Multi-modal intelligent decision-making collaboration system for digital transformation of manufacturing industry

Abstract

The application belongs to the technical field of manufacturing management, and is a multi-modal intelligent decision-making collaboration system for digital transformation of manufacturing, comprising a multi-source industrial state perception collection module, a cross-domain feature fusion module, an equipment health evolution early warning module, a quality full-chain optimization early warning grading module, a supply chain resilience collaborative decision-making module and a production global efficiency evaluation module; the application collects and standardizes the preprocessing of multi-source heterogeneous data of the whole manufacturing process, realizes equipment health state grading early warning, product quality full-chain scoring grading, supply chain resilience evaluation and intelligent scheduling of procurement, inventory and logistics after cross-domain feature fusion, and can comprehensively evaluate the overall performance of the manufacturing line, locate the short board and push optimization instructions, significantly improve the equipment reliability, product qualification rate and supply chain resilience, comprehensively optimize the global efficiency of production, and help the digital transformation of manufacturing.

Description

Technical Field

[0001] This invention relates to the field of manufacturing management technology, specifically a multimodal intelligent decision-making and collaborative system for the digital transformation of the manufacturing industry. Background Technology

[0002] Against the backdrop of the deepening digital transformation of the manufacturing industry, production equipment operation and maintenance, product quality control, supply chain scheduling, and overall production efficiency assessment have become core links for manufacturing enterprises to achieve refined and intelligent management. Relying on multi-source industrial data to carry out cross-link collaborative decision-making is a key support for improving production efficiency, reducing operating costs, and enhancing the resilience of industrial development.

[0003] Currently, most digital management systems in the manufacturing industry adopt a segmented and independent control architecture, which can only carry out local data collection and basic analysis for single business modules such as equipment operation, product quality, and supply chain flow. They cannot unify, integrate, and efficiently utilize multi-source and heterogeneous data generated throughout the entire production process. They have not built an intelligent decision-making system covering the entire production domain. In key links such as equipment health monitoring, product quality control, and supply chain scheduling optimization, they lack multi-dimensional and comprehensive judgment mechanisms and full-process control methods. They cannot achieve accurate early warning of abnormal situations, nor can they complete intelligent scheduling and overall optimization, which is not conducive to the digital transformation of the manufacturing industry.

[0004] Therefore, developing a multimodal intelligent decision-making and collaborative system for the digital transformation of the manufacturing industry, capable of early warning of equipment health evolution, optimization and grading of product quality across the entire chain, collaborative decision-making on supply chain resilience, and comprehensive evaluation of production efficiency across the entire domain, has become an urgent technical problem to be solved in the field of manufacturing management. Summary of the Invention

[0005] The purpose of this invention is to provide a multimodal intelligent decision-making and collaboration system for the digital transformation of the manufacturing industry, in order to solve the problems that the lack of an intelligent decision-making system covering the entire production domain of an enterprise makes it difficult to adapt to the actual needs of efficient, accurate and integrated management and control in the process of digital transformation of the manufacturing industry, and the obvious deficiencies in collaborative operation capabilities and intelligent management and control levels.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a multimodal intelligent decision-making and collaborative system for digital transformation of the manufacturing industry, comprising a multi-source industrial state perception and acquisition module, a cross-domain feature fusion module, an equipment health evolution early warning module, a quality full-chain optimization early warning and grading module, and a supply chain resilience collaborative decision-making module;

[0007] The multi-source industrial state perception and acquisition module is responsible for the full-domain acquisition and standardized preprocessing of multi-source heterogeneous data throughout the entire manufacturing process, and constructs a unified industrial data resource pool. The cross-domain feature fusion module performs feature extraction and unified fusion of multi-modal industrial data, and converts structured and unstructured data into regular feature data.

[0008] The equipment health evolution early warning module monitors and judges the health status of key production equipment and outputs the health status judgment information of each production equipment; the quality full-chain optimization early warning and classification module builds a product quality full-process traceability system and outputs the comprehensive product quality score result; the supply chain resilience collaborative decision-making module uses three core indicators of delivery, inventory and guarantee to build a collaborative efficiency model for analysis, and makes intelligent scheduling decisions for procurement, inventory and logistics based on the analysis results.

[0009] Furthermore, the specific process by which the equipment health evolution early warning module monitors and judges the health status of production equipment is as follows:

[0010] The system receives fused and regularized feature data from the cross-domain feature fusion module. Through performance degradation decision analysis, it obtains the performance degradation coefficient D of the corresponding key production equipment. It also extracts the core parameter data of the corresponding key production equipment from the regularized feature data, constructs the equipment health evolution formula, and calculates the health index HI of the corresponding key production equipment. Based on the HI value, the equipment health status is divided into three levels to accurately judge the operating status of the corresponding key production equipment. Furthermore, it statistically analyzes the HI values ​​of all key production equipment in the product manufacturing line, calculates the arithmetic mean of these values ​​to obtain the overall health rate HIw of the production line equipment, and outputs it.

[0011] Furthermore, the strategy for classifying device health status into three levels based on HI values ​​is as follows:

[0012] If HI≥80, the corresponding key production equipment is judged to be in a stable and healthy state, requiring no additional maintenance and only routine inspections.

[0013] If 60≤HI<80, the corresponding key production equipment is judged to be in a sub-healthy state with slight performance degradation. It is necessary to pay close attention to the changes in the corresponding parameters and arrange regular special inspections.

[0014] If HI < 60, the corresponding critical production equipment is judged to be in an abnormal health state with a high risk of failure, and an early warning should be triggered immediately.

[0015] Furthermore, the specific analysis process for performance degradation decision analysis is as follows:

[0016] Obtain the standard maintenance cycle Tpm for the corresponding key production equipment, mark the time when the actual maintenance cycle exceeds the standard maintenance cycle as the non-maintenance time, and calculate the maintenance cycle over-measurement value Tunm by averaging all non-maintenance times in the historical period.

[0017] In addition, the cumulative effective operating time Trun and the rated total operating life Tlife of the corresponding key production equipment are obtained, and the duration Tenv and the total duration Ttotal of the statistical period of the corresponding key production equipment in the environment with excessive temperature, humidity and dust are obtained in the historical period.

[0018] The performance degradation coefficient D of the corresponding key production equipment is calculated using the formula D=0.4×(Trun / Tlife)+0.3×(Tunm / Tpm)+0.3×(Tenv / Ttotal).

[0019] Furthermore, the operational analysis process of the quality whole-chain optimization early warning and grading module includes:

[0020] The system receives the fused and normalized feature data output from the cross-domain feature fusion module, constructs a quality impact factor map of the produced products, and calculates the product quality using a comprehensive quality scoring formula. After calculating the comprehensive product quality score Qs, the product quality level is quantified based on the Qs value and divided into four levels:

[0021] When Qs≥90, the product quality is considered excellent, all test indicators meet the standards, batch consistency is good, and no additional quality control measures are required.

[0022] When 80≤Qs<90, the product quality is considered acceptable, but there are minor appearance or consistency defects. It is necessary to optimize the production process details to reduce the occurrence rate of minor defects.

[0023] When 70≤Qs<80, the product quality is considered to be basically qualified, but there are general functional defects. It is necessary to trace the root cause of the problem and adjust the relevant links in a timely manner.

[0024] When Qs < 70, the product quality is deemed unqualified, indicating a significant quality hazard. Production on the manufacturing line must be suspended, a comprehensive investigation of quality-influencing factors must be conducted, and production can only resume after rectification is completed and the product meets the required standards.

[0025] Furthermore, the supply chain resilience collaborative decision-making module receives the fused and regularized feature data output by the cross-domain feature fusion module, retrieves the actual inventory and safety stock of the corresponding raw materials, calculates the ratio of the difference between the actual inventory and the safety stock to the safety stock, obtains the material inventory impact coefficient by subtracting the ratio result from the value 1, and calculates the inventory health Ihealth by averaging the inventory impact coefficients of all raw materials.

[0026] Furthermore, the normalized unit supply chain cost Cunit is calculated by comparing the actual unit supply chain cost with the maximum allowable unit supply chain cost of the enterprise; the on-time delivery rate Drate is calculated by comparing the number of on-time delivered orders in the current period with the total number of orders in the current period; and the supply capacity coefficient is calculated by comparing the available supply of the corresponding suppliers with the enterprise's demand; and the supply assurance rate Ssafe is calculated by averaging the supply capacity coefficients of all suppliers.

[0027] A formula for calculating supply chain collaboration efficiency is constructed. After calculating the supply chain collaboration efficiency value Se, the resilience of the supply chain is judged based on the Se value.

[0028] Furthermore, the methods for assessing supply chain resilience are as follows:

[0029] If Se≥0.85: the supply chain is considered to be resilient, with stable delivery, reasonable inventory and no supply risk, and the current scheduling decision can be maintained;

[0030] 0.7≤Se<0.85: The supply chain is in good condition, with minor inventory fluctuations or delivery delays. It can be optimized by slightly adjusting the purchase frequency.

[0031] Se < 0.7: This indicates a risk in the supply chain, such as material shortages, delays, or cost overruns, requiring an immediate switch to alternative suppliers and replenishment of safety stock.

[0032] Furthermore, the equipment health evolution early warning module, the quality full-chain optimization early warning and grading module, and the supply chain resilience collaborative decision-making module are all connected to the production full-domain efficiency assessment module. The production full-domain efficiency assessment module comprehensively evaluates the production performance of the entire manufacturing production line, identifies shortcomings based on the comprehensive assessment results, and pushes optimization instructions back to the corresponding modules.

[0033] Furthermore, the specific evaluation and analysis process of the production-wide efficiency assessment module for comprehensively evaluating the production performance of the entire manufacturing production line is as follows:

[0034] The system retrieves the comprehensive health rate HIw of production line equipment calculated by the equipment health evolution early warning module, the comprehensive product quality score Qs calculated by the quality whole chain optimization early warning and grading module, and the supply chain collaboration efficiency value Se calculated by the supply chain resilience collaboration decision module. It also collects relevant data on production efficiency and cost control to construct a comprehensive evaluation formula for the efficiency of the entire production domain.

[0035] After calculating the overall production efficiency score Pe, the overall production efficiency level is determined based on the overall production efficiency score Pe.

[0036] Furthermore, the specific strategy for determining the overall production efficiency level based on the comprehensive production efficiency score Pe is as follows:

[0037] If Pe≥90: Overall production efficiency is excellent, and all dimensions are well-balanced; maintain the existing strategy.

[0038] If 80≤Pe<90: Overall production efficiency is good, but there is a single weakness. Targeted fine-tuning of the corresponding module parameters is necessary.

[0039] If 70≤Pe<80: The overall production efficiency is average, with obvious shortcomings, and a special optimization plan needs to be developed.

[0040] If Pe < 70: Production efficiency is poor across the board, and multiple dimensions fail to meet standards. A comprehensive investigation and optimization of equipment, quality, and supply chain strategies are needed.

[0041] Compared with the prior art, the beneficial effects of the present invention are:

[0042] 1. In this invention, by standardizing the collection of multi-source data and integrating it across domains, we can achieve equipment health classification and early warning, full-chain management and control of product quality, and intelligent scheduling of the supply chain. This helps to solve the problem of insufficient coordination of equipment, quality and supply chain in the manufacturing industry, and improve equipment reliability, product qualification rate and supply chain resilience.

[0043] 2. In this invention, through comprehensive performance evaluation, the overall performance of the production line can be comprehensively assessed, production shortcomings can be accurately located and optimization instructions can be pushed, so as to realize quantitative control and continuous optimization of the entire production process, comprehensively improve the overall production efficiency, and help the digital transformation of the manufacturing industry. Attached Figure Description

[0044] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings;

[0045] Figure 1 This is a schematic diagram of the overall system structure of the present invention;

[0046] Figure 2 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0047] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0048] Example 1: Refer to Figure 1-2 As shown, the multimodal intelligent decision-making and collaborative system for digital transformation of manufacturing proposed in this invention includes a multi-source industrial state perception and acquisition module, a cross-domain feature fusion module, an equipment health evolution early warning module, a quality full-chain optimization early warning and classification module, and a supply chain resilience collaborative decision-making module.

[0049] The multi-source industrial state perception and acquisition module is responsible for the full-domain acquisition and standardized preprocessing of multi-source heterogeneous data throughout the entire manufacturing process, constructing a unified industrial data resource pool, realizing the full-domain acquisition and standardization of data throughout the entire manufacturing process, significantly improving data integrity and availability, and providing a high-quality data foundation for subsequent integration and decision-making.

[0050] It should be noted that the multi-source industrial state perception and acquisition module collects structured data such as equipment operating parameters, production process data, quality inspection data, supply chain logistics data, and energy consumption data in real time through sensing devices such as industrial sensors, smart meters, industrial cameras, acoustic sensors, and RFID readers deployed on the production site, as well as unstructured data such as equipment vibration audio, product surface images, production environment videos, and operation voice commands.

[0051] The collected raw data is preprocessed, including time-series interpolation filling of missing values, joint detection of outliers using the 3σ criterion and isolated forest, data Z-score standardization, and unified conversion of unstructured data formats. A unified industrial data dictionary and metadata management system are constructed to achieve data classification, storage, indexing and control, forming a standardized multimodal industrial data resource pool.

[0052] The cross-domain feature fusion module extracts and integrates features from multimodal industrial data, transforming structured and unstructured data into regularized feature data. This completes the unified regularization of multimodal data, reduces the data processing pressure on subsequent decision-making modules, ensures that the input data format is uniform and can be directly called, and provides a data foundation for multi-module collaborative decision-making.

[0053] It should be noted that after receiving standardized data from the multi-source industrial state perception and acquisition module, the cross-domain feature fusion module performs conventional feature extraction for different types of data. For structured data, key parameter features are extracted using a dimensionality reduction algorithm. For unstructured data such as images and audio, basic features are extracted using a general deep learning model. Then, the multiple types of features are combined and merged to form feature data of a unified dimension. This data is then directly distributed in parallel to the equipment health evolution early warning module, the quality full-chain optimization early warning and grading module, and the supply chain resilience collaborative decision-making module, serving as the unified data basis for decision analysis in each module.

[0054] The equipment health evolution early warning module monitors and judges the health status of key production equipment, outputs the health status judgment information of each production equipment, realizes real-time monitoring of equipment health status and early warning of faults, improves the accuracy of equipment fault prediction, reduces unplanned downtime and maintenance costs, extends equipment service life, and outputs the comprehensive health rate of production line equipment, which intuitively reflects the overall equipment operation level of the production line.

[0055] Specifically, the equipment health evolution early warning module receives the fused and regularized feature data output by the cross-domain feature fusion module, obtains the standard maintenance cycle Tpm of the corresponding key production equipment, marks the time when the actual maintenance cycle exceeds the standard maintenance cycle as the non-maintenance time, and calculates the maintenance cycle over-measure value Tunm by averaging all the non-maintenance times in the historical period. It should be noted that the larger the value of the maintenance cycle over-measure value Tunm, the more untimely the maintenance of the corresponding key production equipment in the historical period is.

[0056] In addition, the cumulative effective operating time Trun and the rated total operating life Tlife of the corresponding key production equipment are obtained, and the duration Tenv and the total duration Ttotal of the statistical period of the corresponding key production equipment in the environment with excessive temperature, humidity and dust are obtained in the historical period.

[0057] The performance degradation coefficient D of the corresponding key production equipment is calculated using the formula D = 0.4 × (Trun / Tlife) + 0.3 × (Tunm / Tpm) + 0.3 × (Tenv / Ttotal), where D ∈ [0, 0.5]. Furthermore, core parameter data (such as vibration, temperature, pressure, and energy consumption) of the corresponding key production equipment are extracted from the regularized feature data to construct the equipment health evolution formula, as follows:

[0058] ;

[0059] Wherein, HI: the health index of the corresponding key production equipment, HI∈[0,100], the higher the value, the better the health status of the equipment;

[0060] m: The number of core parameters of the corresponding key production equipment;

[0061] sj: The importance weight of the j-th device parameter; sj∈[0,1], and the sum of the importance weights of all device parameters is 1;

[0062] hj: The normalized health value of the j-th parameter, hj∈[0,1]; Calculation method: hj=1-|xj-xjo| / (xjmax-xjmin), where xj is the real-time collected value of the parameter, xjo is the standard operating value of the parameter, and xjmax and xjmin are the upper and lower limits of the parameter respectively;

[0063] After calculating the Health Index (HI) of the corresponding key production equipment, the equipment health status is divided into three levels based on the HI value to accurately determine the operating status of the corresponding key production equipment. The equipment health status classification strategy is as follows:

[0064] If HI≥80, the corresponding key production equipment is judged to be in a stable and healthy state, requiring no additional maintenance and only routine inspections.

[0065] If 60≤HI<80, the corresponding key production equipment is judged to be in a sub-healthy state with slight performance degradation. It is necessary to pay close attention to the changes in the corresponding parameters and arrange regular special inspections.

[0066] If HI < 60, the corresponding critical production equipment is judged to be in an abnormal health state with a high risk of failure, and an early warning needs to be triggered immediately (by combining the fault classification model to predict the fault type, location and time, and generating accurate maintenance suggestions including maintenance time, content and spare parts).

[0067] Furthermore, the equipment health evolution early warning module also counts the HI values ​​of all key production equipment in the product manufacturing line, calculates the arithmetic mean of these values ​​to obtain the overall health rate HIw of the production line equipment and outputs it. It should be noted that the value range of HIw is also [0,100], and its result directly reflects the overall operating level of the equipment in the manufacturing line.

[0068] The quality whole-chain optimization early warning and classification module constructs a product quality full-process traceability system, outputs comprehensive product quality scores, strengthens product quality stability, improves product pass rate, shortens the time for handling quality problems, reduces quality costs and customer complaint rate, and achieves precise and proactive quality control.

[0069] Specifically, the quality whole-chain optimization early warning and classification module receives the fused and normalized feature data output by the cross-domain feature fusion module, constructs a quality impact factor map of the produced products, and calculates the quality of the produced products using a comprehensive quality scoring formula, as follows:

[0070] ;

[0071] Where Qs is the overall product quality score, Qs∈[0,100], and the higher the value, the better the quality level;

[0072] k: Number of quality inspection indicators;

[0073] g: Number of quality problems;

[0074] qi: Score of the i-th quality inspection indicator, qi∈[0,100]; Calculation method: According to industry quality standards, the measured value reaches the mark of 100 points, and 5 points are deducted for every 1% deviation from the standard value, with the lowest score being 0 points;

[0075] ti: The weight coefficient of the i-th indicator, ti∈[0,1], and the sum of the weight coefficients of all indicators is 1; Calculation method: The judgment matrix is ​​constructed by using the analytic hierarchy process, normalized and determined and stored in advance, for example, the weight of key size is 0.4, the weight of performance is 0.3, the weight of appearance is 0.2, and the weight of batch consistency is 0.1.

[0076] lg: The severity coefficient of the j-th quality problem, lg∈[1,5]; for example, a fatal safety defect is assigned a value of 5, a major performance defect is assigned a value of 4, a general functional defect is assigned a value of 3, a minor flaw is assigned a value of 2, and a minor appearance defect is assigned a value of 1.

[0077] rg: the probability of the j-th quality problem occurring, rg∈[0,1]; calculation method: rg=number of products involved in the defect in the last 10 batches / total number of products in the last 10 batches;

[0078] After calculating the overall product quality score Qs, the product quality level is quantified based on the Qs value and divided into four levels, as shown below:

[0079] When Qs≥90, the product quality is considered excellent, all test indicators meet the standards, batch consistency is good, and no additional quality control measures are required.

[0080] When 80≤Qs<90, the product quality is considered acceptable, but there are minor appearance or consistency defects. It is necessary to optimize the production process details to reduce the occurrence rate of minor defects.

[0081] When 70≤Qs<80, the product quality is considered to be basically qualified, but there are general functional defects. It is necessary to trace the root cause of the problem and adjust the relevant links in a timely manner.

[0082] When Qs < 70, the product quality is deemed unqualified, indicating a significant quality hazard. Production on the manufacturing line must be suspended, a comprehensive investigation of quality-influencing factors must be conducted, and production can only resume after rectification is completed and the product meets the required standards.

[0083] The supply chain resilience collaborative decision-making module uses three core indicators—delivery, inventory, and assurance—to construct a collaborative efficiency model for analysis. This enables intelligent scheduling decisions for procurement, inventory, and logistics, shortening supply chain response and product delivery cycles, improving inventory turnover, reducing supply chain costs, and significantly enhancing supply chain resilience and risk resistance.

[0084] Specifically, the supply chain resilience collaborative decision-making module receives the fused and regularized feature data output by the cross-domain feature fusion module, retrieves the actual inventory and safety stock of the corresponding raw materials, calculates the ratio of the difference between the actual inventory and the safety stock to the safety stock, obtains the material inventory impact coefficient by subtracting the ratio result from the value 1, and calculates the inventory health Ihealth by averaging the inventory impact coefficients of all raw materials, where Ihealth∈[0,1].

[0085] Furthermore, the normalized value of the unit product supply chain cost, Cunit, is obtained by calculating the ratio of the actual unit supply chain cost to the maximum unit supply chain cost allowed by the enterprise, where Cunit∈[0,1].

[0086] The on-time delivery rate Drate is calculated by comparing the number of on-time delivered orders in the current period with the total number of orders in the current period, where Drate∈[0,1].

[0087] And collect information on the suppliers involved in the enterprise, calculate the supply capacity coefficient by the ratio of the available supply of the corresponding supplier to the enterprise's demand, and calculate the supply guarantee rate Ssafe by averaging the supply capacity coefficients of all suppliers, and Ssafe∈[0,1].

[0088] It should be noted that Ssafe is a ratio indicator that measures whether the suppliers involved in a company can meet the current production material demand of the production line. It is used to judge whether the company's supply chain will experience material shortages or supply disruptions and whether it can ensure continuous production. It is one of the most intuitive core indicators of supply chain resilience.

[0089] The formula for calculating supply chain collaboration efficiency is as follows:

[0090] ;

[0091] Where Se is the supply chain collaboration efficiency value, normalized to [0,1]. The higher the value, the more stable and efficient the supply chain operation.

[0092] ψ1, ψ2, ψ3, ψ4: Preset weighting coefficients, and the sum of the four is 1; for example, in regular production, ψ1=0.35, ψ2=0.25, ψ3=0.25, ψ4=0.15;

[0093] After calculating the supply chain collaboration efficiency value Se, the supply chain resilience is judged based on the Se value. The method for judging supply chain resilience is as follows:

[0094] If Se≥0.85: the supply chain is considered to be resilient, with stable delivery, reasonable inventory and no supply risk, and the current scheduling decision can be maintained;

[0095] 0.7≤Se<0.85: The supply chain is in good condition, with minor inventory fluctuations or delivery delays. It can be optimized by slightly adjusting the purchase frequency.

[0096] Se < 0.7: This indicates a risk in the supply chain, such as material shortages, delays, or cost overruns, requiring an immediate switch to alternative suppliers and replenishment of safety stock.

[0097] Example 2: Refer to Figure 1-2As shown, the difference between this embodiment and Embodiment 1 is that the equipment health evolution early warning module, the quality full-chain optimization early warning and grading module, and the supply chain resilience collaborative decision-making module are all communicatively connected to the production full-domain efficiency assessment module. The production full-domain efficiency assessment module comprehensively assesses the production performance of the entire manufacturing production line, locates the shortcomings based on the comprehensive assessment results, and pushes the optimization instructions back to the corresponding modules.

[0098] By integrating multi-dimensional data such as equipment health, product quality, and supply chain collaboration, a comprehensive assessment of the entire production line's efficiency is conducted, and weaknesses are accurately identified. The accuracy of weakness identification is high, providing precise guidance for end-to-end optimization and significantly improving the overall production efficiency and digital collaborative management level of manufacturing production lines. The specific assessment and analysis process is as follows:

[0099] The system retrieves the comprehensive health rate HIw of production line equipment calculated by the equipment health evolution early warning module, the comprehensive product quality score Qs calculated by the quality full-chain optimization early warning grading module, and the supply chain collaboration efficiency value Se calculated by the supply chain resilience collaboration decision-making module. It also collects relevant data on production efficiency and cost control to construct a comprehensive evaluation formula for overall production efficiency.

[0100] ;

[0101] Among them, Pe: Comprehensive score of overall production efficiency;

[0102] w1, w2, w3, w4, and w5 are the weighting coefficients for equipment health, product quality, supply chain collaboration, production efficiency, and cost control, respectively, and the sum of the five is 1. For example, for routine production, w1=0.2, w2=0.25, w3=0.2, w4=0.2, and w5=0.15.

[0103] HIw: Overall health rate of key equipment in the production line, which is the arithmetic mean of the health of multiple equipment output by the equipment health evolution early warning module, HIW∈[0,100];

[0104] Qs: Comprehensive quality score, calculated in real time by calling the quality full-chain optimization early warning and classification module, Qs∈[0,100];

[0105] Se: Supply chain collaboration efficiency, which is the normalized result of the supply chain resilience collaboration decision module;

[0106] Ep: Normalized production efficiency, Ep = actual output / theoretical maximum output;

[0107] Cp: ​​Actual production cost per unit of product, Cp = (raw material cost + energy cost + labor cost) / total output of qualified products;

[0108] Cpmax: The upper limit for unit product cost control, the enterprise's budget target value;

[0109] After calculating the overall production efficiency score Pe, the overall production efficiency level is determined based on the overall production efficiency score Pe. The specific strategy is as follows:

[0110] If Pe≥90: the overall production efficiency is excellent, and all dimensions are well-balanced; maintain the existing strategy.

[0111] If 80≤Pe<90: the overall production efficiency is good, but there is a single weakness. The corresponding module parameters should be fine-tuned accordingly.

[0112] If 70≤Pe<80: the overall production efficiency is generally poor, with obvious shortcomings, and a specific optimization plan needs to be developed.

[0113] If Pe < 70: It indicates poor overall production efficiency and failure to meet standards in multiple dimensions. A comprehensive investigation and optimization of equipment, quality, and supply chain strategies are necessary.

[0114] The working principle of this invention is as follows: During use, multi-source heterogeneous data from the entire manufacturing process is collected and standardized preprocessed. Through cross-domain feature fusion, unified and regularized feature data is generated. This enables sequentially achieving graded early warning of equipment health status, graded scoring of product quality across the entire chain, and intelligent scheduling of supply chain procurement, inventory, and logistics. Furthermore, the overall efficiency assessment module comprehensively evaluates the overall production line efficiency, accurately identifies production shortcomings, and pushes optimization instructions. This effectively addresses the pain points of insufficient coordination between equipment, quality, and the supply chain in the manufacturing industry, achieving early warning of equipment failures, full-process control of product quality, and intelligent scheduling of the supply chain. It significantly improves equipment reliability, product qualification rate, and supply chain resilience, comprehensively optimizing overall production efficiency and assisting in the digital transformation of the manufacturing industry.

[0115] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, enabling those skilled in the art to better understand and utilize it. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A multimodal intelligent decision-making and collaborative system for digital transformation of the manufacturing industry, characterized in that: It includes a multi-source industrial state perception and acquisition module, a cross-domain feature fusion module, an equipment health evolution early warning module, a quality full-chain optimization early warning and classification module, and a supply chain resilience collaborative decision-making module; The multi-source industrial state perception and acquisition module is responsible for the full-domain acquisition and standardized preprocessing of multi-source heterogeneous data throughout the entire manufacturing process, and constructs a unified industrial data resource pool. The cross-domain feature fusion module performs feature extraction and unified fusion of multi-modal industrial data, and converts structured and unstructured data into regular feature data. The equipment health evolution early warning module monitors and judges the health status of key production equipment and outputs the health status judgment information of each production equipment. The quality whole-chain optimization early warning and classification module constructs a product quality traceability system throughout the entire process and outputs a comprehensive product quality score. The supply chain resilience collaborative decision-making module uses three core indicators—delivery, inventory, and assurance—to construct a collaborative efficiency model for analysis, and makes intelligent scheduling decisions for procurement, inventory, and logistics based on the analysis results.

2. The multimodal intelligent decision-making and collaborative system for digital transformation of manufacturing industry according to claim 1, characterized in that, The specific process by which the equipment health evolution early warning module monitors and judges the health status of production equipment is as follows: The performance degradation decision analysis is used to obtain the performance degradation coefficient D of the corresponding key production equipment, and the equipment health evolution formula is constructed. After calculating the health index HI of the corresponding key production equipment, the equipment health status is divided into three levels based on the HI value. The HI values ​​of all key production equipment in the product manufacturing line are statistically analyzed, and the arithmetic mean is calculated to obtain the comprehensive health rate HIw of the production line equipment and output.

3. The multimodal intelligent decision-making and collaborative system for digital transformation of manufacturing industry according to claim 2, characterized in that, The strategy for classifying equipment health status into three levels based on the HI value is as follows: HI≥80 indicates that the corresponding key production equipment is in a stable and healthy state; 60≤HI<80 indicates that the corresponding key production equipment is in a sub-healthy state; and HI<60 indicates that the corresponding key production equipment is in an abnormal health state.

4. The multimodal intelligent decision-making and collaborative system for digital transformation of manufacturing industry according to claim 2, characterized in that, The specific analysis process of performance degradation decision analysis is as follows: Obtain the standard maintenance cycle Tpm and maintenance cycle over-measurement value Tunm for the corresponding key production equipment, as well as the cumulative effective running time Trun and rated total operating life Tlife for the corresponding key production equipment, and obtain the duration Tenv and total statistical period Ttotal for the corresponding key production equipment in environments with excessive temperature, humidity, and dust in historical periods. The performance degradation coefficient D of the corresponding key production equipment is obtained through calculation.

5. The multimodal intelligent decision-making and collaborative system for digital transformation of manufacturing industry according to claim 1, characterized in that, The operational analysis process of the quality whole-chain optimization early warning and classification module includes: The quality of manufactured products is calculated using a comprehensive quality scoring formula. After obtaining the comprehensive product quality score Qs, the product quality level is quantified based on the Qs value and divided into four levels: When Qs≥90, the product quality is considered excellent; when 80≤Qs<90, the product quality is considered acceptable; when 70≤Qs<80, the product quality is considered basically acceptable; when Qs<70, the product quality is considered unacceptable.

6. The multimodal intelligent decision-making and collaborative system for digital transformation of manufacturing industry according to claim 1, characterized in that, The supply chain resilience collaborative decision-making module analyzes and obtains inventory health (Ihealth), normalized unit product supply chain cost (Cunit), order delivery timeliness (Drate), and supply assurance rate (Ssafe) to construct a formula for calculating supply chain collaborative efficiency. After calculating the supply chain collaborative efficiency value (Se), the module judges the supply chain resilience based on the Se value.

7. The multimodal intelligent decision-making and collaborative system for digital transformation of manufacturing industry according to claim 6, characterized in that, The method for judging supply chain resilience is as follows: Se≥0.85: the supply chain resilience is excellent; 0.7≤Se<0.85: the supply chain is in good condition. Se < 0.7: Indicates that there is a risk in the supply chain.

8. The multimodal intelligent decision-making and collaborative system for digital transformation of manufacturing industry according to claim 1, characterized in that, The equipment health evolution early warning module, the quality whole-chain optimization early warning and grading module, and the supply chain resilience collaborative decision-making module are all connected to the production whole-domain efficiency assessment module, which comprehensively evaluates the production performance of the entire manufacturing production line.

9. The multimodal intelligent decision-making and collaborative system for digital transformation of manufacturing industry according to claim 8, characterized in that, The specific evaluation and analysis process of the production-wide efficiency assessment module for comprehensively evaluating the production performance of the entire manufacturing production line is as follows: A comprehensive evaluation formula for overall production efficiency is constructed. After calculating the overall production efficiency score Pe, the overall production efficiency level is determined based on the overall production efficiency score Pe.

10. The multimodal intelligent decision-making and collaborative system for digital transformation of manufacturing industry according to claim 9, characterized in that, The specific strategy for determining the overall production efficiency level based on the comprehensive production efficiency score Pe is as follows: Pe≥90: excellent overall production efficiency; 80≤Pe<90: good overall production efficiency; 70≤Pe<80: average overall production efficiency; Pe<70: poor overall production efficiency.

Images