Array type steam heat setting process optimization method based on machine learning

Abstract

The invention discloses an array type steam heat setting process optimization method based on machine learning. The method comprises the following steps: step 1, constructing a disturbance voxel map; 2, constructing a regional disturbance response matrix; 3, generating a candidate micro-intervention control unit set; 4, generating an area control intention vector based on the current cloth operation state; step 5, executing auction scheduling, bidding each region according to the control intention vector, determining a bid-winning region by adopting a Veckie auction mechanism, and constructing a bid-winning control map; sixthly, strategy re-parameter sampling is carried out according to the bid winning control atlas; 7, when the cloth cover is in an abnormal state, backtracking the bid-winning control atlas, and extracting an intervention control unit sequence; and step 8, adjusting the regional control intention vector and the bidding sequence based on the intervention control unit sequence, and forming a new bid winning control atlas. According to the method, voxel modeling and a Vicrie auction mechanism are fused, and intelligent optimization of the array type steam heat setting process is achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of textile heat setting process optimization, and particularly relates to an array type steam heat setting process optimization method based on machine learning. BACKGROUND

[0002] With the continuous growth of market demand for high-end functional fabrics, differentiated fiber products and precision heat treated fabrics, array type steam heat setting equipment is increasingly widely used in high efficiency, wide width, high uniformity heat setting process. However, the existing steam heat setting control system mainly relies on static empirical parameters, regional threshold strategy or traditional single optimization algorithm to adjust the steam behavior in the chamber, and the following problems generally exist under complex working conditions: The existing control method lacks a systematic modeling and feedback adjustment mechanism for the thermal and moisture disturbance coupling effect between multiple chamber steam behaviors, resulting in problems such as uneven tension, heat spot diffusion abnormalities and pattern deformation of the cloth after multiple regional superimposed intervention; the heat disturbance propagation path between steam chambers is complex, and the disturbance response has obvious hysteresis and nonlinear characteristics, making it difficult to achieve global optimization through traditional independent regulation in units of regions; dynamic competition and differentiated response capability metrics are not introduced in the intervention resource allocation process, resulting in insufficient steam response in high demand areas, while intervention in low load areas is excessive, and resource use efficiency is low; the control strategy adjustment relies on fixed rules or optimization objectives constructed based on empirical weights, lacks a data-driven feedback correction mechanism, and is difficult to adapt to actual production scenarios such as batch differences, fabric thickness changes or fluctuations in operating state. In addition, the current method often relies on post-mortem machine shutdown adjustment or experience recovery strategy to handle abnormal conditions, and cannot achieve rapid positioning of abnormal sources and precise tracing of responsible intervention units, thereby affecting the stability of machine running efficiency and setting quality and restricting the development of array type steam heat setting process towards high precision, self-adaptation and intelligence.

[0003] Therefore, how to provide an array type steam heat setting process optimization method based on machine learning is a problem that those skilled in the art need to solve. SUMMARY

[0004] One object of the present application is to provide an array type steam heat setting process optimization method based on machine learning, which integrates voxel modeling and Vickrey auction mechanism to realize intelligent optimization of the whole process of array type steam heat setting process. By constructing a disturbance voxel map to accurately capture the dynamic coupling relationship of steam intervention between multiple chambers, generating an efficient and controllable micro-intervention control unit, and using Vickrey auction to realize differentiated allocation of intervention resources and optimal bidding strategy, the cloth setting quality, control accuracy and system resource utilization efficiency are significantly improved, and the method is suitable for intelligent steam control tasks under complex fabrics and variable working conditions.

[0005] According to an embodiment of the present application, a machine learning-based array steam heat setting process optimization method comprises the following steps: Step 1: divide each steam chamber along the cloth running path in the array steam heat setting equipment into three-dimensional voxel units in space and time dimensions, and construct a perturbation voxel atlas; Step 2: perform sequential perturbation excitation, sequentially activate the steam control unit of each steam chamber, obtain the tension response and thermal image response of adjacent and non-adjacent steam chambers, and construct a regional perturbation response matrix; Step 3: discretize the control behavior of each steam chamber into a plurality of micro-intervention control units, each micro-intervention control unit includes a control time slice, a steam opening degree level, and a corresponding three-dimensional voxel unit index, and generates a candidate micro-intervention control unit set in combination with the perturbation voxel atlas and the regional perturbation response matrix; Step 4: generate a regional control intention vector based on the current cloth running state; Step 5: perform auction scheduling for each candidate micro-intervention control unit, each region bids according to the control intention vector, and determines the winning region using the Vickrey auction mechanism to construct a winning control atlas; Step 6: according to the winning control atlas, perform strategy re-parameter sampling to generate a control strategy for controlling the behavior of the steam chamber; Step 7: when an abnormal state occurs on the cloth surface, according to the index mapping relationship of the abnormal region in the perturbation voxel atlas, backtrack the corresponding winning control atlas, and extract the intervention control unit sequence that causes the abnormality; Step 8: based on the intervention control unit sequence, adjust the regional control intention vector and the corresponding bidding sequence, and re-execute the resource auction scheduling to form a new winning control atlas.

[0006] Optionally, the perturbation voxel atlas is constructed as follows: The steam chambers on the cloth running path in the array steam heat setting equipment are divided into a plurality of regions, each region is divided along the cloth running direction by a set length, and the state of each region is recorded in the time dimension at a set sampling period to form a three-dimensional voxel unit with spatial position index and time index; The data fields in the three-dimensional voxel unit include steam valve opening and closing state, steam opening degree value, tension sensor reading value, infrared thermal imaging temperature value, humidity sensor data, and cloth surface image texture feature value; All three-dimensional voxel units are organized according to spatial region number and time sequence to form a perturbation voxel atlas with three-dimensional coordinate structure.

[0007] Optionally, the step 2 is specifically as follows: The execution sequence perturbation excitation independently activates the steam control unit of each steam chamber in the array type steam heat setting device in turn, and only opens the steam valve of the current target steam chamber and keeps the steam valves of the remaining chambers closed during each activation process. The steam excitation lasts for a preset time; During each steam excitation process, the tension values at a plurality of time points from the current excitation time are collected through the tension sensors arranged at the inlet, center and outlet positions of each steam chamber, and the tension change amounts in the corresponding tension channels of adjacent and non-adjacent steam chambers are calculated to form the multi-region tension response corresponding to the excitation event; The thermal image sequence of the cloth in each region is collected through the infrared thermal imaging device installed above and on the side wall of each steam chamber, the temperature gradient change, thermal spot diffusion range and maximum temperature rise position of each region before and after steam excitation are extracted to form the thermal image response corresponding to adjacent and non-adjacent steam chambers; Based on the tension response and thermal image response data of all regions obtained after each steam chamber excitation, a region perturbation response matrix is constructed according to the index combination of the excitation source steam chamber and the response target steam chamber. Each matrix element of the region perturbation response matrix corresponds to the perturbation influence of one steam chamber on another steam chamber region, which is used to quantify the thermal and humid perturbation coupling relationship between each steam chamber in the array structure.

[0008] Optionally, the step three is specifically: The complete steam control cycle of each steam chamber is divided into a plurality of non-overlapping time segments, and the time length of the time segment is a preset control time interval; For each time segment, a plurality of steam opening degree levels are set, the steam opening degree level is a discrete numerical set divided by 10% increments, including 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90% and 100%, to form a steam action parameter combination set in each time segment; For each steam chamber, under the combination of each time segment and steam opening degree level, combined with the position index of the steam chamber in space, a corresponding micro-intervention control unit is formed. Each micro-intervention control unit binds a unique control time segment, a steam opening degree level and a three-dimensional voxel unit index in the perturbation voxel atlas; For all micro-intervention control units, according to the perturbation influence in the region perturbation response matrix, the tension response and thermal image response generated by each micro-intervention control unit in the corresponding voxel are retrieved, and the perturbation propagation score is generated by weighted linear combination value; Ranking all the micro-intervention control units according to the disturbance propagation scores, eliminating the micro-intervention control units below a preset disturbance influence threshold, and clustering and integrating by three-dimensional voxel unit indexes to form a candidate micro-intervention control unit set, which is used as a resource bidding object in an auction scheduling process.

[0009] Optionally, the step four of generating a regional control intention vector based on the current cloth running state is specifically: For each steam chamber in the array type steam heat setting equipment, a control intention vector of the region is generated based on the current running state data of the cloth of the corresponding region; The current running state data includes a difference value between a real-time tension value collected by a regional tension sensor and a historical tension baseline value, a cloth temperature change gradient value collected by an infrared thermal imaging device, a cloth surface defect proportion detected by a cloth surface image recognition module, and a disturbance lag weight historically executed by the region in a disturbance voxel graph; After standardizing the current running state data, a regional control intention vector is constructed by combination, the control intention vector is a set of real value numerical vectors, including a tension deviation amount, a temperature change gradient value, a cloth surface defect proportion, and a disturbance lag weight.

[0010] Optionally, the auction scheduling is performed on each candidate micro-intervention control unit, each region bids according to the control intention vector, and a Vickrey auction mechanism is used to determine a winning region, and the method is specifically: Each candidate micro-intervention control unit is set as an independent auction unit, each auction unit includes a unique steam chamber number, a control time period, a steam opening degree level, and a three-dimensional voxel unit index; A Vickrey auction execution instance is initialized for each auction unit, the Vickrey auction execution instance is configured as a single round, a closed type, and a second highest price payment rule, after the auction starts, all participating regions submit a bid value of the auction unit in a non-interactive state, and modification or withdrawal is prohibited; The bid value of the auction unit is input into a regional bid calculation module to generate, the regional bid calculation module combines the tension deviation amount, the temperature change gradient value, the cloth surface defect proportion, and the disturbance lag weight into a floating point scalar value according to a preset weight, which is used as a bid input of the region to the current candidate micro-intervention control unit; In each auction unit, the bid values of all auction units are sorted, the region with the highest bid value is determined as the winning region, the bid value of the region with the second highest bid value is used as the payment price of the winning region, and a unique assignment relationship between the winning region and the candidate micro-intervention control unit is recorded.

[0011] Optionally, the winning control graph is constructed, and the method is specifically: Collecting all the candidate micro-intervention control units of the winning regions into a winning control unit set, sorting according to the time index and the voxel space index, and constructing a structured winning control atlas; the winning control atlas is a two-dimensional mapping structure, the time dimension corresponds to the control time period, the space dimension corresponds to the three-dimensional voxel position, and each cell of the winning control atlas includes a steam chamber number, a steam opening degree value and a winning region number.

[0012] Optionally, the strategy reparameterization sampling extracts all control time periods from the winning control atlas, constructs a dynamic control time sequence index, takes each control time period as an independent sampling window, and initializes a control parameter template according to the winning control unit configuration in each sampling window; In each sampling window, the strategy reparameterization sampling process is performed, including: extracting the steam chamber number and the corresponding spatial position of the winning control unit to lock the current control region to be sampled; extracting the winning steam opening degree level as the initial value of the control strength; inputting the disturbance voxel state feature into the strategy adjustment model based on the disturbance voxel state in the current sampling window to adjust the steam opening degree level and the time offset to form the final action parameter, generate a control strategy for controlling the behavior of the steam chamber, and the strategy adjustment model uses a parameter generator based on supervised training, the input is the disturbance voxel state feature, and the output is the disturbance adjustment value of the control action parameter, which is used to disturb the initial value of the winning control strength to realize customized control; The control strategy consists of four tuples of steam chamber number, control time period, actual execution steam opening degree level and corresponding three-dimensional voxel index, and is sent to the execution end of the array type steam heat setting equipment as a control instruction set to drive the corresponding steam chamber to perform steam intervention behavior according to the winning order and the final action parameter.

[0013] Optionally, step seven is specifically: During the operation of the array type steam heat setting equipment, the abnormal state of the cloth surface is detected in real time, and when the cloth surface is abnormal, the specific spatial position of the cloth surface abnormal area is determined through infrared thermal imaging and cloth surface image recognition; According to the spatial position of the cloth surface abnormal area and the time index of the time when the abnormality occurs, the three-dimensional voxel unit index corresponding to the abnormal area in space and time in the disturbance voxel atlas is found; Based on the three-dimensional voxel unit index as a starting point, the candidate micro-intervention control units performed on the abnormal area in the winning control atlas are searched step by step within a preset backtracking period before the abnormality occurs along the time axis in reverse; The retrieved candidate micro-intervention control units are sorted in order from near to far according to the execution time to form an intervention control unit sequence that leads to the abnormal state of the cloth surface.

[0014] Optionally, step eight is specifically: According to the intervention control unit sequence, the number and the spatial position of each candidate micro-intervention control unit in the intervention control unit sequence corresponding to the steam chamber are determined, and the number of intervention executions and the abnormal contribution degree of each area corresponding to the steam chamber before the abnormality occurs are counted; According to the counted intervention execution number and abnormal contribution degree, the tension deviation amount, temperature change gradient value, fabric defect proportion and disturbance lag weight of the corresponding area in the original area control intention vector are modified according to the preset weight; According to the modified area control intention vector, the bid value of each area for each candidate micro-intervention control unit is recalculated, and the Vickrey auction mechanism is re-executed according to the updated area bid value to determine new winning areas; The candidate micro-intervention control units of all new winning areas are summarized into a winning control unit set, sorted according to the time index and voxel space index, and a new winning control map is formed.

[0015] The beneficial effects of the present application are: The present application constructs a dynamic modeling framework combining disturbance voxel map and area disturbance response matrix, aiming at the problems of difficult quantification of multi-chamber coupling disturbance, response lag of control strategy and imbalance of resource allocation in array type steam heat setting equipment, adopts three-dimensional space-time voxel modeling technology to accurately capture the propagation path and time delay characteristics of steam excitation on tension and temperature change; in the micro-intervention control unit generation link, high-impact control candidates are screened through disturbance propagation scores, significantly compressing the control action space and improving the regulation and control accuracy; the resource competition scheduling method based on Vickrey auction mechanism is introduced, the cloth running state is coded into a control intention vector and used as the basis for each area bidding, realizing differentiated competition and adaptive control resource allocation between chambers, and avoiding high-intervention demand areas from being disturbed by low-demand areas; the strategy re-parameter sampling mechanism is introduced in the strategy generation link, the control strength and time execution position are fine-tuned through the disturbance state feature driven parameter generator, realizing dynamic fine-tuning adaptive adjustment of the control strategy; in the fabric abnormality processing process, the reverse tracking of the abnormal area responsibility control unit is realized through index mapping, and the area control intention and bidding strategy are updated retrospectively according to the abnormal contribution degree, forming a closed-loop self-correction mechanism driven by abnormality. Finally, the dynamic optimization control of cross-chamber, multi-period and regional cooperation in the array type steam heat setting process is realized, which ensures the heat setting quality of the fabric, improves the steam resource utilization efficiency and the overall system robustness. BRIEF DESCRIPTION OF DRAWINGS

[0016] The accompanying drawings are included to provide a further understanding of the application, and constitute a part of the specification, together with the embodiments of the application, to explain the application, and do not constitute a limitation of the application. In the drawings: Fig. 1A whole flow chart of an array type steam heat setting process optimization method based on machine learning is provided for the present application; Fig. 2 An auction scheduling flow chart of an array type steam heat setting process optimization method based on machine learning is provided for the present application based on a Vickrey auction mechanism; Fig. 3 An operation flow chart of a strategy re-parameter sampling generation control strategy of an array type steam heat setting process optimization method based on machine learning is provided for the present application. DETAILED DESCRIPTION

[0017] The present application will now be further described in detail with reference to the drawings. These drawings are simplified schematic diagrams and only schematically show the basic structure of the present application, and thus only show the configurations related to the present application.

[0018] REFERENCE Figs. 1-3 An array type steam heat setting process optimization method based on machine learning, comprising the following steps: Step one: divide each steam chamber along the cloth running path in the array type steam heat setting equipment into three-dimensional voxel units in space and time dimensions, and construct a perturbation voxel atlas; Step two: perform sequential perturbation excitation, sequentially activate the steam control unit of each steam chamber, obtain the tension response and thermal map response of adjacent and non-adjacent steam chambers, and construct a regional perturbation response matrix; Step three: discretize the control behavior of each steam chamber into a plurality of micro-intervention control units, each micro-intervention control unit includes a control time slice, a steam opening degree level and a corresponding three-dimensional voxel unit index, and generates a candidate micro-intervention control unit set in combination with the perturbation voxel atlas and the regional perturbation response matrix; Step four: generate a regional control intention vector based on the current cloth running state; Step five: perform auction scheduling for each candidate micro-intervention control unit, each region bids according to the control intention vector, determines the winning region by using the Vickrey auction mechanism, and constructs a winning control atlas; Step six: generate a control strategy for controlling the behavior of the steam chamber according to the winning control atlas; Step seven: when an abnormal state occurs on the cloth surface, backtrack the corresponding winning control atlas according to the index mapping relationship of the abnormal region in the perturbation voxel atlas, and extract the intervention control unit sequence that causes the abnormality; Step eight: adjust the regional control intention vector and the corresponding bidding sequence based on the intervention control unit sequence, and re-execute the resource auction scheduling to form a new winning control atlas.

[0019] In the present embodiment, the construction of the perturbation voxel atlas is specifically: The steam chambers on the fabric running path in the array steam heat setting equipment are divided into several regions, each region is divided along the fabric running direction by a set length, and the state of each region is recorded in the time dimension with a set sampling period to form a three-dimensional voxel unit with spatial position index and time index; The data field in the three-dimensional voxel unit includes steam valve opening and closing state, steam opening degree value, tension sensor reading value, infrared thermal imaging temperature value, humidity sensor data, and fabric image texture characteristic value; All three-dimensional voxel units are organized according to spatial region number and time sequence to form a perturbation voxel atlas with three-dimensional coordinate structure.

[0020] In this embodiment, step two is specifically: Sequential perturbation excitation is performed to independently activate the steam control unit of each steam chamber in the array steam heat setting equipment, in each activation process, only the steam valve of the current target steam chamber is opened, and the steam valves of the remaining chambers remain closed, and the steam excitation lasts for a preset time; In each steam excitation process, the tension sensors arranged at the inlet, center and outlet positions of each steam chamber are used to collect the fabric tension values at a plurality of time points starting from the current excitation time, and the tension change amounts in the tension channels corresponding to adjacent and non-adjacent steam chambers are calculated to form the multi-region tension response corresponding to the excitation event; The infrared thermal imaging devices installed above and on the side walls of each steam chamber are used to collect the thermal image sequences of the fabric in each region, and the temperature gradient change, thermal spot diffusion range and maximum temperature rise position of each region before and after steam excitation are extracted to form the thermal image response corresponding to adjacent and non-adjacent steam chambers; In the specific implementation process, in order to obtain the physical response information of the fabric caused by the steam excitation behavior, in each steam excitation process, the tension sensors arranged at the inlet, center and outlet positions of each steam chamber are used to collect tension change data in real time. Each steam chamber is provided with at least three tension sensors to record the fabric tension values at a plurality of time points from the start of steam excitation to the end of excitation, and the time sampling frequency is not less than 10Hz to ensure the capture of short-time perturbation effect. After the collected data is filtered and normalized, difference calculation is performed with the tension baseline value before steam excitation to obtain the response change amount of the fabric segment of the current excitation chamber and its adjacent and non-adjacent chambers in the tension dimension, thereby forming the tension response data.

[0021] In addition, in order to obtain the heat distribution change in the steam excitation process, an infrared thermal imaging device is installed above and on the side wall of each steam chamber. The device collects a sequence of thermal images of the cloth surface in real time, and the image frame rate is not less than 30 fps. Through inter-frame difference and image enhancement processing technology, the temperature change curve of each cloth area on the time axis is extracted before and after excitation. Further, through image analysis algorithm, the temperature gradient change value, hot spot diffusion area and hot spot maximum temperature rise point coordinates are calculated to construct the thermal image response characteristics, which are used to reflect the thermal disturbance behavior caused by steam excitation on the cloth surface.

[0022] The tension response and thermal image response data are synchronized and bound by time stamp and steam excitation control signal, and are mapped to the voxel index in the disturbance voxel graph according to the corresponding relationship between the chamber number and the cloth position of each sensor or imaging device. Finally, the steam excitation behavior and the tension response and thermal image response data caused by it in different chamber areas form a multi-region intervention response pair, which is used to construct a regional disturbance response matrix. The regional disturbance response matrix can quantify the coupling influence degree of any steam chamber excitation behavior on the physical state of other regions, and reflect the transverse diffusion and longitudinal conduction characteristics of steam disturbance under the array structure.

[0023] Based on the tension response and thermal image response data of all regions obtained after each steam chamber excitation, the regional disturbance response matrix is constructed according to the index combination of the excitation source steam chamber and the response target steam chamber. Each matrix element of the regional disturbance response matrix corresponds to the disturbance influence of one steam chamber on another steam chamber region, which is used to quantify the thermal and humid disturbance coupling relationship between each steam chamber in the array structure.

[0024] In the embodiment, the step three is specifically: The complete steam control cycle of each steam chamber is divided into a plurality of non-overlapping time segments, and the time length of the time segment is a preset control time interval; For each time segment, a plurality of steam opening degree levels are set, the steam opening degree level is a discrete numerical set divided by 10% increment, including 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90% and 100%, to form a steam action parameter combination set in each time segment; For each steam chamber, under the combination of each time segment and steam opening degree level, combined with the spatial position index of the steam chamber, a corresponding micro-intervention control unit is formed, each micro-intervention control unit binds a unique control time segment, steam opening degree level and three-dimensional voxel unit index in the disturbance voxel graph; For all micro-intervention control units, according to the disturbance influence in the area disturbance response matrix, the tension response and thermal map response generated by each micro-intervention control unit in the corresponding voxel are retrieved, and the disturbance propagation score is generated by weighted linear combination value; ; wherein, represents the disturbance propagation score, and a and β represent weight coefficients for adjusting the contribution proportion of the tension response and the thermal map response in the total score, represents the tension response, represents the thermal map response. All micro-intervention control units are sorted according to the disturbance propagation score, micro-intervention control units below a preset disturbance influence threshold are removed, and clustering integration is performed according to three-dimensional voxel unit index to form a candidate micro-intervention control unit set, which is used as a resource bidding object in the auction scheduling process.

[0025] In the present embodiment, the fourth step of generating an area control intention vector based on the current running state of the cloth is specifically: For each steam chamber in the array type steam heat setting device, an area control intention vector is generated based on the current running state data of the corresponding area of the cloth; The current running state data includes the difference between the real-time tension value collected by the area tension sensor and the historical tension baseline value, the cloth temperature change gradient value collected by the infrared thermal imaging device, the cloth surface defect proportion detected by the cloth surface image recognition module, and the disturbance lag weight of the area in the disturbance voxel map; The disturbance lag weight is a measure of the time delay of the physical response (tension change or temperature change) of an area after receiving steam intervention compared to the intervention action, which is used to represent the response sensitivity and control difficulty of the area. The acquisition process is as follows: in the historical running data, for each execution record of the micro-intervention control unit, the time delay experienced by the tension change or temperature change caused in the corresponding area is calculated, all historical intervention response time delays of the area are accumulated, the average lag time is calculated, and the disturbance lag weight is obtained by normalizing according to the preset maximum allowed lag time.

[0026] After standardizing the current running state data, the area control intention vector is constructed by combination, and the control intention vector is a set of real value numerical vectors, including tension deviation, temperature change gradient value, cloth surface defect proportion and disturbance lag weight, which is used to represent the intervention demand degree and urgency of the area to the micro-intervention control unit.

[0027] In the embodiment, the auction scheduling is performed on each candidate micro-intervention control unit, each region bids according to the control intention vector, the winning region is determined by using the Vickrey auction mechanism, and specifically, Each candidate micro-intervention control unit is set as an independent auction unit, each auction unit includes a unique steam chamber number, a control time period, a steam opening degree level and a three-dimensional voxel unit index; A Vickrey auction execution instance is initialized for each auction unit, the Vickrey auction execution instance is configured as a single round, a closed type and a second highest price payment rule, after the auction starts, all participating regions submit the bid value of the auction unit in a non-interactive state, and modification or withdrawal is prohibited; The bid value of the auction unit is input into the region bid calculation module by the corresponding control intention vector, and the tension deviation amount, the temperature change gradient value, the fabric defect proportion and the disturbance lag weight are combined into a floating point scalar value according to a preset weight, which is used as the bid input of the region to the current candidate micro-intervention control unit; In each auction unit, the bid values of all auction units are sorted, the region with the highest bid value is determined as the winning region, the bid value of the region with the second highest bid value is used as the payment price of the winning region, and the unique assignment relationship between the winning region and the candidate micro-intervention control unit is recorded.

[0028] In the traditional array type steam heat setting process, the intervention behavior of each steam chamber is usually controlled by static rules or adjusted by a single optimization target strategy, and the coupling relationship between complex multi-regions and the resource competition state cannot be reasonably reflected, so that the control precision is not high, the disturbance is frequent, and it is difficult to meet the demand of high-quality fabric setting process.

[0029] The present application introduces the Vickrey auction mechanism into the array type steam heat setting control process, forms a dynamic competition mechanism for micro-scale control resources by setting each candidate micro-intervention control unit as an independent auction unit, and breaks the static constraints of traditional control technology on intervention resource allocation.

[0030] Specifically, in the present application, the region control intention vector is quantitatively calculated by the current fabric tension deviation amount, the temperature change gradient value, the fabric defect proportion and the disturbance lag weight, and a variety of physical state information is integrated to form a unified region bid basis. Then, the resource competition is performed in a single round closed Vickrey auction mode, each region independently bids, and the resource use right is determined in a second highest price payment mode, which effectively avoids the problem of strategic overestimation or underestimation of demand between regions, promotes each region to objectively and accurately express its intervention demand, and significantly improves the process optimization efficiency and control stability under complex multi-zone coupling conditions.

[0031] In this embodiment, the winning control map is constructed, specifically: All candidate micro-intervention control units in the winning region are summarized into a winning control unit set, sorted according to the time index and the voxel space index, and a structured winning control map is constructed; the winning control map is a two-dimensional mapping structure, the time dimension corresponds to the control time period, the space dimension corresponds to the three-dimensional voxel position, and each cell of the winning control map includes the steam chamber number, the steam opening value, and the winning region number; The winning control map serves as a template input source for subsequent control strategies, used to define the steam intervention execution behavior of each chamber in the steam setting process and the region correspondence, ensuring that the adjustment behavior and priority order originate from the competitive allocation result under the Vickrey auction mechanism.

[0032] In this embodiment, the strategy re-parameter sampling extracts all control time periods from the winning control map, constructs a dynamic control time sequence index, takes each control time period as an independent sampling window, and initializes the control parameter template according to the winning control unit configuration in each sampling window; The strategy re-parameter sampling process is performed in each sampling window, including: extracting the steam chamber number and corresponding spatial position of the winning control unit to lock the current control region to be sampled; extracting the winning steam opening level as the initial control strength; based on the disturbance voxel state in the current sampling window, inputting the disturbance voxel state features into the strategy adjustment model to adjust the steam opening level and time offset, forming the final action parameters, generating the control strategy for controlling the behavior of the steam chamber, the strategy adjustment model uses a parameter generator based on supervised training, the input is the disturbance voxel state feature, and the output is the disturbance adjustment value of the control action parameter, used to disturb the initial winning control strength to achieve customized control; The strategy adjustment model is a set of lightweight neural network modules trained by supervised learning, partitioned according to the steam chamber number, and the strategy adjustment model structure includes an input layer, at least one hidden layer, and an output layer. The input layer receives the disturbance voxel state feature, which includes the tension gradient, heat map temperature gradient, historical response lag value, and fabric anomaly prediction index of the current voxel after standardization as the model input vector; The output of the strategy adjustment model is two continuous real values: steam opening disturbance correction and time offset adjustment, which are combined and corrected with the steam opening level and execution time of the winning control unit to generate the final parameter configuration of the control action; The control strategy is composed of four tuples: steam chamber number, control time period, actual execution steam opening level, and corresponding three-dimensional voxel index, which is sent to the execution end of the array steam setting device as a control instruction set to drive the corresponding steam chamber to perform steam intervention behavior according to the winning order and the final action parameters.

[0033] In this embodiment, step seven is specifically: In the process of running the array type steam heat setting device, the abnormal state of the cloth surface is detected in real time, and when the cloth surface is abnormal, the specific spatial position of the abnormal area of the cloth surface is determined through infrared thermal imaging and cloth surface image recognition; According to the spatial position of the abnormal area of the cloth surface and the time index when the abnormality occurs, the three-dimensional voxel unit index corresponding to the abnormal area in space and time in the disturbance voxel atlas is searched; Based on the three-dimensional voxel unit index as a starting point, the candidate micro-intervention control units performed on the abnormal area in the control atlas are searched step by step within a preset backtracking period before the abnormality occurs along the time axis in reverse; The candidate micro-intervention control units searched are sorted in order of execution time from near to far to form an intervention control unit sequence that leads to the abnormal state of the cloth surface.

[0034] In this embodiment, step eight is specifically: According to the intervention control unit sequence, the steam chamber number and spatial position corresponding to each candidate micro-intervention control unit in the intervention control unit sequence are determined, and the number of interventions performed before the abnormality occurs and the abnormal contribution degree of each steam chamber corresponding area are counted; According to the number of interventions performed and the abnormal contribution degree counted, the tension deviation amount, temperature change gradient value, cloth surface defect proportion and disturbance lag weight of the corresponding area in the original area control intention vector are modified according to the preset weight; According to the modified area control intention vector, the value of each candidate micro-intervention control unit for each area is recalculated, and the Vickrey auction mechanism is re-executed according to the updated area value to determine new winning areas; The candidate micro-intervention control units of all new winning areas are summarized into a winning control unit set, sorted according to the time index and voxel space index, and a new winning control atlas is formed.

[0035] The abnormal contribution degree is an index for quantifying the influence intensity of each steam chamber corresponding area on the cloth surface, and the specific determination method is as follows: According to the intervention control unit sequence extracted in step seven, the number of steam chambers corresponding to each intervention control unit, the steam opening degree level, the execution time point and the corresponding three-dimensional voxel unit index are obtained; Using the disturbance voxel map constructed in step one, the tension deviation amount, temperature change gradient value and fabric surface defect ratio change value of the corresponding space region after each intervention control unit is executed are traced back and calculated, and the above three numerical value change amplitudes are taken as the single contribution value of the intervention control unit to the fabric surface anomaly; For each steam chamber corresponding region, the single contribution value of all intervention control units is accumulated within the preset time range before the anomaly occurs to obtain the anomaly contribution value of the region; The sum of all region anomaly contribution values is calculated, and the anomaly contribution value of each region is normalized by dividing the sum of all region anomaly contribution values to obtain the anomaly contribution degree corresponding to the region, which represents the contribution proportion of the intervention behavior of the region to the overall anomaly; After the anomaly contribution degree is calculated, the system modifies the tension deviation amount, temperature change gradient value, fabric surface defect ratio and disturbance lag weight in the original region control intention vector of each region according to the intervention execution times and anomaly contribution degree of the region: Specifically, a set of non-negative weight coefficients is preset, and for each region, the original values of the above four indicators are linearly modified by intervention execution times and anomaly contribution degree, that is: Modified index value = original index value x (1 + intervention execution times weight x execution times proportion + anomaly contribution weight x anomaly contribution degree); Wherein, the execution times proportion represents the proportion of the intervention execution times of the region to the total number of all region interventions, and the intervention execution times weight and the anomaly contribution weight are non-negative real numbers preset according to historical data analysis experience, which are used to control the strength of the two factors; Finally, the modified tension deviation amount, temperature change gradient value, fabric surface defect ratio and disturbance lag weight constitute the new region control intention vector of the region, which is used to recalculate the bid value of the candidate micro intervention control unit in the next resource auction scheduling process.

[0036] Example 1: In order to verify the feasibility of the application in implementation, the application is applied to a certain large intelligent textile heat setting production line, which is equipped with 20 steam chambers, the chambers are arranged closely, and independent steam regulation is supported. The processed fabrics are four typical industrial fabrics, which are heavy knitted fabric, light silk fabric, elastic blended fabric and coarse cotton and linen fabric, each with a length of 120 meters. The initial tension state exists unstable fluctuation, and the fabric defect rate is higher than the industry average.

[0037] Before the experiment, the steam heat setting equipment was modified according to the requirements of the present application, including arranging tension sensors and thermal imaging devices, configuring edge computing nodes for voxel modeling and disturbance response analysis, and deploying a server for controlling auction scheduling. In the implementation process, first, the coupling influence of heat and moisture disturbance propagation between different steam chambers was obtained by constructing the disturbance voxel atlas and the regional disturbance response matrix. Subsequently, the regional control intention vector was generated according to the real-time tension of the cloth and the thermal map state.

[0038] The present application introduces the Vickrey auction mechanism to schedule steam resources. Different regions bid for candidate micro-intervention control units according to their own states, and the system automatically determines the most suitable winning region to form a dynamic winning control atlas. The control strategy is generated using a re-parameter sampling method, and the opening degree level and time offset are dynamically adjusted by a supervised training model according to the disturbance voxel state, realizing fine control.

[0039] The traditional uniform distribution strategy evenly distributes steam resources (such as steam opening degree, action time, etc.) to each steam chamber or action area. Through 4 batches of 50 cloth experiments, it is found that under the traditional uniform distribution strategy, the cloth tension fluctuation value is high, the heat setting temperature uniformity is low, and hot spots are prone to accumulate. Under the strategy of the present application, the cloth tension stability is significantly improved, the infrared thermal imaging shows that the temperature distribution is more uniform, and the dimensional stability of the finished cloth is significantly improved.

[0040] Table 1 Comparison of control strategies between traditional method and the present application

[0041] Analyzing the data in Table 1 above, from the tension average deviation data, the tension deviation of all cloth types under the control strategy of the present application is significantly lower than that under the traditional strategy. For example, the tension average deviation of heavy knitted cloth decreases from 18.4N under the traditional strategy to 9.8N, with a reduction of 46.7%; the tension average deviation of elastic blended cloth decreases from 15.6N to 7.9N, with a reduction of nearly 50%. This shows that by constructing the disturbance voxel atlas and the regional control intention vector, the present application effectively alleviates the problem of excessive tension fluctuation in traditional heat setting, and improves the stability of cloth tension distribution.

[0042] In terms of defect rate, the present application also shows superior control effect. Taking thin silk cloth as an example, the average defect rate under the traditional strategy is 7.2%, while after optimization it decreases to 4.1%, a decrease of nearly 43%; the average defect rate of coarse cotton and hemp cloth also decreases from 6.1% to 3.7%. This shows that the fine intervention mechanism based on strategy re-parameter sampling and Vickrey auction can improve the uniformity of heat setting and reduce surface defects such as wrinkles, shrinkage and edge curling caused by uneven steam excitation.

[0043] Table 2 Comparison of regional temperature control uniformity

[0044] Table 2 verifies the advantages of the Vickrey auction mechanism control strategy adopted by the present application in temperature distribution control through comparison of the temperature uniformity of five steam action regions (A~E). The data show that under the traditional uniform scheduling control strategy, the regional temperature uniformity generally maintains between 84.6% and 87.3%, while under the dynamic resource allocation method based on the Vickrey auction mechanism proposed by the present application, the regional temperature uniformity is significantly improved, with the mean value range stabilizing between 92.5% and 95.1%.

[0045] Specifically, the temperature uniformity of region A under the traditional strategy is 86.1%, while it rises to 94.6% after using the strategy of the present application, with an increase of 8.5%; regions B and D achieve increases of 7.7% and 7.9%, respectively; region C performs most outstandingly, with a temperature uniformity of 95.1% under the Vickrey strategy, an increase of 7.8% over the conventional strategy; and region E also has a significant increase of 8.4%. Overall, the temperature control uniformity of all regions increases by more than 7%, and the standard deviation is significantly reduced, showing high consistency and stability.

[0046] This fully demonstrates that the mechanism of bidding based on regional control intention vectors and determining intervention resource allocation strategies through the Vickrey auction mechanism introduced by the present application can more intelligently coordinate the intervention intensity and timing arrangement of the steam chamber, thereby realizing more precise and adaptive dynamic adjustment of the heat energy distribution in the heat setting process of the fabric, solving the problems of heat spots, heat accumulation and insufficient edge heating that are prone to occur in traditional control, and being particularly suitable for setting tasks of heterogeneous fabrics and complex texture fabrics. The implementation of this technology significantly improves the consistency and controllability of the final fabric heat treatment quality, and has extremely high popularization and application value.

[0047] This embodiment realizes intelligent regulation and control of the array type steam heat setting process by introducing the perturbation voxel graph modeling, regional control intention vector expression and Vickrey auction mechanism, effectively improving the tension control stability and temperature distribution uniformity. Compared with the traditional average allocation strategy, the method of the present application can accurately match the micro intervention control unit according to the real-time fabric state, making the steam resource allocation more targeted and flexible, avoiding the problems of heat energy waste and excessive intervention. At the same time, through the abnormal backtracking and dynamic correction mechanism of the control intention vector, it has the ability of rapid response and adaptive optimization to abnormal conditions, significantly enhancing the robustness of the system and the stability of the fabric setting. Overall, this method has the comprehensive advantages of high intelligence, high control precision, strong stability and wide adaptability, and is suitable for complex fabric types and high-quality heat setting demand scenarios.

[0048] The above merely describes preferred specific embodiments of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art, according to the technical solution and inventive concept of the present application, makes equivalent replacement or change within the technical range disclosed by the present application, which should be covered within the protection scope of the present application.

Claims

1. A machine learning-based optimization method for array-type steam thermal setting process, characterized in that, Includes the following steps: Step 1: Divide each steam chamber along the fabric running path in the array-type steam heat setting equipment into three-dimensional voxel units according to the spatial and temporal dimensions, and construct a perturbation voxel map; Step 2: Perform sequential perturbation excitation, activate the steam control unit of each steam chamber in turn, obtain the tension response and thermographic response of adjacent and non-adjacent steam chambers, and construct the regional perturbation response matrix; Step 3: Discretize the control behavior of each steam chamber into several micro-intervention control units. Each micro-intervention control unit includes a control time segment, steam opening level and corresponding three-dimensional voxel unit index. Combine the perturbation voxel map and the regional perturbation response matrix to generate a set of candidate micro-intervention control units. Step 4: Generate a region control intent vector based on the current fabric running status; Step 5: Perform auction scheduling for each candidate micro-intervention control unit. Each region bids according to the control intent vector. The winning region is determined by the Vickrey auction mechanism, and the winning control map is constructed. Step 6: Perform strategy reparameter sampling based on the winning control map to generate a control strategy for controlling the behavior of the steam chamber; Step 7: When an abnormal state occurs on the fabric surface, based on the index mapping relationship of the abnormal area in the perturbation voxel map, backtrack the corresponding indexed control map and extract the sequence of intervention and control units that caused the abnormality; Step 8: Based on the intervention control unit sequence, adjust the regional control intent vector and the corresponding bidding order, re-execute the resource auction scheduling, and form a new winning bid control map.

2. The method for optimizing an array-type steam thermal setting process based on machine learning according to claim 1, characterized in that, The construction of the perturbation voxel map specifically involves: The steam chambers along the fabric running path in the array-type steam heat setting equipment are divided into several regions. Each region is divided along the fabric running direction according to a set length, and the state of each region is recorded in the time dimension at a set sampling period to form a three-dimensional voxel unit with spatial position index and time index. The data fields in the three-dimensional voxel unit include the steam valve opening and closing status, steam opening value, tension sensor reading value, infrared thermal imaging temperature value, humidity sensor data, and fabric image texture feature value. All three-dimensional voxel units are organized according to spatial region numbering and chronological order to form a perturbation voxel map with a three-dimensional coordinate structure.

3. The method for optimizing an array-type steam thermal setting process based on machine learning according to claim 1, characterized in that, Step two specifically involves: The sequential perturbation excitation activates the steam control unit of each steam chamber in the array-type steam heat setting equipment in turn. During each activation process, only the steam valve of the current target steam chamber is opened, while the steam valves of the other chambers remain closed. The steam excitation lasts for a preset time. During each steam activation process, tension sensors placed at the inlet, center, and outlet of each steam chamber collect the fabric tension values ​​at multiple consecutive time points starting from the current activation moment, and calculate the tension changes in the corresponding tension channels of adjacent and non-adjacent steam chambers to form a multi-region tension response corresponding to the activation event. By using infrared thermal imaging devices installed above and on the side walls of each steam chamber, thermal image sequences of the fabric in each region are collected. The temperature gradient changes, hot spot diffusion range and maximum temperature rise location of each region before and after steam excitation are extracted to form thermal response corresponding to adjacent and non-adjacent steam chambers. Based on the tension response and thermographic response data of all regions obtained after each steam chamber excitation, a regional perturbation response matrix is ​​constructed by combining the indices of the excitation source steam chamber and the response target steam chamber. Each element of the regional perturbation response matrix corresponds to the perturbation effect of one steam chamber on another steam chamber region, which is used to quantify the thermal and moisture perturbation coupling relationship between each steam chamber in the array structure.

4. The method for optimizing an array-type steam thermal setting process based on machine learning according to claim 1, characterized in that, Step three specifically involves: The complete steam control cycle of each steam chamber is divided into multiple non-overlapping time segments, the duration of which is a preset control time interval; For each time segment, multiple steam opening levels are set, and the steam opening level is a discrete set of values ​​divided in increments of 10%. For each steam chamber, under the combination of time segments and steam opening level, a corresponding micro-intervention control unit is formed by combining the spatial position index of the steam chamber. Each micro-intervention control unit is bound to a unique control time segment, steam opening level, and three-dimensional voxel unit index in the perturbation voxel map. For all micro-intervention control units, based on the disturbance influence in the regional disturbance response matrix, the tension response and heatmap response generated by each micro-intervention control unit at the corresponding voxel are retrieved, and a disturbance propagation score is generated by weighted linear combination of values. All micro-intervention control units are sorted according to the disturbance propagation score, and those micro-intervention control units below the preset disturbance impact threshold are eliminated. They are then clustered and integrated according to the three-dimensional voxel unit index to form a candidate micro-intervention control unit set, which is used as the resource bidding object in the auction scheduling process.

5. The method for optimizing an array-type steam thermal setting process based on machine learning according to claim 1, characterized in that, Step four: Generating a region control intent vector based on the current fabric running state, specifically: For each steam chamber in the array-type steam heat setting equipment, a control intent vector for the corresponding area is generated based on the current operating status data of the fabric in that area. The current operating status data includes the difference between the real-time tension value collected by the regional tension sensor and the historical tension baseline value, the fabric temperature change gradient value collected by the infrared thermal imaging device, the proportion of fabric defects detected by the fabric image recognition module, and the perturbation hysteresis weight of the region in the perturbation voxel map. After standardizing the current operating status data, a regional control intent vector is constructed by combining the data. The control intent vector is a set of real-valued vectors, including tension deviation, temperature change gradient, fabric defect ratio, and disturbance hysteresis weight.

6. The method for optimizing an array-type steam thermal setting process based on machine learning according to claim 1, characterized in that, The auction scheduling of each candidate micro-intervention control unit, with each region bidding according to its control intent vector, and the winning region determined using the Vickrey auction mechanism, is specifically as follows: Each candidate micro-intervention control unit is set as an independent auction unit, and each auction unit contains a unique steam chamber number, control time period, steam opening level and three-dimensional voxel unit index. A Vickery auction execution instance is initialized for each auction unit. The Vickery auction execution instance is configured with a single round, closed-ended, and second-highest-bid rule. After the auction starts, all participating regions submit the bid value of the auction unit in a non-interactive state, and modification or withdrawal is prohibited. The bid value of the auction unit is generated by inputting the corresponding control intention vector into the regional bid calculation module. The regional bid calculation module combines the tension deviation, temperature change gradient, fabric defect ratio and disturbance hysteresis weight into a floating-point scalar value according to a preset weight, which serves as the region's bid input for the current candidate micro-intervention control unit. Within each auction unit, the bid values ​​of all auction units are sorted, and the area with the highest bid value is determined as the winning bid area. The bid value of the area with the second highest bid value is used as the payment price for the winning bid area, and the unique assignment relationship between the winning bid area and the candidate micro-intervention control unit is recorded.

7. The method for optimizing an array-type steam thermal setting process based on machine learning according to claim 1, characterized in that, The construction of the winning bid control map specifically involves: All candidate micro-intervention control units in the winning areas are aggregated into the winning control unit set, sorted according to time index and voxel spatial index, and a structured winning control map is constructed. The winning control map is a two-dimensional mapping structure, with the time dimension corresponding to the control time period and the spatial dimension corresponding to the three-dimensional voxel position. Each cell of the winning control map includes the steam chamber number, steam opening value and winning area number.

8. The method for optimizing an array-type steam thermal setting process based on machine learning according to claim 1, characterized in that, The strategy re-parameter sampling extracts all control time periods from the winning bid control map, constructs a dynamic control time series index, uses each control time period as an independent sampling window, and initializes the control parameter template in each sampling window according to the configuration of the winning bid control unit. Within each sampling window, a strategy reparameter sampling process is performed, including: extracting the steam chamber number and corresponding spatial location of the winning control unit, and locking the current control area to be sampled; extracting the winning steam opening level as the initial value of the control intensity; based on the fabric disturbance voxel state within the current sampling window, inputting the disturbance voxel state features into the strategy adjustment model, adjusting the steam opening level and time offset to form the final action parameters, and generating a control strategy for controlling the behavior of the steam chamber. The strategy adjustment model adopts a parameter generator based on supervised training, with the disturbance voxel state features as input and the disturbance adjustment value of the control action parameters as output, which is used to disturb the initial value of the winning control intensity to achieve customized control. The control strategy consists of a steam chamber number, a control time period, an actual steam opening level, and a corresponding three-dimensional voxel index quadruple. This set of control instructions is sent to the execution end of the array-type steam heat setting equipment to drive the corresponding steam chambers to perform steam intervention behavior according to the bidding order and the final action parameters.

9. The method for optimizing an array-type steam thermal setting process based on machine learning according to claim 1, characterized in that, Step seven specifically involves: During the operation of the array-type steam heat setting equipment, abnormal conditions of the fabric surface are detected in real time. When an abnormality occurs in the fabric surface, the specific spatial location of the abnormality area is determined by infrared thermal imaging and fabric surface image recognition. Based on the spatial location of the abnormal area on the fabric surface and the time index of the time when the abnormality occurred, find the three-dimensional voxel unit index in the perturbation voxel map that corresponds to the abnormal area in terms of spatial location and time. Starting from the three-dimensional voxel unit index, the process traces back along the time axis to a preset backtracking period before the time of the anomaly occurrence, and searches step by step for candidate micro-intervention control units that have been executed on the abnormal region in the winning control map. The retrieved candidate micro-intervention control units are sorted in order of execution time from most recent to oldest, forming a sequence of intervention control units that cause abnormal fabric conditions.

10. The method for optimizing an array-type steam thermal setting process based on machine learning according to claim 1, characterized in that, Step eight specifically involves: Based on the intervention control unit sequence, determine the steam chamber number and spatial location corresponding to each candidate micro-intervention control unit in the intervention control unit sequence, and count the number of interventions and the degree of contribution to the anomaly in the area corresponding to each steam chamber before the anomaly occurred. Based on the number of interventions and the degree of abnormal contribution, the tension deviation, temperature change gradient, fabric defect ratio and disturbance hysteresis weight of the corresponding area in the original regional control intention vector are corrected according to the preset weights. The bid value for each region for each candidate micro-intervention control unit is recalculated based on the revised regional control intent vector, and the Vickrey auction mechanism is re-executed based on the updated regional bid value to determine the new winning region. All candidate micro-intervention control units in the newly selected regions are aggregated into the set of selected control units, and sorted according to the time index and voxel spatial index to form a new selected control map.

Images