Intelligent spare part warehouse inventory dynamic monitoring and supplementary optimization system

Abstract

The invention discloses an intelligent spare part warehouse inventory dynamic monitoring and supplementary optimization system, and relates to the technical field of inventory management, and the system comprises the steps: collecting environment parameters, equipment operation data, inventory spare part images and historical supply data, and carrying out the preprocessing; analyzing the preprocessed environment parameters, the equipment operation data and the inventory spare part images by using an artificial intelligence algorithm to obtain the spare part demand quantity and the spare part inventory quantity; the inventory state is judged according to the spare part demand quantity and the spare part inventory quantity, and a difference report is generated; constructing a supply cycle model by using the preprocessed historical supply data; performing purchase replenishment according to the difference report, the supply cycle model and the budget to obtain a replenishment record table; and utilizing a block chain to perform uplink storage on the spare part inventory, the spare part demand, the difference report and the replenishment record chart. According to the invention, through data acquisition and algorithm analysis, the spare part demand is dynamically predicted, the replenishment strategy is optimized, and block chain storage is carried out to enhance the data transparency.

Description

Technical Field

[0001] The invention relates to the technical field of inventory management, in particular to an intelligent spare parts warehouse inventory dynamic monitoring and replenishment optimization system. Background Art

[0002] In the existing technology in the field of inventory management technology, due to the wide variety and complex structure of modern manufacturing equipment, the large demand for spare parts and frequent consumption, the high price of spare parts and long supply cycle increase the difficulty of inventory management. Traditional inventory management lacks intelligent means and cannot accurately predict spare parts demand, resulting in production being affected when inventory is insufficient and wasting money and space when inventory is excessive.

[0003] The spare parts management process is complex and has a high error rate. It cannot quickly respond to equipment failures and urgent needs of production lines, which prolongs the equipment maintenance cycle and affects production efficiency. There are problems of information islands and lack of transparency in data management, and the spare parts circulation process cannot be effectively tracked. These problems show that existing technologies cannot meet the needs of modern manufacturing for efficient and accurate spare parts management, and intelligent solutions are urgently needed. Summary of the invention

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

[0005] Therefore, the present invention provides an intelligent spare parts warehouse inventory dynamic monitoring and replenishment optimization system to solve the problem that the existing technology cannot accurately predict spare parts demand and reasonably optimize replenishment strategies.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] The present invention provides an intelligent spare parts warehouse inventory dynamic monitoring and replenishment optimization system, which includes a data acquisition module for collecting environmental parameters, equipment operation data, inventory spare parts images and historical supply data, and performing preprocessing;

[0008] The data analysis module is used to analyze the pre-processed environmental parameters, equipment operation data and spare parts inventory images using artificial intelligence algorithms to obtain spare parts demand and spare parts inventory;

[0009] Inventory status module, used to judge the inventory status by using spare parts demand and spare parts inventory, and generate difference reports;

[0010] The supply cycle module is used to build a supply cycle model using pre-processed historical supply data;

[0011] Spare parts replenishment module, used to purchase and replenish according to the difference report, supply cycle model and budget, and obtain the replenishment record sheet;

[0012] The blockchain module is used to utilize blockchain to store spare parts inventory, spare parts demand, difference reports and replenishment record sheets on the chain.

[0013] As a preferred solution of the automated wafer packaging inspection and sorting system of the present invention, wherein: the environmental parameters include temperature, humidity, particle concentration and noise intensity; the inventory spare parts images include appearance images, packaging images and stacking images of spare parts; the equipment operation data include operating temperature, load, mechanical vibration data, energy consumption data, alarm information and fault records; the historical supply data includes the time, cycle, number of extensions and production capacity of the supplier's supply of spare parts; the preprocessing includes data cleaning, data smoothing and denoising, data normalization and time series synchronization for the environmental parameters, equipment operation data and historical supply data, and format conversion, image enhancement, noise removal and image normalization for the inventory spare parts images.

[0014] As a preferred solution of the automated wafer packaging inspection and sorting system of the present invention, the following specific steps are performed on the image of the spare parts in stock, including format conversion, image enhancement, noise removal and image normalization.

[0015] The inventory spare parts image is converted into a grayscale image format by using the PI L library to obtain the inventory spare parts image in a grayscale image format;

[0016] Gamma transformation and histogram stretching are used to enhance the brightness and contrast of the grayscale image of the spare parts in stock, so as to obtain the enhanced image of the spare parts in stock;

[0017] The enhanced spare parts images are processed by rotating, flipping and cropping to obtain an expanded image dataset;

[0018] The expanded image dataset is optimized using median filtering and histogram equalization, and the pixel values ​​are normalized through numerical mapping to obtain the preprocessed inventory spare parts images.

[0019] As a preferred solution of the automated wafer packaging detection and sorting system of the present invention, the artificial intelligence algorithm is used to analyze the pre-processed environmental parameters to obtain the dynamic influence coefficient of environmental changes on spare parts demand. The specific steps are as follows:

[0020] The preprocessed environmental parameters are divided into an environmental training set, an environmental test set, and an environmental verification set;

[0021] The ARIMA model is used as the artificial intelligence algorithm to analyze the preprocessed environmental parameters, which is expressed as:

[0022] x t =φ1x t-1 +φ2x t-2+…+φ p x t-p +θ1∈ t-1 +θ2∈ t-2 +…+θ q ∈ t-q +

[0023] ∈ t ;

[0024] Among them, x t is the environmental parameter value at time point t, t is the time point, φ1 is the influence weight of the environmental parameter value at time point t-1 on the environmental parameter value at time point t, x t-1 is the environmental parameter value at time point t-1, φ2 is the influence weight of the environmental parameter value at time point t-2 on the environmental parameter value at time point t, x t-2 is the environmental parameter value at time point t-2, φ p is the weight of the environmental parameter value at time point tp on the environmental parameter value at time point t, x t-p is the environmental parameter value at time point tp, θ1 is the influence weight of the random error value at time point t-1 on the environmental parameter value at time point t, ∈ t-1 is the random error value at time point t-1, θ2 is the influence weight of the random error value at time point t-2 on the environmental parameter value at time point t, ∈ t-2 is the random error value at time point t-2, θ q is the influence weight of the random error value at time point tq on the environmental parameter value at time point t, ∈ t-q is the random error value at time point tq, ∈ t is the prediction error value;

[0025] The ARIMA model is trained and optimized using the environmental training set, environmental test set, environmental validation set, and grid search to obtain the changing trend of environmental parameters;

[0026] The dynamic impact analysis algorithm is used to process the changing trend of environmental parameters and obtain the dynamic impact coefficient of environmental parameters on spare parts demand, which is expressed as:

[0027]

[0028] Among them, η is the dynamic influence coefficient, N is the time window length, k is the time step index, e is the base of the natural logarithm, α is the time decay coefficient, X t-k is the environmental parameter value at time point tk, σ x is the standard deviation of environmental parameters, and σ is the standard deviation of random errors.

[0029] As a preferred solution of the automated wafer packaging detection and sorting system of the present invention, the artificial intelligence algorithm is used to analyze the pre-processed equipment operation data and dynamic influence coefficient to obtain the spare parts demand. The specific steps are as follows:

[0030] The pre-processed equipment operation data is processed using the FFT feature extraction algorithm to obtain frequency domain feature data;

[0031] The feature selection algorithm is used to screen the frequency domain feature data to obtain the key frequency domain features;

[0032] The random forest health assessment model is used to analyze the preprocessed equipment operation data and dynamic impact coefficient, which is expressed as:

[0033]

[0034] Where H is the health status evaluation value of the device, W is the total number of decision trees in the random forest, u is the index of the decision tree in the random forest, V u is the health status prediction result of the u-th decision tree for the equipment operation characteristic value, F b is the b-th device running characteristic value, where b is the index of the device running characteristic value;

[0035] Based on the random forest health assessment model, the key frequency domain characteristics and dynamic influence coefficients are analyzed to obtain the equipment health status assessment results, which are expressed as:

[0036]

[0037] Among them, H η is the equipment health status assessment value, ω u is the weight of the decision tree, max is the maximum value function, is the square of the regression weight of the i-th frequency domain feature, is the square of the ith key frequency domain eigenvalue, η is the dynamic influence coefficient;

[0038] The regression model is used to process the equipment health status assessment results to obtain the remaining life of equipment parts, which is expressed as:

[0039]

[0040] Among them, R is the remaining life of equipment parts, b0 is the intercept of the regression model, m is the number of input features, β j is the regression coefficient of the jth feature in the regression model, E j is the jth eigenvalue of the equipment operation status, and j is the index of the eigenvalue of the equipment operation status;

[0041] According to the remaining life of equipment parts, the demand for spare parts in stock is predicted, and the demand for spare parts is obtained, which is expressed as:

[0042]

[0043] Where D is the demand for spare parts, M is the total number of parts, and λ h is the failure rate of the hth component, R is the remaining life of the equipment component, μ is the environmental impact coefficient, S h is the safety stock of the hth component, and c is a very small positive number.

[0044] As a preferred solution of the automated wafer packaging inspection and sorting system of the present invention, wherein: the pre-processed inventory spare parts image is analyzed using an artificial intelligence algorithm to obtain the spare parts inventory quantity, the specific steps are as follows:

[0045] Label the preprocessed inventory spare parts images and divide them into image training set, image test set and image verification set;

[0046] The convolutional neural network is trained using the image training set, image test set, and image verification set to obtain a spare parts image classification model;

[0047] The spare parts image classification model is used to process the inventory spare parts images to obtain the spare parts inventory quantity.

[0048] As a preferred solution of the automated wafer packaging inspection and sorting system of the present invention, the inventory status is judged by using the spare parts inventory and spare parts demand, and a difference report is generated. The specific steps are as follows:

[0049] Determine the insufficient inventory threshold and the excess inventory threshold based on spare parts inventory, spare parts demand and industry experience;

[0050] Use the insufficient inventory threshold and the excessive inventory threshold to compare and analyze the spare parts inventory and spare parts demand;

[0051] When the inventory level falls below the low inventory threshold, it is recorded as needing replenishment in the variance report;

[0052] When the inventory level is between the insufficient inventory threshold and the excessive inventory threshold, it is recorded in the variance report as not requiring replenishment;

[0053] When the inventory level is above the excess inventory threshold, it is recorded as a need for return in the variance report;

[0054] Obtain inventory variance data and generate variance reports.

[0055] As a preferred solution of the automated wafer packaging inspection and sorting system of the present invention, the supply cycle model is constructed by using the pre-processed historical supply data. The specific steps are as follows:

[0056] Divide the preprocessed historical supply data into a supply training set, a supply test set, and a supply verification set;

[0057] The long short-term memory neural network is trained using the supply training set and the supply test set, and the parameters are optimized using the supply verification set to obtain the supply cycle model, which is expressed as:

[0058]

[0059] T = f(S, E, D′);

[0060] Among them, T is the predicted supply cycle, f is the long short-term memory neural network, S is the supply history data, E is the environmental factor, D′ is the demand fluctuation value, and D t is the spare parts demand at the tth moment, is the average value of spare parts demand, and O is the length of the time series.

[0061] As a preferred solution of the automated wafer packaging inspection and sorting system of the present invention, wherein: purchasing and replenishment are performed according to the difference report combined with the supply cycle model and budget to obtain a replenishment record table, the specific steps are as follows:

[0062] According to the difference report, combined with the supply cycle model and budget, purchase and replenishment are carried out to obtain the replenishment record sheet. The specific steps are as follows:

[0063] Calculated based on the variance report and budget, expressed as,

[0064]

[0065] Where C is the total purchase cost, r is the total number of spare parts, a is the number of spare parts, I a is the unit price of the a-th spare part, J a is the demand quantity of the a-th type of spare parts, w a For a is the supply cost function of the a-th spare part, and T is the predicted supply cycle;

[0066] Obtain a list of spare parts that need to be replenished based on the total purchase cost;

[0067] The spare parts list is analyzed using the supply cycle model, which is expressed as:

[0068]

[0069] Among them, P ais the replenishment priority of the a-th spare part, J a is the demand quantity of the ath type of spare parts, T is the predicted supply cycle, U a is the remaining inventory of the a-th spare part, σ a is the standard deviation of the demand for the ath spare part, c is a very small positive number, γ is the inventory elasticity coefficient of the replenishment priority, S′ a is the safety stock of type a spare part;

[0070] Obtain the replenishment order of spare parts based on replenishment priority;

[0071] Use the replenishment sequence to place orders, obtain replenishment order data, and generate a replenishment record table.

[0072] As a preferred solution of the automated wafer packaging inspection and sorting system of the present invention, the spare parts inventory, spare parts demand, difference report and replenishment record table are stored on the blockchain. The specific steps are as follows:

[0073] The Merkle tree structure of the blockchain is used to organize the spare parts inventory, spare parts demand, difference report and replenishment record table data, and generate a unique hash value for each set of data;

[0074] The hashed data is packaged with information such as timestamp, previous block hash value, block number, etc. into a block to obtain block data;

[0075] Use a dynamic voting mechanism to regularly adjust the weight of authorized nodes;

[0076] The PoA consensus mechanism is used to process the block data, and the authorized node verifies the legal status of the block to obtain the block data that has passed the consensus verification;

[0077] The block data verified by consensus is added to the blockchain to form a chain structure.

[0078] The beneficial effects of the present invention are as follows: the present invention realizes accurate analysis of equipment operation data, environmental parameters and inventory spare parts images through data acquisition modules and artificial intelligence algorithms, can dynamically predict spare parts demand and inventory, and improve the scientificity and accuracy of inventory management. Based on the supply cycle model and replenishment priority algorithm, the spare parts procurement and replenishment strategy is optimized to avoid production interruption caused by insufficient inventory or waste of funds and space caused by excess inventory. The present invention introduces blockchain technology to store spare parts inventory, demand, difference reports and replenishment record tables on the chain, thereby enhancing the transparency and traceability of data and effectively solving the problems of information islands and irregular data management in traditional inventory management. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0080] Figure 1 It is a schematic diagram of the intelligent spare parts warehouse inventory dynamic monitoring and replenishment optimization system in Example 1;

[0081] Figure 2 This is a judgment diagram of the inventory status in Example 1. DETAILED DESCRIPTION

[0082] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0083] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0084] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0085] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, provides an intelligent spare parts warehouse inventory dynamic monitoring and replenishment optimization system, including a data acquisition module, a data analysis module, an inventory status module, a supply cycle module, a spare parts replenishment module and a blockchain module.

[0086] The data acquisition module is used to collect and preprocess environmental parameters, equipment operation data, inventory spare parts images and historical supply data; the data analysis module is used to use artificial intelligence algorithms to analyze the preprocessed environmental parameters, equipment operation data and inventory spare parts images to obtain spare parts demand and spare parts inventory; the inventory status module is used to use spare parts demand and spare parts inventory to judge the inventory status and generate a difference report; the supply cycle module is used to use the preprocessed historical supply data to build a supply cycle model; the spare parts replenishment module is used to purchase and replenish according to the difference report, supply cycle model and budget, and obtain a replenishment record sheet; the blockchain module is used to use the blockchain to store the spare parts inventory, spare parts demand, difference report and replenishment record sheet on the chain.

[0087] Specifically, environmental parameters, equipment operation data, inventory spare parts images and historical supply data are collected and preprocessed, including the following steps:

[0088] Environmental parameters include temperature, humidity, particle concentration and noise intensity; inventory spare parts images include appearance images, packaging images and stacking images of spare parts; equipment operation data include operating temperature, load, mechanical vibration data, energy consumption data, alarm information and fault records; historical supply data include the time, cycle, number of extensions and production capacity of suppliers for supplying spare parts; the preprocessing includes data cleaning, data smoothing and noise reduction, data normalization and time series synchronization for environmental parameters, equipment operation data and historical supply data, and format conversion, image enhancement, noise removal and image normalization for inventory spare parts images.

[0089] It should be noted that data collection and preprocessing lay the foundation for the system's intelligent analysis, improve the quality and consistency of data, ensure the efficient operation of subsequent demand forecasting, supply cycle modeling and inventory optimization algorithms, and ultimately achieve precise management and scientific decision-making.

[0090] The inventory spare parts image is converted into a grayscale image format using the PI L library to obtain the inventory spare parts image in a grayscale image format.

[0091] Gamma transformation and histogram stretching are used to enhance the brightness and contrast of the grayscale image of spare parts, and the enhanced image of spare parts is obtained. Furthermore, the color spare parts image is converted to grayscale format through the convert(L) method of the PI L library, the color information is removed, the brightness information of the image is retained, the complexity of the image data is reduced, and a more efficient input data format is provided for subsequent image enhancement and feature extraction.

[0092] The enhanced spare parts images are processed by rotation, flipping and cropping to obtain an expanded image data set. Furthermore, the details of the spare parts images, especially the visibility of dark or highlight areas, are enhanced by adjusting the nonlinear mapping of grayscale values. The image pixel intensity distribution is linearly stretched to expand the grayscale value range from the original narrow range (such as 50-200) to the full range (such as 0-255), improve the contrast, make the image clearer, and facilitate subsequent feature extraction and classification. The image is rotated at random or fixed angles (such as ±90°, 180°) to enhance data diversity and simulate the angle changes that may occur in the stacking of spare parts in actual scenes. Horizontal flipping, vertical flipping or diagonal flipping are performed to increase the symmetry characteristics of the data samples. Random or fixed area cropping of the image generates local images of different sizes, enhances the model's recognition ability of the spare parts partial features, and improves the classification robustness.

[0093] The expanded image dataset is optimized using median filtering and histogram equalization, and the pixel values ​​are normalized through numerical mapping to obtain the preprocessed inventory spare parts image. Furthermore, the image is denoised, and the pixel values ​​are replaced using a median filter algorithm to smooth the noise points in the image while maintaining the image edge details to prevent blurring. The image grayscale distribution is equalized to enhance the global contrast of the image, highlight the details of the feature area of ​​the spare parts, and provide a clearer image structure for feature extraction. The image pixel values ​​are scaled to between 0 and 1 to reduce the impact of the numerical amplitude, which facilitates the rapid convergence of the deep learning model and more efficient calculation.

[0094] It should be noted that data collection and preprocessing are the basis of intelligent analysis of the system, ensuring the efficient operation of subsequent algorithms and improving the accuracy of management and decision-making.

[0095] The artificial intelligence algorithm is used to analyze the pre-processed environmental parameters to obtain the dynamic impact coefficient of environmental changes on spare parts demand, including the following steps:

[0096] The preprocessed environment parameters are divided into an environment training set, an environment test set, and an environment validation set. Furthermore, the ratio of the environment training set, the environment test set, and the environment validation set is 7:2:1.

[0097] The ARIMA model is used as the artificial intelligence algorithm to analyze the preprocessed environmental parameters, which is expressed as:

[0098] x t =φ1x t-1 +φ2x t-2 +…+φ p x t-p +θ1∈ t-1 +θ2∈ t-2 +…+θq ∈ t-q +

[0099] ∈ t ;

[0100] Among them, x t is the environmental parameter value at time point t, t is the time point, φ1 is the influence weight of the environmental parameter value at time point t-1 on the environmental parameter value at time point t, x t-1 is the environmental parameter value at time point t-1, φ2 is the influence weight of the environmental parameter value at time point t-2 on the environmental parameter value at time point t, x t-2 is the environmental parameter value at time point t-2, φ p is the weight of the environmental parameter value at time point tp on the environmental parameter value at time point t, x t-p is the environmental parameter value at time point tp, θ1 is the influence weight of the random error value at time point t-1 on the environmental parameter value at time point t, ∈ t-1 is the random error value at time point t-1, θ2 is the influence weight of the random error value at time point t-2 on the environmental parameter value at time point t, ∈ t-2 is the random error value at time point t-2, θ q is the influence weight of the random error value at time point tq on the environmental parameter value at time point t, ∈ t-q is the random error value at time point tq, ∈ t is the prediction error value.

[0101] The ARIMA model is trained and optimized using the environmental training set, environmental test set, environmental validation set, and grid search to obtain the trend of environmental parameter changes. Furthermore, the environmental training set is input into the ARIMA model, and the model parameters are optimized by the maximum likelihood estimation method. During the training process, the grid search method is used to traverse different combinations of autoregressive order, difference order, and moving average order, and the parameter combination that minimizes the error of the test set is selected.

[0102] The dynamic impact analysis algorithm is used to process the changing trend of environmental parameters and obtain the dynamic impact coefficient of environmental parameters on spare parts demand, which is expressed as:

[0103]

[0104] Among them, η is the dynamic influence coefficient, N is the time window length, k is the time step index, e is the base of the natural logarithm, α is the time decay coefficient, X t-k is the environmental parameter value at time point tk, σ x is the standard deviation of environmental parameters, and σ is the standard deviation of random errors.

[0105] It should be noted that the ARIMA model is used to analyze the time series change trend of environmental parameters and predict the dynamic impact of future environment on spare parts demand. According to the trend of environmental changes, its impact on spare parts demand is quantified and the spare parts demand forecasting model is dynamically adjusted.

[0106] The artificial intelligence algorithm is used to analyze the pre-processed equipment operation data and dynamic influence coefficient to obtain the spare parts demand, including the following steps:

[0107] The pre-processed equipment operation data is processed using the FFT feature extraction algorithm to obtain frequency domain feature data. Furthermore, the pre-processed equipment operation data (such as vibration data, energy consumption data, load data, etc.) is subjected to a fast Fourier transform to convert the time domain signal into a frequency domain signal, thereby extracting the frequency characteristics of the equipment operation status. This can reveal the periodic laws in the equipment operation and the hidden abnormal signal characteristics.

[0108] The feature selection algorithm is used to screen the frequency domain feature data to obtain key frequency domain features. Furthermore, the frequency domain feature data (such as main frequency, spectrum energy, peak frequency, etc.) are screened to retain the key features that have a significant impact on the equipment operation status and health assessment, remove redundant or irrelevant features, reduce computational complexity, and improve model performance.

[0109] The random forest health assessment model is used to analyze the preprocessed equipment operation data and dynamic impact coefficient, which is expressed as:

[0110]

[0111] Where H is the health status evaluation value of the device, W is the total number of decision trees in the random forest, u is the index of the decision tree in the random forest, V u is the health status prediction result of the u-th decision tree for the equipment operation characteristic value, F b Run the characteristic value for the bth device, where b is the index of the device's running characteristic value.

[0112] Based on the random forest health assessment model, the key frequency domain characteristics and dynamic influence coefficients are analyzed to obtain the equipment health status assessment results, which are expressed as:

[0113]

[0114] Among them, H η is the equipment health status assessment value, ω u is the weight of the decision tree, max is the maximum value function, is the square of the regression weight of the i-th frequency domain feature, is the square of the i-th key frequency domain eigenvalue, and η is the dynamic influence coefficient.

[0115] The regression model is used to process the equipment health status assessment results to obtain the remaining life of equipment parts, which is expressed as:

[0116]

[0117] Among them, R is the remaining life of equipment parts, b0 is the intercept of the regression model, m is the number of input features, β j is the regression coefficient of the jth feature in the regression model, E j is the jth eigenvalue of the equipment operation status, and j is the index of the eigenvalue of the equipment operation status.

[0118] According to the remaining life of equipment parts, the demand for spare parts in stock is predicted, and the demand for spare parts is obtained, which is expressed as:

[0119]

[0120] Where D is the demand for spare parts, M is the total number of parts, and λ h is the failure rate of the hth component, R is the remaining life of the equipment component, μ is the environmental impact coefficient, S h is the safety stock of the hth component, and c is a very small positive number.

[0121] It should be noted that the health status of the equipment is predicted based on the random forest algorithm, combined with the equipment operation characteristics and dynamic impact coefficient, and the remaining life of the parts is further estimated through the regression model. The demand for spare parts is predicted based on the remaining life, failure rate and safety stock level of equipment parts.

[0122] The pre-processed spare parts inventory images are analyzed using artificial intelligence algorithms to obtain the spare parts inventory quantity, including the following steps:

[0123] The preprocessed inventory spare parts images are annotated and divided into image training set, image test set and image verification set. Furthermore, the inventory spare parts images are annotated according to the spare parts category, location and quantity, and an annotation file (such as XML or JSON format) containing category labels and quantity information is generated. The image dataset is divided according to the ratio (such as 8:1:1 or 7:2:1).

[0124] The convolutional neural network is trained using the image training set, image test set, and image validation set to obtain a spare parts image classification model. Furthermore, the training set is used to train the convolutional neural network to learn the mapping relationship between image features and categories. The test set is used to test the classification accuracy of the model for unseen images and evaluate the generalization ability. The validation set is used to adjust the network hyperparameters (such as learning rate, number of layers, etc.) during the training process to optimize the model performance.

[0125] The spare parts image classification model is used to process the spare parts images in stock to obtain the spare parts inventory. Furthermore, the trained convolutional neural network is used to extract and classify the spare parts images, identify the spare parts category and quantity, and improve the accuracy and efficiency of inventory identification.

[0126] It should be noted that the data analysis module realizes the correlation analysis between equipment operation and environmental changes, and evaluates spare parts inventory through image processing, providing accurate demand forecasting and inventory management data.

[0127] Use spare parts inventory and spare parts demand to judge the inventory status and generate a difference report, including the following steps:

[0128] Determine the insufficient inventory threshold and the excessive inventory threshold based on spare parts inventory, spare parts demand, and industry experience. Use the insufficient inventory threshold and the excessive inventory threshold to compare and analyze spare parts inventory and spare parts demand. When the inventory is lower than the insufficient inventory threshold, record it in the difference report as requiring replenishment. When the inventory is between the insufficient inventory threshold and the excessive inventory threshold, record it in the difference report as not requiring replenishment. When the inventory is higher than the excessive inventory threshold, record it in the difference report as requiring return. Obtain inventory difference data and generate a difference report.

[0129] It should be noted that the difference report provides data support for inventory optimization, ensuring reasonable inventory levels and avoiding excessive or insufficient spare parts backlogs.

[0130] Using the preprocessed historical supply data, the supply cycle model is constructed, including the following steps:

[0131] The preprocessed historical supply data is divided into supply training set, supply test set and supply verification set. Furthermore, the ratio is 7:2:1.

[0132] The long short-term memory neural network is trained using the supply training set and the supply test set, and the parameters are optimized using the supply verification set to obtain the supply cycle model, which is expressed as:

[0133]

[0134] T = f(S, E, D′);

[0135] Among them, T is the predicted supply cycle, f is the long short-term memory neural network, S is the supply history data, E is the environmental factor, D′ is the demand fluctuation value, and D t is the spare parts demand at the tth moment, is the average value of spare parts demand, and O is the length of the time series.

[0136] It should be noted that the supply cycle model provides a reliable time basis for procurement replenishment planning, reducing the risks caused by delayed supply.

[0137] Purchasing and replenishment are carried out according to the difference report combined with the supply cycle model and budget to obtain the replenishment record sheet, including the following steps:

[0138] Calculated based on the variance report and budget, expressed as,

[0139]

[0140] Where C is the total purchase cost, r is the total number of spare parts, a is the number of spare parts, I a is the unit price of the a-th spare part, J a is the demand quantity of the a-th type of spare parts, w a For a is the supply cost function of the a-th spare part, and T is the predicted supply cycle.

[0141] Based on the total purchase cost, obtain a list of spare parts that need to be replenished.

[0142] The spare parts list is analyzed using the supply cycle model, which is expressed as:

[0143]

[0144] Among them, P a is the replenishment priority of the a-th spare part, J a is the demand quantity of the ath type of spare parts, T is the predicted supply cycle, U a is the remaining inventory of the a-th spare part, σ a is the standard deviation of the demand for the ath spare part, c is a very small positive number, γ is the inventory elasticity coefficient of the replenishment priority, S′ a is the safety stock of type a spare part.

[0145] Based on the replenishment priority, the replenishment order of spare parts is obtained. Furthermore, the replenishment priority of each spare part is calculated by combining the demand quantity and remaining inventory of the spare parts list through the supply cycle model, and its urgency under the current strategy is evaluated.

[0146] Use the replenishment sequence to place orders, obtain replenishment order data, and generate a replenishment record table.

[0147] It should be noted that under budget constraints, supply chain efficiency is maximized by optimizing the replenishment sequence and quantity.

[0148] Using blockchain, spare parts inventory, spare parts demand, difference reports and replenishment records are stored on the chain, including the following steps:

[0149] The Merkle tree structure of the blockchain is used to organize the spare parts inventory, spare parts demand, difference report and replenishment record table data, and generate a unique hash value for each piece of data. Furthermore, the hash values ​​of the spare parts inventory, demand, difference report and replenishment record table are used as leaf nodes, and the hash value of the parent node is recursively calculated to finally generate a unique root hash value to ensure data integrity and non-tamperability. The hash value of each piece of data is generated using the SHA-256 algorithm and stored in the block, providing a basis for data traceability and verification.

[0150] The hashed data is packaged into a block with information such as timestamp, previous block hash value, block number, etc. to obtain block data. A dynamic voting mechanism is used to regularly adjust the weight of authorized nodes. The PoA consensus mechanism is used to process the block data, and the authorized node verifies the legal status of the block to obtain the block data that has passed the consensus verification. The block data that has passed the consensus verification is added to the blockchain to form a chain structure.

[0151] It should be noted that the blockchain module ensures the storage security and traceability of spare parts inventory, demand, difference reports and replenishment records, improves data transparency and security, and provides a reliable traceability mechanism for spare parts management.

[0152] In summary, the present invention realizes accurate analysis of equipment operation data, environmental parameters and inventory spare parts images through data acquisition modules and artificial intelligence algorithms, can dynamically predict spare parts demand and inventory, and improve the scientificity and accuracy of inventory management. Based on the supply cycle model and replenishment priority algorithm, the spare parts procurement and replenishment strategy is optimized to avoid production interruption caused by insufficient inventory or waste of funds and space caused by excess inventory. The present invention introduces blockchain technology to store spare parts inventory, demand, difference reports and replenishment record tables on the chain, thereby enhancing the transparency and traceability of data and effectively solving the problems of information islands and irregular data management in traditional inventory management.

[0153] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. Intelligent spare parts inventory dynamic monitoring and replenishment optimization system, characterized by: include, Data collection module, used to collect environmental parameters, equipment operation data, inventory spare parts images and historical supply data, and perform preprocessing; The data analysis module is used to analyze the pre-processed environmental parameters, equipment operation data and spare parts inventory images using artificial intelligence algorithms to obtain spare parts demand and spare parts inventory; Inventory status module, used to judge the inventory status by using spare parts demand and spare parts inventory, and generate difference reports; The supply cycle module is used to build a supply cycle model using pre-processed historical supply data; Spare parts replenishment module, used to purchase and replenish according to the difference report, supply cycle model and budget, and obtain the replenishment record sheet; The blockchain module is used to utilize blockchain to store spare parts inventory, spare parts demand, difference reports and replenishment record sheets on the chain.

2. The intelligent spare parts warehouse inventory dynamic monitoring and replenishment optimization system according to claim 1 is characterized by: The environmental parameters include temperature, humidity, particle concentration and noise intensity; the inventory spare parts images include appearance images, packaging images and stacking images of spare parts; the equipment operation data include operating temperature, load, mechanical vibration data, energy consumption data, alarm information and fault records; the historical supply data includes the time, cycle, number of extensions and production capacity of suppliers for supplying spare parts; the preprocessing includes data cleaning, data smoothing and denoising, data normalization and time series synchronization for environmental parameters, equipment operation data and historical supply data, and format conversion, image enhancement, noise removal and image normalization for inventory spare parts images.

3. The intelligent spare parts warehouse inventory dynamic monitoring and replenishment optimization system according to claim 2 is characterized by: The inventory spare parts images are format converted, enhanced, denoised and normalized. The specific steps are as follows: The PIL library is used to convert the inventory spare parts image into a grayscale image format to obtain the inventory spare parts image in a grayscale image format; Gamma transformation and histogram stretching are used to enhance the brightness and contrast of the grayscale image of the spare parts in stock, so as to obtain the enhanced image of the spare parts in stock; The enhanced spare parts images are processed by rotating, flipping and cropping to obtain an expanded image dataset; The expanded image dataset is optimized using median filtering and histogram equalization, and the pixel values ​​are normalized through numerical mapping to obtain the preprocessed inventory spare parts images.

4. The intelligent spare parts warehouse inventory dynamic monitoring and replenishment optimization system according to claim 3 is characterized by: The artificial intelligence algorithm is used to analyze the pre-processed environmental parameters to obtain the dynamic impact coefficient of environmental changes on spare parts demand. The specific steps are as follows: The preprocessed environmental parameters are divided into an environmental training set, an environmental test set, and an environmental verification set; The ARIMA model is used as an artificial intelligence algorithm to analyze the preprocessed environmental parameters, which is expressed as, x t =φ1x t-1 +φ2x t-2 +…+φ p x t-p +θ1∈ t-1 +θ2∈ t-2 +…+θ q ∈ t-q +∈ t ; Among them, x t is the environmental parameter value at time point t, t is the time point, φ1 is the influence weight of the environmental parameter value at time point t-1 on the environmental parameter value at time point t, x t-1 is the environmental parameter value at time point t-1, φ2 is the influence weight of the environmental parameter value at time point t-2 on the environmental parameter value at time point t, x t-2 is the environmental parameter value at time point t-2, φ p is the weight of the environmental parameter value at time point tp on the environmental parameter value at time point t, x t-p is the environmental parameter value at time point tp, θ1 is the influence weight of the random error value at time point t-1 on the environmental parameter value at time point t, ∈ t-1 is the random error value at time point t-1, θ2 is the influence weight of the random error value at time point t-2 on the environmental parameter value at time point t, ∈ t-2 is the random error value at time point t-2, θ q is the influence weight of the random error value at time point tq on the environmental parameter value at time point t, ∈ t-q is the random error value at time point tq, ∈ t is the prediction error value; The ARIMA model is trained and optimized using the environmental training set, environmental test set, environmental validation set, and grid search to obtain the changing trend of environmental parameters; The dynamic impact analysis algorithm is used to process the changing trend of environmental parameters and obtain the dynamic impact coefficient of environmental parameters on spare parts demand, which is expressed as: Among them, η is the dynamic influence coefficient, N is the time window length, k is the time step index, e is the base of the natural logarithm, α is the time decay coefficient, X t-k is the environmental parameter value at time point tk, σ x is the standard deviation of environmental parameters, and σ is the standard deviation of random errors.

5. The intelligent spare parts warehouse inventory dynamic monitoring and replenishment optimization system according to claim 4 is characterized by: Artificial intelligence algorithms are used to analyze the pre-processed equipment operation data and dynamic influence coefficients to obtain the spare parts demand. The specific steps are as follows: The pre-processed equipment operation data is processed using the FFT feature extraction algorithm to obtain frequency domain feature data; The feature selection algorithm is used to screen the frequency domain feature data to obtain the key frequency domain features; The random forest health assessment model is used to analyze the preprocessed equipment operation data and dynamic impact coefficient, which is expressed as: Where H is the health status evaluation value of the device, W is the total number of decision trees in the random forest, u is the index of the decision tree in the random forest, V u is the health status prediction result of the u-th decision tree for the equipment operation characteristic value, F b is the b-th device running characteristic value, where b is the index of the device running characteristic value; Based on the random forest health assessment model, the key frequency domain characteristics and dynamic influence coefficients are analyzed to obtain the equipment health status assessment results, which are expressed as: Among them, H η is the equipment health status assessment value, ω u is the weight of the decision tree, max is the maximum value function, is the square of the regression weight of the i-th frequency domain feature, is the square of the ith key frequency domain eigenvalue, η is the dynamic influence coefficient; The regression model is used to process the equipment health status assessment results to obtain the remaining life of equipment parts, which is expressed as: Among them, R is the remaining life of equipment parts, b0 is the intercept of the regression model, m is the number of input features, β j is the regression coefficient of the jth feature in the regression model, E j is the jth eigenvalue of the equipment operation status, and j is the index of the eigenvalue of the equipment operation status; According to the remaining life of equipment parts, the demand for spare parts in stock is predicted, and the demand for spare parts is obtained, which is expressed as: Where D is the demand for spare parts, M is the total number of parts, and λ h is the failure rate of the hth component, R is the remaining life of the equipment component, μ is the environmental impact coefficient, S h is the safety stock of the hth component, and c is a very small positive number.

6. The intelligent spare parts warehouse inventory dynamic monitoring and replenishment optimization system according to claim 5 is characterized by: The artificial intelligence algorithm is used to analyze the preprocessed inventory spare parts images to obtain the spare parts inventory. The specific steps are as follows: Label the preprocessed inventory spare parts images and divide them into image training set, image test set and image verification set; The convolutional neural network is trained using the image training set, image test set, and image verification set to obtain a spare parts image classification model; The spare parts image classification model is used to process the inventory spare parts images to obtain the spare parts inventory quantity.

7. The intelligent spare parts warehouse inventory dynamic monitoring and replenishment optimization system according to claim 6 is characterized by: Use spare parts inventory and spare parts demand to judge the inventory status and generate a difference report. The specific steps are as follows: Determine the insufficient inventory threshold and the excess inventory threshold based on spare parts inventory, spare parts demand and industry experience; Use the insufficient inventory threshold and the excessive inventory threshold to compare and analyze the spare parts inventory and spare parts demand; When the inventory level falls below the low inventory threshold, it is recorded as needing replenishment in the variance report; When the inventory level is between the insufficient inventory threshold and the excessive inventory threshold, it is recorded in the variance report as not requiring replenishment; When the inventory level is above the excess inventory threshold, it is recorded as a need for return in the variance report; Obtain inventory variance data and generate variance reports.

8. The intelligent spare parts warehouse inventory dynamic monitoring and replenishment optimization system according to claim 7 is characterized by: Using the preprocessed historical supply data, the supply cycle model is constructed. The specific steps are as follows: Divide the preprocessed historical supply data into a supply training set, a supply test set, and a supply verification set; The long short-term memory neural network is trained using the supply training set and the supply test set, and the parameters are optimized using the supply verification set to obtain the supply cycle model, which is expressed as: T = f(S, E, D′); Among them, T is the predicted supply cycle, f is the long short-term memory neural network, S is the supply history data, E is the environmental factor, D′ is the demand fluctuation value, and D t is the spare parts demand at the tth moment, is the average value of spare parts demand, and O is the length of the time series.

9. The intelligent spare parts warehouse inventory dynamic monitoring and replenishment optimization system according to claim 8 is characterized by: According to the difference report, combined with the supply cycle model and budget, purchase and replenishment are carried out to obtain the replenishment record sheet. The specific steps are as follows: Calculated based on the variance report and budget, expressed as, Where C is the total purchase cost, r is the total number of spare parts, a is the number of spare parts, I a is the unit price of the a-th spare part, J a is the demand quantity of the a-th spare part, w a For a is the supply cost function of the a-th spare part, and T is the predicted supply cycle; Obtain a list of spare parts that need to be replenished based on the total purchase cost; The spare parts list is analyzed using the supply cycle model, which is expressed as: Among them, P a is the replenishment priority of the a-th spare part, J a is the demand quantity of the ath type of spare parts, T is the predicted supply cycle, U a is the remaining inventory of the a-th spare part, σ a is the standard deviation of the demand for the ath spare part, c is a very small positive number, γ is the inventory elasticity coefficient of the replenishment priority, S′ a is the safety stock of type a spare part; Obtain the replenishment order of spare parts based on replenishment priority; Use the replenishment sequence to place orders, obtain replenishment order data, and generate a replenishment record table.

10. The intelligent spare parts warehouse inventory dynamic monitoring and replenishment optimization system according to claim 9 is characterized in that: Using blockchain, spare parts inventory, spare parts demand, difference reports and replenishment records are stored on the chain. The specific steps are as follows: The Merkle tree structure of the blockchain is used to organize the data of spare parts inventory, spare parts demand, difference reports and replenishment records, and generate a unique hash value for each set of data; The hashed data is packaged with information such as timestamp, previous block hash value, block number, etc. into a block to obtain block data; Use a dynamic voting mechanism to regularly adjust the weight of authorized nodes; The PoA consensus mechanism is used to process the block data, and the authorized node verifies the legal status of the block to obtain the block data that has passed the consensus verification; The block data verified by consensus is added to the blockchain to form a chain structure.

Images