Production bottleneck identification method and system based on multi-dimensional data fusion, and storage medium

Abstract

This invention relates to the field of intelligent manufacturing and industrial data analysis technology, specifically to a method, system, and storage medium for identifying production bottlenecks based on multi-dimensional data fusion. The method includes: acquiring production data and constructing a multi-dimensional data vector; updating the weights of a linear prediction model based on the multi-dimensional data vector; performing multi-dimensional data fusion using the updated weights to obtain a comprehensive index value; obtaining a production bottleneck severity score based on the comprehensive index value; determining the production bottleneck status based on the production bottleneck severity score and the comprehensive index value; and identifying factors influencing the production bottleneck based on weight changes. This invention dynamically adjusts the weights using a gradient descent method to score the bottleneck, and dynamically updates the judgment threshold using an exponentially weighted moving average. Finally, it performs multi-dimensional cause analysis by comprehensively considering the weight contribution, change trend, and data anomaly, achieving adaptive and accurate location and cause identification of production bottlenecks.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing and industrial data analysis technology, specifically to a method, system, and storage medium for identifying production bottlenecks based on multi-dimensional data fusion. Background Technology

[0002] Production bottlenecks are the key factors limiting the overall output of a production system. Accurately identifying these bottlenecks is crucial for improving production efficiency and optimizing resource allocation. With the development of intelligent manufacturing, production systems are becoming increasingly complex, and the dynamic changes in the production environment are becoming more frequent, posing numerous challenges to traditional bottleneck identification methods.

[0003] Existing methods for identifying production bottlenecks mainly include those based on theoretical models, production capacity or utilization rates, data mining, and expert systems. These methods generally suffer from the following problems: most pre-determine and maintain the weights of each indicator, which can lead to inaccurate bottleneck identification results for dynamically changing production systems. Similarly, the threshold for determining bottleneck status is often fixed, which may result in oversensitivity or undersensitivity at different production stages. Furthermore, existing methods often analyze production data across only one or a few dimensions and lack in-depth analytical mechanisms to understand the root causes of bottlenecks.

[0004] As the level of automation and informatization of tobacco industry production systems continues to improve, various sensors and automated equipment generate a large amount of production data. How to effectively utilize this data and establish a more accurate and adaptive method for identifying production bottlenecks is a technical problem that urgently needs to be solved. Summary of the Invention

[0005] The purpose of this invention is to provide a production bottleneck identification method, system, and storage medium based on multi-dimensional data fusion, so as to solve the technical problems of low accuracy and limited practicality of existing technologies in identifying production bottlenecks in dynamic and complex modern production environments.

[0006] To achieve the above objectives, embodiments of the present invention provide a production bottleneck identification method based on multi-dimensional data fusion, comprising: Acquire production data and construct multidimensional data vectors; The weights of the linear prediction model are updated based on the multidimensional data vector; The updated weights are used to fuse multi-dimensional data to obtain a comprehensive index value. Based on the comprehensive index values, a score is obtained to indicate the severity of the production bottleneck. The status of the production bottleneck is determined based on the severity score and comprehensive index value of the production bottleneck. Identification of factors affecting production bottlenecks based on weight changes.

[0007] Optionally, updating the weights of the linear prediction model based on the multidimensional data vector includes: Production efficiency is predicted using a linear prediction model and a multidimensional data vector of the current moment. The prediction error is calculated based on the predicted production efficiency and the actual production efficiency. The weights of each dimension are updated based on the prediction error.

[0008] Optionally, updating the weights of each dimension based on the prediction error includes: Update the weights of each dimension according to formulas (1) to (3). (1) (2) (3) in, For the first Each dimension The momentum component value at time t. Momentum factor For the first Each dimension The momentum component value at time t. For prediction error, For the first Normalized data across multiple dimensions For the first Each dimension Weight of time, For the first Each dimension Weight of time, For linear prediction models in Time offset, For linear prediction models in Time offset, This is the learning rate.

[0009] Optionally, the updated weights are used for multi-dimensional data fusion to obtain comprehensive indicator values, including: The comprehensive index value is obtained according to formula (4). (4) in, for The comprehensive index value at any given time. For the first Each dimension Weight of time, For the first Normalized data across multiple dimensions This represents the number of data dimensions.

[0010] Optionally, obtaining a production bottleneck severity score based on the comprehensive index value includes: The severity score of the production bottleneck is obtained according to formula (5). (5) in, for The severity of production bottlenecks at any given moment is scored. For fuzzy membership function parameters, for The comprehensive index value at any given time. This is the intermediate threshold.

[0011] Optionally, determining the production bottleneck status based on the severity score and comprehensive index value includes: Update the bottleneck determination threshold according to formula (6). (6) in, for The bottleneck determination threshold at any given time. for The comprehensive index value at any given time. for The bottleneck determination threshold at any given time. This is the smoothing coefficient.

[0012] Optionally, determining the production bottleneck status based on the severity score and comprehensive index value includes: If the severity score of the production bottleneck is greater than the highest score or the comprehensive index value is less than the bottleneck judgment threshold, it is judged as a high-level bottleneck. If the severity score of the production bottleneck is greater than the lowest score and less than or equal to the highest score, it is determined to be a medium-severity bottleneck. If the severity score of the production bottleneck is less than or equal to the lowest score and the comprehensive index value is greater than the bottleneck determination threshold, it is determined to be a low-severity bottleneck.

[0013] Optionally, calculate the rate of change of the weights for each dimension from the previous time step to the current time step; Calculate the contribution of each dimension to the overall index value; Calculate the overall importance score for each dimension according to formula (7). (7) in, For the first The overall importance score of each dimension, For the first Contribution in each dimension For the first The rate of change of weights in each dimension For the first Normalized data across multiple dimensions For the first The mean of historical data in each dimension For the first Standard deviation of historical data in each dimension , and These are the weighting coefficients.

[0014] On the other hand, the present invention also provides a production bottleneck identification system based on multidimensional data fusion, the identification system including a processor configured to perform any of the methods described above.

[0015] In another aspect, the present invention also provides a computer-readable storage medium storing instructions that, when executed by a processor, implement any of the methods described above.

[0016] The beneficial effects of this invention are: This invention employs a gradient descent method based on momentum factors to dynamically adjust the weights of each dimension, enabling the system to automatically adapt to dynamic changes in the production process without manual parameter adjustment. Simultaneously, the momentum mechanism effectively avoids excessive impact of short-term data fluctuations on weight updates. Regarding data acquisition, this invention constructs a three-tiered acquisition system at the equipment, process, and workflow levels, comprehensively perceiving the production status at different granularities and ensuring accurate bottleneck location. For bottleneck assessment, the Sigmoid function maps comprehensive indicators to bottleneck severity scores, better reflecting the continuous and gradual changes in bottleneck status during production. Furthermore, the exponentially weighted moving average algorithm dynamically updates the bottleneck judgment threshold, allowing the system to adapt to the fluctuation characteristics of different production stages and effectively avoiding misjudgments caused by fixed thresholds. At the causal analysis level, this invention comprehensively considers three dimensions—weight contribution, trend of change, and degree of data anomaly—to calculate the comprehensive importance score of influencing factors. Compared to single-indicator analysis, this is more comprehensive and robust, not only identifying bottleneck locations but also revealing the root causes of bottleneck formation.

[0017] Other features and advantages of the embodiments of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0018] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings: Figure 1 A flowchart of a production bottleneck identification method based on multidimensional data fusion according to an embodiment of the present invention; Figure 2 This is a flowchart of a method for updating the weights of a linear prediction model based on a multidimensional data vector, according to an embodiment of the present invention. Detailed Implementation

[0019] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.

[0020] It should be noted that the acquisition, transmission, storage, use, and processing of data in the technical solution of this application all comply with relevant laws and regulations. In the embodiments of this application, certain existing industry solutions such as software, components, and models may be mentioned. These should be considered exemplary, intended only to illustrate the feasibility of implementing the technical solution of this application, and do not imply that the applicant has already used or necessarily used such solutions.

[0021] like Figure 1 The diagram shows a flowchart of a production bottleneck identification method based on multi-dimensional data fusion according to an embodiment of the present invention. Figure 1 The detection method may include the following steps: In step S10, production data is acquired and a multidimensional data vector is constructed; In step S11, the weights of the linear prediction model are updated based on the multidimensional data vector; In step S12, the updated weights are used to perform multi-dimensional data fusion to obtain a comprehensive index value; In step S13, a score for the severity of the production bottleneck is obtained based on the comprehensive index value; In step S14, the production bottleneck status is determined based on the production bottleneck severity score and comprehensive index value; In step S15, factors affecting production bottlenecks are identified based on weight changes.

[0022] In such Figure 1In the production bottleneck identification method based on multidimensional data fusion shown, the parameters are first set when the method is executed for the first time. In this example, the parameters can be set as follows: learning rate α = 0.01, used to control the step size of each update; momentum factor μ = 0.9, used to control the proportion of historical gradient information retained; fuzzy membership function parameter κ = 10, used to control the steepness of the Sigmoid function; intermediate threshold (midpoint parameter of fuzzy membership function) θ = 0.5; smoothing coefficient β = 0.3, used to control the degree of influence of new data on the dynamic threshold; initial bottleneck judgment threshold Tth (0) = 0.5, the baseline threshold at the beginning of the method; initial weight value wi (0) of each dimension = 1 / n, where n is the number of data dimensions; initial bias b (0) of the linear prediction model; initial momentum value vi (0) of each dimension.

[0023] Based on historical production data, a normalization benchmark is determined for each dimension of data. The normalization benchmark for each dimension of data includes its maximum and minimum values.

[0024] Step S10 is used to collect production data and perform data preprocessing. In this embodiment, production data is collected at different granularities: equipment level, process level, and workflow level, constructing a multi-dimensional data vector. Equipment-level data includes overall equipment efficiency, equipment failure interval time, machine setup and changeover time, and frequency of abnormal interference. Process-level data includes process parameter fluctuation coefficient, first-pass yield, production cycle time achievement rate, and continuous production time ratio. Workflow-level data includes key process capacity balance index, material turnover time, line-side material supply sufficiency rate, and unit product energy consumption. The collected production data includes: overall equipment efficiency, equipment failure interval time, machine setup and changeover time, frequency of abnormal interference, process parameter fluctuation coefficient, first-pass yield, production cycle time achievement rate, continuous production time ratio, key process capacity balance index, material turnover time, line-side material supply sufficiency rate, and unit product energy consumption.

[0025] In this example, overall equipment efficiency, equipment downtime, turnaround time, and frequency of abnormal interference are collected on a per-equipment basis in the production workshop; that is, data for each piece of equipment needs to be collected. Overall equipment efficiency is the product of equipment availability, performance efficiency, and quality pass rate for a single piece of equipment. Equipment downtime is the average running time between two downtimes for a single piece of equipment. Turnaround time is the time required for a single piece of equipment to transition from stopping production of one product to stable production of another. Frequency of abnormal interference is the number of abnormal events requiring manual intervention per unit time for a single piece of equipment. Process parameter fluctuation coefficient, first-pass yield, production cycle time achievement rate, and continuous production time ratio are collected per process (one process involves multiple pieces of equipment); that is, data for each process needs to be collected. Process parameter fluctuation coefficient is the ratio of the standard deviation to the mean of a key process parameter. First-pass yield is the percentage of products that meet quality standards without any rework or adjustments. Production cycle time achievement rate is the ratio of the actual production cycle time to the planned standard cycle time. Continuous production time ratio is the proportion of actual continuous production time to planned production time. Key process capacity balance index, material turnover time, line-side material supply adequacy rate, and unit product energy consumption are collected by production workshop. The key process capacity balance index is the ratio of actual capacity between adjacent processes. Material turnover time is the average time from material entering the production line to finished product output. Line-side material supply adequacy rate is the ratio of actual inventory of key materials to minimum safety stock. Unit product energy consumption is the total energy consumed in producing one unit of product.

[0026] Based on the collected production data, construct a multidimensional data vector: .

[0027] In this example, the specific method for data preprocessing in step S10 may include: normalizing the collected multidimensional data, detecting and processing outliers, and completing missing data to obtain a normalized multidimensional data vector. Specifically, the collected production data is subjected to Min-Max normalization so that the values ​​of each dimension are within the [0,1] interval. Min-Max normalization is used to unify data of different dimensions and ranges into the [0,1] interval while maintaining the relative relationship of the original data distribution, which is convenient for subsequent fusion and comparison. Normalization is performed using formula (8): (8) in, for Time of the first The normalized values ​​of each dimension of data The original value, For the first The minimum value of each dimension in historical data. For the first The maximum value of each dimension in historical data.

[0028] Then, outlier detection is performed on the data values ​​of each dimension. In this example, the interquartile range method is used for outlier detection, and the formula is as follows: (9) in, , It is the first quartile. The median is the third or quartile. If a data point falls outside this range, it is considered an outlier and replaced with the median of that dimension. Industrial production data often does not conform to a normal distribution and may exhibit skewed or multimodal distributions. The interquartile range method, as a non-parametric method, makes fewer assumptions about the data distribution and is more robust to outliers. It is not significantly affected by extreme values ​​on the mean and standard deviation. At the same time, using the median to replace outliers preserves the effective information of the data and avoids the information loss that may result from simply deleting outliers.

[0029] Next, check for missing data. If missing data is found, fill in the missing data by using the average of the corresponding data in the same dimension from historical production data.

[0030] After outlier detection and missing data detection, the normalized multidimensional data vector is obtained as follows: .

[0031] Step S11 is used to update the weights of the linear prediction model based on the multidimensional data vector. In this embodiment, the specific method for updating the weights of the linear prediction model based on the multidimensional data vector in step S11 can be of various forms known to those skilled in the art. In one example of the present invention, step S11 may include, for example... Figure 2 The steps shown are described in this. Figure 2 In this context, step S11 may include: In step S20, production efficiency is predicted using a linear prediction model and the multidimensional data vector at the current moment; In step S21, the prediction error is calculated based on the predicted production efficiency and the actual production efficiency; In step S22, the weights of each dimension are updated based on the prediction error.

[0032] In such Figure 2 In the method shown, step S20 is used to predict production efficiency using the model weights at the previous time T-1 and the normalized data at the current time T. Specifically, the predicted production efficiency is obtained using formula (10): (10) In the formula, Let T be the predicted production efficiency at time T. For the first The weight values ​​of each dimension at time T-1. The bias value at time T-1. This represents the number of data dimensions.

[0033] Step S21 is used to calculate the prediction error based on the predicted production efficiency and the actual production efficiency. In this example, the actual production efficiency at time T, based on the actual production data collected, is... After normalization, it becomes Then, based on the predicted production efficiency and the actual production efficiency, the prediction error is calculated: .

[0034] Step S22 is used to update the weights of each dimension based on the prediction error. Data in the production environment often has short-term fluctuations and noise, requiring an algorithm that can smooth the weight update process. Therefore, in this example, the gradient descent method with dynamic factors is used to update the weights of each dimension according to formulas (1) to (3): (1) (2) (3) in, For the first Each dimension The momentum component value at time t. Momentum factor For the first Each dimension The momentum component value at time t. For prediction error, For the first Each dimension Weight of time, For the first Each dimension Weight of time, For linear prediction models in Time offset, For linear prediction models in Time offset, This is the learning rate. In this example, the momentum factor... The proportion of historical gradient information retained is usually set to a value close to 1. In this method, it is set to 0.9. The purpose is to make the algorithm maintain the inertia of previous updates as much as possible, while being able to react to new error information to a certain extent, but not being overly sensitive to short-term fluctuations.

[0035] By employing gradient descent with momentum factors, the momentum term retains historical gradient information, giving weight updates a certain inertia and reducing oversensitivity to short-term data fluctuations. Furthermore, gradient descent with momentum factors accelerates the convergence process, avoiding slow convergence in flat regions and making weight updates more efficient.

[0036] Step S12 is used to perform multidimensional data fusion using the updated weights to obtain a comprehensive index value. Specifically, in this example, the updated weights are used to perform weighted fusion on the normalized data at the current time T, and the comprehensive index value is obtained according to formula (4): (4) in, for The comprehensive index value at any given time. For the first Each dimension Weight of time.

[0037] Step S13 is used to obtain the severity score of the production bottleneck based on the comprehensive index value. In this example, the Sigmoid function is used to map the comprehensive index value to the severity score of the production bottleneck, and the severity score of the production bottleneck is obtained according to formula (5): (5) in, for The severity score of the production bottleneck at any given moment, with a value range of [0,1]. This is a parameter for the fuzzy membership function, used to control the steepness of the Sigmoid function. This is the intermediate threshold used to locate the midpoint of the Sigmoid function. The Sigmoid function possesses excellent non-linear characteristics and smooth transition, mapping the comprehensive index to a continuous score value within the [0,1] interval. Changes in the middle region are sensitive, while changes at both ends are gradual. This characteristic is well-suited for assessing bottleneck levels. and The parameters can be adjusted to adapt the function's form to the characteristics of different production environments.

[0038] Step S14 is used to determine the production bottleneck status based on the bottleneck severity score and comprehensive index value. In this example, the Exponential Weighted Moving Average (EWMA) algorithm is used to update the determination threshold. Specifically, the bottleneck determination threshold is updated according to formula (6): (6) in, for The bottleneck determination threshold at any given time. for The bottleneck determination threshold at any given time. EWMA (EasyWay Mapping) is a smoothing coefficient used to control the impact of new data on the dynamic threshold. It is used to dynamically adjust the decision threshold. This determines the degree to which new data affects the threshold. A smaller smoothing coefficient can filter out short-term fluctuations while capturing long-term trends, while a larger value makes the threshold more sensitive to new data.

[0039] Next, the production bottleneck status is determined based on the bottleneck severity score and a threshold. In this example, the production bottleneck status is determined according to the following rules: like ,or If so, it is judged as a high-level bottleneck; like If so, it is determined to be a medium bottleneck; like ,and If so, it is judged as a low-level bottleneck.

[0040] Step S15 is used to identify the factors influencing the production bottleneck. By calculating the weight changes, rates of change, contribution, and overall importance score of each dimension, the main factors influencing the production bottleneck are identified. Specifically, in this example, formulas (11) and (12) are used to calculate the weight changes and rates of change for each dimension from the previous time step to the current time step: (11) (12) in, for Time of the first The change in weight of each dimension, for Time of the first The rate of change of weights in each dimension.

[0041] The contribution of each dimension to the overall index value is calculated using formula (13): (13) in, for Time of the first The contribution of the comprehensive indicator value of each dimension.

[0042] The overall importance score for each dimension is calculated using formula (7): (7) in, For the first The overall importance score of each dimension, For the first The mean of historical data in each dimension For the first Standard deviation of historical data in each dimension , and These are the weighting coefficients. The three components of the overall importance score consider current contribution, trend, and degree of data anomaly, providing a more comprehensive assessment compared to a single indicator. It eliminates the need for pre-setting multiple screening thresholds; instead, it selects the top factors by ranking them, thus avoiding screening bias caused by improper threshold settings.

[0043] according to The values ​​are sorted in descending order across all dimensions, and the top 2-3 factors are selected as the main influencing factors.

[0044] On the other hand, the present invention also provides a production bottleneck identification system based on multidimensional data fusion, the identification system including a processor configured to perform any of the methods described in the production bottleneck identification method based on multidimensional data fusion.

[0045] In another aspect, the present invention also provides a computer-readable storage medium storing instructions that, when executed by a processor, implement any of the methods described in the production bottleneck identification method based on multidimensional data fusion.

[0046] The beneficial effects of this invention are: Existing technologies typically employ fixed or empirically set weights, lacking a dynamic adjustment mechanism. This can easily lead to unreasonable weight settings when production conditions change. In this invention, considering that the production system is dynamically changing and the degree of influence of various factors on the bottleneck varies over time, fixed weights cannot adapt to this change. Therefore, a gradient descent method using momentum factors is employed to dynamically update the weights. Its advantage lies in its ability to automatically adjust the weights of each dimension based on the error between actual and predicted production efficiency, continuously optimizing the assessment of the importance of each dimension. Simultaneously, the addition of momentum factors prevents weight updates from becoming overly sensitive to short-term fluctuations. Secondly, in the bottleneck identification stage, existing technologies typically use a threshold judgment method, comparing the indicator value with a fixed threshold to obtain a binary judgment result of whether or not there is a bottleneck. The method disclosed in this invention uses a sigmoid function for fuzzy mapping, transforming deterministic judgment into a degree-based assessment. Combined with the EWMA algorithm for dynamically updating the judgment threshold, the threshold is automatically adjusted based on historical data trends, enabling the judgment standard to adapt to long-term changes in the production environment. Furthermore, the set smoothing coefficient controls the threshold's sensitivity to short-term fluctuations. Finally, in the stage of identifying influencing factors, the method disclosed in this invention considers three dimensions simultaneously: current contribution, trend of change, and degree of data anomaly. Compared with the conventional approach of ranking influencing factors using a single indicator, the implementation method of this invention provides a more comprehensive assessment of the impact of each factor on the bottleneck.

[0047] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0048] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0049] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0050] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0051] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0052] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, like read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0053] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0054] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0055] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for identifying production bottlenecks based on multi-dimensional data fusion, characterized in that, The identification method includes: Acquire production data and construct multidimensional data vectors; The weights of the linear prediction model are updated based on the multidimensional data vector; The updated weights are used to fuse multi-dimensional data to obtain a comprehensive index value. Based on the comprehensive index values, a score is obtained to indicate the severity of the production bottleneck. The status of the production bottleneck is determined based on the severity score and comprehensive index value of the production bottleneck. Identification of factors affecting production bottlenecks based on weight changes.

2. The identification method according to claim 1, characterized in that, The weights of the linear prediction model updated based on the multidimensional data vector include: Production efficiency is predicted using a linear prediction model and a multidimensional data vector of the current moment. The prediction error is calculated based on the predicted production efficiency and the actual production efficiency. The weights of each dimension are updated based on the prediction error.

3. The identification method according to claim 2, characterized in that, Updating the weights of each dimension based on the prediction error includes: Update the weights of each dimension according to formulas (1) to (3). ，（1） ，（2） ，（3） in, For the first Each dimension The momentum component value at time t. Momentum factor For the first Each dimension The momentum component value at time t. For prediction error, For the first Normalized data across multiple dimensions For the first Each dimension Weight of time, For the first Each dimension Weight of time, For linear prediction models in Time offset, For linear prediction models in Time offset, This is the learning rate.

4. The identification method according to claim 1, characterized in that, The updated weights are used to perform multi-dimensional data fusion to obtain comprehensive indicator values, including: The comprehensive index value is obtained according to formula (4). ，（4） in, for The comprehensive index value at any given time. For the first Each dimension Weight of time, For the first Normalized data across multiple dimensions This represents the number of data dimensions.

5. The identification method according to claim 1, characterized in that, Based on the comprehensive index values, the severity score of the production bottleneck is obtained, including: The severity score of the production bottleneck is obtained according to formula (5). ，（5） in, for The severity of production bottlenecks at any given moment is scored. For fuzzy membership function parameters, for The comprehensive index value at any given time. This is the intermediate threshold.

6. The identification method according to claim 1, characterized in that, Determining the status of a production bottleneck based on the severity score and comprehensive index value includes: Update the bottleneck determination threshold according to formula (6). ，（6） in, for The bottleneck determination threshold at any given time. for The comprehensive index value at any given time. for The bottleneck determination threshold at any given time. This is the smoothing coefficient.

7. The identification method according to claim 6, characterized in that, Determining the status of a production bottleneck based on the severity score and comprehensive index value includes: If the severity score of the production bottleneck is greater than the highest score or the comprehensive index value is less than the bottleneck judgment threshold, it is judged as a high-level bottleneck. If the severity score of the production bottleneck is greater than the lowest score and less than or equal to the highest score, it is determined to be a medium-severity bottleneck. If the severity score of the production bottleneck is less than or equal to the lowest score and the comprehensive index value is greater than the bottleneck determination threshold, it is determined to be a low-severity bottleneck.

8. The identification method according to claim 1, characterized in that, Identifying factors influencing production bottlenecks based on weight changes includes: Calculate the rate of change of the weight of each dimension from the previous time step to the current time step; Calculate the contribution of each dimension to the overall index value; Calculate the overall importance score for each dimension according to formula (7). ，（7） in, For the first The overall importance score of each dimension, For the first Contribution in each dimension For the first The rate of change of weights in each dimension For the first Normalized data across multiple dimensions For the first The mean of historical data in each dimension For the first Standard deviation of historical data in each dimension , and These are the weighting coefficients.

9. A production bottleneck identification system based on multi-dimensional data fusion, characterized in that, The identification system includes a processor configured to perform the method as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed by a processor, implement the method as described in any one of claims 1 to 8.

Images