Supply chain replenishment decision optimization method and system, medium and computer equipment

Abstract

The invention belongs to the technical field of data processing. The invention provides a supply chain replenishment decision optimization method and system, a medium and computer equipment, multi-source features are associated with hyper-parameters of a machine learning model, the hyper-parameters are mapped to different dimensions of a hyper-parameter search space, and each dimension corresponds to a value range of one hyper-parameter; the multi-source data features are input into a pre-trained machine learning model, a demand prediction result of the target product in a certain period of time in the future is obtained, a decision result about whether replenishment is carried out or not is generated according to current inventory data and the demand prediction result, hyper-parameter optimization is carried out on the machine learning model through a fruit fly optimization algorithm, and a decision result about replenishment is generated. And the fruit fly individuals find an optimal solution in the hyper-parameter search space as an optimized hyper-parameter. According to the method, the sensitivity of different hyper-parameters to different features is improved, and higher-precision decision is realized.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a supply chain replenishment decision optimization method, system, medium and computer equipment. Background Art

[0002] The statements in this section merely provide background art related to the present invention and do not necessarily constitute prior art.

[0003] With the development of the Internet of Things and big data technologies, supply chain data has become multi-source, heterogeneous, and dynamic, encompassing both structured and unstructured data, including historical orders, inventory, market trends, logistics, and supplier capacity. To improve data processing capabilities, companies in digital supply chain management often rely on machine learning models to process massive amounts of supply chain business data, enabling dynamic demand forecasting and flexible inventory optimization.

[0004] Existing decision-making machine learning models mostly use pre-collected data for training and hyperparameter optimization. However, even slight deviations in hyperparameters (such as learning rate, tree depth, and time window) can cause demand forecasts to deviate significantly from actual demand fluctuations, leading to erroneous decisions. Existing solutions lack effective integration with supply chain data when optimizing machine learning model hyperparameters. When data exhibits seasonal fluctuations or sudden anomalies, the hyperparameter search strategy cannot be dynamically adjusted based on data changes, resulting in the optimized machine learning model failing to meet predetermined decision accuracy requirements. Summary of the Invention

[0005] To address the shortcomings of the existing technology, the present invention provides a supply chain replenishment decision optimization method, system, medium and computer equipment, which associate multi-source features of different dimensions with hyperparameters of a machine learning model, map the associated hyperparameters to different dimensions of the hyperparameter search space for individual search in a fruit fly optimization algorithm, improve the sensitivity of different hyperparameters to different features, and achieve more accurate decision-making.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions: In a first aspect, the present invention provides a supply chain replenishment decision optimization method.

[0007] A supply chain replenishment decision optimization method includes the following processes: Preprocess the acquired historical order data, user evaluation data, and social media related data of the target product, and perform feature extraction on the preprocessed data to obtain multi-source features; Associating the multi-source features with hyperparameters of the machine learning model, and mapping the hyperparameters to different dimensions of a hyperparameter search space, where each dimension corresponds to a value range of the hyperparameter; The multi-source data features are input into a pre-trained machine learning model to obtain a demand forecast result for the target product in a certain time period in the future, and a decision result on whether to replenish the stock is generated based on the current inventory data and the demand forecast result. The machine learning model uses a fruit fly optimization algorithm to optimize hyperparameters, and fruit fly individuals search for the optimal solution in the hyperparameter search space as the optimized hyperparameter.

[0008] In a second aspect, the present invention provides a supply chain replenishment decision optimization system.

[0009] A supply chain replenishment decision optimization system, comprising: A feature extraction unit is configured to: pre-process the acquired historical order data, user evaluation data, and social media related data of the target product, and perform feature extraction on the pre-processed data to obtain multi-source features; A search space construction unit is configured to: associate multiple sources with hyperparameters of the machine learning model, and map the hyperparameters to different dimensions of a hyperparameter search space, each dimension corresponding to a value range of the hyperparameter; The decision generation unit is configured to: input the multi-source data features into a pre-trained machine learning model to obtain a demand forecast result for the target product in a certain time period in the future, and generate a decision result on whether to replenish the stock based on the current inventory data and the demand forecast result, wherein the machine learning model uses a fruit fly optimization algorithm to optimize hyperparameters, and fruit fly individuals search for the optimal solution in the hyperparameter search space as the optimized hyperparameter.

[0010] In a third aspect, the present invention provides a computer device comprising: a processor and a computer-readable storage medium; a processor adapted to execute a computer program; A computer-readable storage medium having a computer program stored therein, wherein when the computer program is executed by the processor, the supply chain replenishment decision optimization method according to the first aspect of the present invention is implemented.

[0011] In a fourth aspect, the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program is suitable for being loaded by a processor and executing the supply chain replenishment decision optimization method as described in the first aspect of the present invention.

[0012] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention innovatively proposes a supply chain replenishment decision optimization method, which associates multiple sources with the hyperparameters of the machine learning model and maps the hyperparameters to different dimensions of the hyperparameter search space. This can more accurately capture the intrinsic connection between different hyperparameters and multi-source data features, greatly improving the sensitivity of different hyperparameters to different features. For example, in supply chain scenarios that process multi-source data including market trends, supplier delivery times, inventory levels, etc., some hyperparameters may be more sensitive to changes in market trend data, while others may react more strongly to fluctuations in inventory data. By mapping hyperparameters to different dimensions, the algorithm can perceive these differences more finely, thereby more effectively adjusting parameters during model training and improving model performance.

[0013] 2. The present invention uses chaotic map initialization and Gauss-Cauchy hybrid mutation strategy to achieve global wide-area coverage by utilizing the ergodicity of Tent chaotic map in the early stage of search, thus avoiding the distribution deviation of traditional random initialization. In the middle and late stages of iteration, the heavy-tail characteristic of Cauchy mutation is introduced to break through the local extreme value, and combined with the dynamic attenuation step size strategy, a smooth transition from coarse-grained exploration to fine-grained development is achieved.

[0014] 3. The present invention triggers differentiated operations in real time based on the population status, applies dynamic Gaussian mutation to low-fitness individuals, and adopts chaotic perturbation to high-fitness individuals, thereby achieving adaptive matching of parameters and search stages; from the perspective of individual optimization, applying dynamic Gaussian mutation to low-fitness individuals can inject diverse exploration energy into them. Dynamic Gaussian mutation means that the mutation intensity will be adjusted in real time according to the operating status of the algorithm, which enables low-fitness individuals to conduct more extensive and flexible exploration in the search space. It is no longer limited to fixed mutation patterns, but dynamically changes the amplitude and direction of mutation according to the individual's fitness and the overall search progress. In this way, low-fitness individuals have a greater chance of jumping out of the local optimal area they may currently be trapped in and discovering a more promising solution space, thereby improving their own fitness and providing more possibilities for the evolution of the entire population.

[0015] 4. The present invention integrates a multi-criteria stopping strategy, including a maximum number of iterations, an early stopping mechanism, a fitness change rate threshold, and population diversity monitoring, to ensure solution quality while reducing computational time. The maximum number of iterations sets a basic timeframe for the algorithm's operation, ensuring that the algorithm has sufficient opportunities for search and optimization, and avoiding failure to find a satisfactory solution due to too few iterations. Through the early stopping mechanism, when the algorithm shows minimal or no improvement in solution quality over several consecutive iterations, the algorithm is stopped promptly, preventing the algorithm from falling into meaningless repetitive calculations while retaining the better solutions already obtained. The introduction of the fitness change rate threshold enables the algorithm to keenly capture the trend of solution quality changes. When the fitness change rate falls below the set threshold, it indicates that the room for solution improvement is very limited. Stopping the iteration at this time can avoid unnecessary consumption of computing resources without affecting the solution quality. Population diversity monitoring ensures the comprehensiveness of the solution from the perspective of the population. By monitoring the degree of individual differences in the population, it ensures that the algorithm can cover a sufficiently wide range of solution space during the search process, avoiding premature convergence to the local optimal solution, thus providing strong support for obtaining high-quality solutions.

[0016] Advantages of additional aspects of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0018] Figure 1 A flow chart of a supply chain replenishment decision optimization method provided by an exemplary embodiment of the present invention; Figure 2 A schematic diagram of a supply chain replenishment decision optimization system provided by an exemplary embodiment of the present invention; Figure 3 A schematic diagram of a computer device is provided for an exemplary embodiment of the present invention. DETAILED DESCRIPTION

[0019] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0020] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present invention belongs.

[0021] As described in the background technology, in the optimization process of existing machine learning models, the search strategy cannot be dynamically adjusted according to data changes, so that the prediction accuracy of the machine learning model finally optimized cannot meet the predetermined requirements. In addition, there are some other problems with the existing optimization process. For example, methods such as grid search / random search have low computational efficiency in high-dimensional hyperparameter space and are difficult to meet the timeliness requirements of real-time decision-making in the supply chain; swarm intelligence optimization algorithms (such as PSO and GA) are prone to falling into local optimality in complex non-convex search spaces, which will lead to a decrease in the decision accuracy of the final machine learning model. In view of this, this implementation method proposes a supply chain replenishment decision optimization method, such as Figure 1 As shown, the following process is included: S101: Preprocess the acquired historical order data, user evaluation data, and social media-related data (such as discussion and opinion data related to the target product, which are authorized for use) of the target product, and perform feature extraction on the preprocessed data to obtain multi-source features; S102: Associating the multi-source features with hyperparameters of the machine learning model, mapping the hyperparameters to different dimensions of a hyperparameter search space, where each dimension corresponds to a value range of the hyperparameter; S103: Input the multi-source data features into a pre-trained machine learning model (for example, a convolutional neural network model or a long short-term memory neural network model, etc.) to obtain a demand forecast result for the target product in a certain period of time in the future, and generate a decision result on whether to replenish the stock based on the current inventory data and the demand forecast result. The machine learning model uses the fruit fly optimization algorithm to optimize hyperparameters, and the fruit fly individuals search for the optimal solution in the hyperparameter search space as the optimized hyperparameter.

[0022] In S101 of this implementation, specifically, the following steps are included: S101-1: Obtain historical order data for the target product and clean the data to remove missing values, duplicate values, and outliers. Missing values are addressed using mean filling, median filling, or time series interpolation. Duplicate values are directly deleted. Outliers are identified and corrected or deleted using statistical methods (such as the 3σ principle) or the isolation forest algorithm. The cleaned data is normalized to map numerical data to the interval [0, 1] or [-1, 1].

[0023] S101-2: Collect user evaluation data. First, perform text cleaning to remove special characters and stop words (such as "de", "le", "zai"), use lexical analysis in natural language processing technology to segment the text into individual words, and perform word - part tagging. Then, use sentiment analysis algorithms, such as dictionary - based sentiment analysis methods or deep - learning - based sentiment analysis models (such as the LSTM - Attention model), to judge the sentiment polarity (positive, negative, neutral) of user evaluations, and quantify the sentiment analysis results as numerical values. For example, positive is 1, neutral is 0, and negative is - 1.

[0024] S101-3: Collect data related to the target product from social media platforms, use web crawler technology to capture data, and perform data deduplication and format unification. For social media text data, also perform pre - processing operations such as cleaning, word segmentation, and word - part tagging. In addition, extract features such as topic tags, like counts, comment counts, and repost counts from social media associated data, and fuse these features with the text data.

[0025] S101-4: For the pre - processed historical order data, use the principal component analysis (PCA) method to extract key features, retaining the main information while reducing the data dimension. For the text data of user evaluations and social media, use word embedding techniques (such as Word2Vec, BERT) to map words to a low - dimensional vector space to obtain the distributed representation of the text, and then use a convolutional neural network (CNN) or a recurrent neural network (RNN) to extract text features from the text vectors.

[0026] In S102 of this implementation method, specifically, it includes: Determine the hyperparameters of the machine - learning model, such as the maximum depth of the decision - tree model, the number of trees in the random - forest model, the learning rate of the neural - network model, the number of neurons in the hidden layer, etc. Establish an association between the multi - source features and these hyperparameters, and analyze the influence of different features on the hyperparameter values. For example, if the historical order data fluctuates greatly, it may be necessary to adjust the learning rate and the number of neurons in the hidden layer of the neural - network model to better fit the data. Map each hyperparameter to different dimensions of the hyperparameter search space, and set the value range for each dimension. The value range can be determined based on experience, historical experimental data, or the theoretical limitations of the model.

[0027] For continuous parameters, such as learning rates in the range (0.001, 0.1), floating-point encoding and logarithmic sampling are used. Discrete parameters (such as activation function types) are handled through enumeration. The parameter range must be set in a way that balances prior knowledge with dynamic expansion mechanisms. For example, the default range of XGBoost's tree depth parameter is typically set to 3-10, but if the optimal value is found to be close to the upper limit during experimentation, it is expanded to 3-20 to cover potential better solutions. For the curse of dimensionality that can easily arise in high-dimensional spaces, principal component analysis (PCA) can be used for dimensionality reduction, or sensitivity analysis can be used to eliminate low-impact parameters (e.g., if certain regularization terms contribute less than 5% to a particular dataset).

[0028] Optionally, in some other implementations, statistical methods, such as the Pearson correlation coefficient and the Spearman rank correlation coefficient, can be used to calculate the correlation between multi-source features and hyperparameters. For example, analyze the correlation between the fluctuation of historical order data and the learning rate of the neural network model. If the correlation coefficient is close to 1 or -1, it means that there is a strong linear correlation between the two; if the correlation coefficient is close to 0, it means that the linear correlation between the two is weak. In addition to correlation analysis, you can also try to explore the causal relationship between multi-source features and hyperparameters. You can use experimental design methods to fix other factors, change the value of a certain feature, and observe the changes in hyperparameters. For example, in the training of a neural network model, keep other data features unchanged, only change the degree of fluctuation of historical order data, and then observe the changes in hyperparameters such as learning rate and the number of hidden layer neurons to determine whether there is a causal relationship.

[0029] Alternatively, in other implementations, a dynamic hyperparameter adjustment strategy can be developed based on established associations. When multi-source features change, the hyperparameters are automatically adjusted according to the strategy. For example, if a sudden increase in fluctuations in historical order data is detected, the learning rate of the neural network model can be automatically adjusted to a lower value based on the established associations to avoid overfitting during training. At the same time, the number of hidden layer neurons can be appropriately increased to improve the model's ability to fit complex data.

[0030] Alternatively, in other implementations, personalized hyperparameters can be set for each user or scenario, taking into account the differences in multi-source features across different users or scenarios. For example, for users with high purchase frequency and large order amounts, whose historical order data is relatively stable, a higher learning rate and fewer hidden layer neurons can be set for these users; whereas for users with low purchase frequency and large fluctuations in order amounts, a lower learning rate and more hidden layer neurons can be set to improve personalization.

[0031] In S103 of this implementation, the machine learning model uses the fruit fly optimization algorithm to optimize hyperparameters, specifically including: The Tent chaotic map is used to generate the initial solution to address the uneven distribution problem caused by random initialization. The ergodic nature and initial value sensitivity of the chaotic system are exploited to ensure that the population covers potentially high-value areas more evenly within the search space. Each fruit fly individual corresponds to a set of hyperparameter combinations (such as learning rate and number of hidden layer nodes), and its position vector is generated using the Tent chaotic sequence, expressed as follows: (1); in, represents the modulo operation, Representative Generation A random number between Represents the dynamic adjustment parameter related to the current number of iterations. Formula (1) retains the chaotic characteristics through modular operation, while introducing random perturbation terms to enhance the exploration ability. After the initial population is generated, the chaotic variables need to be linearly mapped to the actual definition domain of the hyperparameters.

[0032] In order to further balance global search and local development, the present invention adopts a linear attenuation strategy to dynamically adjust the step size. In the initial iteration stage, a larger dynamic adjustment parameter is set to allow fruit flies to explore extensively in the hyperparameter space and enhance the coverage of unknown areas. As the number of iterations increases, the dynamic adjustment parameter is gradually reduced (e.g. ,in is the maximum number of iterations, is the current iteration number), so that the search focuses on high-potential areas, improving the precision of parameter adjustment, fitness evaluation and search mechanism.

[0033] The design of the fitness function is the key link between the algorithm search mechanism and the actual problem optimization goal. Its core task is to quantify the performance indicators of the hyperparameter combination (such as model accuracy and loss value) into comparable odor concentration values to guide the fruit fly population to evolve towards a better solution. In the single-objective optimization scenario, the inverse mapping is usually used. D i =1 / loss converts the loss value loss into a position determination value from the origin. This method enhances the discrimination of low-loss areas through nonlinear scaling. Substituting the odor concentration determination value into the odor concentration determination function, the odor concentration value of the fruit fly individual is obtained: (2); in, represent and The mapping function between .

[0034] For example, in neural network tuning, when the cross entropy loss value is reduced from 0.5 to 0.2, It will jump from 2 to 5, significantly amplifying the competitive advantage of outstanding individuals.

[0035] For multi-objective optimization problems (such as optimizing both model accuracy and inference speed), a weighted fusion method needs to be introduced: (3); Among them, the weight coefficient 、 It needs to be adjusted dynamically according to the task priority. represents the inference delay, Represents accuracy.

[0036] Through global exploration in the olfactory phase and local development in the visual phase, the algorithm's performance in complex optimization problems is improved. In the olfactory phase, individual fruit flies use a Gaussian perturbation strategy to achieve wide-area exploration. The position update formula for each individual fruit fly is: X i t+1 =X i t +N(0,σ 2 )•R step (t) (4); Among them, the Gaussian noise term N(0,σ 2 ) provides random perturbations and dynamic search step size R step (t) = R max -( R max -R min )•t / T varies with the number of iterations Linear decay, the initial step size is large to cover the unknown area, and then it is reduced to avoid oscillation. i t Represents Drosophila individuals i exist t +1 moment position, X i t+1 Represents Drosophila individuals i exist t The location at the moment, N (0, σ 2 ) is a Gaussian noise term, providing random disturbance, R max Represents the maximum search step length, which specifies the maximum distance a fruit fly individual may move during the search process. R min Represents the minimum search step length, which limits the minimum distance that a fruit fly can move. t Represents the current iteration number, T Represents the total number of iterations.

[0037] After entering the visual stage, the algorithm focuses on the current optimal individual X best In the area where the position is located, Cauchy mutation is used to enhance local search, and the position update formula is: X i t+1 = X best +Cauchy(0,1)• η ( t )(5); in, η ( t ) is the variation intensity, X best Represents the optimal position found in the fruit fly population up to the current iteration, Cauchy(0,1) represents the Cauchy distribution.

[0038] The heavy-tailed nature of the Cauchy distribution allows for a wider range of local jumps. For example, the learning rate may mutate from 0.05 to 0.08 in one iteration, breaking through the limitations of traditional Gaussian mutation. η ( t )= η 0•e -t / T Exponential decay with iteration, allowing larger offsets in the early stage for exploration X best Neighborhood, fine adjustment later, η 0 represents the initial value of mutation intensity.

[0039] The collaboration of the above two stages of the present invention is achieved through an adaptive switching mechanism: when the population fitness variance is lower than the set threshold θ (for example, 0.1), the visual stage is triggered; if the optimal solution for three consecutive generations is not improved, the olfactory stage is restarted.

[0040] Population status discrimination and operation are the core mechanisms of the fruit fly optimization algorithm to dynamically balance global exploration and local development. It triggers differentiated mutation strategies to optimize search efficiency by monitoring the deviation between individual fitness and the average fitness of the group in real time. After each round of iteration, the algorithm first calculates the odor concentration value of each fruit fly individual. smell i and population average concentration smell avg , when the individual concentration is lower than the average (i.e. smell i < smell avg ), it is judged as "aggregation phenomenon", indicating that the individual may fall into a local optimum and needs to be subjected to Gaussian mutation operation: X i ′ = X i + N (0,(( smell max - smellmin ) / (smell avg - smell i ))•0.5 e -t / T )(6); in, X i ′ is the position of the fruit fly individual after mutation, X i Represents the position of the fruit fly individual before mutation, N Represents a random number that follows a Gaussian distribution.

[0041] The mutation strategy dynamically adjusts the variance of Gaussian noise to make the disturbance intensity of low-fitness individuals adaptively match the current iteration stage and population state. The noise intensity is larger in the initial iteration to enhance the ability to escape the local optimal value, and gradually attenuates in the later stage to improve the convergence accuracy. smell i ′ If the solution is better than the original one, the original solution is replaced; otherwise, the original one is retained to avoid invalid perturbations.

[0042] For fruit fly individuals with concentrations above the average ( smell i < smell avg ), it is determined to be a "divergence phenomenon". Logistic chaotic mapping is used to generate a perturbation sequence. The ergodicity of chaotic variables is used to conduct a refined search for high fitness areas. The perturbation formula is: X i ′ = X i +0.1(1-t / T) 2 •(c k -0.5)(7); in, X i ′ Represents the position of the fruit fly individual after the disturbance treatment, X i represents the position of the fruit fly individual after perturbation treatment, c k Represents the chaos intensity coefficient, chaos intensity coefficient c k It decays nonlinearly with the number of iterations. The initial perturbation amplitude is large to cover the potential solutions in the neighborhood, and then gradually shrinks to focus on fine-tuning the optimal solution.

[0043] The termination optimization mechanism of the fruit fly optimization algorithm of the present invention achieves efficient convergence and resource management through a multi-criteria joint judgment and stopping strategy. Its core design includes three parts: basic conditions, dynamic judgment and intelligent monitoring. The basic termination condition sets the maximum number of iterations T as a hard termination line, and introduces an early stopping mechanism: if the optimal fitness is achieved for K consecutive generations (such as 20 generations), Smell best If the improvement is not achieved, the algorithm will be terminated early to avoid invalid calculations. Dynamic convergence is determined by monitoring the fitness change rate. δ and population fitness variance σ 2 Realize intelligent stop judgment, when δ <10 -5 (i.e. the first set threshold) or σ 2 The algorithm is considered converged when T < 0.01 (i.e., the second set threshold). This invention dynamically adjusts the number of iterations by integrating time constraints (e.g., 24 hours) with computational budget constraints (e.g., 105 evaluations): if early convergence is rapid, T is proportionally extended; otherwise, it is shortened to conserve resources.

[0044] S103 of this implementation further includes: obtaining current inventory data, comparing the demand forecast result with the current inventory, setting a replenishment threshold, and generating a replenishment decision when the demand forecast result is greater than the sum of the current inventory and the replenishment threshold; otherwise, no replenishment decision is generated. The replenishment threshold can be set based on factors such as the company's inventory strategy and product sales characteristics.

[0045] Figure 2 A supply chain replenishment decision optimization system is shown, including: The feature extraction unit 201 is configured to: pre-process the acquired historical order data, user evaluation data, and social media related data of the target product, and perform feature extraction on the pre-processed data to obtain multi-source features; A search space construction unit 202 is configured to: associate multiple sources with hyperparameters of the machine learning model, and map the hyperparameters to different dimensions of a hyperparameter search space, each dimension corresponding to a value range of the hyperparameter; The decision generation unit 203 is configured to: input the multi-source data features into a pre-trained machine learning model to obtain a demand forecast result for the target product in a certain time period in the future, and generate a decision result on whether to replenish the stock based on the current inventory data and the demand forecast result, wherein the machine learning model uses a fruit fly optimization algorithm to optimize hyperparameters, and the fruit fly individuals search for the optimal solution in the hyperparameter search space as the optimized hyperparameter.

[0046] It is understandable that each of the above-mentioned units can be separately or completely combined into one or several other units to form a unit, or one (or some) of the units can be further divided into multiple functionally smaller units to form a unit, which can achieve the same operation without affecting the realization of the technical effects of the embodiments of the present application. The above-mentioned units are divided based on logical functions. In actual applications, the functions of one unit can also be implemented by multiple units, or the functions of multiple units can be implemented by one unit. In other embodiments of the present application, the system may also include other units. In actual applications, these functions can also be implemented with the assistance of other units and can be implemented by the collaboration of multiple units.

[0047] According to another embodiment of the present application, the system described in this embodiment can be constructed by running a computer program (including program code) capable of executing the steps involved in the corresponding method of the present invention on a general-purpose computing device such as a computer, which includes processing elements and storage elements such as a central processing unit (CPU), random access memory (RAM), and read-only memory (ROM). The computer program can be recorded on, for example, a computer-readable recording medium, and loaded into the above-mentioned computing device through the computer-readable recording medium and run therein.

[0048] Figure 3 A computer device is shown, which includes a processor 301, a communication interface 302, and a computer-readable storage medium 303. The processor 301, the communication interface 302, and the computer-readable storage medium 303 may be connected via a bus or other means.

[0049] Among them, the communication interface 302 is used to receive and send data, the computer-readable storage medium 303 can be stored in the memory of the electronic device, the computer-readable storage medium 303 is used to store computer programs, the computer programs include program instructions, and the processor 301 is used to execute the program instructions stored in the computer-readable storage medium 303.

[0050] The processor 301 is the computing core and control core of the electronic device, which is suitable for implementing one or more instructions, and is specifically suitable for loading and executing one or more instructions to implement corresponding method processes or corresponding functions.

[0051] The processor 301 is configured to perform the following process: Preprocess the acquired historical order data, user evaluation data, and social media related data of the target product, and perform feature extraction on the preprocessed data to obtain multi-source features; Associating the multi-source features with hyperparameters of the machine learning model, and mapping the hyperparameters to different dimensions of a hyperparameter search space, where each dimension corresponds to a value range of the hyperparameter; The multi-source data features are input into a pre-trained machine learning model to obtain a demand forecast result for the target product in a certain time period in the future, and a decision result on whether to replenish the stock is generated based on the current inventory data and the demand forecast result. The machine learning model uses a fruit fly optimization algorithm to optimize hyperparameters, and fruit fly individuals search for the optimal solution in the hyperparameter search space as the optimized hyperparameter.

[0052] Preprocess the acquired historical order data, user evaluation data, and social media related data of the target product, and perform feature extraction on the preprocessed data to obtain multi-source features; Associating the multi-source features with hyperparameters of the machine learning model, and mapping the hyperparameters to different dimensions of a hyperparameter search space, where each dimension corresponds to a value range of the hyperparameter; The multi-source data features are input into a pre-trained machine learning model to obtain a demand forecast result for the target product in a certain time period in the future, and a decision result on whether to replenish the stock is generated based on the current inventory data and the demand forecast result. The machine learning model uses a fruit fly optimization algorithm to optimize hyperparameters, and fruit fly individuals search for the optimal solution in the hyperparameter search space as the optimized hyperparameter.

[0053] The present invention also provides a computer-readable storage medium, which is a memory device in an electronic device for storing programs and data. It is understood that the computer-readable storage medium herein may include both built-in storage media in the electronic device and, of course, extended storage media supported by the electronic device. The computer-readable storage medium provides storage space that stores the processing system of the electronic device.

[0054] Furthermore, the storage space also stores one or more instructions suitable for being loaded and executed by the processor. These instructions may be one or more computer programs (including program code). It should be noted that the computer-readable storage medium herein may be a high-speed RAM memory or a non-volatile memory, such as at least one disk storage device; alternatively, it may be at least one computer-readable storage medium located remotely from the processor.

[0055] In one embodiment, the computer-readable storage medium stores one or more instructions; the processor loads and executes the one or more instructions stored in the computer-readable storage medium to implement the following process: Preprocess the acquired historical order data, user evaluation data, and social media related data of the target product, and perform feature extraction on the preprocessed data to obtain multi-source features; Associating the multi-source features with hyperparameters of the machine learning model, and mapping the hyperparameters to different dimensions of a hyperparameter search space, where each dimension corresponds to a value range of the hyperparameter; The multi-source data features are input into a pre-trained machine learning model to obtain a demand forecast result for the target product in a certain time period in the future, and a decision result on whether to replenish the stock is generated based on the current inventory data and the demand forecast result. The machine learning model uses a fruit fly optimization algorithm to optimize hyperparameters, and fruit fly individuals search for the optimal solution in the hyperparameter search space as the optimized hyperparameter.

[0056] The present invention also provides a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform the following process: Preprocess the acquired historical order data, user evaluation data, and social media related data of the target product, and perform feature extraction on the preprocessed data to obtain multi-source features; Associating the multi-source features with hyperparameters of the machine learning model, and mapping the hyperparameters to different dimensions of a hyperparameter search space, where each dimension corresponds to a value range of the hyperparameter; The multi-source data features are input into a pre-trained machine learning model to obtain a demand forecast result for the target product in a certain time period in the future, and a decision result on whether to replenish the stock is generated based on the current inventory data and the demand forecast result. The machine learning model uses a fruit fly optimization algorithm to optimize hyperparameters, and fruit fly individuals search for the optimal solution in the hyperparameter search space as the optimized hyperparameter.

[0057] Those skilled in the art will appreciate that the units and algorithmic steps of each example described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional technical personnel may use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0058] The above embodiments can be implemented in whole or in part using software, hardware, firmware, or any combination thereof. When implemented using software, they can be implemented in whole or in part in the form of a computer program product. A computer program product comprises one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions according to the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted via a computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, optical fiber, digital line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer, or a data processing device such as a server or data center that integrates one or more available media. Available media can include magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives).

[0059] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. A supply chain replenishment decision optimization method, characterized in that: The following processes are included: Preprocess the acquired historical order data, user evaluation data, and social media related data of the target product, and perform feature extraction on the preprocessed data to obtain multi-source features; Associating the multi-source features with hyperparameters of the machine learning model, and mapping the hyperparameters to different dimensions of a hyperparameter search space, where each dimension corresponds to a value range of the hyperparameter; The multi-source data features are input into a pre-trained machine learning model to obtain a demand forecast result for the target product in a certain time period in the future, and a decision result on whether to replenish the stock is generated based on the current inventory data and the demand forecast result. The machine learning model uses a fruit fly optimization algorithm to optimize hyperparameters, and fruit fly individuals search for the optimal solution in the hyperparameter search space as the optimized hyperparameter.

2. The supply chain replenishment decision optimization method according to claim 1, characterized in that: Associating order time series features and product category distribution features with the hyperparameters of the machine learning model, including: quantifying the correlation between different features and hyperparameters through mutual information statistics, and optimizing the construction of the hyperparameter search space based on the correlation.

3. The supply chain replenishment decision optimization method according to claim 1, characterized in that: In the fruit fly optimization algorithm, the tent chaotic map is used to generate the initial solution. Each fruit fly individual corresponds to a set of hyperparameter combinations. The position of each fruit fly is: ,in, represents the modulo operation, Representative Generation A random number between Represents the dynamic adjustment parameters related to the current number of iterations.

4. The supply chain replenishment decision optimization method according to claim 1, characterized in that: In the olfactory stage of fruit fly optimization, the position update formula of each fruit fly is: X i t+1 =X i t +N(0,σ 2 )•R step (t) ,in, N(0,σ 2 ) is subject to a mean of 0 and a variance of σ 2 Gaussian distribution, R step (t) is the dynamic search step size, X i t Indicates in t The first iteration i The location of the fruit fly individuals, X i t+1 Indicates in t+1 The first iteration i The location of each fruit fly individual; After entering the visual stage, the position update formula of each fruit fly is: X i t+1 = X best +Cauchy(0,1)• η ( t ),in, η ( t ) is the variation intensity, X best Represents the optimal position found in the fruit fly population up to the current iteration, Cauchy(0,1) represents the Cauchy distribution; The odor concentration value is used as the fitness of the fruit fly individual. When the population fitness variance of the fruit fly population is lower than the set threshold, the visual stage is triggered; if the optimal solution for three consecutive generations is not improved, the olfactory stage is restarted.

5. The supply chain replenishment decision optimization method according to claim 4, characterized in that: After each round of iteration, when the odor concentration value of a fruit fly individual smell i Lower than the population average concentration smell avg When , it is determined to be an aggregation phenomenon, and Gaussian mutation operation is applied to this fruit fly individual: X i ′ = X i + N (0,(( smell max - smell min ) / (smell avg - smell i ))•0.5 e -t / T ),in, X i ′ is the position of the fruit fly individual after mutation, smell max and smell min represent the maximum and minimum odor concentrations in the fruit fly population, t is the current iteration number, T is the total number of iterations, N represents a random number that follows a Gaussian distribution. X i represents the position of the original Drosophila individual; When the odor concentration value of a fruit fly individual smell i Greater than or equal to the population average concentration smell avg , it is determined to be a divergence phenomenon, and the disturbance is applied to this fruit fly individual: X i ′ = X i +0.1(1-t / T) 2 •(c k -0.5), where X i ′ represents the position of the fruit fly individual after the disturbance is applied, c k Represents the chaos intensity coefficient.

6. The supply chain replenishment decision optimization method according to any one of claims 1 to 5, characterized in that: The termination conditions of the fruit fly optimization algorithm include: When the maximum number of iterations T is reached, the fruit fly optimization algorithm terminates. If the optimal fitness of consecutive K generations Smell best If there is no improvement, the fruit fly optimization algorithm terminates; The population fitness change rate and population fitness variance are calculated. When the population fitness change rate is less than a first set threshold or the population fitness variance is less than a second set threshold, the fruit fly optimization algorithm terminates.

7. The supply chain replenishment decision optimization method according to any one of claims 1 to 5, characterized in that: Generate a replenishment decision based on current inventory data and demand forecast results, including: Obtain current inventory data, compare demand forecast results with current inventory, set a replenishment threshold, and generate a replenishment decision when the demand forecast result is greater than the sum of the current inventory and the replenishment threshold; otherwise, no replenishment decision is generated.

8. A supply chain replenishment decision optimization system, characterized by: include: A feature extraction unit is configured to: pre-process the acquired historical order data, user evaluation data, and social media related data of the target product, and perform feature extraction on the pre-processed data to obtain multi-source features; A search space construction unit is configured to: associate multiple sources with hyperparameters of the machine learning model, and map the hyperparameters to different dimensions of a hyperparameter search space, each dimension corresponding to a value range of the hyperparameter; The decision generation unit is configured to: input the multi-source data features into a pre-trained machine learning model to obtain a demand forecast result for the target product in a certain time period in the future, and generate a decision result on whether to replenish the stock based on the current inventory data and the demand forecast result, wherein the machine learning model uses a fruit fly optimization algorithm to optimize hyperparameters, and fruit fly individuals search for the optimal solution in the hyperparameter search space as the optimized hyperparameter.

9. A computer device, characterized in that: include: a processor and a computer-readable storage medium; a processor adapted to execute a computer program; A computer-readable storage medium having a computer program stored therein, wherein when the computer program is executed by the processor, the supply chain replenishment decision optimization method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and the computer program is suitable for being loaded by a processor and executing the supply chain replenishment decision optimization method according to any one of claims 1 to 7.

Images