A data analysis-based embroidery textile production quality detection method

Abstract

The application discloses a kind of embroidery textile production quality detection methods based on data analysis, it is related to intelligent textile technical field, including, real-time acquisition embroidery production line multi-source sensor data, time stamp synchronization and normalization processing are carried out through Internet of Things edge node, output structured process flow data;Structured process flow data is dynamically constructed process causal diagram model based on it, embroider line batch quality, needle speed parameter and environmental temperature and humidity parameter are as node, bayesian conditional probability is as edge weight, the causal effect value of needle speed parameter to broken needle defect is calculated through do-calculus;According to dynamic parameter adjustment instruction, needle speed parameter and embroider line tension value are real-time regulated, and simultaneously, through fuse signal, fault embroidery machine table coordinate is synchronized to operation and maintenance terminal.The application is accurately quantified the causal influence intensity of needle speed parameter and environmental temperature and humidity interaction through bayesian conditional probability and do-calculus framework.

Description

Technical Field

[0001] This invention relates to the field of intelligent textile technology, and in particular to a method for quality inspection of embroidered textile production based on data analysis. Background Technology

[0002] In the field of embroidery textile production, quality inspection technology has gradually transitioned from manual visual inspection to automated monitoring systems based on multi-source sensor data. Existing methods utilize IoT edge nodes to collect data in real time, such as embroidery thread tension, embroidery frame displacement trajectory coordinates, ambient temperature and humidity, and needle speed parameters. This data is then synchronized with timestamps and normalized to output structured process flow data. These technologies, combined with programmable logic controllers (PLCs), enable parameter control and apply statistical models or rule engines to predict defect probabilities, improving the initial responsiveness of the production process. With the integration of tools such as edge computing and Bayesian networks, existing solutions can support a certain degree of data-driven decision-making, providing a basic framework for quality optimization.

[0003] However, existing technologies have core limitations in dynamic causal modeling. Traditional methods cannot accurately analyze the causal dependencies between process parameters, such as the interaction effect of needle speed and ambient temperature and humidity on needle breakage defects, leading to parameter adjustment decisions relying on correlation rather than causation. This reduces the accuracy of defect prevention, introduces the risk of misjudgment, weakens the adaptability of real-time control, and makes it difficult to cope with complex variable disturbances in the production environment. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a data analysis-based method for quality inspection of embroidered textiles to solve the problem of inaccurate prevention of needle breakage defects caused by insufficient dynamic causal modeling in embroidery production.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] This invention provides a data analysis-based method for quality inspection of embroidered textile production. The method includes: real-time acquisition of multi-source sensor data from the embroidery production line; timestamping and normalization processing via IoT edge nodes to output structured process flow data; dynamically constructing a process causal graph model based on the structured process flow data, using embroidery thread batch quality, needle speed parameters, and environmental temperature and humidity parameters as nodes, and Bayesian conditional probabilities as edge weights; calculating the causal effect value of needle speed parameters on needle breakage defects using do-calculus; inputting the causal effect value into a distributed intelligent agent cluster, with each agent bound to an independent embroidery machine; performing collaborative decision-making based on Boids three rules; and outputting dynamic parameter adjustment instructions and a circuit breaker signal; adjusting needle speed parameters and embroidery thread tension values ​​in real time according to the dynamic parameter adjustment instructions, while simultaneously synchronizing the coordinates of the faulty embroidery machine to the maintenance terminal via the circuit breaker signal.

[0008] As a preferred embodiment of the data analysis-based method for detecting the production quality of embroidered textiles described in this invention, the multi-source sensor data of the embroidery production line includes embroidery thread tension value, embroidery frame displacement trajectory coordinates, environmental temperature and humidity parameters, and needle speed parameters.

[0009] As a preferred embodiment of the data analysis-based method for quality inspection of embroidered textile production described in this invention, the specific steps for dynamically constructing a process cause-effect graph model based on structured process flow data are as follows:

[0010] The batch quality of embroidery thread is defined as a discrete variable node in the process cause-effect graph model;

[0011] The needle speed parameter and the ambient temperature and humidity parameter are defined as continuous variable nodes in the process cause-effect graph model;

[0012] By calculating the edge weights between environmental temperature and humidity parameter nodes and needle speed parameter nodes using Bayesian conditional probability, and training the process causal graph structure using historical structured process flow data, a conditional probability table for broken needle defect nodes is established, and a process causal graph model is constructed.

[0013] As a preferred embodiment of the data analysis-based method for quality inspection of embroidered textile production described in this invention, the specific steps for calculating the causal effect value of needle speed parameters on needle breakage defects using do-calculus are as follows:

[0014] For the discrete variable nodes of the process cause-effect graph model, set prior distribution constraints;

[0015] Based on the prior distribution constraints, the needle speed parameter in the continuous variable node is fixed as the intervention value, and the posterior probability distribution of the broken needle defect node is calculated.

[0016] Compare the probability difference between the posterior probability distribution and the observed probability distribution for the broken needle defect;

[0017] When the probability difference exceeds the difference threshold, output the causal effect value of the needle speed parameter on the needle breakage defect.

[0018] As a preferred solution of the method for detecting the production quality of embroidered textiles based on data analysis according to the present invention, wherein: the distributed intelligent agent cluster refers to a set of autonomous decision-making entities bound to independent embroidery machines, and exchanges the embroidery frame displacement trajectory coordinates and the needle speed parameter status through a local communication network.

[0019] As a preferred solution of the method for detecting the production quality of embroidered textiles based on data analysis according to the present invention, wherein: the three Boids rules include an obstacle avoidance rule, a synchronization rule, and a fusing rule;

[0020] The obstacle avoidance rule means that when the monitored probability difference exceeds the obstacle avoidance threshold, generate a needle speed parameter downward adjustment instruction and a lateral displacement instruction;

[0021] The synchronization rule means synchronizing the local quality inspection frequency according to the mean value of the orderliness;

[0022] The fusing rule means that when the mean value of the orderliness drops below the fusing threshold, trigger a cross-region linkage shutdown, and output a dynamic parameter adjustment instruction and a fusing signal.

[0023] As a preferred solution of the method for detecting the production quality of embroidered textiles based on data analysis according to the present invention, wherein: perform collaborative decision-making based on the three Boids rules, and output a dynamic parameter adjustment instruction and a fusing signal. The specific steps are as follows.

[0024] The intelligent agent monitors the probability difference of the needle breakage defect of the embroidery machine in real time. If the probability difference exceeds the obstacle avoidance threshold, generate a needle speed parameter downward adjustment instruction and a lateral displacement instruction;

[0025] Based on the needle speed parameter downward adjustment instruction and the lateral displacement instruction, the intelligent agent calculates the mean value of the orderliness of the distributed intelligent agent cluster, and adjusts the local quality inspection frequency according to the synchronization rule;

[0026] When the mean value of the orderliness is lower than the fusing threshold and the causal effect value exceeds the risk threshold, the intelligent agent broadcasts a fusing signal, and outputs a dynamic parameter adjustment instruction and a fusing signal.

[0027] As a preferred solution of the method for detecting the production quality of embroidered textiles based on data analysis according to the present invention, wherein: adjust the needle speed parameter and the embroidery thread tension value in real time according to the dynamic parameter adjustment instruction, and at the same time synchronize the coordinates of the faulty embroidery machine to the operation and maintenance terminal through the fusing signal. The specific steps are as follows.

[0028] Analyze the lateral displacement instruction, and control the lateral movement of the embroidery frame servo motor to avoid obstacles;

[0029] The needle speed reduction command is parsed, the needle speed adjustment parameters are generated and written into the embroidery machine's programmable logic controller (PLC), and the servo mechanism is controlled by the PLC to synchronously adjust the pressure value of the embroidery thread tension roller and the speed of the embroidery machine's main spindle motor.

[0030] When a circuit breaker signal is received, the location of the faulty embroidery machine is located based on the coordinates of the embroidery frame displacement trajectory and pushed to the maintenance terminal.

[0031] The beneficial effects of this invention are as follows: By employing Bayesian conditional probability and a do-calculus framework, the causal influence of the interaction between needle speed parameters and environmental temperature and humidity is accurately quantified. The process causal graph model construction method supports real-time identification of high-risk parameter combinations and outputs numerical causal effect values ​​as a basis for collaborative decision-making. The conditional probability table is trained based on historical structured process flow data, enhancing the accuracy of calculating the posterior probability distribution of broken needle defects, ensuring robustness and real-time response capability in defect detection, and effectively improving the quality stability of the production process. Attached Figure Description

[0032] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 This is a flowchart of a data analysis-based method for quality inspection of embroidered textile production.

[0034] Figure 2 Build a flowchart for the process cause-effect graph model.

[0035] Figure 3 This is a flowchart of collaborative decision-making for a distributed intelligent agent cluster.

[0036] Figure 4 This is a flowchart of the machine parameter control process. Detailed Implementation

[0037] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0038] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0039] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0040] Reference Figures 1-4 This is one embodiment of the present invention, which provides a method for quality inspection of embroidered textile production based on data analysis, including the following steps:

[0041] S1: Real-time acquisition of multi-source sensor data from the embroidery production line, time-stamp synchronization and normalization processing through IoT edge nodes, and output of structured process flow data.

[0042] Specifically, embroidery thread tension sensors, embroidery frame displacement sensors, ambient temperature and humidity sensors, and needle speed sensors are deployed on the embroidery production line to collect embroidery thread tension values, embroidery frame displacement trajectory coordinates, ambient temperature and humidity parameters, and needle speed parameters in real time as multi-source sensor data.

[0043] Data from multiple sensors is received via IoT edge nodes, and a timestamp synchronization algorithm is executed to assign a unified timestamp to each data point, ensuring that all data points are time-series aligned. The timestamp synchronization algorithm achieves clock synchronization based on the Network Time Protocol (NTP). The synchronized multi-source sensor data is then normalized, mapping the thread tension value, embroidery frame displacement trajectory coordinates, environmental temperature and humidity parameters, and needle speed parameters to a unified numerical range. A data point refers to a single measurement value and its associated timestamp collected in real time from the multi-source sensors (thread tension sensor, embroidery frame displacement sensor, environmental temperature and humidity sensor, and needle speed sensor) on the embroidery production line. Each data point contains the sensor measurement result at a specific moment, such as the thread tension value or embroidery frame displacement trajectory coordinates at a given millisecond.

[0044] It should be noted that the normalization process uses a min-max normalization method, calculating the minimum and maximum values ​​for each data point and converting the original values ​​to values ​​between 0 and 1. The output is structured process flow data, organized in a tabular format, containing fields such as: timestamp, thread tension value, embroidery frame displacement trajectory coordinates, ambient temperature and humidity parameters, and needle speed parameters. The original values ​​refer to the unnormalized sensor measurements at the data points, such as the tension value directly output by the thread tension sensor (unit: Newtons), the coordinate values ​​directly recorded by the embroidery frame displacement sensor (unit: millimeters), the temperature (unit: degrees Celsius) and humidity (unit: percentage) directly collected by the ambient temperature and humidity sensor, and the speed value directly measured by the needle speed sensor (unit: revolutions per minute).

[0045] S2: Based on structured process flow data, a process causal graph model is dynamically constructed. The batch quality of embroidery thread, needle speed parameters, and environmental temperature and humidity parameters are used as nodes, and Bayesian conditional probability is used as edge weights. The causal effect value of needle speed parameters on needle breakage defects is calculated through do-calculus.

[0046] Among them, do-calculus refers to the causal intervention computation framework proposed by Judea Pearl, which includes tools such as the do operator and the backdoor criterion.

[0047] S2.1: Define the batch quality of embroidery thread as a discrete variable node in the process cause-effect graph model.

[0048] Specifically, batch quality data for embroidery thread is obtained from historical databases or real-time input. This batch quality data represents the quality level of the embroidery thread batch and is defined as a discrete variable node in the process causal graph model. Each discrete variable node represents a finite number of possible values, such as high, medium, and low quality levels. The timestamps of the batch quality data are aligned with the structured process flow data to ensure data consistency. The batch quality nodes, as part of the process causal graph model, are used for Bayesian conditional probability calculations.

[0049] S2.2: Define the needle speed parameter and the ambient temperature and humidity parameter as continuous variable nodes in the process cause-effect graph model.

[0050] Specifically, needle speed parameters and ambient temperature and humidity parameters are extracted from structured process flow data and defined as continuous variable nodes in the process cause-effect graph model. Continuous variable nodes represent numerical variables; the needle speed parameter node corresponds to the measurement value of the needle speed sensor (unit: revolutions per minute), and the ambient temperature and humidity parameter node corresponds to the measurement value of the ambient temperature and humidity sensor (temperature unit: degrees Celsius, humidity unit: percentage). Continuous variable nodes serve as continuous input variables in the process cause-effect graph model.

[0051] S2.3: Calculate the edge weights between the environmental temperature and humidity parameter nodes and the needle speed parameter nodes using Bayesian conditional probability, train the process cause-effect graph structure by combining historical structured process flow data, establish a conditional probability table for the broken needle defect node, and construct the process cause-effect graph model.

[0052] Specifically, the broken needle defect node is a discrete variable node in the process cause-effect graph model. The value of the broken needle defect node indicates whether a broken needle defect event occurs on the embroidery machine under specific process conditions (e.g., whether it occurs or not). The occurrence status (occurrence or non-occurrence) of the broken needle defect comes from real-time detection of the embroidery machine's operating status or historical fault records.

[0053] Furthermore, the process causal graph model is a Bayesian network, with nodes including nodes representing embroidery thread batch quality, needle speed parameters, environmental temperature and humidity parameters, and broken needle defects. Edge weights represent the conditional dependencies between nodes, calculated using Bayesian conditional probabilities.

[0054] Bayesian conditional probability, expressed as:

[0055] ;

[0056] In the formula, Indicates in the event Given that the event has already occurred, The conditional probability of occurrence Indicates in the event Given that the event has already occurred, The inverse conditional probability of occurrence Indicates an event The prior probability, Indicates an event The marginal probability;

[0057] It should be noted that the event and events These are node variables in the process cause-effect graph. The edge weights between the ambient temperature and humidity parameter nodes and the needle speed parameter nodes are calculated using the needle speed parameter and the ambient temperature and humidity parameter. The probability distribution is determined and estimated based on historical structured process flow data.

[0058] Furthermore, using historical structured process flow data (including timestamp sequences), Bayesian network learning algorithms (such as maximum likelihood estimation) are applied to learn the graph structure of the process causal graph;

[0059] The process cause-effect graph structure includes node connectivity (e.g., whether environmental temperature and humidity parameter nodes affect needle speed parameter nodes) and edge weights. Historical structured process flow data covers multiple production cycles to ensure the generalization capability of the process cause-effect graph structure.

[0060] The broken needle defect node is defined as a discrete variable node, with possible values ​​such as either occurring or not occurring. This is stored in a conditional probability table (CPT). The parent node includes the embroidery thread batch quality node, the needle speed parameter node, and the ambient temperature and humidity parameter node.

[0061] Conditional probability tables (CPTs) estimate conditions using historical data, such as by using frequency counting to calculate conditions. When data is plentiful, frequency counting is used to calculate conditions; when data is sparse, Laplace smoothing is used to avoid the problem of zero probability.

[0062] The node connection relationship and node attributes (discrete / continuous variable type) are combined into a graph topology, where the embroidery batch quality node and the environmental temperature and humidity parameter node are the root nodes, the needle speed parameter node is the intermediate node, and the broken needle defect node is the leaf node.

[0063] Write the edge weights calculated by Bayesian conditional probability into the weight matrix of the corresponding edge, and embed the conditional probability table (CPT) of the broken needle defect node into the corresponding node attribute to form a computable probability dependency.

[0064] The process causal graph model is tested for the presence of cycles by using a topological sorting algorithm, ensuring that the model satisfies the mathematical properties and causal semantic requirements of a directed acyclic graph (DAG).

[0065] S2.4: Set prior distribution constraints for the discrete variable nodes of the process cause-effect graph model.

[0066] Specifically, the discrete variable nodes in the process cause-effect graph model include the embroidery thread batch quality node and the broken needle defect node. Prior distribution constraints are set for the embroidery thread batch quality node and the broken needle defect node.

[0067] For the embroidery thread batch quality node, the prior distribution (quality grade) is defined as the frequency distribution of the quality grades of historical batches; for the needle breakage defect node, the probability of needle breakage occurring in the prior distribution is defined as the average occurrence rate of historical production data. The prior distribution constraints are solidified into the process causal graph model through Bayesian network parameter learning.

[0068] S2.5: Based on the prior distribution constraints, fix the needle speed parameter in the continuous variable node as the intervention value, and calculate the posterior probability distribution of the broken needle defect node.

[0069] Specifically, a specific value of the needle speed parameter is selected as the intervention target (e.g., fixing the needle speed to 2000 rpm), represented by the do operator as do(needle speed parameter = 2000). Confounding variables such as environmental temperature and humidity parameters are identified and controlled to ensure that the causal relationship between the needle speed parameter and the broken needle defect is not interfered with by other variables. The specific operation is as follows: in the process causal graph model, all backdoor paths pointing to the needle speed parameter node are blocked.

[0070] Based on prior distribution constraints and the conditional probability table of the process causal graph model, the posterior probability distribution of the needle breakage defect node after intervention is calculated. The calculation process utilizes a Bayesian network inference algorithm, combining the joint distribution of needle speed parameters and needle breakage defects in historical data.

[0071] It should be noted that, in the specific calculation, the prior distribution and conditional probability table of each node are first determined according to the process cause-effect graph model. Then, intervention operations are applied to the needle speed parameter, and the influence of confounding factors is eliminated by the do operator. Under the condition that the needle speed parameter is fixed at a specific value, the discrete distribution of the embroidery thread batch quality and the continuous distribution characteristics of the ambient temperature and humidity are combined with the conditional probability table of the broken needle defect node to perform probability propagation calculation.

[0072] Among them, the Monte Carlo sampling method is used to achieve efficient approximate calculation for the processing of continuous variables. The posterior probability distribution of needle breakage defects after intervention is obtained by marginal summation and integration. The posterior probability distribution of needle breakage defects reflects the probability characteristics of needle breakage defects under specific needle speed parameter settings.

[0073] S2.6: Compare the probability difference between the posterior probability distribution and the observed probability distribution for the broken needle defect.

[0074] Specifically, the actual frequency of needle breakage defects is statistically analyzed from the current structured process flow data and recorded as the observed probability (e.g., needle breakage defect); the difference between the posterior probability distribution of the needle breakage defect node after intervention and the observed probability is obtained to obtain the absolute value difference (i.e., the probability difference).

[0075] S2.7: When the probability difference exceeds the difference threshold, output the causal effect value of the needle speed parameter on the broken needle defect.

[0076] Specifically, the difference threshold is set to a preset value to indicate a significant causal effect. The difference threshold is determined based on the fluctuation range of the probability of needle breakage defects in historical production data (for example, the difference threshold is set to 0.1, which means that a probability change of more than 10% is considered significant).

[0077] If the probability difference exceeds the difference threshold, the causal effect value of the needle speed parameter on the broken needle defect is output. The causal effect value is defined as the probability difference itself. The output is a numerical result used in subsequent steps. If the probability difference does not exceed the difference threshold, no causal effect value is output, indicating no significant impact.

[0078] S3: Input the causal effect value into the distributed intelligent agent cluster. Each agent is bound to an independent embroidery machine. Based on the Boids three rules, it performs collaborative decision-making and outputs dynamic parameter adjustment instructions and circuit breaker signals.

[0079] Among them, the distributed intelligent agent cluster refers to a set of autonomous decision-making entities bound to an independent embroidery machine, which exchange the coordinates of the embroidery frame displacement trajectory and the status of needle speed parameters through a local communication network.

[0080] It should be noted that the Boids three rules include obstacle avoidance rules, synchronization rules, and circuit breaker rules. The obstacle avoidance rule refers to generating a needle speed parameter reduction command and a lateral displacement command when the monitoring probability difference exceeds the obstacle avoidance threshold; the synchronization rule refers to synchronizing the local quality inspection frequency according to the orderliness average; the circuit breaker rule refers to triggering cross-regional linkage shutdown when the orderliness average falls below the circuit breaker threshold, outputting a dynamic parameter adjustment command and a circuit breaker signal.

[0081] The obstacle avoidance threshold is set based on the statistical distribution characteristics of the probability difference of needle breakage defects in historical operating data of the embroidery production line. A baseline value is determined by analyzing the fluctuation range of the probability difference under normal production conditions, and then dynamically adjusted in conjunction with the standard deviation of real-time process flow data. The obstacle avoidance threshold reflects the sensitivity of the interaction between needle speed parameters and ambient temperature and humidity to needle breakage defects. A typical example is a value ranging from 0.15 to 0.25 of the absolute value of the probability difference. When the probability difference exceeds the melting threshold, it indicates that the current combination of process parameters has entered a high-risk state.

[0082] The circuit breaker threshold is set based on the critical point of the average orderliness of the distributed intelligent agent cluster in the historical collaborative decision-making process. It is determined by calculating the inflection point value of the decreasing trend of the average orderliness within a continuous production cycle, and its value is negatively correlated with the dynamic adjustment capability of the embroidery machine group. The circuit breaker threshold is used to identify the risk of process runaway. A typical example is a value in the range of 0.35 to 0.45 of the average orderliness. When it is below the circuit breaker threshold, it indicates that the multi-machine collaborative adjustment mechanism has failed and should be immediately suspended.

[0083] S3.1: The agent monitors the probability difference of needle breakage defects on the embroidery machine in real time. If the probability difference exceeds the obstacle avoidance threshold, a needle speed parameter reduction command and a lateral displacement command are generated.

[0084] Specifically, each embroidery machine is equipped with an intelligent agent that continuously reads the probability difference of broken needle defects;

[0085] Real-time comparison of the probability difference between the standby machine and the preset obstacle avoidance threshold: when the probability difference is greater than the obstacle avoidance threshold, execute the needle speed parameter reduction command and the lateral displacement command generation operation.

[0086] Needle speed parameter reduction command: Determines the amount of needle speed parameter reduction. The base reduction coefficient is a preset constant (e.g., 5 revolutions / minute / unit difference). Lateral displacement command: Determines the lateral offset of the embroidery frame displacement trajectory. The offset direction is perpendicular to the embroidery frame displacement trajectory, and the base displacement coefficient is a preset constant (e.g., 0.2 mm / unit difference).

[0087] The instructions to lower the needle speed and the instructions for lateral displacement are sent to the embroidery machine controller for adjustment.

[0088] It should be noted that the obstacle avoidance threshold is set based on the statistical distribution characteristics of the probability difference of needle breakage defects in the historical operation data of the embroidery production line. The benchmark value is determined by analyzing the fluctuation range of the probability difference under normal production conditions. The example value range is 0.12 to 0.18 of the absolute value of the probability difference. When the real-time probability difference exceeds the obstacle avoidance threshold, it indicates that the current combination of process parameters has entered a high-risk state.

[0089] S3.2: Based on the needle speed parameter reduction command and the lateral displacement command, the agent calculates the average orderliness of the distributed agent cluster and adjusts the local quality inspection frequency according to the synchronization rules.

[0090] Specifically, the intelligent agent obtains in real time the data on the adjustment amount of the needle speed parameter and the lateral displacement amount of all members of the distributed intelligent agent cluster.

[0091] The mean orderliness, characterizing the degree of cluster collaboration, is calculated based on the standard deviation of the down-adjustment of needle speed parameters among members and the mean Euclidean distance between them. The expression is:

[0092] ;

[0093] In the formula, This represents the standard deviation of the decrease in needle speed parameter among cluster members. The mean Euclidean distance representing the lateral displacement of cluster members. Indicates the speed-coordination weighting coefficient. This represents the complementary value of the speed coordination weighting coefficient. This indicates the amount by which the needle speed parameter is adjusted downwards. Indicates the lateral displacement;

[0094] It should be noted that, Historical data regression analysis determined that the typical value was 0.6; the mean orderliness was... Positively correlated with the degree of cluster collaboration A higher value indicates greater synergy;

[0095] Based on the mean of orderliness The preset interval threshold is used to adjust the local quality inspection frequency according to the piecewise linear relationship. The preset interval threshold is determined by analyzing the statistical correlation between the mean of orderliness and the broken needle defect rate in historical production data, and the critical point that causes a significant change in the defect rate is selected as the dividing boundary.

[0096] S3.3: When the mean orderliness is lower than the circuit breaker threshold and the causal effect value exceeds the risk threshold, the agent broadcasts a circuit breaker signal and outputs dynamic parameter adjustment instructions and a circuit breaker signal.

[0097] Specifically, the agent continuously monitors the current average degree of order. The two key parameters, needle speed parameter and needle breakage parameter, are used to determine the causal effect value of the broken needle defect. When both of these conditions are met, the agent immediately executes the circuit breaker protocol. The mean orderliness value is lower than the circuit breaker threshold (indicating a severe deterioration in cluster coordination) and the causal effect value exceeds the risk threshold (indicating a high risk in the current combination of process parameters).

[0098] The circuit breaker signal is sent to all members through the distributed intelligent agent cluster communication protocol, and a dynamic parameter adjustment command containing the needle speed adjustment parameter is output. Through the dynamic parameter adjustment command, the needle speed parameter is forcibly set to a preset safety value (such as 200 rpm); the lateral displacement is forcibly set to 0 mm.

[0099] Write the fuse failure signal and the process parameters at the time of triggering into the database and mark it as an abnormal termination event.

[0100] S4: Adjusts needle speed and embroidery thread tension in real time according to dynamic parameter adjustment instructions, and simultaneously synchronizes the coordinates of the faulty embroidery machine to the maintenance terminal through a fuse signal.

[0101] S4.1: Parse the lateral displacement command and control the embroidery frame servo motor to move laterally to avoid obstacles.

[0102] Specifically, the lateral displacement data is extracted from the dynamic parameter adjustment command, converted into a control signal for the embroidery frame servo motor, and driven to move the embroidery frame servo motor laterally a specified distance along the vertical direction of the displacement trajectory to achieve obstacle avoidance.

[0103] S4.2: Parse the needle speed reduction instruction, generate needle speed adjustment parameters and write them into the embroidery machine's programmable logic controller (PLC). The PLC controls the servo mechanism to synchronously adjust the pressure value of the embroidery thread tension roller and the speed of the embroidery machine's main spindle motor.

[0104] Specifically, needle speed adjustment parameters are extracted from dynamic parameter adjustment instructions and written into the programmable logic controller register of the embroidery machine;

[0105] Based on needle speed adjustment parameters Calculate the adjustment amount of embroidery thread tension. The expression is:

[0106] ;

[0107] In the formula, Indicates the tension-needle speed coupling coefficient (unit: The typical value range was determined through historical data regression analysis. , Indicates the reference tension compensation amount (unit: Newtons). ), used to correct inherent deviations in mechanical equipment, usually taken as ;

[0108] The embroidery machine's programmable logic controller (PLC) controls the spindle motor to adjust the needle speed parameters, while simultaneously controlling the embroidery thread tension roller to adjust the embroidery thread tension value.

[0109] S4.3: When a circuit breaker signal is received, the location of the faulty embroidery machine is locked based on the coordinates of the end point of the embroidery frame displacement trajectory and pushed to the maintenance terminal.

[0110] Specifically, when the programmable logic controller of the embroidery machine receives the broadcast fuse signal, it reads the coordinates of the current displacement trajectory endpoint reported by the embroidery frame displacement sensor in real time.

[0111] Based on the coordinates of the endpoint of the embroidery frame displacement trajectory, and combined with the physical layout parameters of the embroidery machine (such as the machine installation position and the movement range of the embroidery frame), the actual physical position of the faulty machine is calculated through coordinate system transformation. The specific method is as follows: superimpose the local coordinates of the embroidery frame onto the global layout coordinate system of the machine; determine the unique identifier and position of the faulty machine based on the preset correspondence between the machine number and the movement area of ​​the embroidery frame.

[0112] The location coordinates (unit: mm) and machine number of the faulty machine are pushed to the monitoring equipment of the operation and maintenance terminal in real time through industrial communication protocols (such as OPC UA), and a high-priority alarm pop-up is triggered to display the location of the faulty machine, the melting time and related process parameters (such as needle speed and tension value), and an operation and maintenance work order is generated.

[0113] This embodiment also provides a computer device applicable to the data analysis-based method for detecting the production quality of embroidered textiles, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the data analysis-based method for detecting the production quality of embroidered textiles as proposed in the above embodiment.

[0114] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0115] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the data analysis-based method for quality inspection of embroidered textile production as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0116] In summary, this invention uses Bayesian conditional probability and a do-calculus framework to accurately quantify the causal influence of the interaction between needle speed parameters and environmental temperature and humidity. The process causal graph model construction method supports real-time identification of high-risk parameter combinations and outputs numerical causal effect values ​​as a basis for collaborative decision-making. By training a conditional probability table based on historical structured process flow data, the accuracy of calculating the posterior probability distribution of broken needle defects is enhanced, ensuring robustness and real-time response capabilities in defect detection and effectively improving the quality stability of the production process.

[0117] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A data analysis-based embroidery textile production quality detection method, characterized by: including, real-time collecting multi-source sensor data of the embroidery production line, performing timestamp synchronization and normalization processing through an Internet of Things edge node, and outputting structured process flow data; the multi-source sensor data of the embroidery production line includes thread tension values, embroidery frame displacement trajectory coordinates, environmental temperature and humidity parameters, and needle speed parameters; dynamically constructing a process causal graph model based on the structured process flow data, taking thread batch quality, needle speed parameters, and environmental temperature and humidity parameters as nodes, and Bayesian conditional probability as the edge weight, and calculating the causal effect value of the needle speed parameter on the broken needle defect through do-calculus. The specific steps are as follows: setting prior distribution constraint conditions for the discrete variable nodes of the process causal graph model; fixing the needle speed parameter in the continuous variable nodes as an intervention value based on the prior distribution constraint conditions, and calculating the posterior probability distribution of the broken needle defect node; comparing the probability difference of the broken needle defect in the posterior probability distribution and the observed probability distribution; when the probability difference exceeds the difference threshold, outputting the causal effect value of the needle speed parameter on the broken needle defect; inputting the causal effect value into the distributed intelligent agent cluster, binding each intelligent agent to an independent embroidery machine, and performing collaborative decision-making based on the Boids three rules, and outputting dynamic parameter adjustment instructions and fusing signals; real-time regulating the needle speed parameter and thread tension value according to the dynamic parameter adjustment instructions, and simultaneously synchronizing the coordinates of the faulty embroidery machine to the operation and maintenance terminal through the fusing signal.

2. The data analysis-based embroidery textile production quality detection method of claim 1, characterized by: The specific steps for dynamically constructing a process causal graph model based on the structured process flow data are as follows: defining the thread batch quality as a discrete variable node of the process causal graph model; defining the needle speed parameter and environmental temperature and humidity parameters as continuous variable nodes of the process causal graph model; calculating the edge weight between the environmental temperature and humidity parameter node and the needle speed parameter node through Bayesian conditional probability, training the process causal graph structure in combination with historical structured process flow data, establishing a conditional probability table for the broken needle defect node, and constructing a process causal graph model.

3. The data analysis-based embroidery textile production quality detection method according to claim 2, characterized in that: The distributed intelligent agent cluster refers to a set of autonomous decision-making entities bound to independent embroidery machines, and exchanging embroidery frame displacement trajectory coordinates and needle speed parameter status through a local communication network.

4. The data analysis-based embroidery textile production quality detection method of claim 3, characterized by: The Boids three rules include an obstacle avoidance rule, a synchronization rule, and a fusing rule; The obstacle avoidance rule means that when the monitored probability difference exceeds the obstacle avoidance threshold, generating a needle speed parameter downward adjustment instruction and a horizontal displacement instruction; The synchronization rule means synchronizing the local quality inspection frequency according to the mean orderliness; The fusing rule means triggering cross-region linkage shutdown when the mean orderliness drops below the fusing threshold, and outputting dynamic parameter adjustment instructions and fusing signals.

5. The data analysis based embroidery textile production quality detection method according to claim 4, characterized in that: The specific steps for performing collaborative decision-making based on the Boids three rules and outputting dynamic parameter adjustment instructions and fusing signals are as follows: real-time monitoring the probability difference of the broken needle defect of the embroidery machine by the intelligent agent. If the probability difference exceeds the obstacle avoidance threshold, generating a needle speed parameter downward adjustment instruction and a horizontal displacement instruction; based on the needle speed parameter downward adjustment instruction and the horizontal displacement instruction, the intelligent agent calculates the mean orderliness of the distributed intelligent agent cluster, and adjusts the local quality inspection frequency according to the synchronization rule; When the mean orderliness is lower than the circuit breaker threshold and the causal effect value exceeds the risk threshold, the agent broadcasts a circuit breaker signal and outputs dynamic parameter adjustment instructions and a circuit breaker signal.

6. The data analysis-based embroidery textile production quality detection method according to claim 5, characterized in that: The process involves real-time adjustment of needle speed and thread tension based on dynamic parameter adjustment commands, while simultaneously synchronizing the coordinates of the faulty embroidery machine to the maintenance terminal via a fuse signal. The specific steps are as follows: Analyze the lateral displacement command and control the embroidery frame servo motor to move laterally to avoid obstacles; The needle speed reduction command is parsed, the needle speed adjustment parameters are generated and written into the embroidery machine's programmable logic controller (PLC), and the servo mechanism is controlled by the PLC to synchronously adjust the pressure value of the embroidery thread tension roller and the speed of the embroidery machine's main spindle motor. When a circuit breaker signal is received, the location of the faulty embroidery machine is located based on the coordinates of the embroidery frame displacement trajectory and pushed to the maintenance terminal.

Images