier's breach of contract causes serious delays in the customer's production plan, an additional aggravated penalty coefficient shall be multiplied when calculating the penalty factor, and the aggravated penalty coefficient shall not be higher than the upper limit of the severe punishment range of the basic penalty coefficient; when the supplier takes effective measures and completes the supply without breach of contract beyond the prescribed period, its penalty factor begins to gradually recover according to the set cycle ratio, and the recovery rate should ensure that the supplier can return to normal levels within a reasonable period of time while continuing to meet the standards; if the supplier can proactively notify the inventory shortage risk in advance and adjust the delivery plan in time, a mitigation coefficient shall be introduced when calculating the penalty factor, and the value of the mitigation coefficient shall be higher than the lower limit of the light punishment range of the basic penalty coefficient; finally, the calculated penalty factor is multiplied by the supplier's initial weight value to obtain the revised weight value after considering the breach of contract record.

[0045] It should be noted that the process of determining the first warning threshold and the second warning threshold is as follows: first, based on the historical inventory coverage data of all suppliers in the industrial park, a dynamic benchmark library is established, and the dynamic benchmark library stores inventory coverage data according to different types of labor protection products; the historical data in the dynamic benchmark library is subjected to time series analysis to extract the long-term trend of inventory coverage; according to the supply characteristics of various types of labor protection products, the supplier's inventory coverage data is divided into peak period data and normal period data, wherein the peak period includes holiday preparation period and production peak season period; the mean and standard deviation of inventory coverage rate in peak period and normal period are calculated respectively, and combined with order fulfillment data, the correlation characteristics between inventory coverage rate and order default risk are identified through cluster analysis; based on the results of the correlation characteristic analysis, the value of the normal period inventory coverage rate mean minus one standard deviation is set as the second warning threshold, and when the supplier's inventory coverage rate is lower than this threshold, it enters a mild warning state; the value of the normal period inventory coverage rate mean minus two standard deviations is set as the first warning threshold, and when the supplier's inventory coverage rate is lower than this threshold, it enters a severe warning state.

[0046] The process of determining the first critical value and the second critical value is as follows: collect the supplier's behavior improvement data under different penalty intensities, and the behavior improvement data includes indicators such as inventory replenishment speed, information feedback timeliness, and emergency plan execution effect; after standardizing each indicator, construct a supplier response curve, and when the penalty intensity continues to increase, record the inflection point where the supplier's behavior improvement effect shows a significant improvement, and set the penalty intensity corresponding to the inflection point as the second critical value, which is the lower limit of the light punishment range; continue to increase the penalty intensity, and when the marginal improvement in the supplier's behavior improvement effect begins to decline significantly, set the penalty intensity corresponding to this point as the first critical value, which is the upper limit of the severe punishment range.

[0047] S4: Perform spatial clustering analysis on the delivery paths in the initial order allocation plan, automatically aggregate orders with similar delivery paths based on the improved isochronous circle division method, and generate the final order scheduling plan through a dynamic spatiotemporal combination optimization algorithm.

[0048] The process of automatically aggregating orders with similar delivery paths based on the improved isochronous circle division method is as follows: first, the latitude and longitude coordinates of the delivery address of each order are obtained, and a two-dimensional coordinate system is established with the supplier warehouse location as the origin; when the number of orders is large, the delivery area is divided into several delivery units using a grid division method, and the size of the delivery unit is adaptively adjusted according to the order density; the straight-line distance between each order and the supplier warehouse is calculated, and a basic isochronous circle in the form of a concentric circle is constructed based on the straight-line distance. When the order distribution is uneven, the radius interval of the isochronous circle is dynamically adjusted; for each order in the basic isochronous circle, the relative distance between the order addresses is calculated. When the relative distance between two orders is less than a first set threshold, the two orders are marked as distance-related orders; the distance is analyzed. The delivery time windows of associated orders are automatically aggregated into the same delivery cluster when the overlap of the delivery time windows of two orders is greater than the second set threshold. If the number of orders in a delivery cluster is less than the minimum order number threshold, the distance threshold is expanded and the orders are re-aggregated until the minimum order number requirement is met. When the total number of orders in a delivery cluster is close to the vehicle load limit, new orders are stopped from being added to the delivery cluster. The geometric center of the order address is calculated for each delivery cluster, and the combination of delivery clusters is optimized according to the distribution law of the geometric center. If the distance between the geometric centers of adjacent delivery clusters is less than the third set threshold and the total order volume does not exceed the vehicle load limit, these delivery clusters are merged. Finally, the orders in the delivery cluster are optimized in time sequence to ensure that the delivery sequence meets the time window constraints of all orders.

[0049] The calculation logic for calculating the geometric center of the order address is as follows: first, obtain the address coordinates of each order in the delivery cluster, and express the address coordinates as two components of longitude and latitude; set a weight coefficient for each order according to its delivery volume. When the order delivery volume is one base unit, the weight coefficient is set to one; when the order delivery volume is multiple base units, the weight coefficient is equal to the ratio of the delivery volume to the base unit; calculate the ratio of the weighted sum of the longitude components of all order coordinates and the weighted sum to obtain the longitude value of the geometric center, and the weighted sum is equal to the sum of the products of the longitude value of each order and its corresponding weight; use the same method to calculate the ratio of the weighted sum of the latitude components of all order coordinates and the weighted sum to obtain the latitude value of the geometric center, and the weighted sum is equal to the sum of the products of the latitude value of each order and its corresponding weight; when the order delivery volume is the same, the geometric center coordinates are directly obtained by calculating the arithmetic mean of the order coordinates, and the arithmetic mean is equal to the sum of the order coordinates divided by the number of orders; if the distance between the coordinates of an order and the geometric center exceeds the set threshold, the weight coefficient of the order is adjusted to the square root of the original value, and the geometric center coordinates are recalculated.

[0050] The process of determining the first set threshold, the second set threshold and the third set threshold is as follows: statistically analyze the delivery efficiency indicators under different delivery distances, different time window combinations, and different delivery cluster merging schemes; group the historical delivery distance data through the hierarchical clustering method to identify the natural separation point of the delivery distance. When the delivery distance is less than the separation point, the delivery efficiency is significantly improved. Based on this, the separation point is set as the initial value of the first set threshold; analyze the delivery cases with overlapping time windows in historical orders, calculate the correlation between the time window overlap and the delivery cost, and when the overlap reaches a certain value, the delivery cost begins to decrease significantly. This value is set as the initial value of the second set threshold; count the completed delivery cluster merging records, extract the corresponding relationship between the distance between the geometric centers of the delivery clusters and the delivery efficiency after the merger, and when the distance exceeds a critical value, the merging effect begins to decrease significantly. This critical value is set as the initial value of the third set threshold.

[0051] Furthermore, the process of generating the final order scheduling plan through the dynamic spatiotemporal combination optimization algorithm is as follows: first, construct an order spatiotemporal feature matrix, which contains attribute information such as the delivery time window, delivery address, order priority and order weight of each order; based on the rolling time domain division method, the delivery period is divided into multiple scheduling cycles, and the length of each scheduling cycle is adaptively adjusted according to the order arrival frequency; when entering a new scheduling cycle, orders with urgent time window constraints are processed first, and they are preferentially allocated to the currently available delivery resources; if the delivery time window of an order spans multiple scheduling cycles, the delivery cost of the order in different scheduling cycles is calculated, and the scheduling cycle with the lowest cost is selected for allocation; for the formed delivery cluster, the time window distribution of its internal orders is analyzed, and when there is an interval with concentrated time windows, the interval is set as the core delivery period of the delivery cluster; if the delivery cluster If the time windows of orders within a delivery cluster are discretely distributed, a time window negotiation mechanism is used to try to adjust the delivery time of some orders to move them closer to the core delivery period. The delivery time and driving distance of each delivery cluster are calculated. When the execution cost of a delivery cluster exceeds a set threshold, the delivery cluster is split into multiple sub-delivery clusters. The degree of spatiotemporal overlap between adjacent delivery clusters is analyzed. When the core delivery periods of two delivery clusters are similar and the delivery areas are adjacent, the two delivery clusters are considered for merging. An initial delivery sequence is generated for all delivery clusters, and the delivery order is optimized using an improved local search algorithm. When a more optimal delivery sequence is found, the plan is automatically updated. If new urgent orders appear during the delivery process, they are dynamically inserted according to the remaining delivery capacity, and existing delivery clusters are reorganized if necessary. Finally, a detailed delivery scheduling schedule is generated, which contains information such as the delivery vehicle, departure time, and estimated arrival time for each order.

[0052] The process of optimizing the delivery sequence by the improved local search algorithm is as follows: first, the initial delivery sequence is decoded, and a neighborhood structure of the order delivery sequence is constructed, which includes three basic operations: order exchange, order insertion, and order reversal; the target value of the current delivery sequence is calculated, and when the search optimization starts, a group of orders are randomly selected in the neighborhood of the current solution for basic operations. If the basic operation causes the time window constraint to be violated, it automatically falls back to the last feasible solution; for each newly generated delivery sequence, its evaluation function value is calculated, and when the evaluation value of the new sequence is better than the current optimal solution, it is updated to the current optimal solution; if a better solution cannot be found after multiple consecutive searches, the neighborhood search range is expanded, and a composite movement strategy is constructed by combining multiple basic operations; when a localization occurs in the search process, the evaluation function value is calculated, and when the evaluation value of the new sequence is better than the current optimal solution, the new sequence ... is not found after multiple consecutive searches, the neighborhood search range is expanded, and a composite movement strategy is constructed by combining multiple basic operations; when a localization occurs in the search process, the evaluation function value is calculated, and when a localization occurs in the search process, the evaluation function value is calculated, and when a localization occurs in the search process, the evaluation function value is calculated, and when a localization occurs in the search process, the evaluation function value is calculated, and when a localization occurs in the search process, the evaluation function value is calculated, and when a localization occurs in the search process, the evaluation function value is calculated, and when a localization occurs in the search process, the evaluation function value is calculated, and when a localization occurs in the search process, the evaluation function value is calculated, and when a localization occurs in the search process, the evaluation function value is calculated, and when a localization occurs in When the local optimal state is reached, a perturbation mechanism is introduced to break the current local optimal state, and the perturbation mechanism includes operations such as random segment rearrangement and local inversion. A dynamic adjustment mechanism for the search step is set. When a better solution is continuously found, the search step is reduced to refine the search. When there is no improvement for a long time, the search step is increased to jump out of the local optimal state. For delivery sequences of different sizes, the generation strategy of the neighborhood structure is automatically adjusted. When the sequence is short, the precise search is preferred, and when the sequence is long, the heuristic search is preferred. The search history is regularly analyzed to identify the characteristic patterns of high-quality solutions, and these characteristic patterns are used as the priority direction of the search. Finally, the termination condition is used to determine whether to end the search. When the maximum number of iterations is reached or several consecutive searches fail to improve the target value, the current optimal delivery sequence is output.

[0053] The calculation of the evaluation function value is shown in the following formula:

[0054]

[0055] in, is the evaluation function value, is the default degree of the time window of the i-th order (completely satisfied is 0, partial default is a value between 0 and 1), is the delivery distance between adjacent orders i and i+1, is the load balance adjustment factor, is the variance between the vehicle’s current load and the average load, is the current delivery sequence length, is the maximum allowed delivery sequence length, is the number of orders in the delivery sequence.

[0056] In summary, the method for dynamic scheduling and inventory management of labor protection supplies orders based on the industrial Internet according to the embodiment of the present invention is explained. It obtains the order demand information of each enterprise in the industrial park and the real-time inventory data and historical delivery records of the supplier, dynamically constructs the probability density function of the supplier's delivery capacity, and combines the gravity model to calculate the compatibility between the supplier and the order demand, thereby generating an accurate order allocation plan. In this way, the rationality and efficiency of order allocation can be improved, and order delays and resource waste can be reduced. In addition, through the improved isochronous circle division method and dynamic spatiotemporal combination optimization algorithm, cluster analysis and scheduling optimization of the distribution path can be carried out, which can significantly improve the timeliness of distribution and the rationality of the path, thereby reducing logistics costs and improving the overall logistics management efficiency of the industrial park.

[0057] Here, those skilled in the art will understand that the specific operations of each step in the above-mentioned method for dynamic scheduling and inventory management of labor protection products orders based on the industrial Internet have been referred to above. Figure 1 It has been introduced in detail in the description of the dynamic scheduling and inventory management method of labor protection supplies orders based on industrial Internet, and therefore, its repeated description will be omitted.

Claims

1. The method for dynamic scheduling and inventory management of labor protection supplies orders based on the Industrial Internet is characterized by: include: Obtain labor protection supplies order demand information reported by enterprises within the industrial park, and collect real-time inventory data and historical delivery records of suppliers within the park; Based on the historical delivery records, calculate the order fulfillment time distribution curve of each supplier to obtain the delivery capacity probability density function of each supplier, and use the probability density function as the basic parameter for the supplier dynamic scoring; Based on the delivery capability probability density function and the supplier's real-time inventory data, an order allocation weight matrix is ​​constructed. Based on the geographic spatial relationship between suppliers and enterprises, a gravity model is used to calculate each supplier's suitability for each order demand, and an initial order allocation plan is generated. Based on the improved isochronous circle division method, orders with similar delivery paths are automatically aggregated, and the final order scheduling plan is generated through a dynamic spatiotemporal combination optimization algorithm.

2. The method for dynamic scheduling and inventory management of labor protection products orders based on the Industrial Internet according to claim 1 is characterized in that: The order fulfillment time distribution curve is obtained by extracting the fulfillment time dataset from the historical delivery records, constructing a Gaussian kernel function centered on the fulfillment time records of the fulfillment time dataset, superimposing the Gaussian kernel function and normalizing it.

3. The method for dynamic scheduling and inventory management of labor protection products orders based on the Industrial Internet according to claim 2 is characterized in that: A Gaussian kernel function centered on the performance time record is constructed, and a weight-adjusted symmetrical bell-shaped distribution curve is constructed according to the number of repetitions of the performance time record. The function value is then calculated through the time interval, and after normalization, a standardized Gaussian kernel function with an integral value of one is obtained.

4. The method for dynamic scheduling and inventory management of labor protection products orders based on the Industrial Internet according to claim 1 is characterized in that: The dynamic score of the supplier is obtained by constructing an expected delivery time window, calculating the cumulative probability value within it, and dynamically adjusting the judgment interval in combination with the order type. Finally, the cumulative probability value is weightedly combined with the historical order size to obtain a scoring indicator to measure the supplier's comprehensive delivery capability.

5. The method for dynamic scheduling and inventory management of labor protection products orders based on the Industrial Internet according to claim 4 is characterized in that: The order allocation weight matrix is ​​constructed by combining the basic rating value and inventory coverage of the supplier, calculating the initial weight value of the supplier and order demand combination, and introducing a penalty factor to correct the initial weight value, and finally generating the order allocation weight matrix.

6. The method for dynamic scheduling and inventory management of labor protection products orders based on the Industrial Internet according to claim 5 is characterized in that: The penalty factor is set based on whether the supplier has breached the contract; When a supplier defaults for the first time, a basic penalty coefficient is determined based on the inventory coverage ratio at the time of default, where the basic penalty coefficient is set by a piecewise function: When the inventory coverage ratio is lower than the first warning threshold, the basic penalty coefficient is taken from the severe penalty range, and the upper limit of the severe penalty range does not exceed the first critical value; When the inventory coverage ratio is between the first warning threshold and the second warning threshold, the basic penalty coefficient is taken from the light penalty range, and the lower limit of the light penalty range is not lower than the second critical value; When a supplier defaults multiple times, the penalty factor is calculated using a cumulative multiplication method.

7. The method for dynamic scheduling and inventory management of labor protection products orders based on the Industrial Internet according to claim 6 is characterized in that: The gravity model uses the probability density function value of the supplier's delivery capability as the subject quality item in the gravity formula, the enterprise's order demand as the object quality item in the gravity formula, and the geographical distance between the two points as the distance item in the gravity formula.

8. The method for dynamic scheduling and inventory management of labor protection products orders based on the Industrial Internet according to claim 1 is characterized in that: Orders with similar delivery paths are automatically aggregated based on the supplier warehouse as the origin, and delivery clusters are constructed through grid division, adaptive adjustment of delivery units and dynamically adjusted time circles. The delivery clusters are gradually optimized in combination with restrictions and the geometric center of the order address.

9. The method for dynamic scheduling and inventory management of labor protection products orders based on the Industrial Internet according to claim 1 is characterized in that: The order scheduling scheme is generated by constructing an order spatiotemporal feature matrix and rolling time domain division, allocating orders to scheduling cycles according to priority, time window and delivery cost, and combining the optimization of delivery clusters to generate a detailed delivery schedule including vehicle arrangement and time arrangement.

10. The method for dynamic scheduling and inventory management of labor protection products orders based on the Industrial Internet according to claim 9 is characterized in that: The optimization of the delivery cluster optimizes the delivery sequence through neighborhood operations, uses a combination of dynamic step size, perturbation mechanism and feature guidance to jump out of the local optimum, and finally outputs the optimal solution that meets the termination conditions.

Images