An integrated platform and method based on production materials, MRO and temporary procurement

Abstract

The present invention belongs to the technical field of data processing. The present invention discloses an integrated platform and method based on production materials, MRO, and temporary procurement, including a data acquisition module for acquiring production supply chain data; a data processing module for processing the production supply chain data to obtain a supply chain comprehensive feature data set; a demand forecasting module for using the supply chain comprehensive feature data as the input of a material demand forecasting model to forecast the material demand within the next n time periods; an inventory optimization module for using RFID technology to obtain real-time inventory data, obtaining an inventory gap in combination with the predicted material demand within the next n time periods, and determining whether inventory adjustment is required based on the inventory gap; if inventory adjustment is required, an optimal inventory adjustment plan is obtained through a spectral particle optimization algorithm in combination with a spectrum perturbation and momentum factor restriction mechanism, ensuring the smooth operation of the production line while improving the inventory turnover rate.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and more specifically, to an integrated platform and method based on production materials, MRO, and temporary procurement. Background Art

[0002] A patent with the publication number CN104281902A discloses a method for formulating material requirements suggestions. The method includes: obtaining a demand source, where the demand source is one or more of the latest master production plan, the latest sales order, or a temporary plan; obtaining all the demanded items involved in the demand source and the demand quantity of each demanded item; obtaining the static available quantity of each item, where the static available quantity includes: the inventory quantity of each item, the available supply quantity of each item involved in each planning element before obtaining the demand source; the planning elements include: one or more of production suggestions, procurement suggestions, subcontracting suggestions, production plans, procurement plans, production orders, procurement orders, and subcontracting orders; using the static available quantity of each item to offset the demand quantity of the demanded item with the same name as the item, and the result is displayed as the material requirements suggestion. The present invention can quickly and accurately adjust the demand quantity of each demanded item when the demand changes.

[0003] The existing integrated platforms and methods for production materials, MRO, and temporary procurement have the following main problems:

[0004] It is unable to effectively manage and optimize the key cost parts in the inventory (ordering cost, holding cost, shortage cost), resulting in unreasonable cost control, unable to effectively balance these costs in actual operations, and affecting the maximization of profits; due to the lack of accurate identification and control of costs, unnecessary resource waste and inventory backlog are caused; the dynamic adaptability of multi-variety inventory management is not considered, which may lead to the system being only applicable to the management of a single material and unable to flexibly respond to the demand fluctuations of multiple material categories; it is unable to dynamically adjust the inventory according to the actual demand situation of different materials, thereby increasing the complexity and uncertainty of inventory management; there is no effective inventory adjustment plan, which may lead to a lack of a clear optimization path and difficulty in effectively optimizing the inventory during the solution process of the inventory optimization problem;

[0005] Lack of flexible and efficient optimization strategies. The search space of traditional inventory optimization methods may be too narrow, resulting in the optimization process being unable to fully explore the solution space and easily falling into local optimal solutions at an early stage, reducing the efficiency and accuracy of the algorithm; for large-scale, multi-variety inventory optimization problems, there is a lack of adaptive and flexible adjustment mechanisms, making the algorithm unable to adapt to the ever-changing demands and complex material management strategies.

[0006] In view of this, the present invention proposes an integrated platform and method based on production materials, MRO, and temporary procurement to solve the above problems. Summary of the Invention

[0007] To overcome the above-mentioned defects of the prior art and to achieve the above object, the present invention provides the following technical solution: An integrated platform based on production materials, MRO, and temporary procurement, comprising:

[0008] A data acquisition module for acquiring production supply chain data;

[0009] A data processing module for processing the production supply chain data to obtain a supply chain comprehensive feature data set;

[0010] A demand forecasting module for using the supply chain comprehensive feature data as the input of a material demand forecasting model to forecast the material demand within the next n time periods;

[0011] An inventory optimization module that uses RFID technology to obtain real-time inventory data, combines the predicted material demand within the next n time periods to obtain an inventory gap, and determines whether inventory adjustment is required based on the inventory gap; if inventory adjustment is required, then through a spectral particle optimization algorithm combined with a spectrum perturbation and momentum factor limiting mechanism, an optimal inventory adjustment plan is obtained;

[0012] A decision feedback module for transmitting the optimal inventory adjustment plan to an intelligent monitoring terminal, issuing an adjustment instruction through the intelligent monitoring terminal, and dynamically adjusting the inventory level; each module is connected by wired and / or wireless means.

[0013] Further, the production supply chain data includes production material data, MRO parameter data, and temporary procurement data; the production material data includes material name, material specification, inventory quantity, usage cycle, supplier information, and procurement cost; the MRO parameter data includes production equipment code, production equipment model, maintenance cycle, failure rate, maintenance record, and MRO material inventory; the temporary procurement data includes temporary procurement quantity, applying department, temporary procurement material demand, temporary procurement supplier information, and temporary procurement cost.

[0014] Further, the method for processing the production supply chain data includes:

[0015] Adopting the Z-score algorithm to detect outliers in the production supply chain data and removing the identified outliers; filling the missing values in the production supply chain data after removing outliers by linear interpolation method to obtain production supply chain parameter data; performing standard deviation normalization on the production supply chain parameter data to convert it into a standard normal distribution with a mean of 0 and a standard deviation of 1, and then obtaining the normalized production supply chain parameter data.

[0016] Further, the method for obtaining the supply chain comprehensive feature data set includes;

[0017] Perform data dimensionality reduction on the normalized production supply chain parameter data through principal component analysis to obtain a production supply chain feature dataset; use the Apriori algorithm to mine association rules for the normalized production supply chain parameter data to obtain an association feature dataset; fuse the association feature dataset and the production supply chain feature dataset through a weighted model to further obtain a comprehensive supply chain feature dataset; denote the association feature dataset as , and denote the production supply chain feature dataset as ;

[0018] The weighted model is: ; where is the weight coefficient of the association feature dataset; is the weight coefficient of the production supply chain feature dataset.

[0019] Furthermore, the method for obtaining the association feature dataset includes:

[0020] S51. Discretize the normalized production supply chain parameter data and convert it into the form of transaction data; preset the minimum support threshold as , and the minimum confidence threshold as ; initialize the frequent item set as an empty set;

[0021] S52. For each parameter combination, calculate the support in the normalized production supply chain parameter data, retain the parameter combinations with support greater than or equal to the preset minimum support threshold , generate -item frequent item sets, and repeat the iteration until no new frequent item sets are generated; collect the frequent item sets generated in each iteration to further obtain a comprehensive frequent item set;

[0022] S53. For each frequent item set in the comprehensive frequent item set, generate corresponding association rules; for each association rule, calculate its confidence and retain the association rules with confidence greater than or equal to the minimum confidence threshold;

[0023] S54. Collect all association rules with confidence greater than or equal to the minimum confidence threshold to further obtain the association feature dataset.

[0024] Furthermore, the training method of the material demand prediction model includes:

[0025] Divide the dataset into a training set, a validation set, and a test set, train the model and evaluate the model performance; the sample set is a subset of the dataset, and each sample set includes the historical comprehensive supply chain feature dataset and the corresponding material demand within the next n time periods;

[0026] Build a material demand prediction model using the deep learning library TensorFlow; the material demand prediction model includes an input layer, an LSTM layer, and an output layer; the input layer of the model is used to input the historical supply chain comprehensive feature dataset; the output layer of the model is used to output the material demand within the next n periods; the material demand prediction model is an LSTM model;

[0027] Define the loss function of the model, and use the L2-regularized mean squared error loss function to measure the difference between the predicted value and the true value of the model; use the training set to train the material demand prediction model, and update the model parameters through the backpropagation algorithm to minimize the loss function; use the validation set to evaluate the performance of the material demand prediction model by calculating the accuracy metric;

[0028] Select the Adam optimization algorithm as the optimizer, tune the model according to the performance feedback of the validation set, and stop adjusting the model parameters until the performance no longer improves or reaches the preset number of iterations; use the test set to evaluate the performance of the model in the prediction task, and use the trained material demand prediction model to predict the current supply chain comprehensive feature dataset to obtain the material demand within the next n periods.

[0029] Furthermore, the method for determining whether inventory adjustment is needed based on the inventory gap includes:

[0030] Use RFID technology to obtain real-time inventory data, and the real-time inventory data includes the current inventory quantity, storage location, and material status;

[0031] Subtract the current inventory quantity from the predicted material demand within the next n periods to obtain the inventory gap; preset an inventory gap threshold, and compare the inventory gap with the preset inventory gap threshold;

[0032] If the inventory gap is less than the preset inventory gap threshold, it is determined that no inventory adjustment is needed; if the inventory gap is greater than or equal to the preset inventory gap threshold, it is determined that inventory adjustment is needed.

[0033] Furthermore, the method for obtaining the optimal inventory adjustment plan includes:

[0034] S81. Preset the goal of the inventory adjustment plan to minimize the inventory cost, and define the objective function as: ; where, is the objective function; is the unit ordering cost; is the adjustment quantity of the th type of material; is the unit inventory holding cost; is the unit shortage cost; is the demand for the th type of material; is the index of material categories; is the total number of material categories;

[0035] S82. Define that the position of each spectral particle represents an inventory adjustment plan, and the spectral particle ; where is the adjustment amount of the th type of material; is the total number of adjustment amounts of materials;

[0036] S83. Initialize the positions of spectral particles and the corresponding spectral frequencies , and initialize the spectral particles: ; where is the initial position of the th spectral particle; is the lower bound of the material inventory adjustment amount; is the upper bound of the material inventory adjustment amount; is the random number of the initial spectral particle position;

[0037] Initialize the spectral frequencies: ; where is the initial frequency of the th spectral particle; is the minimum value of the spectral frequencies; is the maximum value of the spectral frequencies; is the random number of the initial spectral particle frequency;

[0038] S84. Update the positions and velocities of spectral particles through the position update formula and the velocity update formula based on the energy propagation law of the frequency spectrum; the position update formula is: ; where is the position of the spectral particle at the th iteration; is the position of the spectral particle at the th iteration; is the velocity of the spectral particle at the th iteration;

[0039] The velocity update formula is: ; where is the velocity of the spectral particle at the th iteration; is the velocity of the spectral particle at the th iteration; represents the momentum factor; is the learning factor; is the gradient of the objective function; is the frequency spectrum perturbation factor; is the spectral oscillation term; is the temperature of the spectral particle at the th iteration;

[0040] S85. The temperature of the spectral particle during iteration is restricted by the temperature decay formula, and the temperature decay formula is: ; where is the initial temperature of the spectral particle; is the temperature control factor; is the current iteration number;

[0041] The spectral perturbation factor is dynamically adjusted by the spectral perturbation adjustment formula, and the spectral perturbation adjustment formula is: ; where is the initial maximum spectral perturbation factor; is the adjustment parameter that controls the decay rate of the perturbation factor with the iteration number ; is the total number of spectral particles; the momentum factor is restricted by the momentum factor restriction formula;

[0042] S86. According to the fitness of the spectral particle, the spectral frequency is updated by the spectral update formula; the spectral update formula is: ; where is the current global optimal fitness; is the spectral adjustment factor; is at the th spectral particle at the th iteration of the spectral frequency; is at the th spectral particle at the th iteration of the spectral frequency;

[0043] S87. In each iteration, calculate the fitness of each spectral particle and update the global optimal solution , preset the global optimal solution threshold, and stop when the spectral optimization algorithm reaches the maximum iteration number or the global optimal solution is less than the preset global optimal solution threshold; the global optimal solution at this time is the optimal inventory adjustment plan.

[0044] Furthermore, the method for restricting and constraining the momentum factor by the momentum factor restriction formula includes:

[0045] The momentum factor restriction formula is: ; where is the maximum value of the momentum factor; is the maximum iteration number; is the absolute value of the gradient of the objective function; It is a regulation parameter for balancing the influence of gradient and particle number on the momentum factor.

[0046] An integrated method based on production materials, MRO, and temporary procurement, comprising:

[0047] S1. Obtain production supply chain data;

[0048] S2. Process the production supply chain data to obtain a comprehensive supply chain feature dataset;

[0049] S3. Use the comprehensive supply chain feature data as the input of the material demand prediction model to predict the material demand within the next n periods;

[0050] S4. Use RFID technology to obtain real-time inventory data, combine with the predicted material demand within the next n periods to obtain the inventory gap, and judge whether inventory adjustment is needed based on the inventory gap; if inventory adjustment is needed, use the spectral particle optimization algorithm combined with the spectrum perturbation and momentum factor restriction mechanism to obtain the optimal inventory adjustment plan;

[0051] S5. Transmit the optimal inventory adjustment plan to the intelligent monitoring terminal, and issue an adjustment instruction through the intelligent monitoring terminal to dynamically adjust the inventory level.

[0052] The technical effects and advantages of the integrated platform and method of the present invention based on production materials, MRO, and temporary procurement:

[0053] In the present invention, the objective function clearly decomposes the inventory cost into three parts: ordering cost, holding cost, and shortage cost, which are the core cost sources of inventory management; by considering the adjustment quantity and demand of each type of material, it is ensured that the objective function can dynamically adapt to multi-variety inventory management; the position of each spectral particle represents an inventory adjustment plan, and this mapping method is intuitive and easy to implement the optimization algorithm; the upper and lower bounds and random numbers are used to generate the initial solution to ensure the diversity of the search space and avoid falling into local optimal solutions; the position update formula combines the current position, velocity, and gradient of the objective function of the particle, making the particle have directivity when searching the solution space;

[0054] The speed update formula introduces momentum factor, learning factor and spectrum perturbation factor to control inertia, learning ability and dynamic perturbation respectively, enhancing the exploration and exploitation capabilities of the algorithm; temperature controls the randomness of particles, and in an exponentially decaying manner, ensures that the algorithm widely searches the solution space in the early stage and gradually focuses on the vicinity of the global optimal solution in the later stage; by dynamically adjusting the spectrum perturbation factor, it adapts to particle swarms of different scales while retaining the dynamic adjustment characteristics of the algorithm during iteration; it avoids the influence of too large or too small momentum factor on the search trajectory and ensures the stability of particle movement; through the dynamic adjustment of the global optimal fitness and spectrum adjustment factor, the spectral particles can approach the optimal solution faster; it enhances the search ability of the algorithm and the convergence performance of the global optimal solution, and has high practical value for solving complex inventory optimization problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 FIG. is a schematic structural diagram of an integrated platform based on production materials, MRO and temporary procurement according to the present invention;

[0056] Figure 2 FIG. is a schematic flowchart of an integrated method based on production materials, MRO and temporary procurement according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0057] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0058] Embodiment 1

[0059] Please refer to Figure 1 shown. An integrated platform based on production materials, MRO and temporary procurement in this embodiment includes:

[0060] A data acquisition module for acquiring production supply chain data;

[0061] A data processing module for processing the production supply chain data to obtain a supply chain comprehensive feature data set;

[0062] A demand forecasting module for using the supply chain comprehensive feature data as the input of a material demand forecasting model to forecast the material demand in the next n periods of time;

[0063] Inventory optimization module, which uses RFID technology to obtain real-time inventory data, combines the predicted material requirements in the next n periods of time to obtain the inventory gap, and determines whether inventory adjustment is needed based on the inventory gap; if inventory adjustment is needed, it uses the spectral particle optimization algorithm combined with the spectrum perturbation and momentum factor limit mechanism to obtain the optimal inventory adjustment plan;

[0064] Decision feedback module, which is used to transmit the optimal inventory adjustment plan to the intelligent monitoring terminal, issue adjustment instructions through the intelligent monitoring terminal, and dynamically adjust the inventory level; each module is connected by wired and / or wireless means.

[0065] Production supply chain data includes production material data, MRO parameter data, and temporary procurement data; production material data includes material name, material specification, inventory quantity, usage cycle, supplier information, and procurement cost; MRO parameter data includes production equipment code, production equipment model, maintenance cycle, failure rate, maintenance record, and MRO material inventory; temporary procurement data includes temporary procurement quantity, application department, temporary procurement material requirements, temporary procurement supplier information, and temporary procurement cost.

[0066] The method for processing production supply chain data includes:

[0067] Adopt the Z-score algorithm to detect outliers in the production supply chain data and eliminate the identified outliers; fill in the missing values in the production supply chain data after eliminating outliers through linear interpolation to obtain the production supply chain parameter data; perform standard deviation normalization on the production supply chain parameter data and convert it into a standard normal distribution with a mean of 0 and a standard deviation of 1, and then obtain the normalized production supply chain parameter data.

[0068] The method for obtaining the comprehensive supply chain feature dataset includes;

[0069] Perform data dimensionality reduction on the normalized production supply chain parameter data through principal component analysis to obtain the production supply chain feature dataset; use the Apriori algorithm to mine association rules from the normalized production supply chain parameter data to obtain the association feature dataset; fuse the association feature dataset and the production supply chain feature dataset through a weighted model, and then obtain the comprehensive supply chain feature dataset; denote the association feature dataset as The production supply chain feature dataset is denoted as ;

[0070] The weighted model is: ; where Is the weight coefficient of the association feature dataset; The weight coefficient of the production supply chain feature dataset.

[0071] The method for obtaining the associated feature dataset includes:

[0072] S51. Discretize the normalized production supply chain parameter data and convert it into the form of transaction data; preset the minimum support threshold as , and the minimum confidence threshold as ; initialize the frequent item set as an empty set;

[0073] Transaction data consists of transactions, items, and boolean values. Each transaction represents an independent record or an event. Each "item" in a transaction represents a feature, parameter, or item. In association rule mining, each item is usually extracted from an attribute (such as temperature, humidity, product, behavior, etc.) or an event (such as purchased goods, purchase frequency, etc.) in the dataset. In the form of transaction data, each item is usually represented by a boolean value (True or False) to indicate whether it appears in the transaction or meets a certain condition. True indicates that the item appears in the transaction or meets a certain condition (for example, the value is greater than a certain threshold), and False indicates that the item does not appear in the transaction or does not meet the condition;

[0074] The minimum support threshold is a very crucial parameter in the Apriori algorithm, used to determine which item sets are "frequent"; during the process of association rule mining, support measures the appearance frequency of a certain item set in the transaction dataset, and the minimum support threshold is used to filter out those item sets with too low appearance frequencies; the minimum confidence threshold is used to filter out association rules with too low confidence. Only when the confidence of an association rule is greater than or equal to the set minimum confidence threshold, the association rule is considered valid and can be used for subsequent analysis and applications;

[0075] S52. For each parameter combination, calculate the support in the normalized production supply chain parameter data, and retain the parameter combinations with support greater than or equal to the preset minimum support threshold to generate item frequent item sets, and repeat the iteration until no new frequent item sets are generated; collect the frequent item sets generated in each iteration to obtain the comprehensive frequent item set;

[0076] In the context of association rule mining, "each parameter combination" refers to the set of different parameters that may appear in the normalized production supply chain parameter data; specifically, the forms of these parameter combinations include single - parameter combinations and multi - parameter combinations; single - parameter combinations are basic items, specifically referring to individual parameters in the normalized production supply chain parameter data, such as a specific material name, material specification, production equipment model, etc.; multi - parameter combinations are item sets, which are sets composed of two or more basic items. For example, a parameter combination may include "material name ", "material specification ” and “supplier ”;

[0077] In data mining and association rule learning, an item frequent itemset is an itemset that contains items. Taking 1-item frequent itemset as an example: “1-item frequent itemset” is also called singleton frequent itemset or frequent single itemset, specifically referring to a single element that appears in a dataset more frequently than the minimum support threshold preset by the user; 1-item refers to a single element in the dataset. If the dataset is transaction records, then each individually sold item is an “item” here;

[0078] S53. For each frequent itemset in the comprehensive frequent itemset, generate corresponding association rules; for each association rule, calculate its confidence and retain the association rules whose confidence is greater than or equal to the minimum confidence threshold; association rules are mainly used to mine the association relationships between data items, such as “if happens, then may happen”. When applied to the normalized production supply chain parameter data, potential relationships in the data can be mined, such as “if the purchased material , then supplier may be selected”;

[0079] S54. Collect all association rules whose confidence is greater than or equal to the minimum confidence threshold, and then obtain the association feature dataset.

[0080] The training method of the material demand prediction model includes:

[0081] Divide the dataset into a training set, a validation set, and a test set, train the model and evaluate the model performance; the sample set is a subset of the dataset, and each sample set includes the historical supply chain comprehensive feature dataset and the corresponding material demand within the next n time periods;

[0082] Use the deep learning library TensorFlow to build a material demand prediction model; the material demand prediction model includes an input layer, an LSTM layer, and an output layer; the input layer of the model is used to input the historical supply chain comprehensive feature dataset; the output layer of the model is used to output the material demand within the next n time periods; the material demand prediction model is an LSTM model;

[0083] Define the loss function of the model, and use the L2 regularization mean square error loss function to measure the difference between the predicted value and the true value of the model; use the training set to train the material demand prediction model, and update the model parameters through the backpropagation algorithm to minimize the loss function; use the validation set to evaluate the performance of the material demand prediction model by calculating the accuracy metric;

[0084] The Adam optimization algorithm is selected as the optimizer, and the model is tuned according to the performance feedback of the validation set. The model parameters are adjusted until the performance no longer improves or reaches the preset number of iterations. The performance of the model in the prediction task is evaluated using the test set, and the trained material demand prediction model is used to predict the current supply chain comprehensive feature dataset to obtain the material demand in the next n periods.

[0085] The method for judging whether inventory adjustment is needed based on the inventory gap includes:

[0086] The RFID technology is used to obtain real-time inventory data, and the real-time inventory data includes the current inventory quantity, storage location, and material status.

[0087] Subtract the current inventory quantity from the predicted material demand in the next n periods to obtain the inventory gap. A preset inventory gap threshold is set, and the inventory gap is compared with the preset inventory gap threshold.

[0088] If the inventory gap is less than the preset inventory gap threshold, it is judged that no inventory adjustment is needed. If the inventory gap is greater than or equal to the preset inventory gap threshold, it is judged that inventory adjustment is needed.

[0089] The method for obtaining the optimal inventory adjustment plan includes:

[0090] S81. The goal of the preset inventory adjustment plan is to minimize the inventory cost (including ordering cost, holding cost, and shortage cost). The objective function is defined as: ; where, is the objective function; is the unit ordering cost; is the adjustment quantity of the th type of material; is the unit inventory holding cost; is the unit shortage cost; is the demand for the th type of material; is the index of the material category; is the total number of material categories;

[0091] S82. Define the position of each spectral particle to represent an inventory adjustment plan. The spectral particle ; where, is the adjustment quantity of the th type of material; is the total number of adjustment quantities of the materials;

[0092] S83. Initialize the positions of the spectral particles and the corresponding spectral frequencies , and initialize the spectral particles: ; where, is the initial position of the th spectral particle, and the initial position corresponds to the inventory adjustment decision; is the lower bound of the material inventory adjustment amount, representing the minimum possible value of the adjustment amount (e.g., the minimum reduced inventory); is the upper bound of the material inventory adjustment amount, representing the maximum possible value of the adjustment amount (e.g., the maximum increased inventory); is the random number of the initial spectral particle position, with a value range of , which is used to randomly initialize the position of the spectral particle between the upper and lower bounds;

[0093] Initialize the spectral frequency: ; where is the initial frequency of the th spectral particle; is the minimum value of the spectral frequency, which is used to limit the range of the initial frequency of the spectral particle; is the maximum value of the spectral frequency, which is used to limit the range of the initial frequency of the spectral particle; is the random number of the initial spectral particle frequency, with a value range of , which is used to randomly initialize the frequency of the spectral particle between the upper and lower bounds;

[0094] S84. Update the position and velocity of the spectral particle through the position update formula and the velocity update formula based on the spectral energy propagation law; the position update formula is: ; where is the position (inventory adjustment amount) of the spectral particle at the th iteration; is the position of the spectral particle at the th iteration; is the velocity of the spectral particle at the th iteration;

[0095] The velocity update formula is: ; where is the velocity of the spectral particle at the th iteration; is the velocity of the spectral particle at the th iteration; represents the momentum factor, which controls the contribution of the spectral particle inertia to the velocity and smooths the movement trajectory of the spectral particle; is the learning factor, which controls the velocity of the spectral particle moving in the gradient direction (the descending direction of the objective function); is the gradient of the objective function, representing the change trend of the current solution in the solution space and guiding the spectral particle to move towards a better solution; is the spectral perturbation factor, which controls the influence of spectral oscillation on velocity; is the spectral oscillation term, which simulates the spectral propagation law, introduces dynamic perturbation, and enhances the diversity of spectral particles; is the temperature of the spectral particle at the -th iteration;

[0096] S85. The temperature of the spectral particle during iteration is restricted by the temperature decay formula, and the temperature decay formula is: ; where is the initial temperature of the spectral particle; is the temperature control factor, which represents the control parameter of the temperature decay rate. A larger value makes the temperature drop rapidly, while a smaller value delays the temperature drop; is the current iteration number;

[0097] Through exponential decay, the temperature can naturally decrease gradually with the increase of the iteration number, so as to realize a smooth transition of the search behavior from "randomness" to "stability";

[0098] For example, the initial temperature of the spectral particle is 100; the temperature control factor is 0.1, and the current iteration number is 5, then ;

[0099] The spectral perturbation factor is dynamically adjusted by the spectral perturbation adjustment formula, and the spectral perturbation adjustment formula is: ; where is the initial maximum spectral perturbation factor; is the adjustment parameter that controls the decay rate of the perturbation factor with the iteration number ; is the total number of spectral particles;

[0100] The combination of the number of spectral particles and time. While the spectral perturbation adjustment formula adapts to spectral particle swarms of different scales, it also retains the dynamic adjustment characteristics of iteration;

[0101] For example, the initial maximum spectral perturbation factor is 1; the adjustment parameter that controls the decay rate of the perturbation factor with the iteration number is 0.1; the total number of spectral particles is 25; the current iteration number is 10, then the dynamically adjusted spectral perturbation factor .

[0102] The momentum factor is restricted by the momentum factor restriction formula;

[0103] S86. According to the fitness of the spectral particles, that is, the objective function , update the spectral frequency through the spectral update formula; the spectral update formula is: ; where is the current global optimal fitness; is the spectral adjustment factor, which controls the sensitivity of the spectral particles to the fitness; is the -th spectral particle's spectral frequency at the -th iteration; is the -th spectral particle's spectral frequency at the -th iteration;

[0104] S87. In each iteration, calculate the fitness of each spectral particle and update the global optimal solution , preset the global optimal solution threshold, and stop when the spectral optimization algorithm reaches the maximum number of iterations or the global optimal solution is less than the preset global optimal solution threshold; the global optimal solution at this time is the optimal inventory adjustment plan.

[0105] The method for restricting the momentum factor by the momentum factor restriction formula includes:

[0106] The momentum factor restriction formula is: ; where is the maximum value of the momentum factor; is the maximum number of iterations; is the absolute value of the gradient of the objective function; is the adjustment parameter that balances the influence of the gradient and the number of particles on the momentum factor;

[0107] The gradient guides the optimization direction, and the number of particles adjusts the search intensity. The combination of the two improves the overall optimization efficiency. At high gradients, reduce the momentum to avoid oscillation. At low gradients, moderately increase the momentum to increase the search ability. For example, the maximum value of the momentum factor is 0.9; the maximum number of iterations is 200; the current iteration number is 50, the absolute value of the gradient of the objective function is 0.02; the adjustment parameter that balances the influence of the gradient and the number of particles on the momentum factor is 0.1; the total number of spectral particles is 30; then the restricted momentum factor is .

[0108] The preset inventory gap threshold is set by the staff. Different inventory gaps are collected through the intelligent monitoring terminal, and the average value of multiple inventory gaps is taken as the preset inventory gap threshold.

[0109] In this embodiment, the objective function clearly decomposes the inventory cost into three parts: ordering cost, holding cost, and shortage cost, which are the core cost sources of inventory management. By considering the adjustment quantity and demand quantity of each type of material, it ensures that the objective function can dynamically adapt to multi-variety inventory management. The position of each spectral particle represents an inventory adjustment plan, and this mapping method is intuitive and easy to implement optimization algorithms. The initial solution is generated using upper and lower bounds and random numbers to ensure the diversity of the search space and avoid falling into local optimal solutions. The position update formula combines the current position, velocity, and gradient of the objective function of the particle, making the particle have a guiding direction when searching the solution space.

[0110] The velocity update formula introduces a momentum factor, a learning factor, and a spectral perturbation factor to control inertia, learning ability, and dynamic perturbation respectively, enhancing the exploration and development ability of the algorithm. The temperature controls the randomness of the particle, and in an exponentially decaying manner, it ensures that the algorithm widely searches the solution space in the early stage and gradually focuses near the global optimal solution in the later stage. By dynamically adjusting the spectral perturbation factor, it adapts to particle swarms of different scales while retaining the dynamic adjustment characteristics of the algorithm during the iteration process. It avoids the influence of the momentum factor being too large or too small on the search trajectory and ensures the stability of particle movement. Through the dynamic adjustment of the global optimal fitness and the spectral adjustment factor, the spectral particles can approach the optimal solution faster. It enhances the search ability of the algorithm and the convergence performance of the global optimal solution, and has high practical value for solving complex inventory optimization problems.

[0111] Embodiment 2

[0112] Please refer to Figure 2 As shown, for the parts not described in detail in this embodiment, refer to the description content of Embodiment 1. A method for integrating production materials, MRO, and temporary procurement is provided, including:

[0113] S1. Obtain production supply chain data;

[0114] S2. Process the production supply chain data to obtain a supply chain comprehensive feature data set;

[0115] S3. Use the supply chain comprehensive feature data as the input of the material demand prediction model to predict the material demand in the next n periods of time;

[0116] S4. Use RFID technology to obtain real-time inventory data, combine it with the predicted material demand in the next n periods of time to obtain the inventory gap, and determine whether inventory adjustment is needed based on the inventory gap; if inventory adjustment is needed, use the spectral particle optimization algorithm combined with the spectrum perturbation and momentum factor limit mechanism to obtain the optimal inventory adjustment plan;

[0117] S5. Transmit the optimal inventory adjustment plan to the intelligent monitoring terminal, issue an adjustment instruction through the intelligent monitoring terminal, and dynamically adjust the inventory level.

[0118] Since the electronic device introduced in this embodiment is the electronic device used in the integrated platform and method based on production materials, MRO, and temporary procurement in the embodiments of the present application, based on the integrated platform and method based on production materials, MRO, and temporary procurement introduced in the embodiments of the present application, those skilled in the art can understand the specific implementation manners and various variations of the electronic device in this embodiment. Therefore, the specific implementation of how this electronic device implements the method in the embodiments of the present application will not be described in detail here. As long as those skilled in the art implement the electronic device used in the integrated platform and method based on production materials, MRO, and temporary procurement in the embodiments of the present application, it falls within the protection scope of the present application.

[0119] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to obtain a formula that is closest to the actual situation. The preset parameters and threshold selection in the formulas are set by those skilled in the art according to the actual situation.

[0120] The above are only the preferred implementation manners of the present invention. The protection scope of the present invention is not limited to the above embodiments. All technical solutions falling within the concept of the present invention belong to the protection scope of the present invention. It should be noted that for ordinary technical users in this technical field, several improvements and refinements made without departing from the principle of the present invention should also be regarded as within the protection scope of the present invention.

Claims

1. An integrated platform based on production materials, MRO, and temporary procurement, characterized in that Including: A data acquisition module, used to acquire production supply chain data; A data processing module, used to process the production supply chain data to obtain a comprehensive supply chain feature dataset; A demand forecasting module, used to take the comprehensive supply chain feature data as the input of the material demand forecasting model, and forecast the material demand within the next n periods of time; An inventory optimization module, which uses RFID technology to obtain real-time inventory data, combines the predicted material demand within the next n periods of time to obtain the inventory gap, and determines whether inventory adjustment is required based on the inventory gap; If inventory adjustment is required, an optimal inventory adjustment plan is obtained through a spectral particle optimization algorithm combined with a spectrum perturbation and momentum factor limitation mechanism; The method for obtaining the optimal inventory adjustment plan includes: The goal of the preset inventory adjustment plan is to minimize the inventory cost, and the objective function is defined as: where f(x) is the objective function; k is the unit ordering cost; x i is the adjustment quantity of the i-th type of material; h is the unit inventory holding cost; p is the unit shortage cost; d i is the demand for the i-th type of material; i is the index of the material category; m is the total number of material categories; Defining the position of each spectral particle represents an inventory adjustment plan; initializing the positions \(x\) of \(N\) spectral particles i and the corresponding spectral frequencies \(\lambda\) i ; updating the positions and velocities of the spectral particles through the position update formula and the velocity update formula based on the energy propagation law of the spectrum; The position update formula is: x i (t + 1) = x i (t) + v i (t); where x i (t + 1) is the position of the spectral particle at the (t + 1)-th iteration; x i (t) is the position of the spectral particle at the t-th iteration; v i (t) is the velocity of the spectral particle at the t-th iteration; the velocity update formula is: where v i (t + 1) is the velocity of the spectral particle at the (t + 1)-th iteration; v i (t) is the velocity of the spectral particle at the t-th iteration; β represents the momentum factor; γ is the learning factor; is the gradient of the objective function; η is the spectral perturbation factor; sin(λ i (t)) is the spectral oscillation term; T(t) is the temperature of the spectral particle at the t-th iteration; Limit the temperature of spectral particles during iteration through the temperature decay formula; dynamically adjust the spectral perturbation factor η through the spectral perturbation adjustment formula; update the spectral frequency according to the fitness of spectral particles through the spectral update formula; the spectral update formula is: λ i (t + 1) = λ i (t) + θ · (f best - f(x i (t))); where f best is the current global optimal fitness; θ is the spectral adjustment factor; λ i (t + 1) is the spectral frequency of the i-th spectral particle at the (t + 1)-th iteration; λ i (t) is the spectral frequency of the i-th spectral particle at the t-th iteration; f(x i (t)) is the objective function value corresponding to the position of the i-th spectral particle at the t-th iteration; In each iteration, the fitness of each spectral particle is calculated and the global optimal solution x is updated. best , a global optimal solution threshold is preset. When the spectral optimization algorithm reaches the maximum number of iterations or the global optimal solution x best is less than the preset global optimal solution threshold, it stops; the global optimal solution x best at this time is the optimal inventory adjustment plan. A decision feedback module, used to transmit the optimal inventory adjustment plan to the intelligent monitoring terminal, issue an adjustment instruction through the intelligent monitoring terminal, and dynamically adjust the inventory level; each module is connected by wired and / or wireless means.

2. The integrated platform based on production materials, MRO and temporary procurement according to claim 1, characterized in that The production supply chain data includes production material data, MRO parameter data, and temporary procurement data; the production material data includes material name, material specification, inventory quantity, usage cycle, supplier information, and procurement cost; the MRO parameter data includes production equipment code, production equipment model, maintenance cycle, failure rate, maintenance record, and MRO material inventory; the temporary procurement data includes temporary procurement quantity, applying department, temporary procurement material demand, temporary procurement supplier information, and temporary procurement cost.

3. The integrated platform based on production materials, MRO and temporary procurement according to claim 2, characterized in that The method for processing the production supply chain data includes: Using the Z-score algorithm to detect outliers in the production supply chain data, and eliminating the identified outliers; filling the missing values in the production supply chain data after eliminating outliers through linear interpolation to obtain production supply chain parameter data; performing standard deviation normalization on the production supply chain parameter data to convert it into a standard normal distribution with a mean of 0 and a standard deviation of 1, and then obtaining the normalized production supply chain parameter data.

4. The integrated platform based on production materials, MRO and temporary procurement according to claim 3, wherein The method for obtaining the comprehensive supply chain feature dataset includes; Performing data dimensionality reduction on the normalized production supply chain parameter data through principal component analysis to obtain a production supply chain feature dataset; Using the Apriori algorithm to mine association rules from the normalized production supply chain parameter data to obtain an association feature dataset; fusing the association feature dataset and the production supply chain feature dataset through a weighted model, and then obtaining a comprehensive supply chain feature dataset; Denote the association feature dataset as F1 and the production supply chain feature dataset as F2; The weighted model is: FJ = α1·F1 + α2·F2; where, α1 is the weight coefficient of the association feature dataset; α2 is the weight coefficient of the production supply chain feature dataset.

5. The integrated platform based on production materials, MRO and temporary procurement according to claim 4, wherein The method for obtaining the association feature dataset includes: S51. Discretize the normalized production supply chain parameter data and convert it into the form of transaction data; preset the minimum support threshold as E and the minimum confidence threshold as D; initialize the frequent item set as an empty set; S52. For each parameter combination, calculate the support in the normalized production supply chain parameter data, retain the parameter combinations with support greater than or equal to the preset minimum support threshold E, generate K-item frequent item sets, and repeat the iteration until no new frequent item sets are generated; collect the frequent item sets generated in each iteration, and then obtain the comprehensive frequent item sets; S53. For each frequent item set in the comprehensive frequent item sets, generate the corresponding association rules; for each association rule, calculate its confidence and retain the association rules with confidence greater than or equal to the minimum confidence threshold; S54. Collect all the association rules with confidence greater than or equal to the minimum confidence threshold, and then obtain the association feature data set.

6. The integrated platform based on production materials, MRO, and temporary procurement according to claim 5, wherein The training method of the material demand prediction model includes: Divide the data set into a training set, a validation set and a test set, train the model and evaluate the model performance; the sample set is a subset of the data set, and each sample set includes the historical supply chain comprehensive feature data set and the corresponding material demand in the next n time periods; Build a material demand prediction model using the deep learning library TensorFlow; the material demand prediction model includes an input layer, an LSTM layer and an output layer; the input layer of the model is used to input the historical supply chain comprehensive feature data set; the output layer of the model is used to output the material demand in the next n time periods; the material demand prediction model is an LSTM model; Define the loss function of the model, use the L2 regularization mean square error loss function to measure the difference between the predicted value and the true value of the model; use the training set to train the material demand prediction model, update the model parameters through the backpropagation algorithm to minimize the loss function; use the validation set to evaluate the performance of the material demand prediction model by calculating the accuracy index; Select the Adam optimization algorithm as the optimizer, tune the model according to the performance feedback of the validation set, and adjust the model parameters until the performance no longer improves or reaches the preset number of iterations; use the test set to evaluate the performance of the model in the prediction task, and use the trained material demand prediction model to predict the current supply chain comprehensive feature data set to obtain the material demand in the next n time periods.

7. The integrated platform based on production materials, MRO and temporary procurement according to claim 6, wherein The method for judging whether inventory adjustment is needed based on the inventory gap includes: Use RFID technology to obtain real-time inventory data, and the real-time inventory data includes the current inventory quantity, storage location and material status; Subtract the current inventory quantity from the predicted material demand in the next n time periods to obtain the inventory gap; preset the inventory gap threshold, and compare the inventory gap with the preset inventory gap threshold; If the inventory gap is less than the preset inventory gap threshold, it is judged that no inventory adjustment is needed; if the inventory gap is greater than or equal to the preset inventory gap threshold, it is judged that inventory adjustment is needed.

8. The integrated platform based on production materials, MRO, and temporary procurement according to claim 7, wherein The method for obtaining the optimal inventory adjustment plan further includes: Spectral particle x = [x1, x2,..., x i ,..., x N ; where x N is the adjustment amount of the Nth type of material; N is the total number of adjustment amounts of materials; Initialize the spectral particle: x i (0) = x min + r1·(x max - x min ); where x i (0) is the initial position of the ith spectral particle; x min is the lower bound of the material inventory adjustment amount; x max is the upper bound of the material inventory adjustment amount; r1 is the random number of the initial spectral particle position; Initialize the spectral frequency: λ i (0) = λ min + r2 · (λ max - λ min ); where λ i (0) is the initial frequency of the i-th spectral particle; λ min is the minimum value of the spectral frequency; λ max is the maximum value of the spectral frequency; r2 is the random number of the initial spectral particle frequency; The temperature decay formula is: T(t) = T0·e -ζ·t ; where, T0 is the initial temperature of the spectral particles; ζ is the temperature control factor; t is the current iteration number; the spectral perturbation adjustment formula is: where, η max is the initial maximum spectral perturbation factor; δ is the adjustment parameter that controls the decay rate of the perturbation factor with the iteration number t; N is the total number of spectral particles; the momentum factor is restricted by the momentum factor constraint formula.

9. The integrated platform based on production materials, MRO and temporary procurement according to claim 8, wherein The method for restricting and constraining the momentum factor through the momentum factor restriction formula includes: The momentum factor limit formula is as follows: where β max is the maximum value of the momentum factor; t max is the maximum number of iterations; is the absolute value of the gradient of the objective function; ω is the adjustment parameter that balances the influence of the gradient and the number of particles on the momentum factor.

10. An integrated method based on production materials, MRO, and temporary procurement, which is implemented based on the integrated platform for production materials, MRO, and temporary procurement described in any one of claims 1 to 9, and is characterized in that, including: S1. Obtain production supply chain data; S2. Process the production supply chain data to obtain the supply chain comprehensive feature data set; S3. Use the comprehensive supply chain characteristic data as the input of the material demand prediction model to predict the material demand in the next n periods of time; S4. Use RFID technology to obtain real-time inventory data, combine it with the predicted material demand in the next n periods of time to obtain the inventory gap, and judge whether inventory adjustment is needed based on the inventory gap; If inventory adjustment is needed, use the spectral particle optimization algorithm combined with the spectrum perturbation and momentum factor limitation mechanism to obtain the optimal inventory adjustment plan; S5. Transmit the optimal inventory adjustment plan to the intelligent monitoring terminal, and issue an adjustment instruction through the intelligent monitoring terminal to dynamically adjust the inventory level.

Images