Manufacturing process optimization decision-making method, device and equipment based on big data analysis

Abstract

The invention relates to the technical field of manufacturing optimization, and discloses a manufacturing process optimization decision-making method, device and equipment based on big data analysis, and the method comprises the steps: collecting data and quality evaluation information of each link of a target product manufacturing process, forming reference data, continuously recording the reference data through a big data platform, and aligning the reference data with key characteristics; according to the method, a plurality of manufacturing reference models are obtained, key factor analysis is performed on the models, optimization factors are extracted, an objective function containing quality, efficiency and cost constraints is constructed, iterative optimization solution is performed, and an optimization result is verified through a big data platform until an optimal decision scheme is obtained. The production efficiency is improved, the cost is reduced, the stability of the product quality is ensured, and the problem that in the prior art, multi-target optimization is difficult to carry out on the multi-link manufacturing process is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of manufacturing optimization, and in particular to a manufacturing process optimization decision-making method, device and equipment based on big data analysis. Background Art

[0002] In modern manufacturing, the production process of products is becoming increasingly complex, requiring effective coordination of factors in different links to optimize product quality, production efficiency and production costs. With the rapid development of information technology, the application of big data technology in the manufacturing field has become an important force in promoting intelligent manufacturing. Traditional manufacturing process optimization often relies on manual experience and simple mathematical models, which makes optimization decisions lack data support and difficult to cope with rapidly changing market demands and production conditions. The manufacturing process optimization decision-making method based on big data analysis can provide more accurate and dynamic optimization decision-making solutions by collecting and analyzing large amounts of production data, combined with machine learning and optimization algorithms, and has significant advantages.

[0003] With the widespread adoption of the Industrial Internet of Things (IIoT), sensor technology, and smart devices, manufacturing companies can now obtain detailed data on every aspect of the production process in real time, including equipment operating status, product quality monitoring, resource consumption, etc. These massive amounts of data are stored, processed, and analyzed through big data platforms, providing companies with a foundation for making accurate decisions and optimizing production processes.

[0004] In modern manufacturing, with the increasing variety of products and complexity of production processes, traditional production optimization methods often cannot meet the requirements of high efficiency, low cost and high quality. In order to improve production efficiency, reduce costs and ensure product quality, manufacturing process optimization has become an important goal pursued by enterprises. However, manufacturing process optimization usually involves many factors, such as production efficiency, product quality, resource consumption, etc. These factors are interrelated and influence each other, and traditional technical means are difficult to dynamically adjust and optimize. Summary of the Invention

[0005] The purpose of the present invention is to provide a manufacturing process optimization decision-making method, device and equipment based on big data analysis, aiming to solve the problem in the prior art that it is difficult to perform multi-objective optimization on multi-link manufacturing processes.

[0006] The present invention is implemented as follows: In a first aspect, the present invention provides a manufacturing process optimization decision-making method based on big data analysis, comprising: Collecting link element data of each link in the manufacturing process of the target product and quality assessment information of the target product, and combining the link element data of each link and the quality assessment information as reference data; Continuously record reference data of target products through a big data platform, and perform several forms of key feature alignment processing on each reference data to obtain several manufacturing reference models; Performing key factor analysis on the manufacturing reference model to obtain manufacturing element optimization factors fed back by the manufacturing reference model; Based on the manufacturing element optimization factors of each manufacturing reference model, constructing an objective function including quality constraints, efficiency constraints and cost constraints for the target product; The objective function is iteratively optimized and solved, and the solution is verified using the big data platform as a supervision condition until an optimized decision-making solution is obtained.

[0007] In a second aspect, the present invention provides a manufacturing process optimization decision-making method device based on big data analysis, which is used to implement the manufacturing process optimization decision-making method based on big data analysis described in any one of the first aspects, including: A data acquisition module is used to collect link element data of each link in the manufacturing process of the target product and quality assessment information of the target product, and combine the link element data of each link and the quality assessment information into reference data; A data alignment module is used to continuously record reference data of target products through a big data platform and perform several forms of key characteristic alignment processing on each reference data to obtain several manufacturing reference models; A factor analysis module, configured to perform key factor analysis on the manufacturing reference model to obtain manufacturing element optimization factors fed back by the manufacturing reference model; a function construction module, configured to construct an objective function including quality constraints, efficiency constraints, and cost constraints for the target product based on the manufacturing element optimization factors of each manufacturing reference model; The optimization decision module is used to iteratively optimize and solve the objective function, and at the same time use the big data platform as a supervision condition to verify the solution results until an optimized decision solution is obtained.

[0008] In a third aspect, the present invention provides a manufacturing process optimization decision-making device based on big data analysis, comprising a memory and a processor, wherein the memory stores a manufacturing process optimization decision-making program based on big data analysis that can be run on the processor, and when the processor executes the manufacturing process optimization decision-making program based on big data analysis, it implements a manufacturing process optimization decision-making method based on big data analysis as described in any one of the first aspects.

[0009] The present invention provides a manufacturing process optimization decision-making method based on big data analysis, which has the following beneficial effects: The present invention collects data and quality assessment information from each link in the manufacturing process of the target product to form reference data, and continuously records and aligns them with key characteristics through a big data platform to obtain multiple manufacturing reference models. The key factors of the models are analyzed, the optimization factors are extracted, and an objective function containing quality, efficiency and cost constraints is constructed. The optimization function is solved through iterative optimization and the optimization results are verified through a big data platform until the best decision-making solution is obtained. This method realizes the intelligent optimization of the manufacturing process, improves production efficiency, reduces costs, and ensures the stability of product quality, solving the problem in the existing technology that it is difficult to perform multi-objective optimization on multi-link manufacturing processes. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 1 is a schematic diagram of the steps of a manufacturing process optimization decision-making method based on big data analysis provided by an embodiment of the present invention; Figure 2 It is a structural schematic diagram of a manufacturing process optimization decision-making method device based on big data analysis provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0011] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0012] The implementation of the present invention is described in detail below with reference to specific embodiments.

[0013] Reference Figure 1 、 Figure 2 As shown, a preferred embodiment of the present invention is provided.

[0014] In a first aspect, the present invention provides a manufacturing process optimization decision-making method based on big data analysis, comprising: S1: collecting link element data of each link in the manufacturing process of the target product and quality assessment information of the target product, and combining the link element data of each link and the quality assessment information as reference data; S2: Continuously record reference data of the target product through the big data platform, and perform several forms of key feature alignment processing on each reference data to obtain several manufacturing reference models; S3: performing key factor analysis on the manufacturing reference model to obtain manufacturing element optimization factors fed back by the manufacturing reference model; S4: constructing an objective function including quality constraints, efficiency constraints, and cost constraints for the target product based on the manufacturing element optimization factors of each manufacturing reference model; S5: Iteratively optimize and solve the objective function, and use the big data platform as a supervision condition to verify the solution results until an optimized decision-making solution is obtained.

[0015] Specifically, in step S1 of the embodiment provided by the present invention, each link in the manufacturing process of the target product is first clarified, including all links from raw material preparation, production process, equipment operation, quality inspection to product packaging. Different links determine the quality, efficiency and cost of the target product. Therefore, accurately identifying and dividing each link is a prerequisite for data collection and subsequent optimization analysis. Clear link division helps to accurately collect relevant data of each link and ensure the comprehensiveness and integrity of the data. This step provides a structured framework for subsequent data association and analysis.

[0016] More specifically, at each stage, data on factors influencing the manufacturing process is collected. These factors include, but are not limited to: material type, quantity, state, and composition; production equipment configuration, operating status, and maintenance records; and process parameter settings (such as temperature, pressure, and speed). This data is a core variable in the manufacturing process, directly impacting product quality and production efficiency. This data is collected to provide the necessary input for subsequent analysis and optimization. By accurately collecting this data, we ensure that the control parameters of each stage can be quantified, enabling effective optimization of the entire manufacturing process. This data provides the foundation for subsequent quality assessment and cost-benefit analysis.

[0017] More specifically, quality assessment data for target products is collected, primarily including: finished product quality inspection data (such as dimensional accuracy, appearance inspection, and functional testing); quality indicators such as defect rate and pass rate; and customer feedback and return data. Quality assessment information directly reflects the final quality of the target product. This data is potentially correlated with each operating parameter (link element) in the manufacturing process. Therefore, accurately recording product quality assessment data is crucial. Quality assessment information provides a quantitative standard for key decisions in subsequent optimization processes. By combining it with link element data, it is possible to identify which links have a significant impact on quality, thus providing direction for optimization decisions.

[0018] More specifically, the element data of each link is matched and combined with the quality assessment information corresponding to the link. The specific methods include: data mapping the element data of each link and the corresponding quality assessment information to generate a multi-dimensional reference data set, standardizing data from different sources and formats to ensure that they can be effectively compared and analyzed, and converting unstructured data (such as image data, text data, etc.) into structured data through coding methods. Finally, the link element data and quality assessment information are matched by link to form a complete reference data set; the link element data and quality assessment information are combined to explore the potential correlation between the two and reveal the impact of link elements on quality. In manufacturing optimization, quality is not just the terminal result, it is closely related to every link in the production process; by combining the link element data and quality assessment information, a comprehensive and quantifiable reference data set is obtained. These data provide direct input for subsequent model construction and optimization analysis, which can help identify key links and factors affecting quality and provide support for optimization decisions.

[0019] More specifically, the collected reference data is verified and cleaned to ensure the accuracy and consistency of the data, including: identifying and correcting outliers or erroneous values ​​in the data, ensuring consistency between different data sources, preventing data conflicts, interpolating or removing missing data, and ensuring data integrity. High-quality reference data is the basis for optimization decisions, so the accuracy and integrity of the data must be guaranteed. Cleaned data helps reduce noise and improve the accuracy of subsequent analysis and optimization models. Data cleaning and verification help ensure that subsequent data analysis results will not be interfered with by inaccurate data, thereby improving the reliability and effectiveness of the entire optimization process.

[0020] More specifically, the collected reference data is stored in a big data platform or data warehouse, and it is ensured that the data can be accessed and transmitted to the optimization model in real time. The big data platform or data warehouse can provide powerful data processing capabilities to ensure that the collected data can be used at any time for subsequent analysis, model training, optimization calculations, etc. At the same time, the high efficiency of data transmission can also ensure real-time optimization decisions. Data storage and transmission ensure the long-term availability and real-time nature of the data, providing necessary support for continuous optimization decisions. The timeliness and accuracy of the data facilitate real-time optimization adjustments.

[0021] Specifically, in step S2 of the embodiment provided by the present invention, various reference data (such as process element data and quality assessment data) of the target product during the manufacturing process are recorded in real time or periodically through a big data platform. This process includes: real-time collection of sensor data, operation records, environmental monitoring data, etc. from each link in the production process; automatic synchronization and storage of quality inspection data, production efficiency data, raw material usage data, etc.; use of the big data platform for data backup and version control to ensure data integrity; and continuous recording of reference data for the target product is the basis for optimizing decision-making. The big data platform enables efficient data storage and management, ensuring that various types of data can be extracted and analyzed at any time; by continuously recording data, ensuring the real-time and integrity of the data, the data can be accumulated over time, providing more information for subsequent analysis and optimization, which also ensures that the data used in decision-making is the most up-to-date and comprehensive.

[0022] More specifically, key characteristic alignment is performed on the collected reference data. Various data sources (such as sensor data, quality data, and process data) are cleaned, redundant data is removed, and the data is standardized into a unified format. This helps ensure data consistency. Key characteristics that most influence the manufacturing process (such as temperature, pressure, and speed) are extracted from the data at each stage and calibrated and standardized. Key data from different stages are aligned chronologically to ensure that data at the same time point corresponds to the same manufacturing state. Data from different sources (such as quality inspection data and production process data) are fused, and correlations between different data sources are determined through algorithms to create a unified characteristic representation. Reference data for the manufacturing process may come from multiple stages or systems, and different data may have different time scales, measurement units, and data formats. Therefore, key characteristic alignment is a prerequisite for ensuring effective data comparison, analysis, and modeling. Aligned data is consistent and comparable, enabling the extraction of valuable information from data from different sources. Alignment ensures the accuracy and usability of data in subsequent steps and reduces the potential for misleading conclusions due to data inconsistencies.

[0023] More specifically, based on the aligned reference data, multiple manufacturing reference models are generated through data analysis and modeling techniques. Statistical analysis, machine learning, or other data modeling techniques are then used to construct multiple predictive models based on the aligned reference data. For example, regression analysis, decision trees, neural networks, and other methods are used to predict indicators such as product quality and production efficiency based on process factor data. After constructing the initial model, the model is validated using actual production data and optimized based on the validation results. The reference models are categorized or layered according to different manufacturing scenarios or product categories, allowing for the use of different models for different needs. Multiple manufacturing reference models help the system select the most appropriate optimization solution based on different production environments, product requirements, and process requirements. Each model focuses on different characteristics (such as quality, efficiency, and cost). Therefore, generating multiple models facilitates a more comprehensive solution to the optimization problem. By constructing multiple manufacturing reference models, the most appropriate optimization strategy can be selected based on different production situations, improving flexibility and adaptability within the production process. Different models can support different optimization objectives (such as reducing costs, improving quality, and increasing production efficiency).

[0024] More specifically, the real-time monitoring capabilities of the big data platform continuously monitor various indicators in the production process (such as production line status, equipment efficiency, and quality control), and dynamically update the manufacturing reference model based on real-time data. Real-time data is collected through sensors and IoT devices and compared and analyzed with historical data to monitor for possible anomalies in the production process. Based on the comparison results of real-time and historical data, the reference model is adjusted regularly or on demand. For example, if the production efficiency of a particular link decreases, the model is automatically updated to facilitate corresponding optimization decisions. The actual situation of the manufacturing process is constantly changing (such as equipment aging, raw material changes, etc.), so the manufacturing reference model needs to be adjusted based on real-time monitoring data to ensure that it always reflects the current production status. Real-time monitoring and model updates ensure the continued validity and accuracy of the reference model. Adjusting the model based on the latest data helps to respond quickly in a changing environment and improve the flexibility and adaptability of the production process.

[0025] More specifically, based on the constructed manufacturing reference model, verification and application in actual production are carried out. The constructed reference model is applied to the production process, and simulation tests are carried out to verify the deviation between the model's predicted results and the actual production results. According to the feedback from actual production, the parameters in the reference model are adjusted to improve the accuracy and practicality of the model. The theoretical model may not be able to fully adapt to the complex actual production environment, so verification and application are needed to ensure the reliability and effectiveness of the model. Through verification and application, the practicality and reliability of the manufacturing reference model can be ensured, which can help make manufacturing process optimization decisions more accurate and enable rapid response to changes in production.

[0026] Specifically, in step S3 of the embodiment provided by the present invention, based on the constructed manufacturing reference model, the key factors in the model (such as production process parameters, equipment performance, personnel operation, material quality, etc.) are clarified. These factors are usually variables that directly affect product quality, production efficiency and cost: determine the factors in the model that affect the most important output indicators (such as product qualification rate, production speed, cost, etc.), analyze the relationship between each factor, and identify which factors are key to the optimization target; in a manufacturing process, different parameters and factors may affect the final result, so it is necessary to clarify the key factors in order to focus on the variables that have the greatest impact on the optimization. This step is to ensure that subsequent analysis can concentrate on processing the most critical factors. By determining the key factors in the manufacturing reference model, it can help focus on the variables that most significantly affect the results, avoid wasting resources on unimportant factors, and thus improve optimization efficiency.

[0027] More specifically, sensitivity analysis is performed on the identified key factors to analyze the degree of impact of each factor change on the model output. Common methods include: single-factor analysis, changing the value of a single factor, observing the changes in the model output, and determining which factors have a significant impact on the production results; multi-factor analysis, considering the impact of multiple factors changing together on the model output, and exploring the interaction effects between factors; variance analysis, statistically analyzing the contribution of the variance of different factors to the final result; sensitivity analysis helps to identify the factors that have the greatest impact on the manufacturing process. By understanding the sensitivity of each factor, optimization decisions can be guided more accurately; sensitivity analysis helps to determine which factors should be given more attention during the optimization process, thereby optimizing the model and improving production efficiency and quality. It provides a scientific basis for subsequent decision-making and avoids excessive attention to insignificant factors.

[0028] More specifically, through the results of sensitivity analysis, we extract optimization factors that have a significant impact on the results of the manufacturing process, and quantify their impact on the results. Through regression analysis, weighted average and other methods, we quantify the contribution of each factor to the model output. For the interaction effects of multiple factors, we use machine learning algorithms (such as decision trees, random forests, etc.) to model the relationship between factors and quantify the interaction effects of each factor. Quantifying the impact of each factor into a numerical value can provide a clear basis for subsequent optimization decisions. Through quantification, we can know which factors contribute the most to the optimization goals (such as cost, quality, efficiency, etc.) and make targeted adjustments to them. The quantified optimization factors provide clearer guidance for decision-making. In the actual production process, we can more efficiently adjust and optimize the most influential factors, thereby achieving the expected optimization goals (for example, improving production efficiency, reducing scrap rates, etc.).

[0029] More specifically, after quantifying the impact of the optimization factors, a mathematical model is established to describe the relationship between the optimization factors and the manufacturing process results. Common modeling methods include: regression analysis: through linear regression or nonlinear regression, a mathematical expression between the factors and the target variables is established; optimization algorithm: applying optimization methods such as linear programming, nonlinear programming or genetic algorithms to optimize the objective function (such as cost, quality, etc.); establishing a mathematical model helps to clearly describe the relationship between factors and results, and provides a clear path for subsequent optimization decisions. Through the model, the results under different factor values ​​can be predicted and effective optimization can be carried out; through the establishment of a mathematical model, the impact of factor adjustments on the final results can be accurately predicted, thereby providing guidance for optimization. The application of the model can help make the manufacturing process more automated and intelligent, and reduce errors and deviations in human decision-making.

[0030] More specifically, based on the optimization factors and their impacts as fed back by the model, adjustments are implemented and their effects verified. Based on the recommendations of the model and optimization factors, adjustments are made to the production process, equipment configuration, and raw material selection. After optimization is implemented, key indicators in the production process are monitored in real time to examine the effectiveness of the optimization measures. The model is further adjusted based on actual data. By comparing production data before and after optimization, it is verified whether the optimization adjustments have achieved the desired results (such as increased production efficiency, reduced costs, and improved product quality). The implementation of optimization adjustments requires actual production verification to ensure the adaptability and accuracy of the theoretical model in the real world. Real-time monitoring and feedback ensure that optimization measures are adjusted in a timely manner to avoid adverse consequences. By implementing optimization adjustments and verifying their effects, the match between the theoretical optimization model and the actual production environment can be ensured, improving the effectiveness of the optimization. At the same time, feedback adjustments can ensure continuous improvement and optimization, further enhancing the stability and sustainability of production.

[0031] More specifically, optimization is an ongoing process, requiring regular evaluation and re-optimization of optimization factors and models. Production process data must be regularly collected, and the relationships between factors and influences in the model must be updated. Based on new production conditions, optimization factors or model structures must be adjusted and re-optimized. Optimization results must be continuously tracked to ensure continued improvement in production efficiency. Manufacturing process conditions and technologies may change over time, necessitating continuous monitoring and optimization based on new circumstances. This helps maintain long-term competitiveness and flexibility in production. Continuous improvement and re-optimization ensure that the production process remains optimal, avoiding issues caused by environmental changes or technological advancements. Through continuous optimization, long-term improvements in production efficiency can be achieved.

[0032] Specifically, in step S4 of the embodiment provided by the present invention, the core optimization objectives of the objective function are clarified. These objectives are usually: quality: ensuring that the final product meets the quality standards, including dimensional accuracy, surface quality, strength, etc.; efficiency: improving the overall efficiency of the production process, reducing production cycles, equipment idle time, etc.; cost: reducing the inputs required in the production process, and controlling production costs (such as raw materials, labor, energy, etc.); to construct a reasonable objective function, it is necessary to clarify the direction of optimization, that is, the balance between product quality, production efficiency and cost control. Different optimization objectives may contradict each other (for example, improving quality may lead to increased costs). Therefore, the priority of each objective must be reasonably defined. By clarifying the optimization objectives, a clear direction can be provided for subsequent modeling and constraint setting to ensure that the final model meets actual production needs and strategic goals.

[0033] More specifically, based on the identified optimization factors for manufacturing elements, mathematical models or formulas are established for quality, efficiency, and cost. These formulas are typically based on production data, historical experience, or derived using optimization methods. Quality functions can be based on indicators such as product quality standards, process capability indices, and yield rates. For example, quality can be mathematically expressed by how raw material selection and process control affect the yield rate of the final product. Efficiency can be measured through factors such as production cycle time, machine speed, and product output. Cost functions typically consider factors such as raw materials, labor, and energy. Cost formulas can be used to describe the input-output relationship of each process, or the unit production cost can be directly used. The mathematical expression of each constraint forms the basis for constructing the objective function, ensuring that it accurately reflects the key factors in the production process. Quantifying these factors facilitates precise control over the impact of different factors during actual optimization. Mathematical expression provides a clear numerical basis for the objective function, facilitating subsequent adjustment and solution using optimization algorithms. This allows for comprehensive optimization of quality, efficiency, and cost.

[0034] More specifically, quality constraints are set based on the product's quality standards. Quality constraints generally include: setting a minimum pass rate standard, such as requiring a pass rate of no less than 95%, setting quality control standards for each production process, such as each dimensional error not exceeding 0.5mm, setting the maximum allowable range of surface defects (such as scratches, bubbles, etc.), and setting reliability requirements based on the product's usage environment or industry standards. Quality constraints are the core conditions for ensuring that products meet user needs and industry specifications. Substandard quality will directly affect the product's market competitiveness and brand reputation, so it must be strictly controlled. Quality constraints ensure that during the optimization process, product quality will not be sacrificed due to the pursuit of other goals (such as reducing costs and improving efficiency), avoiding the production of unqualified products, thereby reducing rework and scrap rates.

[0035] More specifically, set the shortest production cycle or production time. For example, set the production time from raw materials to final finished products to no more than 48 hours, require the utilization rate of production equipment to reach a certain standard (such as above 80%) to ensure the efficient operation of the production line, set the minimum production volume per hour or day, and ensure that the production line operates at high efficiency. Efficiency constraints can ensure that the utilization of equipment and personnel is maximized during the production process, thereby improving production capacity and shortening delivery cycles, and enhancing the company's market responsiveness. By setting production efficiency constraints, it can ensure that there will be no excessive production bottlenecks and resource waste in the production process, improve overall production efficiency, and shorten product delivery time.

[0036] More specifically, cost constraints are set based on cost analysis during the production process. Cost constraints may include: setting the production cost per unit of product to no more than a certain limit (e.g., the production cost per product does not exceed RMB 10), setting reasonable raw material procurement costs based on market price fluctuations, setting labor cost limits, such as the average wage cost of production line workers not exceeding 30% of the total cost, setting energy consumption standards to avoid cost overruns caused by excessive energy consumption. Cost constraints are the core of a manufacturing company's competitiveness in the market. Cost control can not only increase profit margins, but also enhance the price competitiveness of products in the market. By setting cost constraints, we can ensure that while pursuing quality and efficiency, we will not exceed the budget, prevent cost waste due to excessive investment of resources, and ensure the economy of production activities.

[0037] More specifically, after establishing the mathematical expressions of quality, efficiency and cost, they are integrated into an objective function and corresponding constraints are added. In the objective function, quality constraints, efficiency constraints and cost constraints are added. The construction of the objective function integrates the most important indicators in the production process (quality, efficiency, cost) and ensures that they are balanced in actual optimization. The constraints ensure that the optimization plan does not violate the basic requirements of production. The combination of objective function and constraints makes the optimization problem practical and avoids deviation from actual production requirements during the optimization process. Through mathematical models, the weights and constraints between objectives can be accurately adjusted to achieve the best overall effect.

[0038] More specifically, the constructed objective function is solved through mathematical optimization methods (such as linear programming, nonlinear programming, genetic algorithms, etc.) to obtain the optimal solution. Through the solution process, the optimal product production plan that meets all constraints is found, including the optimal production process, equipment configuration, personnel arrangement, etc. Through mathematical optimization methods, the optimal production plan can be found under the premise of meeting quality, efficiency and cost constraints, maximizing the company's production efficiency. The optimization results provide the company with the most cost-effective production plan, which not only improves production efficiency, but also ensures product quality and cost control, and enhances market competitiveness.

[0039] Specifically, in step S5 of the embodiment provided by the present invention, the established objective function and the associated quality, efficiency and cost constraints are confirmed, and the parameters of the optimization problem are initialized, including the weight coefficients in the objective function, the specific values ​​of each constraint, and related production parameters (such as machine production capacity, worker working hours, raw material prices, etc.). This is the starting point of the optimization problem, ensuring that the defined objective function and constraints can reflect the actual production scenario. Accurate initialization helps to ensure the rationality of the optimization solution, avoid the invalidation of the solution result due to incorrect initialization, and ensure that the optimization process starts from a reasonable starting point, laying the foundation for subsequent optimization work.

[0040] More specifically, according to the nature of the objective function (linear or nonlinear), the form of the constraints and the scale of the problem, an appropriate optimization algorithm should be selected. Commonly used optimization algorithms include: Linear Programming (LP) / Nonlinear Programming (NLP): used to solve linear or nonlinear objective functions, Genetic Algorithm (GA): suitable for complex nonlinear optimization problems and can handle multi-constraint and multi-objective situations, Particle Swarm Optimization (PSO): suitable for problems with large-scale search spaces, Simulated Annealing (SA): suitable for optimization problems that avoid local optimal solutions; different optimization algorithms are suitable for different types of problems. Reasonable selection of optimization algorithms can speed up the solution process and avoid falling into local optimal solutions, thereby improving computational efficiency and optimization effects. The choice of optimization algorithm directly affects the speed of solution and the quality of the results, ensuring that efficient and accurate optimization solutions can be obtained.

[0041] More specifically, the selected optimization algorithm is used to preliminarily solve the objective function and obtain a preliminary optimization solution. This stage mainly uses the algorithm to optimize the objective function and obtain a solution that meets the constraints. The preliminary solution is a quick test of the objective function and constraints to help check whether there are obvious errors or unreasonable solutions, ensure that the optimization path is moving in the right direction, and obtain a preliminary optimization result, which can provide a basis for subsequent data verification and iterative optimization.

[0042] More specifically, the preliminary optimization results are submitted to the big data platform for data verification, and the real-time data analysis capabilities of the data platform are used to verify the rationality and practicality of the current optimization plan. The specific verification content includes: comparing with historical production data to verify whether the current plan is consistent with past production conditions, using real-time data to monitor the production process, checking whether the objective function can be adjusted in real time to adapt to possible changes in the production process (such as fluctuations in raw material costs, equipment failures, etc.), analyzing hidden patterns in historical data, predicting the performance of the optimization plan in actual production, and checking whether it meets the expected goals. Through verification on the big data platform, historical data and real-time data can be fully utilized to ensure the feasibility and stability of the optimization plan, and avoid the deviation of results based on model calculations alone from actual production needs. Data verification provides data support for the optimization plan, making optimization decisions closer to the actual production environment and reducing decision-making errors due to inaccurate assumptions or incomplete data.

[0043] More specifically, based on the verification results of the big data platform, the optimization plan is fed back and corrected, and the objective function, constraints or parameters of the optimization algorithm are readjusted. This process usually includes the following operations: According to the feedback from the big data platform, the weight coefficient in the objective function is adjusted to better balance the relationship between quality, efficiency and cost, and the constraints are adjusted to ensure that the optimization plan has high adaptability and stability in the actual production process. If necessary, the parameters of the optimization algorithm are changed, and the solution process is optimized to converge to the optimal solution faster. Iterative optimization is a crucial step in optimization solution, because there may be gaps between the actual production environment and the theoretical model. The iterative process can gradually eliminate errors, refine the optimization plan, and make it more practical. After multiple iterations and adjustments, the accuracy and practicality of the optimization plan are gradually improved, and ultimately an optimization decision-making plan that better meets actual production needs can be provided.

[0044] More specifically, after the optimization plan is implemented, key parameters in the production process (such as production cycle, cost and quality, etc.) are monitored in real time and synchronized with the big data platform. This step includes: monitoring every link involved in the production process, collecting real-time data such as production, quality, and cost, and making rapid adjustments and optimizations based on real-time data to ensure that the production process always remains in the best state. The production process is often affected by various unexpected factors, such as equipment failures, supply chain fluctuations, etc., so continuous monitoring and dynamic adjustments are required to ensure that production goals are always met. Through real-time monitoring and dynamic adjustments, the optimization plan can be guaranteed to maintain efficient operation in a changing production environment, further improving the company's production flexibility and response speed.

[0045] More specifically, after multiple iterations of optimization and data verification, the optimal production plan is finally confirmed and transformed into an actual operational process. When implementing the plan, detailed implementation rules also need to be formulated, including: resource allocation: ensuring the rational allocation of human, material, and financial resources; production scheduling: adjusting production plans and scheduling according to the optimization plan; quality control: formulating corresponding quality control standards and processes to ensure that the production process meets the optimization goals; the implementation of the final decision is a key step to ensure that the optimization plan can be applied in practice. A successful optimization plan does not just remain at the theoretical level, but also requires effective implementation to ensure its success. Through effective implementation, the final decision plan will improve production efficiency, reduce costs, and improve product quality, creating higher economic benefits for the company.

[0046] More specifically, the effects after implementation need to be continuously monitored and evaluated, which includes regular checks on production data, analysis of production efficiency, quality pass rate and cost control effects to ensure the long-term effectiveness of the optimization plan. The production environment and market demand are constantly changing, so it is necessary to track the results after implementation over the long term to ensure that the optimization plan can continue to be effective under different circumstances. Through continuous monitoring and evaluation, the shortcomings of the optimization plan can be discovered in time, and adjustments can be made quickly to ensure that the production process is always in the best state.

[0047] The present invention provides a manufacturing process optimization decision-making method based on big data analysis, which has the following beneficial effects: The present invention collects data and quality assessment information from each link in the manufacturing process of the target product to form reference data, and continuously records and aligns them with key characteristics through a big data platform to obtain multiple manufacturing reference models. The key factors of the models are analyzed, the optimization factors are extracted, and an objective function containing quality, efficiency and cost constraints is constructed. The optimization function is solved through iterative optimization and the optimization results are verified through a big data platform until the best decision-making solution is obtained. This method realizes the intelligent optimization of the manufacturing process, improves production efficiency, reduces costs, and ensures the stability of product quality, solving the problem in the existing technology that it is difficult to perform multi-objective optimization on multi-link manufacturing processes.

[0048] Preferably, the step of collecting link element data of each link in the manufacturing process of the target product and quality assessment information of the target product, and combining the link element data of each link and the quality assessment information as reference data includes: S11: Obtain manufacturing process information of the target product, and construct a product manufacturing information chain consisting of several process links based on the manufacturing process information; wherein the process links include a material preparation link, an equipment configuration link, and a process parameter link, and each process link includes several sub-links; S12: collecting original record data of each process link of the manufacturing process corresponding to the target product based on the product manufacturing information chain, and performing quality assessment on the target product to obtain quality assessment information; S13: performing feature encoding of a specified format on the original recorded data to convert the original recorded data into a corresponding feature vector matrix; S14: Using the tensor decomposition method, the eigenvector matrix corresponding to each sub-link in each process link is used to capture the potential correlation features across sub-links to generate potential correlation factors between the sub-links; S15: performing vector expansion processing on the eigenvector matrix corresponding to each sub-link according to the position relationship of each sub-link in the product manufacturing information chain and the potential correlation factor, so as to obtain an enhanced vector matrix corresponding to each sub-link; S16: Aggregate and splice the enhanced vector matrices corresponding to each sub-step according to the time sequence and logical order, and extract key semantics from each enhanced vector matrix; S17: taking each enhanced vector matrix as deep information, taking the extracted key semantics as surface information corresponding to the deep information, and combining the deep information, the surface information and the quality assessment information to obtain reference data.

[0049] Specifically, information on the target product during the manufacturing process is collected, including raw materials, production equipment, process parameters, etc., and a product manufacturing information chain consisting of several process links is constructed based on this information; material preparation link: including the procurement, inspection, and preparation of raw materials; equipment configuration link: including equipment installation, debugging, status detection, etc.; process parameter link: including the setting of production processes, such as temperature, pressure, time, etc.; different links in the manufacturing process (materials, equipment, processes) are key factors affecting product quality. By establishing an information chain, comprehensive data collection for each link can be ensured, ensuring the acquisition of comprehensive manufacturing process data, providing a basis for subsequent data analysis and optimization decisions.

[0050] More specifically, based on the collected manufacturing process information, it is decomposed into several process links, and each process link can be further divided into multiple sub-links. The construction of the information chain should take into account the time, sequence and correlation between each link; the manufacturing process of the product is complex and multi-link. By building an information chain, the role and influence of each link and sub-link in the entire process can be clearly described, which is convenient for subsequent data analysis and optimization; and provides a clear and structured framework for further analysis, data collection and feature extraction.

[0051] More specifically, based on the established manufacturing information chain, raw data records are collected for each process link and its sub-links, including production data (such as equipment operating status, raw material consumption, and production speed) and quality data (such as defect rate, pass rate, and product quality inspection results). This raw data provides the necessary information source for subsequent analysis. Collecting this data helps to fully understand the production status and quality performance of each link, ensuring its comprehensiveness and accuracy, laying the foundation for subsequent data processing and feature coding.

[0052] More specifically, quality assessment is conducted on the target products, and data related to product quality is collected, such as the pass rate, the proportion of unqualified products, the type of defects and their distribution, and test results. Quality assessment information reflects the impact of each link in the manufacturing process on the quality of the final product. Combined with quality data, it can help to deeply analyze which links affect quality. It provides data support for subsequent optimization, so that the optimization of the production process can clearly target the links with quality problems.

[0053] More specifically, the collected raw record data is feature encoded and converted into the corresponding feature vector matrix. Feature encoding is the process of converting numerical values, text and other information into vector representations, such as using one-hot encoding, standardization and other methods. Feature encoding is the core step of data preprocessing. It can convert raw data into a numerical form that can be understood by the machine learning model, facilitating subsequent analysis and modeling, and converting complex raw data into structured feature vectors to facilitate subsequent modeling and analysis.

[0054] More specifically, tensor decomposition technology (such as CP decomposition or Tucker decomposition) is used to capture the potential correlation features across sub-links of the eigenvector matrix corresponding to the sub-links in each process link. This can help identify potential correlation factors between different links and sub-links, such as process parameters and equipment configuration, raw materials and process parameters, etc. The tensor decomposition method can capture the potential correlation between different links, which is crucial for discovering the key factors affecting product quality. It can reveal the patterns hidden in the data. By capturing potential correlations, it is possible to find deep connections between different links and provide guidance for optimizing the production process.

[0055] More specifically, based on the positional relationship and potential correlation factors of the sub-links in the manufacturing information chain, the vector expansion method is used to expand the eigenvector matrix corresponding to each sub-link. Through expansion, the potential connection between each sub-link can be strengthened, and richer information can be provided for subsequent analysis. Vector expansion processing can enhance the correlation characteristics between different links, making the subsequent analysis more comprehensive and accurate, ensuring that in the subsequent aggregation and analysis process, the potential connection between each link can be fully reflected, thereby improving the accuracy of the analysis.

[0056] More specifically, the enhanced vector matrices corresponding to each sub-link are aggregated and spliced ​​according to the time sequence and logical order. The purpose of aggregation and splicing is to integrate the characteristic information of different links to form complete product manufacturing process data. The manufacturing process is continuous and has a time and logical order. The aggregation and splicing steps can ensure that the data is merged in sequence, reflecting the full process information from raw material preparation to the final product. Through aggregation and splicing, a complete and time-series enhanced vector matrix is ​​obtained, which is convenient for subsequent analysis and modeling.

[0057] More specifically, key semantics are extracted from the merged enhanced vector matrix, and the most important features or information for the manufacturing process and quality are extracted through clustering, dimensionality reduction or feature selection methods. Key semantic extraction can identify the most representative and influential factors, thereby simplifying the complexity of data analysis, improving the interpretability and accuracy of the model, extracting the most critical features for production quality, reducing data redundancy, and improving the effectiveness of subsequent optimization and decision-making.

[0058] More specifically, the extracted key semantics are combined with deep information (enhanced vector matrices) and quality assessment information to form a complete reference data set. This dataset can be used for subsequent analysis and decision support. By combining deep information, surface information, and quality assessment information, the reference data can be ensured to contain comprehensive information about the manufacturing process and provide feedback on actual production quality, making it suitable for decision support systems. The resulting reference data not only encompasses every detail of the manufacturing process but is also closely linked to product quality, providing a reliable basis for subsequent optimization decisions.

[0059] Preferably, the steps of continuously recording reference data of the target product through a big data platform and performing several forms of key characteristic alignment processing on each of the reference data to obtain several manufacturing reference models include: S21: continuously recording reference data of the target product through the big data platform, and simultaneously obtaining production environment information of the target product, and combining the production environment information with the reference data; S22: Setting key characteristics to be aligned on the big data platform, and performing data matching on the big data platform according to the key characteristics to obtain a reference data set that meets the key characteristics; wherein the key characteristics include excellent quality characteristics, various types of defect quality characteristics, various types of material condition characteristics, various types of equipment condition characteristics, and various types of process parameter characteristics; S23: using the key characteristic as an alignment target, performing data separation and analysis on the deep information of each reference data in the reference data set according to the surface information of each reference data, so as to separate directly related data and indirectly related data corresponding to the alignment target from the reference data; S24: performing cluster analysis on the directly associated data and the indirectly associated data separated from the reference data, respectively, to extract cluster features and discrete point features of each directly associated data and each indirectly associated data; S25: Construct a direct association reference model based on the cluster characteristics and discrete point characteristics of each directly associated data item, construct an indirect association reference model based on the cluster characteristics and discrete point characteristics of each indirectly associated data item, and perform information mapping processing on the direct association reference model and the indirect association reference model with the alignment target to obtain a manufacturing reference model of the alignment target.

[0060] Specifically, the reference data of the target product is continuously recorded through the big data platform, and the production environment information of the target product is obtained at the same time, which includes data in the manufacturing process (such as equipment status, raw material status, process parameters, etc.) and production environment information (such as temperature, humidity, etc.). The reference data is closely related to the production environment information. Changes in the production environment will directly affect the manufacturing process and the quality of the final product. Combining this information can form more comprehensive reference data, provide more valuable input for subsequent analysis, ensure the collection of comprehensive and real-time manufacturing process data and production environment information, and provide sufficient data support for subsequent data alignment and analysis.

[0061] More specifically, key characteristics to be aligned are set on the big data platform, including: excellent quality characteristics: such as product durability, appearance, precision, etc.; various types of defect quality characteristics: such as cracks, deformation, surface flaws, etc.; various types of material condition characteristics: such as material quality grade, strength, hardness, etc.; various types of equipment condition characteristics: such as equipment operating status, performance, maintenance cycle, etc.; various types of process parameter characteristics: such as temperature, pressure, humidity, processing time, etc.; these key characteristics are the main factors affecting product quality, and setting these characteristics will help in the subsequent precise alignment and analysis of reference data; ensuring that all reference data can be matched and aligned around these key characteristics, thereby forming a manufacturing reference model with quality guidance significance.

[0062] More specifically, data in the big data platform is matched according to established key characteristics. This data matching process yields a reference data set that meets key characteristic requirements. This process allows the massive amount of collected data to be filtered for data that is relevant to quality and meets production requirements, ensuring more accurate and targeted data analysis. This effectively filters out valuable data and eliminates irrelevant or noisy data, providing a high-quality reference data set for subsequent analysis and modeling.

[0063] More specifically, each data item in the reference data set is subjected to deep data separation and analysis based on its surface information, and the direct and indirect related data of each reference data are analyzed. Direct related data is usually directly related to product quality, while indirect related data indirectly affects quality through multiple levels. Deep data separation can more clearly identify the key factors and influencing factors in the data, thereby improving the accuracy of the analysis. By separating directly and indirectly related data, it is helpful to understand which factors have a direct impact on product quality and which are indirect influencing factors. After data separation, it is possible to clearly distinguish which data directly affects quality and which indirectly affects quality through other factors, facilitating targeted optimization.

[0064] More specifically, cluster analysis is performed on the directly related data and indirectly related data after deep data separation to extract their cluster characteristics and discrete point characteristics. Directly related data: such as the direct relationship between equipment status and product quality, the direct relationship between process parameters and quality results, etc.; indirectly related data: such as the indirect impact of temperature on material properties, the impact of raw material status on process results, etc. Through cluster analysis, samples with similar characteristics in the data can be identified, which helps to identify different types of production models, quality issues or potential improvement points. Discrete point characteristics help to discover anomalies and potential risks in the data. Through cluster analysis, potential patterns in the data can be more deeply explored and help identify key factors affecting quality. Discrete point characteristics can help to discover potential causes of quality fluctuations.

[0065] More specifically, a direct association reference model is constructed based on the cluster characteristics and discrete point characteristics of directly associated data; an indirect association reference model is constructed based on the cluster characteristics and discrete point characteristics of indirectly associated data. The direct association reference model is used to describe factors that directly affect product quality, such as equipment status, process parameters, etc., and the indirect association reference model is used to describe factors that indirectly affect product quality, such as material conditions, environmental conditions, etc. Constructing two different types of reference models can analyze and optimize the product manufacturing process from different dimensions. These two models can help production teams perform quality management and optimization at different levels. Constructing two reference models that can respectively reflect direct and indirect impacts on quality provides a more comprehensive analysis tool for the optimization of the production process.

[0066] More specifically, the direct association reference model and the indirect association reference model are information mapped with the key characteristic alignment target to generate a final manufacturing reference model, which integrates the influence of direct and indirect association factors to form a complete reference model. The information mapping step integrates the two models (direct and indirect association) to provide a comprehensive and complete manufacturing reference model, which can better guide production optimization and quality control, and obtain a comprehensive and complete manufacturing reference model, which can help to perform real-time quality control and process optimization during the production process.

[0067] It can be understood that through the above steps, a comprehensive manufacturing reference model is finally obtained, which can reflect the key quality characteristics, production environment factors and direct and indirect influencing factors of the target product in the manufacturing process, improve the accuracy of the data, and make the reference data more in line with actual production needs through the alignment of key characteristics and in-depth analysis of the data. Through cluster analysis and model construction, the key factors affecting quality can be identified, and the establishment of a reference model can provide guidance for the optimization of the production process. Information mapping ultimately generates a comprehensive manufacturing reference model with real-time application value, which can support quality control and decision optimization. This model plays an important role in quality monitoring and optimization in the production process, and can ensure the continuous improvement of the quality and production efficiency of the target product in actual applications.

[0068] Preferably, the step of performing key factor analysis on the manufacturing reference model to obtain the manufacturing element optimization factor fed back by the manufacturing reference model includes: S31: performing principal component analysis of cluster features on a directly associated reference model in the manufacturing reference model to generate a number of core dominant factors relative to the alignment target; S32: performing discrete probability summarization of discrete point features on the directly associated reference model in the manufacturing reference model to generate fluctuation probability information of each of the core dominant factors; S33: performing principal component analysis of cluster characteristics on the indirectly associated reference model in the manufacturing reference model to generate a plurality of latent conduction factors relative to the alignment target; S34: performing a discrete probability summary of discrete point features on the indirect correlation reference model in the manufacturing reference model to generate fluctuation probability information of each of the implicit transmission factors; S35: performing quality-oriented evaluation on each of the core dominant factors and each of the implicit conduction factors according to the manufacturing reference model, so as to extract core dominant optimization factors and implicit conduction optimization factors with positive quality orientation from each of the core dominant factors; S36: performing factor confidence analysis on the core dominant optimization factor and the implicit transmission optimization factor based on the fluctuation probability information to generate factor confidences of the core dominant optimization factor and the implicit transmission optimization factor; S37: Combining the core dominant optimization factor and the implicit conduction optimization factor with the corresponding factor confidence levels to serve as manufacturing factor optimization factors.

[0069] Specifically, principal component analysis (PCA) is performed on the cluster features of the directly associated reference models within the manufacturing reference model to extract several core dominant factors relative to the alignment target. Cluster features reflect the combination of different factors in the data. PCA can transform multiple variables into several core factors, reducing complexity and highlighting key influencing factors. PCA can help identify the dominant factors that have a significant impact on product quality and extract core dominant factors. These factors represent the key factors that directly affect product quality, facilitating subsequent analysis and optimization.

[0070] More specifically, a discrete probability summary is performed on the discrete point features in the directly associated reference model to generate fluctuation probability information for each core dominant factor. The discrete point features represent outliers and extreme cases in the data. The discrete probability summary helps to quantify the fluctuation range of each factor and its impact, which can reveal the change pattern of key factors and their potential impact on product quality. Through the discrete probability summary, a fluctuation range is assigned to each core dominant factor, providing a quantitative risk assessment to help better understand the causes of quality fluctuations.

[0071] More specifically, principal component analysis is performed on the cluster characteristics of the indirect correlation reference model in the manufacturing reference model to generate several implicit transmission factors relative to the alignment target. The indirect correlation reference model reflects the indirect factors affecting quality, such as materials, environmental conditions, etc. Through principal component analysis, these implicit transmission factors can be identified. Although these factors do not directly affect product quality, they indirectly affect it through other factors. The extracted implicit transmission factors can reveal potential risks and opportunities hidden in the manufacturing process, which is helpful to take preventive measures in product quality management.

[0072] More specifically, a discrete probability summary is performed on the discrete point features in the indirect correlation reference model to generate fluctuation probability information for each implicit transmission factor. Similar to the directly correlated factors, the volatility of indirect factors will also affect the stability of quality. The discrete probability summary can reveal the fluctuation range of these factors and their possible impact, thereby providing important decision-making support for production optimization and generating fluctuation probability information for implicit transmission factors, so that the risks of indirect influencing factors can be quantified, further enhancing the ability to predict and manage quality fluctuations.

[0073] More specifically, a quality-oriented evaluation is conducted on each core dominant factor and implicit transmission factor based on the manufacturing reference model. Through the evaluation, core dominant optimization factors with positive quality orientation and implicit transmission optimization factors are extracted from the core dominant factors. The quality-oriented evaluation helps identify factors that help improve product quality by quantifying the quality impact of each factor. These optimization factors may include improving stability, reducing fluctuations, improving consistency, etc., extracting optimization factors that have a positive effect on product quality and providing specific optimization directions. These factors will become key driving factors for the optimization of manufacturing elements.

[0074] More specifically, factor confidence analysis is performed based on the fluctuation probability information of the core dominant optimization factors and the implicit transmission optimization factors. This analysis helps confirm the reliability and stability of each optimization factor. Factor confidence analysis quantifies the risk and uncertainty of each optimization factor by evaluating the fluctuation probability. Factors with high confidence indicate that the factor has higher stability and lower risk in the optimization process. The confidence value generated for each optimization factor provides data support for production decisions and helps evaluate which factors have more reliable optimization effects.

[0075] More specifically, the core dominant optimization factors and implicit conduction optimization factors extracted from the quality-oriented assessment are combined with their respective factor confidence levels as the final manufacturing factor optimization factors. Combining the optimization factors with the confidence levels can ensure that the stability and reliability of each factor are fully considered during the optimization process. This combination provides more practical guidance for actual operations and derives the final manufacturing factor optimization factors. These factors can serve as important decision-making basis in the production process, and are used to optimize the production process, improve product quality, and reduce production risks.

[0076] It is understandable that through this series of steps, the final manufacturing factor optimization factors will include core dominant factors that directly affect quality and implicit transmission factors that are transmitted through indirect influences. By extracting core dominant factors and implicit transmission factors, key factors affecting quality can be identified, thereby achieving more accurate quality control and management. Through factor confidence analysis and quality-oriented evaluation, reliability analysis of optimization factors is provided to ensure that the implementation of optimization measures has a high probability of success. By combining fluctuation probability information and optimization factors, the final generated manufacturing factor optimization factors provide strong data support for the optimization of the production process, helping to achieve more efficient and stable production. This process provides a data-driven solution for continuous optimization of the manufacturing process, ensuring continuous improvement of manufacturing factors and continuous improvement of quality.

[0077] Preferably, the step of constructing an objective function including quality constraints, efficiency constraints and cost constraints for the target product based on the manufacturing element optimization factors of each manufacturing reference model includes: S41: Constructing a manufacturing process simulation function corresponding to the target product based on a product manufacturing information chain consisting of a plurality of process links; wherein the manufacturing process simulation function includes a plurality of process calculation units, each of which is used to correspond to each sub-link in each process link; S42: Allocating each of the manufacturing element optimization factors to a corresponding process calculation unit in the manufacturing process simulation function to serve as a candidate calculation value for each of the process calculation units; S43: performing an evaluation method analysis of quality evaluation, efficiency evaluation, and cost evaluation on the target product according to the selected calculation values ​​of each process calculation unit in the manufacturing process simulation function, so as to generate a quality evaluation unit, an efficiency evaluation unit, and a cost evaluation unit of the manufacturing process simulation function for the target product; S44: The manufacturing process simulation function and the quality evaluation unit, efficiency evaluation unit and cost evaluation unit based on the manufacturing process simulation function are jointly constructed as the target function of the target product.

[0078] Specifically, the manufacturing process of the target product is analyzed, and a simulation function is constructed based on the manufacturing information chain of the product. This manufacturing process simulation function is composed of multiple process calculation units, and each calculation unit is responsible for processing a specific link or sub-link in the manufacturing process. The manufacturing process is composed of multiple links and sub-links, and each link may have an impact on quality, efficiency and cost. By splitting the entire process into multiple calculation units, the impact of each link on the performance of the final product can be simulated in detail, which helps to grasp the manufacturing process and effect of the product as a whole. A detailed manufacturing process simulation function is constructed, which can accurately represent the various manufacturing links of the target product and provide a basic framework for subsequent optimization analysis.

[0079] More specifically, the optimization factors of each manufacturing element obtained through the above steps are assigned to the corresponding process calculation units in the manufacturing process simulation function. Each process calculation unit will accept the corresponding optimization factor as a candidate calculation value based on the links it involves. The manufacturing element optimization factor reflects the key optimization factors in the production process, such as quality, stability, efficiency, etc. By combining these factors with specific process links, it can be ensured that the optimization effect of each link in the simulation process is accurately reflected, thereby helping to identify which optimization factors are most critical to the quality, efficiency and cost of the target product, ensuring that each manufacturing link can take into account the relevant optimization factors during the simulation process, thereby improving the accuracy and effect of the objective function construction.

[0080] More specifically, based on the candidate calculation values ​​of each process calculation unit in the manufacturing process simulation function, the target product is evaluated in terms of quality, efficiency and cost. These evaluation methods analyze the performance of the product under different constraints and generate corresponding quality evaluation units, efficiency evaluation units and cost evaluation units. Quality, efficiency and cost are the three most important constraints in product manufacturing. By simulating these evaluations, it is possible to clarify how different process links affect these three factors, and then optimize the manufacturing process. This helps to construct an objective function that can fully reflect the manufacturing goals. Through the evaluation of quality, efficiency and cost, the corresponding evaluation units are generated, which fully considers the impact of product quality, production efficiency and manufacturing cost, which provides basic data and evaluation basis for the formulation of subsequent optimization objective functions.

[0081] More specifically, the manufacturing process simulation function is combined with the corresponding quality evaluation unit, efficiency evaluation unit and cost evaluation unit to construct the objective function of the target product. This objective function will comprehensively consider the constraints of product quality, production efficiency and cost-effectiveness. The construction of the objective function requires the integration of all relevant factors so as to find the best solution that meets the three major constraints of quality, efficiency and cost during the optimization process. Combining the manufacturing process simulation function with the evaluation unit can optimize these factors at the same time under a unified framework to ensure that the target product can be optimally manufactured under various constraints. An objective function that comprehensively considers quality, efficiency and cost is generated, which provides a theoretical basis and decision-making support for subsequent manufacturing process optimization.

[0082] It can be understood that this process constructs an objective function that includes quality, efficiency, and cost constraints through the following steps, and fully utilizes the role of manufacturing factor optimization factors. By modeling the manufacturing process of the target product and combining the optimization factors with each link, a detailed simulation of the manufacturing process is achieved. The quality, efficiency, and cost evaluation units ensure that the objective function takes into account these three key constraints. Integrating all factors into one objective function provides theoretical and data support for manufacturing process optimization. The objective function can comprehensively consider the quality, efficiency, and cost of the product, optimize the manufacturing process under various constraints, and ensure that the performance of the final product meets the requirements. Through quantitative analysis of the objective function, specific optimization directions can be provided and decision makers can be assisted in making adjustments in different manufacturing links. By evaluating efficiency and cost, inefficient and high-cost links in the production process can be discovered and improved, thereby optimizing resource allocation, improving production efficiency, and reducing production costs. The quality evaluation unit in the objective function can help ensure that product quality is always within an acceptable range, thereby enhancing the market competitiveness of the product. This method provides a refined analysis and decision-making basis for the optimization of manufacturing products, and can help manufacturers balance efficiency and cost while ensuring product quality to achieve the best manufacturing effect.

[0083] Preferably, the step of iteratively optimizing and solving the objective function, and simultaneously verifying the solution results using the big data platform as a supervision condition until an optimized decision solution is obtained includes: S51: Randomly combining the candidate calculation values ​​of the process calculation units in the objective function to obtain an initial intelligent population consisting of a plurality of intelligent individuals; wherein each intelligent individual corresponds to one candidate calculation value; S52: Performing a multi-dimensional evaluation of the candidate calculated values ​​of the initial intelligent population using the quality evaluation unit, the efficiency evaluation unit, and the cost evaluation unit in the objective function, and assigning a target value orientation to each intelligent individual based on the multi-dimensional evaluation results; S53: Based on the target value orientations obtained for each intelligent individual, the initial intelligent population is adjusted and selected to obtain a reconstructed intelligent population. The reconstructed intelligent population is then treated as the initial intelligent population for multi-dimensional evaluation and target value orientation assignment. This step is repeated to obtain several candidate optimal populations. S54: Matching and collecting related reference data from the big data platform according to each of the candidate optimal populations to verify the feasibility of each of the candidate optimal populations, and selecting the candidate optimal population to be implemented from each of the candidate optimal populations based on the feasibility verification results to convert them into an optimization decision plan.

[0084] Specifically, the candidate calculation values ​​of each process calculation unit in the objective function are randomly combined to form an initial intelligent population composed of several intelligent individuals. Each intelligent individual corresponds to a candidate calculation value, which represents the specific choice of the objective function in a certain link. By randomly combining the candidate calculation values, a diverse initial intelligent population can be generated, increasing the diversity of exploring different solutions during the optimization process. The diversity of the initial population helps to avoid falling into local optimal solutions and increases the chance of finding the global optimal solution. By generating a diverse initial population, a good foundation is laid for subsequent iterative optimization and the breadth of the search process is ensured.

[0085] More specifically, the quality evaluation unit, efficiency evaluation unit, and cost evaluation unit in the objective function are used to conduct a multi-dimensional evaluation of each intelligent individual in the initial intelligent population. Based on the evaluation results, a target value tendency is assigned to each intelligent individual to reflect its contribution to the optimization of the objective function. The objective function includes multiple evaluation dimensions such as quality, efficiency, and cost. Multiple evaluation units can comprehensively reflect the performance of each intelligent individual in these dimensions, thereby providing clear guidance for subsequent optimization. The target value tendency is a quantitative evaluation of the performance of each intelligent individual. It can help select the optimal individual and guide the optimization direction. By conducting multi-dimensional evaluation and value tendency allocation for each intelligent individual, the advantages and disadvantages of each individual are accurately assessed, providing a basis for subsequent adjustment and selection of intelligent individuals.

[0086] More specifically, the initial intelligent population is adjusted and selected based on the target value orientation of each intelligent individual. During the selection process, excellent intelligent individuals are retained, while poorer individuals are eliminated or replaced to generate a new, optimized intelligent population (i.e., a reconstructed intelligent population). Then, the reconstructed intelligent population is regarded as the new initial population, and multi-dimensional evaluation and target value orientation allocation are continued. By selecting excellent individuals and eliminating poorly performing individuals, the most suitable solutions can be gradually screened out. This selection mechanism simulates the "survival competition" process in nature, allowing the excellent individuals in the population to continue to reproduce and the optimization results to be gradually improved. The reconstructed intelligent population can achieve more precise optimization based on the adjustment of the target value orientation, avoiding the inefficiency or incomprehensibility that may be caused by early random combinations. The optimization process gradually converges and the quality of the solution continues to improve.

[0087] More specifically, the reconstructed intelligent population undergoes a multi-dimensional evaluation and target value orientation assignment. Based on the evaluation results, intelligent individuals are adjusted and selected to form a new population. This process continues through multiple iterations until a number of candidate optimal populations are obtained, each of which exhibits the best objective function optimization results. Through continuous iterative optimization, the search space is gradually narrowed and the quality of the solution is improved. Each iteration improves certain aspects, ultimately reaching a global optimal solution or one close to it. Over multiple iterations, the optimization algorithm continuously approaches the optimal solution, ensuring that the quality of the candidate populations gradually improves, ultimately providing multiple high-quality solutions that meet the constraints of the objective function.

[0088] More specifically, based on the obtained candidate optimal population, the big data platform is used to match and collect relevant reference data to verify the feasibility of these candidate optimal populations. During the verification process, it is checked whether the candidate solutions meet the constraints in actual production and whether they are operational. The big data platform provides rich real data and historical data, which can provide verification of the optimization results in a practical context. This step helps to ensure that the optimization results are not only optimal in theory, but also feasible and practical in actual applications. Through data verification on the big data platform, the mismatch between theoretical optimization results and actual conditions can be effectively avoided, ensuring the actual feasibility of the optimization plan.

[0089] More specifically, based on the feasibility verification results, the verified candidate optimal populations are screened out as the final optimization decision-making plans. These decision-making plans will become the implementation plans in the actual manufacturing process. Through feasibility verification, the final selected plan not only meets the theoretical optimization requirements, but also meets the actual production and operation conditions. The final selection of the optimization decision-making plan is efficient and feasible. Through this step, the optimization results are finally verified and converted into implementable optimization decision-making plans, providing clear guidance for the manufacturing process.

[0090] It can be understood that this process combines iterative optimization and big data verification, providing a systematic framework for solving the objective function. The random generation of the initial intelligent population ensures the breadth of the search space and increases the diversity of optimization. Multi-dimensional evaluation and target value tendency allocation accurately evaluate the pros and cons of each intelligent individual, helping the optimization process to develop towards the global optimal solution. The adjustment selection and iterative optimization of the intelligent population gradually eliminate unsuitable individuals, optimize the quality of the solution, and finally obtain the optimal solution. The feasibility verification of the big data platform ensures that the optimization results are not only theoretically optimal, but also practical, which increases the practicality of the solution. After multi-dimensional analysis and verification, an implementable and optimal optimization decision plan was finally determined. Through the evolution and iterative process of the intelligent population, the optimal solution that meets the constraints can be found efficiently. The actual data of the big data platform is used for verification to ensure that the optimization results are practical, providing manufacturing companies with a systematic optimization decision-making tool to help make reasonable decisions under multiple target constraints, improve production efficiency, reduce costs and maintain high quality.

[0091] Preferably, the step of matching and collecting relevant reference data from the big data platform according to each candidate optimal population includes: S541: performing feature decomposition on the candidate optimal population to obtain a quantifiable feature vector corresponding to the candidate optimal population; S542: performing similarity redundancy expansion processing based on the quantifiable feature vector to construct a correlation data matching model of the candidate optimal population; S543: Performing a layered gradient reference data correlation analysis on the big data platform through the correlation data matching model to obtain correlation reference data of a complete matching gradient, an overall matching gradient, and a partial matching gradient.

[0092] Specifically, all relevant features are extracted from each candidate optimal population. For each intelligent individual, its important attributes in the objective function, such as quality, efficiency, and cost, are identified and quantified. These features can be numerical (e.g., cost, efficiency, etc.) or categorical (e.g., technology category, product type, etc.). These features are then converted into quantifiable feature vectors for comparison and analysis in the subsequent matching process. Feature decomposition is intended to convert complex optimization results into actionable numerical or categorical features to facilitate subsequent matching with actual data in the big data platform. The quantified data has the advantages of being quantifiable and standardized, facilitating subsequent analysis and calculation. By converting the candidate optimal population into quantifiable feature vectors, it is possible to effectively compare the differences between different populations and provide a clear data structure for subsequent matching analysis.

[0093] More specifically, the information in the feature vector is used for algorithmic processing to perform similarity redundancy expansion. This process calculates the similarity between each candidate optimal population and the existing data in the big data platform, and adds redundant features to construct a correlation data matching model for the candidate optimal population. Redundancy expansion can include enhancement of some influencing factors, feature combination or expansion to ensure that the model captures the correlation between the population and big data more comprehensively and accurately. Redundancy expansion can improve the robustness and accuracy of the model. Similarity redundancy expansion processing can ensure that the model can still provide effective correlation analysis in the face of missing or incomplete data matching. This process helps to establish a more flexible and reliable matching relationship between the candidate optimal population and the big data platform. Through similarity redundancy expansion, a more complete correlation data matching model is constructed. This model not only improves the accuracy of matching, but also ensures the processing capability of possible feature incompleteness or data differences during the matching process, thereby enhancing the practicality of the system.

[0094] More specifically, a hierarchical gradient correlation analysis of reference data of the big data platform is performed through the associated data matching model, and the constructed associated data matching model is used to carry out a correlation analysis of the existing data in the big data platform. The data of the big data platform will be divided into different gradient levels according to similarity, usually including complete matching gradient, overall matching gradient and partial matching gradient: Complete matching gradient: refers to the fact that the characteristics of the candidate optimal population are completely consistent or highly consistent with the data in the big data platform; Overall matching gradient: refers to the fact that the characteristics of the candidate optimal population are mostly similar to the data in the big data platform. Although there are a few differences, the overall matching degree is relatively high; Partial matching gradient: refers to the fact that the characteristics of the candidate optimal population match the data in the big data platform in some key features, but the overall matching degree is low; Through hierarchical gradient analysis, the degree of matching between the candidate optimal population and the data in the big data platform can be clearly judged, providing a basis for subsequent feasibility verification and optimization decision-making. Hierarchical gradient analysis helps to distinguish data types with different matching degrees, thereby accurately screening the most appropriate reference data. Hierarchical gradient correlation analysis provides a clear matching assessment between the candidate optimal population and the data in the big data platform. This analysis result not only helps improve the accuracy of the optimization process, but also provides specific data support for subsequent feasibility verification.

[0095] More specifically, based on the results of hierarchical gradient analysis, data with different matching degrees in the big data platform are classified into fully matching gradients, overall matching gradients, and partially matching gradients. These data will serve as reference data for candidate optimal populations and will be further used for feasibility verification and generation of decision plans. By classifying the reference data according to the matching degree, data that meets the actual conditions can be selected more accurately. The data with fully matching gradients represents the optimal reference data, while the overall matching and partial matching data can be used as supplements or alternatives to enhance the flexibility and robustness of the model. The classified and organized reference data provides rich data support for the candidate optimal population. These data can effectively verify the actual feasibility of the population and help optimize the generation process of decision plans, ensuring that the final optimization results have high practical adaptability.

[0096] It is understandable that this process ensures that the optimization results are not only optimal in theory, but also have good matching and feasibility in practical applications through multiple steps such as feature decomposition, similarity redundancy expansion, hierarchical gradient analysis and matching data classification. The specific implementation effects are as follows: through feature decomposition and similarity redundancy expansion, it is ensured that the candidate optimal population is highly matched with the data in the big data platform. The redundant expansion processing improves the adaptability of the model when facing incomplete matching. Through hierarchical gradient analysis, it is clarified which data best meets the actual needs of the candidate optimal population, providing clear guidance for subsequent feasibility verification. This process provides comprehensive data support for the formulation of the final optimization decision plan, ensuring that the decision plan not only meets the optimization requirements of the objective function, but can also be effectively implemented in practical applications.

[0097] Reference Figure 2 As shown, in a second aspect, the present invention provides a manufacturing process optimization decision-making method device based on big data analysis, which is used to implement the manufacturing process optimization decision-making method based on big data analysis described in any one of the first aspects, including: A data acquisition module is used to collect link element data of each link in the manufacturing process of the target product and quality assessment information of the target product, and combine the link element data of each link and the quality assessment information into reference data; A data alignment module is used to continuously record reference data of target products through a big data platform and perform several forms of key characteristic alignment processing on each reference data to obtain several manufacturing reference models; A factor analysis module, configured to perform key factor analysis on the manufacturing reference model to obtain manufacturing element optimization factors fed back by the manufacturing reference model; a function construction module, configured to construct an objective function including quality constraints, efficiency constraints, and cost constraints for the target product based on the manufacturing element optimization factors of each manufacturing reference model; The optimization decision module is used to iteratively optimize and solve the objective function, and at the same time use the big data platform as a supervision condition to verify the solution results until an optimized decision solution is obtained.

[0098] In this embodiment, for the specific implementation of each module in the above device embodiment, please refer to the above method embodiment, which will not be repeated here.

[0099] In a third aspect, the present invention provides a manufacturing process optimization decision-making device based on big data analysis, comprising a memory and a processor, wherein the memory stores a manufacturing process optimization decision-making program based on big data analysis that can be run on the processor, and when the processor executes the manufacturing process optimization decision-making program based on big data analysis, it implements a manufacturing process optimization decision-making method based on big data analysis as described in any one of the first aspects.

[0100] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A manufacturing process optimization decision-making method based on big data analysis, characterized in that: include: Collecting link element data of each link in the manufacturing process of the target product and quality assessment information of the target product, and combining the link element data of each link and the quality assessment information as reference data; Continuously record reference data of target products through a big data platform, and perform several forms of key feature alignment processing on each reference data to obtain several manufacturing reference models; Performing key factor analysis on the manufacturing reference model to obtain manufacturing element optimization factors fed back by the manufacturing reference model; Based on the manufacturing element optimization factors of each manufacturing reference model, constructing an objective function including quality constraints, efficiency constraints and cost constraints for the target product; The objective function is iteratively optimized and solved, and the solution is verified using the big data platform as a supervision condition until an optimized decision-making solution is obtained.

2. The manufacturing process optimization decision-making method based on big data analysis according to claim 1, characterized in that: The steps of collecting link element data of each link in the manufacturing process of the target product and quality assessment information of the target product, and combining the link element data of each link and the quality assessment information into reference data include: Obtaining manufacturing process information of the target product, and constructing a product manufacturing information chain consisting of several process links based on the manufacturing process information; wherein the process links include a material preparation link, an equipment configuration link, and a process parameter link, and each process link includes several sub-links; Collecting original record data of each process link of the manufacturing process corresponding to the target product based on the product manufacturing information chain, and performing quality assessment on the target product to obtain quality assessment information; Performing feature encoding of a specified format on the original recorded data to convert the original recorded data into a corresponding feature vector matrix; The tensor decomposition method is used to capture the potential correlation features across sub-links of the eigenvector matrix corresponding to each sub-link in each process link to generate the potential correlation factors between each sub-link; Performing vector expansion processing on the eigenvector matrix corresponding to each sub-link according to the positional relationship of each sub-link in the product manufacturing information chain and the potential correlation factor to obtain an enhanced vector matrix corresponding to each sub-link; Aggregate and splice the enhanced vector matrices corresponding to each sub-link according to the time sequence and logical order, and extract the key semantics of each enhanced vector matrix; Each enhanced vector matrix is ​​used as deep information, the extracted key semantics are used as surface information corresponding to the deep information, and the deep information, the surface information and the quality assessment information are combined to obtain reference data.

3. The manufacturing process optimization decision-making method based on big data analysis according to claim 2, characterized in that: The steps of continuously recording reference data of target products through a big data platform and performing several forms of key characteristic alignment processing on each reference data to obtain several manufacturing reference models include: Continuously record reference data of the target product through the big data platform, and simultaneously obtain production environment information of the target product, and combine the production environment information with the reference data; Key characteristics to be aligned are set for the big data platform, so as to perform data matching on the big data platform according to the key characteristics, so as to obtain a reference data set that meets the key characteristics; wherein the key characteristics include excellent quality characteristics, various types of defect quality characteristics, various types of material condition characteristics, various types of equipment condition characteristics, and various types of process parameter characteristics; Taking the key characteristic as an alignment target, performing data separation and analysis on the deep information of each reference data in the reference data set based on the surface information of each reference data, so as to separate directly related data and indirectly related data corresponding to the alignment target from the reference data; Perform cluster analysis on the direct related data and indirect related data separated from each reference data to extract the cluster characteristics and discrete point characteristics of each direct related data and each indirect related data; A direct association reference model is constructed based on the cluster characteristics and discrete point characteristics of each directly associated data item, and an indirect association reference model is constructed based on the cluster characteristics and discrete point characteristics of each indirectly associated data item. The direct association reference model and the indirect association reference model are information mapped to the alignment target to obtain a manufacturing reference model of the alignment target.

4. The manufacturing process optimization decision-making method based on big data analysis according to claim 3, characterized in that: The step of performing key factor analysis on the manufacturing reference model to obtain manufacturing element optimization factors fed back by the manufacturing reference model includes: performing principal component analysis of cluster features on a directly associated reference model in the manufacturing reference model to generate a number of core dominant factors relative to the alignment target; Performing discrete probability summarization of discrete point features on a directly associated reference model in the manufacturing reference model to generate fluctuation probability information of each of the core dominant factors; performing principal component analysis of cluster features on the indirectly associated reference model in the manufacturing reference model to generate a plurality of latent transmission factors relative to the alignment target; Performing discrete probability summarization of discrete point features on the indirect correlation reference model in the manufacturing reference model to generate fluctuation probability information of each of the implicit conduction factors; Performing a quality-oriented evaluation on each of the core dominant factors and each of the implicit conduction factors according to the manufacturing reference model, so as to extract a core dominant optimization factor and an implicit conduction optimization factor with positive quality orientation from each of the core dominant factors; performing a factor confidence analysis on the core dominant optimization factor and the implicit transmission optimization factor based on the volatility probability information to generate factor confidences of the core dominant optimization factor and the implicit transmission optimization factor; The core dominant optimization factor and the implicit conduction optimization factor are respectively combined with the corresponding factor confidence levels to serve as manufacturing factor optimization factors.

5. The manufacturing process optimization decision-making method based on big data analysis according to claim 1, characterized in that: The steps of constructing an objective function including quality constraints, efficiency constraints, and cost constraints for the target product based on the manufacturing element optimization factors of each manufacturing reference model include: Constructing a manufacturing process simulation function corresponding to the target product based on a product manufacturing information chain consisting of a plurality of process links; wherein the manufacturing process simulation function includes a plurality of process calculation units, each of which is used to correspond to each sub-link in each process link; Allocating each of the manufacturing element optimization factors to a corresponding process calculation unit in the manufacturing process simulation function to serve as a candidate calculation value for each of the process calculation units; An evaluation method for performing quality evaluation, efficiency evaluation, and cost evaluation on the target product according to the selected calculation values ​​of each process calculation unit in the manufacturing process simulation function is analyzed to generate a quality evaluation unit, an efficiency evaluation unit, and a cost evaluation unit of the manufacturing process simulation function for the target product; The manufacturing process simulation function and the quality evaluation unit, efficiency evaluation unit and cost evaluation unit based on the manufacturing process simulation function are jointly constructed as the target function of the target product.

6. The manufacturing process optimization decision-making method based on big data analysis according to claim 5, characterized in that: The steps of iteratively optimizing and solving the objective function and verifying the solution results using the big data platform as a supervision condition until an optimized decision solution is obtained include: Randomly combining the candidate calculation values ​​of the process calculation units in the objective function to obtain an initial intelligent population consisting of a plurality of intelligent individuals; wherein each intelligent individual corresponds to one candidate calculation value; Performing a multi-dimensional evaluation of the candidate calculated values ​​of the initial intelligent population through the quality evaluation unit, efficiency evaluation unit, and cost evaluation unit in the objective function, and assigning a target value tendency to each intelligent individual based on the multi-dimensional evaluation results; Based on the target value orientations obtained by each intelligent individual, the initial intelligent population is adjusted and selected to obtain a reconstructed intelligent population. The reconstructed intelligent population is then treated as the initial intelligent population for multi-dimensional evaluation and target value orientation assignment. This step is repeated to obtain several candidate optimal populations. According to each of the candidate optimal populations, relevant reference data is matched and collected from the big data platform to verify the feasibility of each of the candidate optimal populations. Based on the feasibility verification results, a candidate optimal population to be implemented is selected from each of the candidate optimal populations to convert into an optimization decision plan.

7. The manufacturing process optimization decision-making method based on big data analysis according to claim 6, characterized in that: The step of matching and collecting related reference data from the big data platform according to each candidate optimal population includes: Performing feature decomposition on the candidate optimal population to obtain a quantifiable feature vector corresponding to the candidate optimal population; Performing similarity redundancy expansion processing based on the quantifiable feature vector to construct a correlation data matching model for the candidate optimal population; The association data matching model is used to perform a hierarchical gradient reference data association analysis on the big data platform to obtain association reference data of a complete matching gradient, an overall matching gradient, and a partial matching gradient.

8. A manufacturing process optimization decision-making method and device based on big data analysis, characterized in that: A manufacturing process optimization decision-making method based on big data analysis for implementing any one of claims 1 to 7, comprising: A data acquisition module is used to collect link element data of each link in the manufacturing process of the target product and quality assessment information of the target product, and combine the link element data of each link and the quality assessment information into reference data; A data alignment module is used to continuously record reference data of target products through a big data platform and perform several forms of key characteristic alignment processing on each reference data to obtain several manufacturing reference models; A factor analysis module, configured to perform key factor analysis on the manufacturing reference model to obtain manufacturing element optimization factors fed back by the manufacturing reference model; a function construction module, configured to construct an objective function including quality constraints, efficiency constraints, and cost constraints for the target product based on the manufacturing element optimization factors of each manufacturing reference model; The optimization decision module is used to iteratively optimize and solve the objective function, and at the same time use the big data platform as a supervision condition to verify the solution results until an optimized decision solution is obtained.

9. A manufacturing process optimization decision-making device based on big data analysis, comprising a memory and a processor, wherein the memory stores a manufacturing process optimization decision-making program based on big data analysis that can be run on the processor, characterized in that: When the processor executes the manufacturing process optimization decision-making program based on big data analysis, it implements the manufacturing process optimization decision-making method based on big data analysis described in any one of claims 1-7.

Images