Tool material inventory quantity management early warning threshold prediction method and system
By combining XGBoost, EEMD and BiLSTM models, the early warning threshold of the material inventory of dynamically predicting the tool dynamically has solved the problems of strong dependence on manual experience and lack of dynamic adjustment mechanism in the existing technology, improving the accuracy and efficiency of inventory management, and reducing operating costs and environmental pollution.
Patent Information
- Application Number
- CN202411837313.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-13
- Publication Date
- 2025-05-06
AI Technical Summary
In the existing tool material management system, the manual experience is highly dependent and the lack of a dynamic adjustment mechanism, which leads to unreasonable setting of inventory warning thresholds and the inability to adapt to the dynamic changes in maintenance needs, which increases the complexity and cost of inventory management.
By collecting historical inventory data of tool materials and various inventory-influencing conditions, data cleaning and standardization are carried out, feature screening and EEMD decomposition are used using XGBoost, BiLSTM time series prediction model is constructed, and early warning thresholds for tool materials are dynamically predicted.
It significantly improves the accuracy of inventory forecasting, reduces the dependence of manual intervention, improves the scientificity and efficiency of inventory management, reduces the production of inventory backlog and expired materials, and reduces operating costs and environmental pollution.
Smart Images
Figure CN119941112A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of tool material inventory management, and in particular to a tool material inventory quantity management warning threshold prediction method and system. Background Art
[0002] The maintenance of subway vehicles is one of the key tasks to ensure their safe operation. In the process of vehicle maintenance, a large number of various tools and maintenance materials are needed. In order to ensure the smooth progress of maintenance work, these tools and materials are usually kept in reserve at the maintenance site. One of the main management goals of the existing tool and material management system is to minimize the number of tools and materials in reserve to achieve "just enough". This can not only reduce the reserve of spare materials, but also reduce fixed assets, storage land and space, and reduce the workload of material inventory, thereby improving management efficiency.
[0003] In order to achieve the goal of "just enough", the key is to determine a reasonable warning threshold for the number of tools and materials. When the reserve materials are lower than this limit, the material management department can be reminded to replenish the materials in time. At present, there are still some problems in the management methods of various management systems for this problem, mainly including manually setting a "value" based on experience, but the rationality of this "value" is based only on the subjective experience of the management personnel, lacking scientific basis, which can easily lead to unreasonable threshold setting; fixed warning thresholds cannot adapt to the dynamic changes in maintenance needs, and it is difficult to accurately control the inventory level. When the inventory is lower than the warning threshold, manual intervention is required for procurement and replenishment, which is inefficient; due to the wide variety of inventory, manual inventory work is cumbersome and time-consuming; storage costs are high: in order to cope with demand fluctuations, a high level of safety inventory has to be maintained, which takes up a lot of funds and storage space. Therefore, it is necessary to design a new warning threshold determination method model to overcome the above defects. Summary of the invention
[0004] The present application provides a tool material inventory quantity management warning threshold prediction method and system to solve the problems in the prior art such as strong reliance on manual experience, lack of dynamic adjustment mechanism, and large inventory workload.
[0005] According to the first aspect, an embodiment provides a method for predicting a tool material inventory quantity management warning threshold, the method comprising:
[0006] Collect historical inventory data of tool materials and various factors affecting inventory conditions, and perform data cleaning and standardization;
[0007] Extract and screen effective features of the processed time series data, and further perform EEMD decomposition on the screened effective features;
[0008] The screened effective features and the features obtained by further EEMD decomposition are used as input to train the constructed time series prediction model, and the trained time series prediction model is used to predict the tool material quantity warning threshold.
[0009] Furthermore, historical inventory data of tool materials and various inventory influencing factors are collected, including:
[0010] Collect various conditions and factors that affect the quantity of tool materials used, including maintenance work plans, historical usage data, seasonal factors, equipment operation and maintenance conditions, and process changes.
[0011] Furthermore, effective feature extraction and screening are performed on the processed time series data, specifically including:
[0012] The constructed XGBoost model is trained, and the trained XGBoost model is used to perform preliminary feature extraction and score the importance of the extracted features, and effective features are screened out based on the scores.
[0013] Furthermore, effective features are screened out based on the scores, including:
[0014] Features whose importance scores exceed the preset threshold are selected as valid features.
[0015] Furthermore, the screened effective features are further decomposed by EEMD, including:
[0016] After multiple EMD decompositions, several intrinsic mode functions (IMFs) and trend components are extracted. Each IMFs represents a signal component with different frequency characteristics. The remaining signal after multiple decompositions is used as the trend component.
[0017] Furthermore, the constructed prediction model is trained, including:
[0018] The prediction model adopts the BiLSTM model.
[0019] Furthermore, the BiLSTM model specifically includes: an input layer for feature input; a bidirectional LSTM layer for processing input sequences through forward and reverse LSTM to capture time dependencies; a fully connected layer for connecting LSTM outputs to an output layer; and an output layer for predicting warning thresholds.
[0020] Furthermore, the constructed time series prediction model is trained, including:
[0021] During the training process, the model parameters are optimized by the back propagation algorithm. After the training is completed, the performance of the model is evaluated using the test set, including mean square error (MSE), root mean square error (RMSE), R 2The evaluation index of the coefficient is evaluated.
[0022] According to the second aspect, an embodiment provides a tool material inventory quantity management warning threshold prediction system, the system comprising:
[0023] Data collection module, used to collect historical inventory data of tool materials and various inventory influencing factors, and perform data cleaning and standardization;
[0024] The feature extraction module is used to extract and filter effective features of the processed time series data, and further perform EEMD decomposition on the filtered effective features;
[0025] The prediction module is used to take the screened effective features and the features obtained by further EEMD decomposition as input, train the constructed time series prediction model, and use the trained time series prediction model to predict the tool material quantity warning threshold.
[0026] According to a third aspect, an embodiment provides an electronic device, the device comprising: a processor and a memory;
[0027] The memory is used to store one or more program instructions;
[0028] The processor is used to run one or more program instructions to execute the steps of a tool material inventory quantity management early warning threshold prediction method as described in any of the above items.
[0029] According to the fourth aspect, an embodiment provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of a tool material inventory quantity management warning threshold prediction method as described in any one of the above items are implemented.
[0030] The present application provides a tool material inventory quantity management warning threshold prediction method and system, which has the following beneficial effects:
[0031] (1) By combining XGBoost for feature screening and EEMD decomposition, more valuable information can be extracted, thereby significantly improving the accuracy of inventory forecasting. The dynamic prediction capability of the model enables inventory management to be adjusted at any time according to changes in market demand, reducing the risk of excess or shortage inventory.
[0032] (2) This method reduces the reliance on manual experience, reduces the interference of human factors in inventory management, and improves the scientificity and systematicness of management. Through the automated prediction and early warning mechanism, the workload of manual inventory counting and data processing is reduced, and the efficiency of inventory management is improved. Accurate inventory management can effectively reduce the waste of raw materials and energy consumption, reduce the operating costs of enterprises, and improve resource utilization.
[0033] (3) Through the automated processing of the model, the operational process of inventory management is simplified, allowing managers to focus more on decision-making and strategy adjustment. Accurate inventory management reduces inventory backlogs and the generation of expired materials, helps reduce environmental pollution, and promotes sustainable development.
[0034] (4) Real-time data storage and visualization functions enable managers to quickly obtain inventory status and make timely decisions when necessary, thereby improving response speed. BRIEF DESCRIPTION OF THE DRAWINGS
[0035] Figure 1 A flowchart of a tool material inventory quantity management warning threshold prediction method provided by an embodiment of the present invention;
[0036] Figure 2 A flowchart for the specific implementation of a tool material inventory quantity management early warning threshold prediction method provided by one embodiment of the present invention;
[0037] Figure 3 A schematic diagram of the composition structure of a tool material inventory quantity management early warning threshold prediction system provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0038] The present invention is further described in detail below by specific embodiments in conjunction with the accompanying drawings. Wherein similar elements in different embodiments adopt associated similar element numbers. In the following embodiments, many detailed descriptions are for making the present application better understood. However, those skilled in the art can easily recognize that some features can be omitted in different situations, or can be replaced by other elements, materials, methods. In some cases, some operations related to the present application are not shown or described in the specification, this is to avoid the core part of the present application being overwhelmed by too much description, and for those skilled in the art, it is not necessary to describe these related operations in detail, and they can fully understand the related operations according to the description in the specification and the general technical knowledge in the art.
[0039] In addition, the features, operations or characteristics described in the specification can be combined in any appropriate manner to form various implementations. At the same time, the steps or actions in the method description can also be interchanged or adjusted in a manner that is obvious to those skilled in the art. Therefore, the various sequences in the specification and the drawings are only for the purpose of clearly describing a certain embodiment and are not meant to be a required sequence, unless otherwise specified that a certain sequence must be followed.
[0040] The first embodiment of the present invention provides a tool material inventory quantity management warning threshold prediction method based on XGboost-EEMD-BiLSTM, which can comprehensively consider various factors affecting the tool material usage quantity, reasonably determine the tool material shortage warning threshold, and achieve the goal of scientifically managing the tool material inventory quantity. Figure 1 and Figure 2 Provide detailed explanation.
[0041] like Figure 1 As shown, in step S100, historical inventory data of tool materials and various inventory influencing factors are collected, and data cleaning and standardization are performed.
[0042] Specifically, various factors that affect the number of tool materials used are collected, such as maintenance work plans, historical usage data, seasonal factors, equipment operation and maintenance, process changes, etc., and a data set is created. This step is the basis of the entire method. It is necessary to collect various factors that may affect the number of tool materials used. First, it is necessary to collect production plan information, including product types, product quantities, production cycles, etc. This information can help predict future tool material needs; secondly, it is necessary to collect historical usage data, including the specific usage of various types of tool materials, such as usage time, usage frequency, usage quantity, etc. These data can reflect the usage patterns and trends of tool materials. In addition, it is also necessary to collect some external factors, such as seasonality, changes in market demand, etc., because these factors will also affect the usage of tool materials. By comprehensively collecting the characteristics of these influencing thresholds and creating a data set for the time series prediction model.
[0043] like Figure 1 As shown, in step S200, effective features are extracted and screened for the processed time series data, and the screened effective features are further decomposed by EEMD.
[0044] Specifically, in this embodiment, the constructed XGBoost model is first trained, and the trained XGBoost model is used to perform preliminary feature extraction and the extracted features are scored for importance, and effective features are screened out based on the scores.
[0045] The constructed dataset is divided into a training set and a test set. The training set is used to train the model, while the test set is used to evaluate the performance of the model. The dataset is divided into 80% for training and 20% for testing. During the training process, the model will learn the relationship between the features and the target variable, and reduce the prediction error through continuous optimization. The overall objective function of XGboost can be expressed as:
[0046]
[0047] in ∑σ(fk ) represents the complexity of k trees, and L(φ) is the expression in linear space;
[0048] because:
[0049]
[0050] Then L(φ) can be expressed as:
[0051]
[0052] After training, the XGBoost model can calculate the importance score of each feature. These importance scores reflect the contribution of each feature to the prediction target. The XGBoost regression tree structure will calculate the importance (importants) every time it grows, that is, it calculates the value of the decrease in label impurity in this growth and accumulates it to the importance of the feature corresponding to this growth. According to the importance score of the feature, a threshold must be set to filter out features with importance higher than the threshold. These effective features will be used for subsequent model training and prediction.
[0053] Then, the filtered effective features are further decomposed by EEMD to extract multiple intrinsic mode functions and trend components, providing richer information for subsequent models and improving the accuracy and generalization of the models.
[0054] EEMD is a signal processing method used to analyze nonlinear and non-stationary time series data. It overcomes the mode aliasing problem in the traditional empirical mode decomposition (EMD) method by adding random noise to the original signal and performing multiple empirical mode decompositions. EEMD can decompose complex signals into multiple intrinsic mode functions (IMFs) and a trend component. First, an appropriate amount of white noise is added to the original time series data. This process helps to improve the stability of the decomposition and reduce mode aliasing. The signal with added noise is subjected to multiple EMD decompositions to extract several intrinsic mode functions (IMFs). Each IMF represents a signal component with different frequency characteristics. After completing multiple decompositions, the remaining signal part is usually regarded as a trend component. The trend component reflects the long-term trend of the time series. Each IMF represents the characteristics of different frequencies in the signal. The intrinsic mode function can be expressed as:
[0055]
[0056] From the steps of the EEMD algorithm, it can be seen that the effect of the algorithm decomposition depends on the number of integrations N and the amplitude A of the added white noise. The values of N and A should satisfy the following relationship:
[0057]
[0058] Where ε is the final standard deviation of the error, which is calculated by the difference between the original signal time series and the sum of the intrinsic mode functions obtained by the EEMD algorithm.
[0059] Low-frequency IMFs may reflect long-term trends, while high-frequency IMFs may capture short-term fluctuations and noise. The trend component provides the overall direction of the time series and helps the model understand the long-term change pattern. The extracted IMFs and trend components are added to the original dataset as new features.
[0060] like Figure 1 As shown, in step S300, the screened effective features and the features obtained by further EEMD decomposition are used as input to train the constructed time series prediction model, and the trained time series prediction model is used to predict the tool material quantity warning threshold.
[0061] In this embodiment, the time series model used is the BiLSTM model.
[0062] BiLSTM is a bidirectional long short-term memory network that can capture contextual information in time series. Convert the integrated data set into a three-dimensional array format suitable for BiLSTM input. The format of the array can be set to: [number of features, time step, feature value]. Sequence data can be created by sliding window method so that the model can capture the dynamic changes of time series.
[0063] The steps of building the model include: input layer: receiving feature input; bidirectional LSTM layer: processing input sequence through forward and backward LSTM to capture time dependencies; fully connected layer: connecting LSTM output to output layer; output layer: predicting warning threshold. For each time step t:
[0064]
[0065] Where f t For the forget gate, i t is the input gate, With C t is the cell update state, h t In hidden state.
[0066] The BiLSTM model is trained using the training set. During training, the warning thresholds are obtained through manual annotation of the training samples and the model parameters are optimized through the back propagation algorithm. After the training is completed, the performance of the model is evaluated using the test set. The mean square error (MSE), root mean square error (RMSE), R 2 The evaluation indexes such as coefficient are used for evaluation.
[0067] Then, set the inventory warning threshold according to the prediction results of BiLSTM and actual business needs, and dynamically adjust the inventory management strategy. After training, the BiLSTM model will generate a set of optimal weights, which represent the degree of influence of the input features on the predicted output. Use the trained BiLSTM model to predict the latest collected data, and input the new data into the trained BiLSTM model after feature screening and extraction to obtain the predicted warning threshold. The model will generate a predicted value for future inventory based on the patterns learned from historical data. According to the prediction results of the BiLSTM model, set the inventory warning threshold to respond to inventory changes in a timely manner. According to historical data and business needs, flexibly adjust the threshold and set the inventory warning mechanism. When the predicted inventory quantity is lower than the set threshold, the warning is triggered to remind relevant personnel to take measures. Regularly monitor the accuracy of the prediction and the effectiveness of the inventory management strategy. Continuously optimize the model and management strategy through the feedback mechanism to ensure the continuous improvement of the inventory management system.
[0068] Finally, save the forecast data to a local or cloud server for real-time viewing. For the traceability and reproducibility of inventory management methods, all kinds of data need to be properly saved. First of all, it is necessary to save all kinds of input data, forecast results, dynamic adjustment process and other information in a local database to ensure the security and integrity of the data, and to facilitate future access and analysis. At the same time, these data need to be synchronized to the cloud server for multi-party sharing and collaboration. Cloud storage can achieve cross-regional and cross-departmental data sharing, improve the transparency and collaboration efficiency of the entire inventory management. In addition, a sound data backup mechanism needs to be established to ensure data security. Local and cloud data can be backed up regularly to avoid the risk of data loss due to unexpected situations.
[0069] The present invention combines the two models of XGBoost and BiLSTM, utilizes the feature screening ability of XGBoost and the time series prediction ability of BiLSTM, realizes the effective integration of multiple models, and improves the accuracy and robustness of prediction. The EEMD (combined empirical mode decomposition) technology is introduced to conduct in-depth analysis of the time series data of inventory quantity. This innovation enables the model to extract multi-level features and trends and provide richer information input. Through real-time analysis of the prediction results, the inventory warning threshold is dynamically adjusted. This flexible management strategy can better adapt to changes in market demand and avoid inventory backlogs and insufficient supply. The feature importance scoring mechanism of XGBoost is adopted to systematically screen out the features that have the greatest impact on the prediction, ensure the effectiveness of the model input, and thus improve the prediction effect. The automated process of data collection, cleaning, standardization and feature extraction is realized, which significantly reduces the complexity of manual operation and improves the efficiency and consistency of data processing. A variety of factors that affect the number of tool materials used, such as seasonality, historical trends, etc., are fully considered to form a more comprehensive prediction model, which enhances the applicability and reliability of the system.
[0070] It should also be noted that some structural features or method features of the present invention can be expanded or replaced by other means to solve the same technical problems and achieve similar or better effects. For example, GRU can be used as a substitute for BiLSTM. GRU has a simpler structure and higher computational efficiency, and is suitable for processing sequence data. Alternatively, a model based on a self-attention mechanism can be selected to better capture long-distance dependencies and is suitable for processing large-scale data. At the same time, other feature selection techniques can be selected, such as genetic algorithms or particle swarm optimization algorithms. This method can be used for feature selection to replace the feature importance score of XGBoost, find the optimal feature subset, and improve model performance. Alternatively, principal component analysis (PCA) is used to extract the main features through dimensionality reduction technology, reduce data dimensions, and reduce computational complexity.
[0071] Corresponding to the above-disclosed tool material inventory quantity management warning threshold prediction method, the embodiment of the present invention also discloses a tool material inventory quantity management warning threshold prediction system, such as Figure 3 As shown, it specifically includes:
[0072] Data collection module, used to collect historical inventory data of tool materials and various inventory influencing factors, and perform data cleaning and standardization;
[0073] The feature extraction module is used to extract and filter effective features of the processed time series data, and further perform EEMD decomposition on the filtered effective features;
[0074] The prediction module is used to take the screened effective features and the features obtained by further EEMD decomposition as input, train the constructed time series prediction model, and use the trained time series prediction model to predict the tool material quantity warning threshold.
[0075] It should be noted that the detailed description of a tool material inventory quantity management early warning threshold prediction system provided in an embodiment of the present invention can refer to the relevant description of a tool material inventory quantity management early warning threshold prediction method provided in an embodiment of the present application, which will not be repeated here.
[0076] In addition, an embodiment of the present invention also provides an electronic device, comprising: a processor and a memory; the memory is used to store one or more program instructions; the processor is used to run one or more program instructions to execute the steps of a tool material inventory quantity management early warning threshold prediction method as described in any of the above items.
[0077] It should be noted that, for the detailed description of an electronic device provided in an embodiment of the present invention, reference can be made to the relevant description of a tool material inventory quantity management warning threshold prediction method provided in an embodiment of the present application, which will not be repeated here.
[0078] In addition, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of a tool material inventory quantity management warning threshold prediction method as described in any of the above items are implemented.
[0079] It should be noted that the detailed description of a computer-readable storage medium provided in an embodiment of the present invention can refer to the relevant description of a tool material inventory quantity management warning threshold prediction method provided in an embodiment of the present application, which will not be repeated here.
[0080] Those skilled in the art will appreciate that all or part of the functions of the various methods in the above-mentioned embodiments can be implemented by hardware or by computer programs. When all or part of the functions in the above-mentioned embodiments are implemented by computer programs, the program can be stored in a computer-readable storage medium, and the storage medium can include: read-only memory, random access memory, disk, optical disk, hard disk, etc., and the program is executed by a computer to implement the above-mentioned functions. For example, the program is stored in the memory of the device, and when the program in the memory is executed by the processor, all or part of the above-mentioned functions can be implemented. In addition, when all or part of the functions in the above-mentioned embodiments are implemented by computer programs, the program can also be stored in a storage medium such as a server, another computer, disk, optical disk, flash disk or mobile hard disk, and can be downloaded or copied and saved in the memory of the local device, or the system of the local device is updated, and when the program in the memory is executed by the processor, all or part of the functions in the above-mentioned embodiments can be implemented.
[0081] The above specific examples are used to illustrate the present invention, which is only used to help understand the present invention and is not intended to limit the present invention. For those skilled in the art, according to the concept of the present invention, some simple deductions, modifications or substitutions can be made.
Claims
1. A tool material inventory quantity management warning threshold prediction method, characterized in that: The method comprises: Collect historical inventory data of tool materials and various factors affecting inventory conditions, and perform data cleaning and standardization; Extract and screen effective features of the processed time series data, and further perform EEMD decomposition on the screened effective features; The screened effective features and the features obtained by further EEMD decomposition are used as input to train the constructed time series prediction model, and the trained time series prediction model is used to predict the tool material quantity warning threshold.
2. A tool material inventory quantity management warning threshold prediction method as claimed in claim 1, characterized in that: Collect historical inventory data of tool materials and various inventory influencing factors, including: Collect various conditions and factors that affect the quantity of tool materials used, including maintenance work plans, historical usage data, seasonal factors, equipment operation and maintenance conditions, and process changes.
3. A tool material inventory quantity management warning threshold prediction method as claimed in claim 1, characterized in that: Effective feature extraction and screening of the processed time series data, including: The constructed XGBoost model is trained, and the trained XGBoost model is used to perform preliminary feature extraction and score the importance of the extracted features, and effective features are screened out based on the scores.
4. A tool material inventory quantity management warning threshold prediction method as claimed in claim 3, characterized in that: Filter out effective features based on the scores, including: Features whose importance scores exceed the preset threshold are selected as valid features.
5. A tool material inventory quantity management warning threshold prediction method as claimed in claim 1, characterized in that: The screened effective features are further decomposed by EEMD, including: After multiple EMD decompositions, several intrinsic mode functions (IMFs) and trend components are extracted. Each IMFs represents a signal component with different frequency characteristics. The remaining signal after multiple decompositions is used as the trend component.
6. A tool material inventory quantity management warning threshold prediction method as claimed in claim 1, characterized in that: Training the constructed prediction model includes: The prediction model adopts the BiLSTM model.
7. A tool material inventory quantity management warning threshold prediction method as claimed in claim 6, characterized in that: The BiLSTM model specifically includes: an input layer for feature input; a bidirectional LSTM layer for processing input sequences through forward and reverse LSTM to capture time dependencies; a fully connected layer for connecting LSTM outputs to an output layer; and an output layer for predicting warning thresholds.
8. A tool material inventory quantity management warning threshold prediction method as claimed in claim 6, characterized in that: Training the constructed time series prediction model includes: During the training process, the model parameters are optimized by the back propagation algorithm. After the training is completed, the performance of the model is evaluated using the test set, including mean square error (MSE), root mean square error (RMSE), R 2 The evaluation index of the coefficient is evaluated.
9. A tool material inventory quantity management warning threshold prediction system, characterized in that: The system comprises: Data collection module, used to collect historical inventory data of tool materials and various inventory influencing factors, and perform data cleaning and standardization; The feature extraction module is used to extract and filter effective features of the processed time series data, and further perform EEMD decomposition on the filtered effective features; The prediction module is used to take the screened effective features and the features obtained by further EEMD decomposition as input, train the constructed time series prediction model, and use the trained time series prediction model to predict the tool material quantity warning threshold.
10. An electronic device, characterized in that: The device comprises: a processor and a memory; The memory is used to store one or more program instructions; The processor is used to run one or more program instructions to execute the steps of a tool material inventory quantity management early warning threshold prediction method as described in any one of claims 1 to 8.