Textile production man-machine efficiency optimization method and system fusing multiple AI algorithms

Abstract

The invention provides a multi-AI algorithm fused spinning production man-machine efficiency optimization method and system in the technical field of spinning production management. The method comprises the steps of S1, collecting core spinning production data; s2, performing anomaly detection and filtering on the core textile production data through an anomaly detection model; s3, inputting each piece of core textile production data into the productivity prediction model to obtain a textile productivity prediction result, and generating a textile production scheduling scheme list based on the textile productivity prediction result and a production scheduling configuration rule; s4, inputting the textile production scheduling scheme list into the production scheduling optimization model to obtain an optimal production scheduling scheme; and S5, executing the optimal production scheduling scheme, and generating a textile production report based on the statistical indexes on the basis of the statistical indexes of the core textile production data generated after execution. The method has the advantages that intelligent production scheduling, automatic report generation and data closed-loop management are realized, so that the equipment utilization rate is effectively improved, the report generation time is shortened, and a data barrier is broken through.

Description

Technical Field

[0001] This invention relates to the field of textile production management technology, and in particular to a method and system for optimizing human-machine efficiency in textile production by integrating multiple AI algorithms. Background Technology

[0002] As a traditional manufacturing sector, the textile industry has long faced core management challenges such as fragmented orders, diverse equipment types, and massive amounts of production data. While existing production management systems possess basic data collection capabilities, significant shortcomings remain in key areas such as production scheduling and report generation. Specifically: 1. Insufficient level of intelligent production scheduling: The current production scheduling process mainly relies on manual experience and fails to effectively integrate the real-time operating status of equipment (such as downtime rate and operating speed) and the dynamic changes in material inventory. This leads to frequent occurrences of high-yield equipment being idle and low-priority orders occupying core resources, resulting in unbalanced capacity allocation, delayed order delivery, and consequently affecting customer satisfaction.

[0003] 2. Low report generation efficiency: Production reports rely on manual data screening and manual input into fixed templates. A single monthly report takes 2 to 3 hours and is prone to errors due to human operation, affecting the accuracy of management decisions. In addition, report generation is severely delayed (e.g., daily production reports can only be completed the next day), the analysis dimensions are limited, and it can only provide basic data statistics, making it difficult to support real-time and dynamic production decisions.

[0004] 3. Data value not fully realized: Production traceability information and personnel efficiency data accumulated in the system have not been effectively linked with production scheduling and reporting processes, forming data silos. This hinders the construction of a closed-loop management mechanism of "data collection - analysis and decision-making - production execution", limits the mining of potential data value, and restricts the enterprise's transformation and upgrading to intelligent manufacturing.

[0005] Therefore, how to provide a method and system for optimizing human-machine efficiency in textile production by integrating multiple AI algorithms, so as to realize intelligent production scheduling, automated report generation and data closed-loop management, thereby effectively improving equipment utilization, shortening report generation time and breaking down data barriers, has become an urgent technical problem to be solved. Summary of the Invention

[0006] The technical problem to be solved by this invention is to provide a method and system for optimizing human-machine efficiency in textile production by integrating multiple AI algorithms, so as to realize intelligent production scheduling, automated report generation and data closed-loop management, thereby effectively improving equipment utilization, shortening report generation time and breaking down data barriers.

[0007] In a first aspect, the present invention provides a method for optimizing human-machine efficiency in textile production by integrating multiple AI algorithms, comprising the following steps: Step S1: Collect core textile production data in real time, including textile order data, textile equipment data, and employee data, and preprocess each of the core textile production data. Step S2: Perform anomaly detection and filtering on the pre-processed core textile production data using a pre-trained anomaly detection model; Step S3: Input the filtered core textile production data into the pre-trained capacity prediction model to obtain textile capacity prediction results, and generate a textile production scheduling scheme list based on the textile capacity prediction results and the preset production scheduling configuration rules. Step S4: Input the list of textile production scheduling schemes into the pre-trained production scheduling optimization model to obtain the optimal production scheduling scheme; Step S5: Execute the optimal production scheduling plan, calculate statistical indicators based on the core textile production data generated after execution, and generate and display a textile production report based on each statistical indicator using a Word template engine.

[0008] Furthermore, step S1 specifically includes: Real-time collection of core textile production data, including textile order data, textile equipment data, and employee data; the textile order data includes at least customer order information and product BOM data, the customer order information includes at least product model, production quantity, and delivery date, and the product BOM data includes at least cotton yarn usage; the textile equipment data includes at least machine speed, actual running time, downtime records, and material loss rate; the employee data includes at least employee clock-in time, production meters, and employee operating efficiency. The core textile production data are preprocessed, including at least data cleaning, data deduplication, and feature extraction.

[0009] Furthermore, in step S2, the anomaly detection model is created based on the isolated forest algorithm to identify abnormal data that exceeds the preset configuration range, and to mark the abnormal data as special operation data and filter it.

[0010] Furthermore, in step S3, the capacity prediction model is built based on an LSTM network and is used to output a textile capacity prediction result carrying the machine output and equipment failure probability for the next 24 hours based on the input machine speed, actual running time, material loss rate and employee operating efficiency. The capacity forecasting model is trained based on historical data from the past six months; The textile production scheduling plan list contains a preset number of textile production scheduling plans; each textile production scheduling plan includes at least a machine allocation plan, a material requirement list, a production sequence, and employee binding relationships.

[0011] Furthermore, in step S4, the production scheduling optimization model is constructed based on a genetic algorithm; The fitness function of the production scheduling optimization model is formulated as follows: F = W1 × F1 + W2 × F2 + W3 × F3; F1 = Σ(Actual delivery time - Required delivery time) / Total number of orders; F2 = 1 - Σ(running time of a single machine - average running time)² / total number of machines; F3 = Σ(Available inventory quantity - Production demand quantity) / Total material types; Where F represents adaptability; F1 represents order delay rate; F2 represents machine load balance; F3 represents material matching degree; W1, W2, and W3 are all weighting coefficients, with values ​​of 40%, 30%, and 30%, respectively. The optimization process of the production scheduling optimization model is as follows: the textile production scheduling scheme list is initialized as a population, the fitness is calculated based on the fitness function, a preset proportion of particles are selected from the population based on the fitness using the roulette wheel algorithm, crossover and mutation operations are performed on each selected particle in sequence to generate a new population, and it is determined whether the preset termination condition is met. If yes, the optimal production scheduling scheme is output; otherwise, the population iteration continues. The particle refers to the textile production scheduling scheme; the optimal production scheduling scheme is the best textile production scheduling scheme. The termination condition is that the number of iterations is ≥100, or the fitness is <0.12; Step S5 specifically involves: The optimal production scheduling plan is executed, and statistical indicators, including at least the total daily output, machine efficiency, year-on-year production difference data, and month-on-month production difference data, are calculated based on the core textile production data generated after execution. Trend charts and optimization suggestions are generated based on each of the statistical indicators using a Word template engine. A textile production report is generated and displayed based on each of the statistical indicators, trend charts, and optimization suggestions.

[0012] Secondly, this invention provides a textile production human-machine efficiency optimization system that integrates multiple AI algorithms, comprising the following modules: The core textile production data acquisition module is used to collect core textile production data in real time, including textile order data, textile equipment data, and employee data, and to preprocess each of the core textile production data. An anomaly detection and filtering module is used to perform anomaly detection and filtering on the pre-processed core textile production data using a pre-trained anomaly detection model. The textile production scheduling module is used to input the filtered core textile production data into a pre-trained capacity prediction model to obtain textile capacity prediction results, and generate a list of textile production scheduling schemes based on the textile capacity prediction results and preset scheduling configuration rules. The production scheduling optimization module is used to input the list of textile production scheduling schemes into a pre-trained production scheduling optimization model to obtain the optimal production scheduling scheme; The textile production report generation module is used to execute the optimal production scheduling plan, calculate statistical indicators based on the core textile production data generated after execution, and generate and display a textile production report based on each statistical indicator using a Word template engine.

[0013] Furthermore, the core textile production data acquisition module is specifically used for: Real-time collection of core textile production data, including textile order data, textile equipment data, and employee data; the textile order data includes at least customer order information and product BOM data, the customer order information includes at least product model, production quantity, and delivery date, and the product BOM data includes at least cotton yarn usage; the textile equipment data includes at least machine speed, actual running time, downtime records, and material loss rate; the employee data includes at least employee clock-in time, production meters, and employee operating efficiency. The core textile production data are preprocessed, including at least data cleaning, data deduplication, and feature extraction.

[0014] Furthermore, in the anomaly detection and filtering module, the anomaly detection model is created based on the isolated forest algorithm, which is used to identify abnormal data that exceeds the preset configuration range, mark the abnormal data as special operation data, and filter it.

[0015] Furthermore, in the textile production scheduling module, the capacity prediction model is built based on an LSTM network. It is used to output a textile capacity prediction result carrying the machine output and equipment failure probability for the next 24 hours based on the input machine speed, actual running time, material loss rate and employee operating efficiency. The capacity forecasting model is trained based on historical data from the past six months; The textile production scheduling plan list contains a preset number of textile production scheduling plans; each textile production scheduling plan includes at least a machine allocation plan, a material requirement list, a production sequence, and employee binding relationships.

[0016] Furthermore, in the production scheduling optimization module, the production scheduling optimization model is constructed based on a genetic algorithm; The fitness function of the production scheduling optimization model is formulated as follows: F = W1 × F1 + W2 × F2 + W3 × F3; F1 = Σ(Actual delivery time - Required delivery time) / Total number of orders; F2 = 1 - Σ(running time of a single machine - average running time)² / total number of machines; F3 = Σ(Available inventory quantity - Production demand quantity) / Total material types; Where F represents adaptability; F1 represents order delay rate; F2 represents machine load balance; F3 represents material matching degree; W1, W2, and W3 are all weighting coefficients, with values ​​of 40%, 30%, and 30%, respectively. The optimization process of the production scheduling optimization model is as follows: the textile production scheduling scheme list is initialized as a population, the fitness is calculated based on the fitness function, a preset proportion of particles are selected from the population based on the fitness using the roulette wheel algorithm, crossover and mutation operations are performed on each selected particle in sequence to generate a new population, and it is determined whether the preset termination condition is met. If yes, the optimal production scheduling scheme is output; otherwise, the population iteration continues. The particle refers to the textile production scheduling scheme; the optimal production scheduling scheme is the best textile production scheduling scheme. The termination condition is that the number of iterations is ≥100, or the fitness is <0.12; The textile production report generation module is specifically used for: The optimal production scheduling plan is executed, and statistical indicators, including at least the total daily output, machine efficiency, year-on-year production difference data, and month-on-month production difference data, are calculated based on the core textile production data generated after execution. Trend charts and optimization suggestions are generated based on each of the statistical indicators using a Word template engine. A textile production report is generated and displayed based on each of the statistical indicators, trend charts, and optimization suggestions.

[0017] The advantages of this invention are: 1. The system collects core textile production data in real time, including textile order data, textile equipment data, and employee data, and preprocesses each core data point. Then, an anomaly detection model is used to detect and filter anomalies in the preprocessed core textile production data. The filtered core textile production data is then input into a capacity prediction model to obtain textile capacity prediction results. Based on the textile capacity prediction results and preset production scheduling rules, a textile production scheduling plan list is generated. This list is then input into a pre-trained production scheduling optimization model to obtain the optimal production scheduling plan. Finally, the optimal production scheduling plan is executed, and statistical indicators are calculated based on the core textile production data generated after execution. A textile production report is then generated and displayed using a Word template engine based on these statistical indicators. In short, by collecting and preprocessing order, equipment, and employee data in real time, the system first uses the isolated forest algorithm to detect anomalies. The data is filtered to ensure quality; then, the clean data is input into the LSTM capacity forecasting model to proactively estimate equipment output and failure risks; based on this forecast, a genetic algorithm with order on-time performance, equipment load balancing, and material matching as core objectives is used to optimize the production scheduling plan for multiple objectives, generating the optimal production scheduling plan and executing it; finally, based on the execution feedback data, a Word template engine is used to quickly generate a comprehensive production report containing trend analysis and optimization suggestions; this process realizes a complete closed loop from data perception, intelligent decision-making to execution feedback, not only upgrading the production scheduling basis from static experience to dynamic data-driven, but also transforming report generation from a time-consuming manual operation into an instantaneous automated process, ultimately achieving intelligent production scheduling, automated report generation, and closed-loop data management, thereby effectively improving equipment utilization, shortening report generation time, and breaking down data barriers.

[0018] 2. By collecting textile order data, equipment data, and employee data in real time and performing preprocessing (such as data cleaning and feature extraction), the comprehensiveness and real-time nature of the data sources are ensured. By integrating multi-dimensional data (such as order delivery dates, machine operating status, and employee efficiency), a high-precision production data foundation can be built, thus providing reliable input for subsequent AI analysis. This fusion processing method is superior to traditional single-data source methods, reducing information silos and improving the accuracy and response speed of overall decision-making. In particular, it can effectively avoid errors caused by data lag in dynamic textile production environments.

[0019] 3. An anomaly detection model based on the isolated forest algorithm is adopted to automatically identify and filter abnormal data that exceeds the preset range (such as equipment downtime or employee operation deviations). This algorithm is good at processing high-dimensional data and can quickly detect rare but high-impact anomalies, thereby improving data cleanliness. Compared with manual inspection or simple threshold methods, it reduces the need for human intervention, improves the automation level of production efficiency monitoring, ensures that subsequent prediction and optimization models run on high-quality data, and reduces the risk of decision-making errors caused by data noise.

[0020] 4. The capacity forecasting model is built on an LSTM network and uses historical data (such as machine speed and employee efficiency over the past six months) to predict output and equipment failure probability for the next 24 hours. LSTM is good at processing time series data and can capture long-term dependencies, making the forecast results more in line with actual production fluctuations. This helps companies adjust resources in advance to avoid overcapacity or undercapacity, thereby optimizing inventory and delivery cycles. Compared with traditional statistical methods, it improves the robustness and adaptability of forecasts and provides a scientific basis for generating production scheduling plans.

[0021] 5. A production scheduling optimization model is constructed using a genetic algorithm. The fitness function comprehensively considers order delay rate, machine load balance, and material matching degree (with weights of 40%, 30%, and 30%, respectively), and searches for the optimal solution iteratively. This multi-objective optimization method can balance conflicting objectives (such as delivery time and resource utilization) and generate a globally optimal production scheduling plan. The parallel search characteristic of the genetic algorithm avoids local optima, improves the quality of solutions and convergence speed. Compared with manual scheduling or single-rule optimization, it significantly improves the flexibility and efficiency of production scheduling and reduces operating costs.

[0022] 6. Based on the execution results, the system automatically calculates statistical indicators (such as daily output and machine efficiency) and generates trend charts and optimization suggestions through a Word template engine. This end-to-end automated process reduces the time spent on manual summarization and report preparation, while providing visual insights to help managers quickly identify problems and adjust strategies. This reflects the system's practicality and user-friendliness, transforming AI output into actionable knowledge and enhancing the transparency and continuous improvement capabilities of production management.

[0023] 7. Seamlessly integrating multiple AI algorithms (such as Isolation Forest, LSTM, and Genetic Algorithm) into a single process, forming a closed loop from data collection to report presentation, this modular design allows for adjusting model parameters or adding new data sources according to different production needs, exhibiting excellent scalability. For example, the weights of the fitness function are customizable, easily adapting to the priorities of different textile enterprises. This systematic innovation not only improves human-machine efficiency but also lays the foundation for future technological upgrades (such as the introduction of deep learning or the Internet of Things), enhancing the long-term value of the solution.

[0024] 8. By integrating real-time acquisition of multi-source data, anomaly detection based on the isolated forest algorithm, capacity prediction using LSTM networks, and multi-objective production scheduling optimization using genetic algorithms, a closed-loop intelligent decision-making system from data preprocessing to report generation was constructed. Its core advantage lies in significantly improving the automation level and resource utilization efficiency of textile production: Through the collaboration of high-quality data foundation and AI model, the system achieves accurate capacity prediction, load-balanced production scheduling, and rapid response to anomalies. Ultimately, it drives continuous optimization with automated reports, thereby reducing the cost of human intervention, effectively shortening the delivery cycle, and improving the overall human-machine collaboration efficiency.

[0025] 9. Through a full-chain design of "data-algorithm-configuration-execution," significant optimizations are achieved compared to existing technologies: Production scheduling efficiency is greatly improved: Relying on LSTM capacity prediction and genetic algorithm optimization, production scheduling time is reduced from 2 hours / time of traditional manual scheduling to 5 minutes / time, improving efficiency by 91.7%; Simultaneously, because the textile production scheduling solution considers both equipment status and order priority, the on-time delivery rate of orders is improved by 30%, effectively solving the problems of capacity waste and delivery delays; Report timeliness is significantly improved: Real-time data collection and automatic report generation enable daily report generation in real time, with a lag time of ≤10 minutes, completely solving the lag problem of traditional "next-day generation" reports, supporting management to dynamically adjust production strategies and improve decision-making response speed; Data value is fully released: By linking order, equipment, and employee data through AI algorithms (breaking data silos), and combining trend analysis to accurately identify production bottlenecks (such as a 15% decrease in efficiency for a certain machine model every Wednesday), it can directly guide equipment maintenance planning and employee skills training, forming a "data-decision-execution" closed loop, driving the transformation of production management from "experience-driven" to "data-driven." Attached Figure Description

[0026] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0027] Figure 1 This is a flowchart of a method for optimizing human-machine efficiency in textile production that integrates multiple AI algorithms, according to the present invention.

[0028] Figure 2 This is a schematic diagram of the structure of a textile production human-machine efficiency optimization system that integrates multiple AI algorithms, according to the present invention. Detailed Implementation

[0029] The technical solution in this application embodiment follows the general idea as follows: By collecting and preprocessing order, equipment, and employee data in real time, the isolated forest algorithm is first used to filter abnormal data to ensure data quality. Then, the clean data is input into the LSTM capacity prediction model to proactively estimate equipment output and failure risk. Based on this prediction result, a genetic algorithm with order on-time rate, equipment load balancing, and material matching degree as core objectives is used to optimize the production scheduling plan for multiple objectives, generate the optimal production scheduling plan, and execute it. Finally, based on the execution feedback data, a three-dimensional production report containing trend analysis and optimization suggestions is quickly generated using a Word template engine. This process realizes a complete closed loop from data perception, intelligent decision-making to execution feedback, not only upgrading the production scheduling basis from static experience to dynamic data-driven, but also transforming report generation from a time-consuming manual operation into an instantaneous automated process. This achieves intelligent production scheduling, automated report generation, and closed-loop data management, thereby effectively improving equipment utilization, shortening report generation time, and breaking down data barriers.

[0030] Please refer to Figures 1 to 2 As shown, a preferred embodiment of the present invention, a method for optimizing human-machine efficiency in textile production by integrating multiple AI algorithms, includes the following steps: Step S1: Collect core textile production data in real time, including textile order data, textile equipment data, and employee data, and preprocess each of the core textile production data. Step S2: Perform anomaly detection and filtering on the pre-processed core textile production data using a pre-trained anomaly detection model; Step S3: Input the filtered core textile production data into the pre-trained capacity prediction model to obtain textile capacity prediction results. Based on the textile capacity prediction results and the preset production scheduling rules (such as "machine A daily running time ≤ 16 hours", "order priority: rush order > regular order"), generate a textile production scheduling plan list. Step S4: Input the list of textile production scheduling schemes into the pre-trained production scheduling optimization model to obtain the optimal production scheduling scheme; Step S5: Execute the optimal production scheduling plan, calculate statistical indicators based on the core textile production data generated after execution, and generate and display a textile production report based on each statistical indicator using a Word template engine.

[0031] Step S1 specifically involves: Real-time collection of core textile production data, including textile order data, textile equipment data, and employee data; the textile order data includes at least customer order information and product BOM data, the customer order information includes at least product model, production quantity, and delivery date, and the product BOM data includes at least cotton yarn usage (calculated using the formula: weight = meters × width × grammage / 1000); the textile equipment data includes at least machine speed (unit: revolutions per minute), actual running time, downtime records (including classification labels such as spinning, yarn threading, and maintenance), and material loss rate; the employee data includes at least employee clock-in time (bound to the machine), production meters (automatically accumulated across shifts), and employee operating efficiency; The core textile production data are preprocessed, including at least data cleaning, data deduplication, and feature extraction.

[0032] In step S2, the anomaly detection model is created based on the isolated forest algorithm to identify abnormal data that exceeds the preset configuration range (such as yarn threading time > 30 minutes, machine speed < 500 rpm). The abnormal data is marked as special operation data and filtered to avoid abnormal data affecting production scheduling and reporting accuracy.

[0033] In step S3, the capacity prediction model is built based on an LSTM network and is used to output a textile capacity prediction result carrying the machine output (unit: meters) and equipment failure probability for the next 24 hours based on the input machine speed, actual running time, material loss rate and employee operating efficiency. The capacity prediction model is trained based on historical data from the past six months (trained by machine type to ensure prediction accuracy). The textile production scheduling plan list contains a preset number of textile production scheduling plans; each textile production scheduling plan includes at least a machine allocation plan (machine-order correspondence table, including production start / end time), a material requirement list (if material inventory is insufficient, an inventory warning will be automatically triggered), a production sequence, and employee binding relationships (employee shift schedule).

[0034] In step S4, the production scheduling optimization model is constructed based on a genetic algorithm; The fitness function (multi-objective optimization function) of the production scheduling optimization model is formulated as follows: F = W1 × F1 + W2 × F2 + W3 × F3; F1 = Σ(Actual delivery time - Required delivery time) / Total number of orders; F2 = 1 - Σ(running time of a single machine - average running time)² / total number of machines; F3 = Σ(Available inventory quantity - Production demand quantity) / Total material types; Where F represents adaptability; F1 represents order delay rate; F2 represents machine load balance; F3 represents material matching degree; W1, W2, and W3 are all weighting coefficients, with values ​​of 40%, 30%, and 30%, respectively. The optimization process of the production scheduling optimization model is as follows: the textile production scheduling scheme list is initialized as a population, the fitness is calculated based on the fitness function, and a preset proportion of particles (e.g., retaining the top 30% of schemes / particles in fitness) are selected from the population using a roulette wheel algorithm based on the fitness. For each selected particle, crossover operation (50% probability of exchanging machine allocation segments, such as machine 1 of order A and machine 2 of order B exchanging production times) and mutation operation (10% probability of randomly adjusting machine allocation or production time ±1 hour) are performed sequentially to generate a new population (merging the retained parent schemes with the crossover and mutation-induced child schemes). It is then determined whether the preset termination condition is met. If yes, the optimal production scheduling scheme is output; otherwise, the population iteration continues. The particle refers to the textile production scheduling scheme; the optimal production scheduling scheme is the best textile production scheduling scheme (for example, allocating rush orders to machines with a failure probability of <5%, and simultaneously matching the yarn threading worker's shift time to ensure human-machine collaboration). The termination condition is that the number of iterations is ≥100, or the fitness is <0.12; Step S5 specifically involves: The optimal production scheduling plan is executed, and based on the core textile production data generated after execution, statistical indicators are calculated, including at least the total daily output (total meters = ∑ cumulative meters produced by each shift), machine efficiency (efficiency = actual running time / working hours), year-on-year production difference data (compared to the same day last week), and month-on-month production difference data (compared to the previous day). A Word template engine is used to generate trend charts and optimization suggestions based on these statistical indicators (e.g., "Machine B efficiency decreased by 8% year-on-year; it is recommended to check the equipment lubrication status"). An editable textile production report is generated and displayed based on these statistical indicators, trend charts, and optimization suggestions. In specific implementation, the textile production report is optimized based on report configuration rules; for example, the textile production report should include the average machine efficiency for the day, the top 3 employee outputs, and month-on-month data on yarn threading time.

[0035] A preferred embodiment of the textile production human-machine efficiency optimization system integrating multiple AI algorithms according to the present invention includes the following modules: The core textile production data acquisition module is used to collect core textile production data in real time, including textile order data, textile equipment data, and employee data, and to preprocess each of the core textile production data. An anomaly detection and filtering module is used to perform anomaly detection and filtering on the pre-processed core textile production data using a pre-trained anomaly detection model. The textile production scheduling module is used to input the filtered core textile production data into a pre-trained capacity prediction model to obtain textile capacity prediction results. Based on the textile capacity prediction results and preset scheduling configuration rules (such as "machine A daily running time ≤ 16 hours", "order priority: rush order > regular order"), a list of textile production scheduling schemes is generated. The production scheduling optimization module is used to input the list of textile production scheduling schemes into a pre-trained production scheduling optimization model to obtain the optimal production scheduling scheme; The textile production report generation module is used to execute the optimal production scheduling plan, calculate statistical indicators based on the core textile production data generated after execution, and generate and display a textile production report based on each statistical indicator using a Word template engine.

[0036] The core textile production data acquisition module is specifically used for: Real-time collection of core textile production data, including textile order data, textile equipment data, and employee data; the textile order data includes at least customer order information and product BOM data, the customer order information includes at least product model, production quantity, and delivery date, and the product BOM data includes at least cotton yarn usage (calculated using the formula: weight = meters × width × grammage / 1000); the textile equipment data includes at least machine speed (unit: revolutions per minute), actual running time, downtime records (including classification labels such as spinning, yarn threading, and maintenance), and material loss rate; the employee data includes at least employee clock-in time (bound to the machine), production meters (automatically accumulated across shifts), and employee operating efficiency; The core textile production data are preprocessed, including at least data cleaning, data deduplication, and feature extraction.

[0037] In the anomaly detection and filtering module, the anomaly detection model is created based on the isolated forest algorithm. It is used to identify abnormal data that exceeds the preset configuration range (such as yarn threading time > 30 minutes, machine speed < 500 rpm), mark the abnormal data as special operation data and filter it to avoid abnormal data affecting production scheduling and reporting accuracy.

[0038] In the textile production scheduling module, the capacity prediction model is built based on an LSTM network. It is used to output a textile capacity prediction result carrying the machine output (unit: meters) and equipment failure probability for the next 24 hours based on the input machine speed, actual running time, material loss rate and employee operating efficiency. The capacity prediction model is trained based on historical data from the past six months (trained by machine type to ensure prediction accuracy). The textile production scheduling plan list contains a preset number of textile production scheduling plans; each textile production scheduling plan includes at least a machine allocation plan (machine-order correspondence table, including production start / end time), a material requirement list (if material inventory is insufficient, an inventory warning will be automatically triggered), a production sequence, and employee binding relationships (employee shift schedule).

[0039] In the production scheduling optimization module, the production scheduling optimization model is constructed based on a genetic algorithm; The fitness function (multi-objective optimization function) of the production scheduling optimization model is formulated as follows: F = W1 × F1 + W2 × F2 + W3 × F3; F1 = Σ(Actual delivery time - Required delivery time) / Total number of orders; F2 = 1 - Σ(running time of a single machine - average running time)² / total number of machines; F3 = Σ(Available inventory quantity - Production demand quantity) / Total material types; Where F represents adaptability; F1 represents order delay rate; F2 represents machine load balance; F3 represents material matching degree; W1, W2, and W3 are all weighting coefficients, with values ​​of 40%, 30%, and 30%, respectively. The optimization process of the production scheduling optimization model is as follows: the textile production scheduling scheme list is initialized as a population, the fitness is calculated based on the fitness function, and a preset proportion of particles (e.g., retaining the top 30% of schemes / particles in fitness) are selected from the population using a roulette wheel algorithm based on the fitness. For each selected particle, crossover operation (50% probability of exchanging machine allocation segments, such as machine 1 of order A and machine 2 of order B exchanging production times) and mutation operation (10% probability of randomly adjusting machine allocation or production time ±1 hour) are performed sequentially to generate a new population (merging the retained parent schemes with the crossover and mutation-induced child schemes). It is then determined whether the preset termination condition is met. If yes, the optimal production scheduling scheme is output; otherwise, the population iteration continues. The particle refers to the textile production scheduling scheme; the optimal production scheduling scheme is the best textile production scheduling scheme (for example, allocating rush orders to machines with a failure probability of <5%, and simultaneously matching the yarn threading worker's shift time to ensure human-machine collaboration). The termination condition is that the number of iterations is ≥100, or the fitness is <0.12; The textile production report generation module is specifically used for: The optimal production scheduling plan is executed, and based on the core textile production data generated after execution, statistical indicators are calculated, including at least the total daily output (total meters = ∑ cumulative meters produced by each shift), machine efficiency (efficiency = actual running time / working hours), year-on-year production difference data (compared to the same day last week), and month-on-month production difference data (compared to the previous day). A Word template engine is used to generate trend charts and optimization suggestions based on these statistical indicators (e.g., "Machine B efficiency decreased by 8% year-on-year; it is recommended to check the equipment lubrication status"). An editable textile production report is generated and displayed based on these statistical indicators, trend charts, and optimization suggestions. In specific implementation, the textile production report is optimized based on report configuration rules; for example, the textile production report should include the average machine efficiency for the day, the top 3 employee outputs, and month-on-month data on yarn threading time.

[0040] In summary, the advantages of this invention are as follows: 1. The system collects core textile production data in real time, including textile order data, textile equipment data, and employee data, and preprocesses each core data point. Then, an anomaly detection model is used to detect and filter anomalies in the preprocessed core textile production data. The filtered core textile production data is then input into a capacity prediction model to obtain textile capacity prediction results. Based on the textile capacity prediction results and preset production scheduling rules, a textile production scheduling plan list is generated. This list is then input into a pre-trained production scheduling optimization model to obtain the optimal production scheduling plan. Finally, the optimal production scheduling plan is executed, and statistical indicators are calculated based on the core textile production data generated after execution. A textile production report is then generated and displayed using a Word template engine based on these statistical indicators. In short, by collecting and preprocessing order, equipment, and employee data in real time, the system first uses the isolated forest algorithm to detect anomalies. The data is filtered to ensure quality; then, the clean data is input into the LSTM capacity forecasting model to proactively estimate equipment output and failure risks; based on this forecast, a genetic algorithm with order on-time performance, equipment load balancing, and material matching as core objectives is used to optimize the production scheduling plan for multiple objectives, generating the optimal production scheduling plan and executing it; finally, based on the execution feedback data, a Word template engine is used to quickly generate a comprehensive production report containing trend analysis and optimization suggestions; this process realizes a complete closed loop from data perception, intelligent decision-making to execution feedback, not only upgrading the production scheduling basis from static experience to dynamic data-driven, but also transforming report generation from a time-consuming manual operation into an instantaneous automated process, ultimately achieving intelligent production scheduling, automated report generation, and closed-loop data management, thereby effectively improving equipment utilization, shortening report generation time, and breaking down data barriers.

[0041] 2. By collecting textile order data, equipment data, and employee data in real time and performing preprocessing (such as data cleaning and feature extraction), the comprehensiveness and real-time nature of the data sources are ensured. By integrating multi-dimensional data (such as order delivery dates, machine operating status, and employee efficiency), a high-precision production data foundation can be built, thus providing reliable input for subsequent AI analysis. This fusion processing method is superior to traditional single-data source methods, reducing information silos and improving the accuracy and response speed of overall decision-making. In particular, it can effectively avoid errors caused by data lag in dynamic textile production environments.

[0042] 3. An anomaly detection model based on the isolated forest algorithm is adopted to automatically identify and filter abnormal data that exceeds the preset range (such as equipment downtime or employee operation deviations). This algorithm is good at processing high-dimensional data and can quickly detect rare but high-impact anomalies, thereby improving data cleanliness. Compared with manual inspection or simple threshold methods, it reduces the need for human intervention, improves the automation level of production efficiency monitoring, ensures that subsequent prediction and optimization models run on high-quality data, and reduces the risk of decision-making errors caused by data noise.

[0043] 4. The capacity forecasting model is built on an LSTM network and uses historical data (such as machine speed and employee efficiency over the past six months) to predict output and equipment failure probability for the next 24 hours. LSTM is good at processing time series data and can capture long-term dependencies, making the forecast results more in line with actual production fluctuations. This helps companies adjust resources in advance to avoid overcapacity or undercapacity, thereby optimizing inventory and delivery cycles. Compared with traditional statistical methods, it improves the robustness and adaptability of forecasts and provides a scientific basis for generating production scheduling plans.

[0044] 5. A production scheduling optimization model is constructed using a genetic algorithm. The fitness function comprehensively considers order delay rate, machine load balance, and material matching degree (with weights of 40%, 30%, and 30%, respectively), and searches for the optimal solution iteratively. This multi-objective optimization method can balance conflicting objectives (such as delivery time and resource utilization) and generate a globally optimal production scheduling plan. The parallel search characteristic of the genetic algorithm avoids local optima, improves the quality of solutions and convergence speed. Compared with manual scheduling or single-rule optimization, it significantly improves the flexibility and efficiency of production scheduling and reduces operating costs.

[0045] 6. Based on the execution results, the system automatically calculates statistical indicators (such as daily output and machine efficiency) and generates trend charts and optimization suggestions through a Word template engine. This end-to-end automated process reduces the time spent on manual summarization and report preparation, while providing visual insights to help managers quickly identify problems and adjust strategies. This reflects the system's practicality and user-friendliness, transforming AI output into actionable knowledge and enhancing the transparency and continuous improvement capabilities of production management.

[0046] 7. Seamlessly integrating multiple AI algorithms (such as Isolation Forest, LSTM, and Genetic Algorithm) into a single process, forming a closed loop from data collection to report presentation, this modular design allows for adjusting model parameters or adding new data sources according to different production needs, exhibiting excellent scalability. For example, the weights of the fitness function are customizable, easily adapting to the priorities of different textile enterprises. This systematic innovation not only improves human-machine efficiency but also lays the foundation for future technological upgrades (such as the introduction of deep learning or the Internet of Things), enhancing the long-term value of the solution.

[0047] 8. By integrating real-time acquisition of multi-source data, anomaly detection based on the isolated forest algorithm, capacity prediction using LSTM networks, and multi-objective production scheduling optimization using genetic algorithms, a closed-loop intelligent decision-making system from data preprocessing to report generation was constructed. Its core advantage lies in significantly improving the automation level and resource utilization efficiency of textile production: Through the collaboration of high-quality data foundation and AI model, the system achieves accurate capacity prediction, load-balanced production scheduling, and rapid response to anomalies. Ultimately, it drives continuous optimization with automated reports, thereby reducing the cost of human intervention, effectively shortening the delivery cycle, and improving the overall human-machine collaboration efficiency.

[0048] 9. Through a full-chain design of "data-algorithm-configuration-execution," significant optimizations are achieved compared to existing technologies: Production scheduling efficiency is greatly improved: Relying on LSTM capacity prediction and genetic algorithm optimization, production scheduling time is reduced from 2 hours / time of traditional manual scheduling to 5 minutes / time, improving efficiency by 91.7%; Simultaneously, because the textile production scheduling solution considers both equipment status and order priority, the on-time delivery rate of orders is improved by 30%, effectively solving the problems of capacity waste and delivery delays; Report timeliness is significantly improved: Real-time data collection and automatic report generation enable daily report generation in real time, with a lag time of ≤10 minutes, completely solving the lag problem of traditional "next-day generation" reports, supporting management to dynamically adjust production strategies and improve decision-making response speed; Data value is fully released: By linking order, equipment, and employee data through AI algorithms (breaking data silos), and combining trend analysis to accurately identify production bottlenecks (such as a 15% decrease in efficiency for a certain machine model every Wednesday), it can directly guide equipment maintenance planning and employee skills training, forming a "data-decision-execution" closed loop, driving the transformation of production management from "experience-driven" to "data-driven."

[0049] While specific embodiments of the present invention have been described above, those skilled in the art should understand that the specific embodiments described are merely illustrative and not intended to limit the scope of the present invention. Equivalent modifications and variations made by those skilled in the art in accordance with the spirit of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for optimizing human-machine efficiency in textile production by integrating multiple AI algorithms, characterized in that: Includes the following steps: Step S1: Collect core textile production data in real time, including textile order data, textile equipment data, and employee data, and preprocess each of the core textile production data. Step S2: Perform anomaly detection and filtering on the pre-processed core textile production data using a pre-trained anomaly detection model; Step S3: Input the filtered core textile production data into the pre-trained capacity prediction model to obtain textile capacity prediction results, and generate a textile production scheduling scheme list based on the textile capacity prediction results and the preset production scheduling configuration rules. Step S4: Input the list of textile production scheduling schemes into the pre-trained production scheduling optimization model to obtain the optimal production scheduling scheme; Step S5: Execute the optimal production scheduling plan, calculate statistical indicators based on the core textile production data generated after execution, and generate and display a textile production report based on each statistical indicator using a Word template engine.

2. The method for optimizing human-machine efficiency in textile production by integrating multiple AI algorithms as described in claim 1, characterized in that: Step S1 specifically involves: Real-time collection of core textile production data, including textile order data, textile equipment data, and employee data; the textile order data includes at least customer order information and product BOM data, the customer order information includes at least product model, production quantity, and delivery date, and the product BOM data includes at least cotton yarn usage; the textile equipment data includes at least machine speed, actual running time, downtime records, and material loss rate; the employee data includes at least employee clock-in time, production meters, and employee operating efficiency. The core textile production data are preprocessed, including at least data cleaning, data deduplication, and feature extraction.

3. The method for optimizing human-machine efficiency in textile production by integrating multiple AI algorithms as described in claim 1, characterized in that: In step S2, the anomaly detection model is created based on the isolated forest algorithm and is used to identify abnormal data that exceeds the preset configuration range, mark the abnormal data as special operation data, and filter it.

4. The method for optimizing human-machine efficiency in textile production by integrating multiple AI algorithms as described in claim 1, characterized in that: In step S3, the capacity prediction model is built based on an LSTM network and is used to output a textile capacity prediction result carrying the machine output and equipment failure probability for the next 24 hours based on the input machine speed, actual running time, material loss rate and employee operating efficiency. The capacity forecasting model is trained based on historical data from the past six months; The textile production scheduling plan list contains a preset number of textile production scheduling plans; each textile production scheduling plan includes at least a machine allocation plan, a material requirement list, a production sequence, and employee binding relationships.

5. The method for optimizing human-machine efficiency in textile production by integrating multiple AI algorithms as described in claim 1, characterized in that: In step S4, the production scheduling optimization model is constructed based on a genetic algorithm; The fitness function of the production scheduling optimization model is formulated as follows: F = W1 × F1 + W2 × F2 + W3 × F3; F1 = Σ(Actual delivery time - Required delivery time) / Total number of orders; F2 = 1 - Σ(running time of a single machine - average running time)² / total number of machines; F3 = Σ(Available inventory quantity - Production demand quantity) / Total material types; Where F represents adaptability; F1 represents order delay rate; F2 represents machine load balance; F3 represents material matching degree; W1, W2, and W3 are all weighting coefficients, with values ​​of 40%, 30%, and 30%, respectively. The optimization process of the production scheduling optimization model is as follows: the textile production scheduling scheme list is initialized as a population, the fitness is calculated based on the fitness function, a preset proportion of particles are selected from the population based on the fitness using the roulette wheel algorithm, crossover and mutation operations are performed on each selected particle in sequence to generate a new population, and it is determined whether the preset termination condition is met. If yes, the optimal production scheduling scheme is output; otherwise, the population iteration continues. The particle refers to the textile production scheduling scheme; the optimal production scheduling scheme is the best textile production scheduling scheme. The termination condition is that the number of iterations is ≥100, or the fitness is <0.12; Step S5 specifically involves: The optimal production scheduling plan is executed, and statistical indicators, including at least the total daily output, machine efficiency, year-on-year production difference data, and month-on-month production difference data, are calculated based on the core textile production data generated after execution. Trend charts and optimization suggestions are generated based on each of the statistical indicators using a Word template engine. A textile production report is generated and displayed based on each of the statistical indicators, trend charts, and optimization suggestions.

6. A textile production human-machine efficiency optimization system integrating multiple AI algorithms, characterized in that: Includes the following modules: The core textile production data acquisition module is used to collect core textile production data in real time, including textile order data, textile equipment data, and employee data, and to preprocess each of the core textile production data. An anomaly detection and filtering module is used to perform anomaly detection and filtering on the pre-processed core textile production data using a pre-trained anomaly detection model. The textile production scheduling module is used to input the filtered core textile production data into a pre-trained capacity prediction model to obtain textile capacity prediction results, and generate a list of textile production scheduling schemes based on the textile capacity prediction results and preset scheduling configuration rules. The production scheduling optimization module is used to input the list of textile production scheduling schemes into a pre-trained production scheduling optimization model to obtain the optimal production scheduling scheme; The textile production report generation module is used to execute the optimal production scheduling plan, calculate statistical indicators based on the core textile production data generated after execution, and generate and display a textile production report based on each statistical indicator using a Word template engine.

7. The textile production human-machine efficiency optimization system integrating multiple AI algorithms as described in claim 6, characterized in that: The core textile production data acquisition module is specifically used for: Real-time collection of core textile production data, including textile order data, textile equipment data, and employee data; the textile order data includes at least customer order information and product BOM data, the customer order information includes at least product model, production quantity, and delivery date, and the product BOM data includes at least cotton yarn usage; the textile equipment data includes at least machine speed, actual running time, downtime records, and material loss rate; the employee data includes at least employee clock-in time, production meters, and employee operating efficiency. The core textile production data are preprocessed, including at least data cleaning, data deduplication, and feature extraction.

8. The textile production human-machine efficiency optimization system integrating multiple AI algorithms as described in claim 6, characterized in that: In the anomaly detection and filtering module, the anomaly detection model is created based on the isolated forest algorithm and is used to identify abnormal data that exceeds the preset configuration range, mark the abnormal data as special operation data, and filter it.

9. The textile production human-machine efficiency optimization system integrating multiple AI algorithms as described in claim 6, characterized in that: In the textile production scheduling module, the capacity prediction model is built based on an LSTM network. It is used to output a textile capacity prediction result carrying the machine output and equipment failure probability for the next 24 hours based on the input machine speed, actual running time, material loss rate and employee operating efficiency. The capacity forecasting model is trained based on historical data from the past six months; The textile production scheduling plan list contains a preset number of textile production scheduling plans; each textile production scheduling plan includes at least a machine allocation plan, a material requirement list, a production sequence, and employee binding relationships.

10. The textile production human-machine efficiency optimization system integrating multiple AI algorithms as described in claim 6, characterized in that: In the production scheduling optimization module, the production scheduling optimization model is constructed based on a genetic algorithm; The fitness function of the production scheduling optimization model is formulated as follows: F = W1 × F1 + W2 × F2 + W3 × F3; F1 = Σ(Actual delivery time - Required delivery time) / Total number of orders; F2 = 1 - Σ(running time of a single machine - average running time)² / total number of machines; F3 = Σ(Available inventory quantity - Production demand quantity) / Total material types; Where F represents adaptability; F1 represents order delay rate; F2 represents machine load balance; F3 represents material matching degree; W1, W2, and W3 are all weighting coefficients, with values ​​of 40%, 30%, and 30%, respectively. The optimization process of the production scheduling optimization model is as follows: the textile production scheduling scheme list is initialized as a population, the fitness is calculated based on the fitness function, a preset proportion of particles are selected from the population based on the fitness using the roulette wheel algorithm, crossover and mutation operations are performed on each selected particle in sequence to generate a new population, and it is determined whether the preset termination condition is met. If yes, the optimal production scheduling scheme is output; otherwise, the population iteration continues. The particle refers to the textile production scheduling scheme; the optimal production scheduling scheme is the best textile production scheduling scheme. The termination condition is that the number of iterations is ≥100, or the fitness is <0.12; The textile production report generation module is specifically used for: The optimal production scheduling plan is executed, and statistical indicators, including at least the total daily output, machine efficiency, year-on-year production difference data, and month-on-month production difference data, are calculated based on the core textile production data generated after execution. Trend charts and optimization suggestions are generated based on each of the statistical indicators using a Word template engine. A textile production report is generated and displayed based on each of the statistical indicators, trend charts, and optimization suggestions.

Images