Self-adaptive control method for needle tooth density of wool spinning carding machine based on fiber fineness monitoring

Abstract

The invention discloses a wool spinning carding machine needle tooth density self-adaptive control method and system based on fiber fineness monitoring, and relates to the field of mechanical intelligent control, and the method comprises the steps: obtaining fiber fineness monitoring data of a target scene and carding area operation characteristic data in real time; acquiring the needle tooth density of the target working condition through the fiber fineness monitoring data; constructing a working condition density distribution model of the needle tooth density of the target carding machine, and initializing the working condition density distribution model; in combination with the working condition density distribution model, the carding area operation characteristic data and a pre-constructed second relation model, calculating expected working condition pin tooth density, and adjusting the working condition density distribution model; and based on the working condition density distribution model, outputting an optimal apparent needle tooth density set value, generating a control instruction and driving a needle tooth density execution mechanism of the carding machine to act. The problems that in the prior art, manual sampling and off-line detection are dependent, information is seriously lagged, the needle tooth density adaptability is poor, and the dynamic operation working condition of the carding machine is not considered are solved.

Description

Technical Field

[0001] This invention relates to the field of intelligent mechanical control, specifically to an adaptive control method for the needle density of a wool spinning carding machine based on fiber fineness monitoring. Background Technology

[0002] Wool combing is the core process in wool sliver preparation. Its task is to open, comb, and remove impurities from the wool fibers using the high-speed rotating needles of the combing machine, providing a wool sliver with uniform structure and good fiber orientation for subsequent spinning. The needle density is a key process parameter that determines the combing quality, the degree of fiber damage, and the shedding rate.

[0003] In traditional carding processes, the needle density is typically preset by operators based on the type of raw material. Firstly, the static control mode during production relies on manual sampling and offline testing, resulting in significant information lag. It cannot dynamically adjust the needle density according to the real-time fiber fineness, impacting sliver quality and yield. Secondly, it fails to consider the dynamic changes in speed and temperature during the high-speed operation of the carding machine, making it difficult to guarantee process quality stability. Furthermore, it neglects the random fluctuations in fiber needle density during actual operation, as well as the slow time-varying characteristics of the carding machine, leading to a gradual deterioration in process performance. Summary of the Invention

[0004] This application provides an adaptive control method for the needle density of a wool combing machine based on fiber fineness monitoring, which addresses the problems of existing technologies that rely on manual sampling and offline detection, have serious information lag, poor adaptability of needle density, and do not consider the dynamic operating conditions of the combing machine.

[0005] In view of the above problems, this application provides an adaptive control method for the needle density of a wool combing machine based on fiber fineness monitoring.

[0006] This application provides an adaptive control method for the needle density of a wool combing machine based on fiber fineness monitoring, the method comprising:

[0007] Real-time acquisition of fiber fineness monitoring data and combing zone operation characteristic data of the target scene;

[0008] The fiber fineness monitoring data is input into a pre-constructed first relationship model to obtain the matching target working condition needle density;

[0009] Construct a working condition density distribution model of the needle tooth density of the target combing machine, and initialize the working condition density distribution model based on the target working condition needle tooth density;

[0010] Combining the working condition density distribution model, the combing zone operation characteristic data and the pre-constructed second relationship model, the desired working condition needle tooth density is calculated, and the working condition density distribution model is iteratively solved and adjusted with the desired working condition needle tooth density approaching the target working condition needle tooth density as the optimization objective.

[0011] Based on the working condition density distribution model that satisfies the optimization objective, the optimal apparent needle density setpoint is output, control commands are generated, and the needle density actuator of the combing machine is driven to perform actions.

[0012] Optionally, the operating characteristic data of the combing zone includes at least the operating speed data of the combing roller and the characteristic temperature data of the combing zone.

[0013] Optionally, real-time acquisition of fiber fineness monitoring data and combing zone operation characteristic data of the target scene, including:

[0014] The fiber fineness monitoring data is collected in real time by an online fiber fineness monitoring component configured at the feeding mechanism of the target carding machine, wherein the fiber fineness monitoring data includes at least the average diameter of the fed fiber;

[0015] By using an encoder assembly and a non-contact temperature measurement assembly configured at the combing roller of the target combing machine, the running speed data of the combing roller and the characteristic temperature data of the combing zone are obtained respectively, and the output is the running characteristic data of the combing zone.

[0016] Optionally, the construction of the first relational model includes:

[0017] Obtain historical production data of the target combing machine, wherein the historical production data includes historical fiber fineness data and corresponding needle density records of high-quality output slivers;

[0018] Data from the historical production data that are identical to the target combing machine and the real-time raw material type are selected as sample data.

[0019] Based on the sample data, analysis and training are performed to establish the first relational model with fiber fineness as input, wherein the first relational model is constructed based on either a mapping function or a rule base.

[0020] Optionally, a working condition density distribution model of the needle density of the target combing machine is constructed, and the working condition density distribution model is initialized based on the target working condition needle density, including:

[0021] Based on the historical production data of the target combing machine, multiple sets of sample apparent needle density and sample working condition needle density distribution are extracted, and the distribution basis model of the working condition density distribution model is determined by matching accordingly.

[0022] By combining the aforementioned distribution base model, the apparent needle density of multiple sets of samples, and the needle density distribution under operating conditions of the samples, the distribution parameters of the operating condition density distribution model are analyzed and determined.

[0023] The target working condition needle density is used as the distribution center of the working condition density distribution model, and initialization is performed in combination with the distribution base model and the distribution parameters.

[0024] Optionally, the construction of the second relational model includes:

[0025] Based on a prior knowledge graph, a first mechanism model of the target combing machine is constructed, wherein the first mechanism model takes the combing operation characteristic data as input and outputs a first equivalent density adjustment coefficient to characterize the influence of centrifugal force.

[0026] Based on a prior knowledge graph, a second mechanism model of the target combing machine is constructed. The second mechanism model takes the combing operation characteristic data and the working condition density distribution model as inputs and outputs a second equivalent density adjustment coefficient to characterize the influence of frictional temperature rise.

[0027] The second relationship model is constructed by combining the first mechanism model and the second mechanism model.

[0028] Optionally, by combining the working condition density distribution model, the combing operation characteristic data, and the pre-constructed second relationship model, the desired working condition needle density is calculated. Then, with the goal of approximating the target working condition needle density to the desired working condition needle density, the working condition density distribution model is iteratively solved and adjusted, including:

[0029] Input the working condition density distribution model and the combing operation characteristic data of the current iteration step into the second relationship model;

[0030] The first equivalent density adjustment coefficient and the second equivalent density adjustment coefficient under the current iteration step are calculated and output through the second relationship model, and the distribution center of the working condition density distribution model is adjusted accordingly.

[0031] Based on the adjusted working condition density distribution model, the mathematical expectation is calculated as the expected working condition needle tooth density, and the difference between the expected working condition needle tooth density and the target working condition needle tooth density is calculated.

[0032] With the goal of minimizing the difference, the distribution center of the working condition density distribution model is iteratively adjusted using an optimization algorithm until the difference is less than a preset convergence threshold.

[0033] Optionally, based on the working condition density distribution model that satisfies the optimization objective, an optimal apparent needle density setpoint is output, control commands are generated, and the needle density actuator of the combing machine is driven to perform actions, including:

[0034] From the working condition density distribution model that satisfies the optimization objective, extract the distribution center setting value and output the distribution center setting value as the optimal apparent needle density setting value.

[0035] Based on the optimal apparent needle density setting value, the control command based on machine language is generated and sent to the needle density actuator to perform needle density control.

[0036] The needle density actuator includes either a needle cloth mechanical mechanism with dynamically adjustable needle density or a needle cloth unit group with differentiated preset needle density.

[0037] Optionally, generating control commands and driving the needle density actuator of the combing machine to perform actions, and then further including:

[0038] Record the operation log of the target carding machine, and combine the operation log with preset posterior constraints to obtain the quality feedback data of the output sliver.

[0039] Based on the quality feedback data and the corresponding optimal apparent needle density setting value, posterior feedback data is analyzed and screened to obtain the posterior feedback data.

[0040] Based on the posterior feedback data, the distribution parameters in the working condition density distribution model are updated using Bayesian methods.

[0041] Optionally, the posterior constraints include at least one of the following:

[0042] Since the last Bayes update, the target comber has been running continuously for the preset time period.

[0043] The quality feedback data of the output yarn, which is monitored in real time, exceeds the preset quality fluctuation threshold.

[0044] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0045] This application first transforms the discrete and lagging offline detection in the carding process into online data by synchronously monitoring the fiber fineness of the raw materials and the core operating conditions of the equipment, providing accurate and timely multi-dimensional input for subsequent intelligent decision-making. Second, by constructing a first relational model using fiber fineness monitoring data, the theoretically optimal needle density target value can be determined based on real-time raw material fineness, providing a high-quality optimization starting point for the entire adaptive control process and effectively avoiding blind parameter setting. Third, a working condition density distribution model of the target carding machine's needle density is constructed and initialized using a probability distribution model centered on the target working condition needle density, improving the system's robustness, quantifying the deviation between the apparent setpoint and the working condition needle density caused by factors such as wear and vibration during equipment execution, determining the direction of subsequent optimization processes, and enhancing the control strategy's adaptability to disturbances during production.

[0046] Simultaneously, by combining the working condition density distribution model, the operational characteristic data of the combing zone, and the pre-constructed second relationship model, the dynamic impact of the current operating conditions on the combing effect is evaluated. The desired working condition needle density is calculated, and the optimal distribution center, i.e., the optimal apparent setpoint, is automatically found through an iterative algorithm. This ensures that the desired working condition needle density under dynamic operating conditions can accurately approximate the target working condition needle density, thereby compensating for the combing intensity fluctuations caused by changes in equipment operating status in real time and automatically, achieving precise compensation for the working condition density distribution. Finally, based on the working condition density distribution model that meets the optimization objective, the optimal probability distribution is transformed into the optimal apparent setpoint, resulting in and generating control commands that can directly drive the actuators. This ensures that intelligent decisions can be executed accurately and reliably, ultimately achieving dynamic adjustment of the needle density control and steadily improving the output quality of the wool combing machine. Attached Figure Description

[0047] Figure 1 This is a flowchart illustrating the adaptive control method for needle density of a wool combing machine based on fiber fineness monitoring, as described in this application.

[0048] Figure 2 This is a flowchart illustrating the iterative solution of the working condition density distribution model in the adaptive control method for the needle tooth density of a wool spinning carding machine based on fiber fineness monitoring, as described in this application. Detailed Implementation

[0049] This application provides an adaptive control method for the needle density of a wool combing machine based on fiber fineness monitoring, which specifically solves the problems of existing technologies that rely on manual sampling and offline detection, have serious information lag, poor adaptability of needle density, and do not consider the dynamic operating conditions of the combing machine.

[0050] The present invention will now be described in detail with reference to the accompanying drawings.

[0051] In the embodiments, such as Figure 1 As shown, this application provides an adaptive control method for the needle density of a wool spinning carding machine based on fiber fineness monitoring. The method includes:

[0052] S10: Real-time acquisition of fiber fineness monitoring data and combing zone operation characteristic data of the target scene;

[0053] In this embodiment, fiber fineness monitoring data refers to fiber geometric morphology parameters measured in real time by sensors or online detection devices, which are usually fiber fineness; combing zone operation characteristic data are dynamic state parameters of the combing zone, the core working area of ​​the combing machine, during operation, such as operating speed and operating temperature.

[0054] Specifically, the process begins by using sensors and other detection equipment to collect real-time fiber fineness monitoring data (i.e., real-time fiber fineness) and combing zone operation characteristic data during the production process through continuous sampling.

[0055] In step S10 of the method provided in this application embodiment, the combing zone operation characteristic data includes at least the running speed data of the combing roller and the characteristic temperature data of the combing zone.

[0056] In this embodiment, the combing roller is a high-speed rotating roller with a surface covered with carding cloth, and its rotation directly impacts, separates, and combs the fibers; the running speed data of the combing roller is the rotational speed of the combing roller; the characteristic temperature data of the combing zone is a representative temperature value measured within the combing zone.

[0057] Specifically, during the combing process, intense friction and repeated bending between fibers and the needles, as well as between fibers themselves, generate heat, leading to a localized temperature increase in the combing zone. Characteristic temperature data is a key indicator reflecting the thermodynamic state and tribological effects of the combing process. It affects the mechanical properties of the fibers, the surface friction coefficient, and potential thermal damage, thus altering the ease with which the fibers are combed and the perceived intensity of the combing action. Therefore, real-time measurement of the characteristic temperature data in the combing zone is necessary. The operating speed of the combing roller is the fundamental factor determining the mechanical and dynamic intensity of the combing action. Speed ​​changes directly affect the impact force of the needles on the fibers, the residence time of the fibers on the needle surface, and the centrifugal force effect, significantly altering the combing intensity, the degree of fiber damage, and the shedding rate. Therefore, real-time measurement of the combing roller's operating speed is also required. The operating speed data of the combing roller and the characteristic temperature data of the combing zone are core data during the combing process; therefore, data acquisition equipment such as sensors will be used to collect the characteristic operating data of the combing zone in real time.

[0058] Step S10 in the method provided in this application embodiment includes:

[0059] The fiber fineness monitoring data is collected in real time by an online fiber fineness monitoring component configured at the feeding mechanism of the target carding machine, wherein the fiber fineness monitoring data includes at least the average diameter of the fed fiber;

[0060] By using an encoder assembly and a non-contact temperature measurement assembly configured at the combing roller of the target combing machine, the running speed data of the combing roller and the characteristic temperature data of the combing zone are obtained respectively, and the output is the running characteristic data of the combing zone.

[0061] In this embodiment, the feeding mechanism is a mechanical device in the carding machine that continuously and uniformly feeds fiber raw materials into the carding zone for processing; the fiber fineness online monitoring component is an instrument or sensing component on the production line that can continuously measure fiber fineness without interrupting the production process; the average diameter of the fed fiber is the arithmetic mean or weighted average of the diameters of all individual fibers in the measured fiber sample; the carding roller is a high-speed rotating cylindrical roller in the carding machine, with its surface covered with a specific density of carding cloth; the encoder component is a precision sensor that converts the position, angle, or speed of the rotational motion into electrical or digital signals, and is installed on the carding roller shaft; the non-contact temperature measurement component is a device that can measure the temperature of the object without contacting its surface.

[0062] Specifically, firstly, a fiber fineness online monitoring component configured at the feeding mechanism of the target carding machine is used to collect fiber fineness monitoring data in real time. This data includes at least the average diameter of the fed fibers, and may also include length distribution and impurity content information. The fiber fineness online monitoring component includes at least one of optical projection, laser scattering, and digital image analysis components.

[0063] A fiber fineness online monitoring component is installed above or to the side of the target carding machine. As the fiber bundle is fed forward, it passes through the detection window of the component. The component continuously captures and analyzes images of the passing fiber bundle, calculating the average diameter of the currently flowing fibers in real time as the fiber fineness monitoring data. For example, at a certain moment, several fiber images are acquired and processed, and the average diameter of the fiber bundle is calculated to be 20.5 micrometers. If the raw material of the fiber bundle changes, for example, switching to an average diameter of 22.0 micrometers, the component will immediately detect the change and update the data, thereby recalculating and adjusting the fiber fineness monitoring data.

[0064] Secondly, by using an encoder assembly and a non-contact temperature measurement assembly configured at the combing roller of the target combing machine, the operating speed data of the combing roller and the characteristic temperature data of the combing zone are acquired respectively. The encoder assembly collects the operating speed data of the combing roller during operation, and the non-contact temperature measurement assembly collects the characteristic temperature data of the combing zone during operation. These two types of data are then combined and output as the operating characteristic data of the combing zone. Through the aforementioned separately configured encoder assembly and non-contact temperature measurement assembly, the operating speed and characteristic temperature data are independently and synchronously acquired simultaneously without interference, and used as input data for the subsequent first relational model.

[0065] In this embodiment of the application, by synchronously monitoring the core properties of raw materials and the core operating conditions of equipment online, the discrete and lagging offline detection and experience judgment in the traditional combing process are transformed into continuous and real-time data acquisition, thereby obtaining the operating characteristic data of the combing zone and providing accurate data input for subsequent intelligent decision-making.

[0066] S20: Input the fiber fineness monitoring data into the pre-constructed first relationship model to obtain the matching target working condition needle density;

[0067] In this embodiment of the application, the first relationship model is a pre-established data model or rule set used to construct a mapping relationship between fiber fineness and recommended needle density; the target working condition needle density is an initial reference value of needle density obtained from historical data based on the current raw material fiber fineness.

[0068] Specifically, the fiber fineness monitoring data collected above is input into the pre-constructed first relational model. After mapping or reasoning calculations within the model, the matching target working condition needle density is obtained. The working condition needle density can reflect the actual combing intensity that the fiber bears during the combing process, which is convenient for optimization and adjustment in actual dynamic operation.

[0069] In step S20 of the method provided in this application embodiment, the construction of the first relation model includes:

[0070] Obtain historical production data of the target combing machine, wherein the historical production data includes historical fiber fineness data and corresponding needle density records of high-quality output slivers;

[0071] Data from the historical production data that are identical to the target combing machine and the real-time raw material type are selected as sample data.

[0072] Based on the sample data, analysis and training are performed to establish the first relational model with fiber fineness as input, wherein the first relational model is constructed based on either a mapping function or a rule base.

[0073] In this embodiment, historical production data is a multi-dimensional data set related to the production process and quality results stored in the system during the historical production process of the target carding machine; historical fiber fineness data is the average diameter data of raw material fibers obtained and recorded through online monitoring or offline detection during the historical production process; and the needle density record of high-quality output tops is the needle density of the carding machine during the production process of high-quality finished tops in the historical data.

[0074] Specifically, firstly, the historical production data of high-quality output tops is obtained from the historical production database of the target carding machine. The average diameter data of the raw fiber is extracted from the historical production data as historical fiber fineness data. At the same time, the carding machine tooth density from the high-quality finished top production data is obtained as the tooth density record of high-quality output tops. After obtaining the historical production data of the target carding machine, the obtained historical fiber fineness data and the corresponding high-quality output top tooth density record are combined to form the corresponding data: [Historical Fiber Fineness Data - Tooth Density Record].

[0075] For example, a record of historical production data obtained from the database could be: a batch of wool was processed, and the historical fiber fineness was recorded as 19.8 micrometers. Simultaneously, the produced wool tops were tested and rated as high-quality, and the carding machine's needle density was set to 410 teeth / square inch.

[0076] Secondly, data from historical production data that are of the same origin as the target combing machine and the real-time raw material type are selected as sample data.

[0077] Specifically, historical production data with the same equipment model and raw material type as the target carding machine are selected from historical production data to form sample data. The same raw material type includes, for example, the variety of processed fibers. Through this selection, data similar to the target carding machine and the raw material type used by it are obtained, constructing sample data. This reduces systematic errors between equipment and errors caused by the type of production raw material, ensuring the internal consistency of the sample data set and maximizing its relevance to the current task. This lays the data foundation for building a highly targeted and accurate model.

[0078] For example, suppose the historical data retrieved from the database contains mixed records of Merino wool and cashmere from three different models of carding machines (Model A, Model B, and Model C). Currently, the target carding machine is Model A, and the real-time raw material type to be processed is Merino wool. First, only the data from carding machine A is retained; second, within the data from carding machine A, only the data labeled as the raw material type Merino wool is retained. After these two filtering steps, a set of data might be obtained: {(19.8,410),(21.2,380),(20.1,400),(22,375)}, which will be used as sample data for constructing the first relational model.

[0079] Next, based on the sample data, analysis and training are performed to establish a first relational model with fiber fineness as input. The first relational model is constructed based on either a mapping function or a rule base. The analysis and training uses mathematical or computer methods to discover patterns, fit relationships, or extract rules from the provided sample data. The mapping function is a function that can map input values ​​to output values. The rule base is a collection containing logical judgment rules.

[0080] Specifically, the sample data mentioned above is analyzed and trained, and a first relational model can be constructed based on a mapping function or a rule base. A suitable function can be selected from linear functions, polynomial functions, etc., to construct the mapping function. Assuming that the least squares method is used, with fiber fineness as input x and recommended needle density as output y, a linear function is constructed, resulting in the fitted function y = b - ax, where a and b are the coefficients of the linear function. If a rule base is constructed, assuming it is built using clustering and interval partitioning, the rules are: IF fineness ∈ [n1, n2) THEN density = y1; IF fineness ∈ [n2, n4) THEN density = y2; IF fineness ∈ [n4, n5) THEN density = y3. According to the logical judgment rules of the rule base, the corresponding target working condition needle density is output.

[0081] For example, if a mapping function is established, and the fitted function is y = b - ax, then if a and b are 5 and 500 respectively, then y = 500 - 5x. Therefore, if the wool fiber fineness is 20.5 micrometers, the linear function calculates the target working condition needle density as 500 - 5 × 20.5 ≈ 398 teeth / square inch. If a rule base is established: IF fineness ∈ [19.5, 20.5) THEN density = 400; IF fineness ∈ [20.5, 21.5) THEN density = 385; IF fineness ∈ [21.5, 22.5) THEN density = 375. Since the wool fiber fineness of 20.5 micrometers falls within the second interval [20.5, 21.5), the target working condition needle density output by the rule base is 385 teeth / square inch.

[0082] In this embodiment, by acquiring high-quality historical production data of a specific carding machine that processes a specific type of raw material, and selecting data from the same source as the target carding machine and the real-time raw material type as sample data based on the historical production data, a first relational model is constructed through a mapping function or rule base. This achieves a high degree of personalization, conforms to the unique mechanical characteristics of the equipment and the unique processing requirements of the raw material, and makes the output target working condition needle density more accurate and reliable.

[0083] S30: Construct a working condition density distribution model of the needle tooth density of the target combing machine, and initialize the working condition density distribution model based on the target working condition needle tooth density;

[0084] In this embodiment of the application, the working condition density distribution model describes the distribution of the needle density in the combing zone around a certain central value under actual production and operation conditions; the apparent needle density is the theoretical needle cloth density value set manually or by the control system command; the working condition needle density is the actual value of the needle density that actually participates in the combing process due to the influence of the above-mentioned random factors; initialization is the process of assigning initial parameter values ​​to the newly established mathematical model.

[0085] Specifically, in actual production, the fluctuation of needle density is investigated by constructing a working condition density distribution model of the needle density of the target combing machine, using the target working condition needle density to initialize the model center, and statistically analyzing the fluctuation of needle density through the working condition density distribution model to obtain the working condition density distribution of needle density with the standard deviation parameter as the variance, thereby quantifying the uncertainty of needle density.

[0086] Step S30 in the method provided in this application embodiment includes:

[0087] Based on the historical production data of the target combing machine, multiple sets of sample apparent needle density and sample working condition needle density distribution are extracted, and the distribution basis model of the working condition density distribution model is determined by matching accordingly.

[0088] By combining the aforementioned distribution base model, the apparent needle density of multiple sets of samples, and the needle density distribution under operating conditions of the samples, the distribution parameters of the operating condition density distribution model are analyzed and determined.

[0089] The target working condition needle density is used as the distribution center of the working condition density distribution model, and initialization is performed in combination with the distribution base model and the distribution parameters.

[0090] In this embodiment, the apparent needle density of the sample is the theoretical value of the needle density reached by the target combing machine in a past production operation extracted from historical data; the needle density distribution of the sample working condition is the probability distribution of the effective needle density that actually participates in the combing action in actual production operation, corresponding to the apparent needle density of the sample; the distribution basis model is a mathematical probability distribution type used to describe the needle density distribution of the sample working condition.

[0091] Specifically, firstly, based on the historical production data of the target combing machine, multiple sets of sample apparent needle density are extracted. Then, based on the sample apparent needle density, quality inspection is performed on the output of the historical production data for a certain period to obtain the corresponding sample working condition needle density. Similarly, the same processing is performed on multiple sets of sample apparent needle density to obtain multiple sets of corresponding sample working condition needle density. Then, from distribution types such as normal distribution, uniform distribution, and triangular distribution, the distribution type that best describes the fluctuation pattern of historical needle density data is selected as the distribution base model of the working condition density distribution model.

[0092] For example, for a target carding machine processing merino wool, the apparent needle density of samples produced during a historical period is extracted from the database and set to 400 teeth / square inch. High-frequency quality testing is performed on the wool tops produced during this period, and the results are back-calculated to obtain a set of estimated needle density values ​​for sample working conditions: [398, 402, 399, 397, 403, 400, 401, 396, 399, 402] (teeth / square inch). Multiple sets of sample working condition needle density data are obtained, and through goodness-of-fit testing, it is determined that the multiple sets of sample working condition needle density data all conform well to a normal distribution. Therefore, the normal distribution is determined as the distribution basis model describing the working condition needle density distribution of the target carding machine.

[0093] Secondly, by combining the distribution base model, the apparent needle density of multiple samples and the needle density distribution under sample working conditions, the distribution parameters of the working condition density distribution model are analyzed and determined. Among them, the distribution parameters are numerical constants used to specifically define a particular distribution in the selected distribution base model.

[0094] Specifically, by combining the distributed basis model, the apparent pin density of multiple samples, and the pin density distribution under various operating conditions, the standard deviation of the pin density distribution under each sample operating condition is calculated. Then, the average or median of the standard deviations of all groups is taken to obtain the standard deviation representing the inherent fluctuation level of the equipment under various settings. A smaller standard deviation indicates that the equipment executes accurately and with minimal fluctuation; a larger standard deviation indicates that the control effect is discrete and susceptible to interference. In subsequent optimization, this will directly affect the optimization algorithm's evaluation of the setpoint adjustment step size and risk.

[0095] For example, analyzing three sets of historical data extracted from a target combing machine processing Merino wool: The first set has an apparent needle density of 400, with historical data of [398, 402, 399, 397, 403, 400, 401, 396, 399, 402]. The standard deviation is calculated to be σ1≈2.2. The second set has an apparent needle density of 380, with historical data of [378, 383, 379, 377, 382, ​​380, 381, 376, 378, 384]. The standard deviation is calculated to be σ2≈2.5. The third set has an apparent needle density of 420, with historical data of [417, 423, 418, 416, 422, 419, 421, 415, 417, 424]. The standard deviation is calculated to be σ3≈2.3. Mean standard deviation: (2.2+2.5+2.3) / 3≈2.3 teeth / square inch. This means that for a target carding machine processing merino wool, the actual density has approximately a 95% probability of falling within the set value ±4.6 teeth / square inch.

[0096] Finally, the needle density of the target working condition is used as the distribution center of the working condition density distribution model. The distribution base model and distribution parameters are combined to perform initialization. In the symmetric distribution, the distribution center refers to the mean, which is the position where the distribution is most likely to take a value, representing the point with the highest probability density.

[0097] Specifically, the target working condition needle density is initialized using the distribution center, distribution basis model, and distribution dispersion parameter of the working condition density distribution model, generating a probability distribution function, such as a normal distribution N(μ,σ). 2 ).

[0098] For example, the probability distribution function is obtained as: normal distribution N(395,2.3²).

[0099] In this embodiment, historical data analysis quantifies the random deviation between the apparent needle tooth density of the target working condition and the needle tooth density distribution of the sample working condition. Historical production data is used to extract the apparent needle tooth density of the target working condition and the needle tooth density distribution of the sample working condition, and the distribution basis model and parameters are determined. The distribution basis model describes the inherent fluctuation level of a specific device when executing density commands, quantifying the execution accuracy. Subsequently, the target working condition needle tooth density, based on the fineness of the raw material, is initialized as the distribution center. The deterministic target working condition apparent needle tooth density and the distribution basis model of the sample working condition needle tooth density are combined. Through the distribution basis model, the optimal probability distribution is obtained, thereby reducing fluctuations caused by various factors during production, i.e., reducing errors, and enhancing the stability and adaptability of the system.

[0100] S40: Combining the working condition density distribution model, the combing zone operation characteristic data and the pre-constructed second relationship model, the desired working condition needle tooth density is calculated, and the working condition density distribution model is iteratively solved and adjusted with the desired working condition needle tooth density approaching the target working condition needle tooth density as the optimization objective.

[0101] In this embodiment, the second relational model is a model constructed based on physicochemical mechanisms, used to quantify the influence of the combing machine's operating state on the effective combing effect; the desired working condition needle density is a needle density value calculated based on the first and second relational models; the iterative solution is a process of repeatedly iterating, calculating the corresponding output, comparing it with the target value, and deciding how to adjust it in the next step based on the comparison results, until the parameters that make the two closest are found; approximation is a process of optimizing and adjusting to make the calculated desired working condition needle density infinitely close to or equal to the target value.

[0102] Specifically, the working condition density distribution model, the combing zone operation characteristic data, and the second relationship model are used to output the expected working condition needle tooth density after integrating dynamic influences. Subsequently, the working condition density distribution model is iteratively optimized by adjusting the center of the distribution model, with the optimization objective of the expected working condition needle tooth density approaching the target working condition needle tooth density.

[0103] In step S40 of the method provided in this application embodiment, the construction of the second relation model includes:

[0104] Based on a prior knowledge graph, a first mechanism model of the target combing machine is constructed, wherein the first mechanism model takes the combing operation characteristic data as input and outputs a first equivalent density adjustment coefficient to characterize the influence of centrifugal force.

[0105] Based on a prior knowledge graph, a second mechanism model of the target combing machine is constructed. The second mechanism model takes the combing operation characteristic data and the working condition density distribution model as inputs and outputs a second equivalent density adjustment coefficient to characterize the influence of frictional temperature rise.

[0106] The second relationship model is constructed by combining the first mechanism model and the second mechanism model.

[0107] In this embodiment, the prior knowledge graph is a set of domain knowledge that already exists before the model is constructed. In this application, the prior knowledge graph is a connection relationship of the physical, chemical, and mechanical principles of wool spinning and combing processes, including but not limited to: force analysis of fibers on the high-speed rotating needle surface, fiber material mechanics, tribothermodynamics, etc.; the first equivalent density adjustment coefficient is a correction coefficient calculated by the first mechanism model.

[0108] Specifically, the prior knowledge graph is a structural map of the physical, chemical, and mechanical principles of wool combing technology that is pre-constructed before building the model. Before constructing the first mechanism model, the pre-constructed prior knowledge graph is first invoked, and the combing operation characteristic data is used as the model input to construct the first mechanism model of the target combing machine. This model can calculate the first equivalent density adjustment coefficient using the formula k1=1-α×(v2×r) / F, where v is the surface linear velocity of the combing roller, r is the radius of the combing roller, F is the typical gripping force between the fiber and the needle teeth, and α is an empirical coefficient related to the fiber type. If the first equivalent density adjustment coefficient is 1, it indicates no effect; less than 1 indicates a reduction in effective density; and greater than 1 indicates a positive effect under certain special conditions. However, under normal circumstances, the first equivalent density adjustment coefficient is less than 1.

[0109] For example, in a target combing machine for processing Merino wool, the diameter of the combing roller is 0.3 meters. Based on a prior knowledge graph, a first mechanistic model is constructed, in which the rotational speed of the combing roller is 850 rpm, and the surface linear velocity v = π × diameter × rotational speed = 3.14 × 0.3 × 14.17 ≈ 13.35 m / s. k1 = 1 - 0.00012 × (13.352) = 0.98, indicating that at the current high speed of 850 rpm, the centrifugal force effect makes the actual combing density approximately 98% of the set density.

[0110] Secondly, based on the prior knowledge graph, a second mechanism model of the target carding machine is constructed. The second mechanism model takes the carding operation characteristic data and the working condition density distribution model as inputs and outputs a second equivalent density adjustment coefficient to characterize the influence of frictional temperature rise. The second mechanism model is a mathematical model used to describe the temperature rise in the carding zone caused by friction, which in turn affects the physical state of the fiber and the carding effect. The second equivalent density adjustment coefficient is another dimensionless correction coefficient calculated by the second mechanism model.

[0111] Specifically, based on the prior knowledge graph containing fiber glass transition temperature, the relationship between friction coefficient and temperature, and heat conduction equations, a second mechanistic model of the target carding machine needs to be constructed, taking into account the equipment's heat capacity and heat dissipation conditions. Carding operation characteristic data and the working condition density distribution model are used as inputs, utilizing the carding operation characteristic temperature data and carding operation speed data from the carding operation characteristic data. A comprehensive evaluation is conducted on the net impact of changes in fiber state on the carding effect under the current temperature and density distribution, and compensation is made for the current temperature and density distribution using a second equivalent density adjustment coefficient.

[0112] The second equivalent density adjustment coefficient k2 = 1 + β × (T - T0) × (z / D), where T is the measured characteristic temperature, T0 is the reference temperature, μ is the mean of the density distribution model under the current working condition, D is the reference equivalent density, and β is the gain coefficient, which is usually 0.005. If the temperature is too low and the fiber becomes stiff, k2 may be less than 1, indicating that a higher apparent density is needed to overcome the combing resistance. If the temperature is in the optimal range, k2 > 1. If the temperature is too high, k2 may decrease rapidly or even be less than 1, indicating that the density needs to be reduced or cooling measures need to be taken to prevent damage.

[0113] For example, based on a prior knowledge graph, it is determined that Merino wool fibers soften moderately at 40-50°C, which facilitates combing, while the risk of damage increases above 60°C. A second mechanism model is constructed, where the second equivalent density adjustment coefficient k2 = 1 + β × (T - T0) × (μ / D). If the current measured temperature T = 42°C, the mean μ = 395 is obtained from the initialized working condition density distribution model N(395, 2.3²). Then, k2 = 1 + 0.005 × (42 - 35) × (395 / 400) ≈ 1.03, indicating that the current temperature of 42°C and the expected effective density of approximately 395 result in slight thermal softening, increasing the effective combing density by approximately 3%.

[0114] Finally, by combining the first mechanism model and the second mechanism model, a second relationship model is constructed. The combination involves integrating two sub-models that represent different physical influences into a holistic model that can comprehensively output decision information through certain mathematical rules or logical structures.

[0115] Specifically, the first mechanism model and the second mechanism model are connected in parallel to structurally merge the two models and construct a second relationship model that integrates the effects of centrifugal force and frictional heat on the combing effect. After inputting the working condition density distribution model and the combing operation characteristic data into the second relationship model, the first mechanism model and the second mechanism model are calculated simultaneously, and the final fusion coefficient is obtained by merging the calculation results.

[0116] like Figure 2 As shown, step S40 in the method provided in this application embodiment includes:

[0117] Input the working condition density distribution model and the combing operation characteristic data of the current iteration step into the second relationship model;

[0118] The first equivalent density adjustment coefficient and the second equivalent density adjustment coefficient under the current iteration step are calculated and output through the second relationship model, and the distribution center of the working condition density distribution model is adjusted accordingly.

[0119] Based on the adjusted working condition density distribution model, the mathematical expectation is calculated as the expected working condition needle tooth density, and the difference between the expected working condition needle tooth density and the target working condition needle tooth density is calculated.

[0120] With the goal of minimizing the difference, the distribution center of the working condition density distribution model is iteratively adjusted using an optimization algorithm until the difference is less than a preset convergence threshold.

[0121] In this embodiment, the current iteration step is a specific loop in the execution of the iterative optimization algorithm. The algorithm starts from the initial value, and each attempt to update the variable to be optimized constitutes an iteration step; the working condition density distribution model of the current iteration step is the state of the needle density probability distribution model to be optimized at the beginning of this iteration.

[0122] Specifically, the working condition density distribution model of the current iteration step and the combing operation characteristic data are first input into the second relationship model.

[0123] Specifically, in the first iteration, the initialized working condition density distribution model N(μ,σ) is... 2 The current proposed distribution center is μ. The real-time collected combing operation characteristic data ({velocity, temperature}) are input into the pre-built second relational model, which then prepares to begin calculations based on these inputs. Similarly, in the Mth iteration, the data from the operating condition density distribution model and the corresponding combing operation characteristic data are input into the corresponding second relational model, which then prepares to begin calculations based on these inputs.

[0124] Secondly, the first equivalent density adjustment coefficient and the second equivalent density adjustment coefficient under the current iteration step are calculated and output through the second relational model, and the distribution center of the working condition density distribution model is adjusted accordingly.

[0125] Specifically, the second relational model contains a comprehensive adjustment coefficient K, which is the product of the first equivalent density adjustment coefficient and the second equivalent density adjustment coefficient, i.e., K = k1 × k2. For the initial operating condition density distribution model and real-time operational characteristic data, the second relational model first calls the first mechanism model to calculate k1, then calls the second mechanism model to calculate k2, and outputs the comprehensive coefficient K. Subsequently, using the adjustment coefficient calculated by the second relational model, the distribution center and mean μ of the input operating condition density distribution model are corrected once, resulting in an adjusted distribution center μ' = μ / K.

[0126] For example, under the current operating conditions, the first mechanism model outputs: k1=0.98; the second mechanism model outputs: k2=1.03. The calculation result of the second relational model for the current input, i.e. the comprehensive adjustment coefficient, is: K=0.98×1.03≈1.01, μ=395. The adjusted distribution center μ'=395 / 1.01≈391.3 is obtained, and the adjusted operating condition density distribution model is N(391.3,2.3²).

[0127] Next, based on the adjusted working condition density distribution model, the expected value is calculated as the desired working condition needle tooth density, and the difference between the desired working condition needle tooth density and the target working condition needle tooth density is calculated. The expected value is the weighted average of all possible values ​​of a random variable multiplied by their probabilities. For a normal distribution N(μ,σ²), the expected value is equal to its mean μ. Calculating the expected value involves reading the distribution center value of the adjusted working condition density distribution model. The desired working condition needle tooth density is the level of needle tooth density expected to be achieved under the current dynamic working conditions after the physical compensation adjustment in this iteration, and is equal to the expected value of the adjusted distribution model. The target working condition needle tooth density is the ideal target value based on the fineness of the raw material obtained from the first relational model. The difference is the arithmetic difference between the desired working condition needle tooth density and the target working condition needle tooth density.

[0128] The distribution center obtained after physical compensation correction is directly extracted as the output prediction value for this iteration, i.e., the desired working condition needle tooth density. The desired working condition needle tooth density is compared with the target working condition needle tooth density, and the difference between the two is calculated. The smaller the difference, the better the current proposed solution. Therefore, the next step is to find an optimal distribution center that minimizes this difference.

[0129] For example, based on the adjusted working condition density distribution model N(391.3,2.3²), its mathematical expectation is calculated, that is, the expected working condition needle tooth density = 391.3 teeth / square inch, the target working condition needle tooth density = 395 teeth / square inch, and the difference between the two is calculated as |391.3-395| = ​​3.7 teeth / square inch.

[0130] Ultimately, with the goal of minimizing the difference, the distribution center of the working condition density distribution model is iteratively adjusted using an optimization algorithm until the difference is less than a preset convergence threshold. The optimization algorithm determines how to modify the distribution center of the working condition density distribution model based on the difference in the current iteration step and possible gradient information to begin the next iteration. Iterative adjustment involves the optimization algorithm generating a new distribution center value according to its rules, updating the working condition density distribution model, and performing an optimization loop. The convergence threshold is a pre-set criterion for stopping iterations; when the difference calculated in a certain iteration is less than this threshold, the current distribution center is considered sufficiently close to the optimal solution, and the optimization process terminates.

[0131] For example, using a one-dimensional search optimization algorithm, after receiving the difference from the first iteration, the algorithm determines whether the trial distribution center needs to be increased based on the difference. If the difference between the desired working condition needle density and the target working condition needle density is less than a preset convergence threshold, the iteration continues. First, the optimization algorithm generates a new distribution center value according to its rules, updates the working condition density distribution model, and begins the second iteration. In the second iteration, the updated working condition density distribution model and new combing operation feature data are input into the second relation model. The model calculates a new adjustment coefficient, then adjusts the distribution center, recalculates the difference, and the algorithm again judges based on the difference and the preset convergence threshold. After multiple such iterations, the difference is made less than the preset convergence threshold, satisfying the stopping condition. Finally, a working condition density distribution model that satisfies the optimization objective is obtained, and its distribution center is the optimal apparent needle density setting value to be output.

[0132] In this embodiment, a first mechanism model and a second mechanism model are constructed respectively. Then, the centrifugal force and frictional temperature rise mechanisms are integrated and coupled in parallel to obtain a second relational model. This second relational model can calculate comprehensive adjustment parameters based on the first and second mechanism models, combining the predictive power and dynamic compensation capabilities of physical principles to quantify the impact of operational state changes on the combing effect in real time. Subsequently, the distribution center of the working condition density distribution model is adjusted according to the comprehensive adjustment coefficient using the second relational model. The difference between the desired working condition needle density and the target working condition needle density is then calculated, with the goal of minimizing the difference, until the difference is less than a preset convergence threshold, at which point the working condition density distribution model is updated.

[0133] S50: Based on the working condition density distribution model that satisfies the optimization objective, output the optimal apparent needle density setpoint, generate control commands, and drive the needle density actuator of the combing machine to perform actions.

[0134] In this embodiment, the optimal apparent needle density setting value is the needle density value that is finally determined after optimization and iteration and needs to be implemented by the physical actuator; the control command is a digital command generated by the control system that can be recognized and executed by the downstream actuator; the needle density actuator is a mechanical device that can physically change the needle density in the combing area.

[0135] Specifically, the distribution center, i.e., the optimal apparent needle tooth density setpoint, is extracted from the optimal operating condition density distribution model. This optimal apparent needle tooth density setpoint is the globally optimal process parameter setpoint determined by the current raw material characteristics and the current equipment operating state. Subsequently, the optimal apparent needle tooth density setpoint is converted into control commands that the equipment can understand and execute. After the control commands are issued, the actuator begins to operate, completing the adjustment of the needle tooth density.

[0136] Step S50 in the method provided in this application embodiment includes:

[0137] From the working condition density distribution model that satisfies the optimization objective, extract the distribution center setting value and output the distribution center setting value as the optimal apparent needle density setting value.

[0138] Based on the optimal apparent needle density setting value, the control command based on machine language is generated and sent to the needle density actuator to perform needle density control.

[0139] The needle density actuator includes either a needle cloth mechanical mechanism with dynamically adjustable needle density or a needle cloth unit group with differentiated preset needle density.

[0140] In this embodiment of the application, after iterative optimization, the distribution center setting value, i.e. the mean parameter of the normal distribution, is extracted from the final working condition density distribution model and used as the optimal apparent needle density setting value for this control cycle, representing the ideal combing effect obtained under the current fiber fineness monitoring data and combing zone operation characteristic data.

[0141] For example, the final working condition density distribution model that satisfies the optimization objective is N(400, 2.3²). Extracting the distribution center setting value μ'=400 from it, the optimal apparent needle density setting value is obtained as 400 teeth / square inch.

[0142] Secondly, based on the optimal apparent needle density setting value, machine language-based control instructions are generated and sent to the needle density actuator to execute needle density control. The sending is achieved through physical communication links such as industrial fieldbus and I / O ports, reliably transmitting the generated digital instructions from the upper-level controller to the local controller or driver of the designated needle density actuator. Execution of needle density control involves the actuator's internal drive unit starting to work after receiving the sent control instructions, physically changing the density configuration of the card cloth in the combing area so that the actual needle density reaches or approaches the optimal apparent needle density setting value required by the instructions.

[0143] Specifically, the control system calculates the target position based on the optimal apparent needle density setpoint and then generates control commands based on machine language. The control system sends these control command messages to the needle density actuators, such as servo drives, via a physical communication link. Upon receiving the commands, the drive unit of the needle density actuator adjusts the electromechanical components accordingly, generating precise mechanical displacement or state switching, thus physically changing the density configuration of the card fabric in the combing zone and completing the needle density control.

[0144] For example, the optimal apparent tooth density setting of 400 teeth / square inch is translated into control instructions that the actuator can execute. For instance, writing data 4000 to the target position register at address 1 of the servo driver, the control system sends the above instruction to the servo driver at address 1. This drives the servo motor to rotate, causing the lead screw to move precisely to a position of 4.00 millimeters, thus adjusting the actual tooth density to approximately 400 teeth / square inch.

[0145] The needle density actuator includes either a needle cloth mechanical mechanism with dynamically adjustable needle density or a needle cloth unit group with differentiated preset needle density. The needle density actuator is a collection of electromechanical devices or components that can receive control commands and directly or indirectly change the physical state of the needle cloth density in the combing zone of the combing machine. The needle cloth mechanical mechanism is a mechanical system that can realize continuous or high-precision step adjustment of the needle density. The needle cloth unit group consists of multiple needle cloth units with different fixed needle density that are pre-prepared.

[0146] Specifically, the needle cloth mechanism uses a needle cloth with dynamically adjustable needle tooth density, for example, by changing the lateral spacing of the needle teeth using a micro servo motor. However, this method is difficult to implement mechanically, costly, and its stability needs further verification. The alternative approach, which switches needle cloth units without changing the density of a single piece of needle cloth, utilizes multiple sets of needle cloth units with different preset needle tooth densities. Based on the fiber condition, it automatically selects and switches to the most suitable set of needle cloths for operation. This results in lower mechanical complexity, lower cost, and greater reliability, making it more feasible. By employing these two approaches, this patented method can adapt to wool mills with different technical approaches, cost budgets, and levels of automation, enhancing its practical value and market coverage.

[0147] Step S50 in the method provided in this application embodiment further includes:

[0148] Record the operation log of the target carding machine, and combine the operation log with preset posterior constraints to obtain the quality feedback data of the output sliver.

[0149] Based on the quality feedback data and the corresponding optimal apparent needle density setting value, posterior feedback data is analyzed and screened to obtain the posterior feedback data.

[0150] Based on the posterior feedback data, the distribution parameters in the working condition density distribution model are updated using Bayesian methods.

[0151] In this embodiment, the operation log is the time-series data about the production process that the system continuously records outside the main control cycle; the posterior constraint is the prerequisite or rule that must be met to trigger the system to switch from the control execution mode to the model learning mode; the quality feedback data is the online quality monitoring sensor installed at the output end or the offline test report of the laboratory, which is the objective data for evaluating the optimal apparent needle density set value.

[0152] Specifically, the operation log of the target combing machine is first recorded, and the quality feedback data of the output sliver is obtained by combining the operation log with the preset posterior constraints.

[0153] Outside the main control closed loop, a learning optimization closed loop is established to continuously record time-series data about the production process, obtain preset posterior constraints, and compare the cumulative duration in the current operation log with the preset posterior constraints. If the update conditions are met, subsequent actions are triggered to obtain quality feedback data within the corresponding time period, thereby ensuring that the data samples used for learning are sufficient and representative, and avoiding incorrect learning due to accidental fluctuations or transient processes.

[0154] Secondly, based on the quality feedback data and the corresponding optimal apparent needle density setting value, posterior feedback data is obtained through analysis and screening. The analysis and screening involves jointly processing and judging the obtained raw quality feedback data and setting value to select data points or data pairs that are valuable and reliable for model updates. The posterior feedback data are high-quality data pairs that are retained after analysis and screening.

[0155] Determine whether the acquired quality feedback data meets the optimal apparent needle density setting. If not, it indicates quality failure and the data is discarded. Filter all quality feedback data to obtain the updated posterior feedback data.

[0156] Finally, based on the posterior feedback data, the distribution parameters in the working condition density distribution model are updated using Bayesian methods. Bayesian updating treats the initial understanding of the model parameters as a prior distribution. After obtaining new observation data, the prior distribution is combined with the new evidence according to Bayes' theorem to obtain an updated posterior distribution, which serves as a new understanding of the parameters.

[0157] Based on posterior feedback data, high-quality data pairs were obtained: (optimal apparent needle density setpoint, observed needle density). Based on historical data, the standard deviation σ is assumed to follow a normal distribution N(μ,σ). 2 The new observed needle density provides a new sample of deviations between the setpoint and the actual value. First, the posterior distribution of the standard deviation is re-estimated. If the new sample shows deviations within the expected range of fluctuation as predicted by the prior standard deviation, the update to the standard deviation is minimal. If multiple consecutive new samples show deviations significantly smaller than the prior expectation, the posterior standard deviation may decrease, indicating improved equipment execution accuracy. Conversely, if the deviation increases and persists, the posterior standard deviation will increase, suggesting potential equipment wear and increased execution uncertainty. Subsequently, the updated standard deviation is used in the initialization of the operating condition density distribution model for all subsequent control cycles.

[0158] For example, assume the prior point estimate of the standard deviation σ of the current load condition density distribution model is 2.3. From the posterior feedback data, the observed needle density corresponding to the setpoint 400 is 398, i.e., the single observation bias is x = 398 - 400 = -2. After calculation using the Bayesian update formula, the new observation bias x = -2 is incorporated. The posterior expected value of the variance changes slightly after calculation, and the posterior point estimate of the standard deviation of the distribution parameter is updated to σ' = 2.28.

[0159] In step S50 of the method provided in this application embodiment, the posterior constraint includes at least one of the following:

[0160] Since the last Bayes update, the target comber has been running continuously for the preset time period.

[0161] The quality feedback data of the output yarn, which is monitored in real time, exceeds the preset quality fluctuation threshold.

[0162] In this embodiment, the continuous running time is the cumulative running time of the target combing machine in a stable production state from the last Bayes update to the current time; the preset time period is a time threshold set in the system by the system administrator based on factors such as production cycle, raw material batch size, and equipment status change rate; the preset quality fluctuation threshold is a pre-set numerical standard used to define the boundary between normal and abnormal quality fluctuations, representing the quality fluctuation limit that the process control can tolerate.

[0163] Specifically, since the last Bayesian update, the cumulative running time of the target comber in stable production mode reaches a preset period, at which point the posterior constraint is deemed satisfied. This is equivalent to a constraint on the running time dimension of the target comber. Every time a certain amount of running data is accumulated, the distribution parameter σ needs to be re-evaluated and calibrated to ensure the regularity and planning of model updates, preventing the model from becoming out of sync with the actual equipment state due to prolonged lack of learning.

[0164] Alternatively, the real-time monitored quality feedback data of the output felt may exceed a preset quality fluctuation threshold. In industrial processes, many factors can cause sudden quality fluctuations that are not necessarily related to equipment wear. When the quality feedback data exceeds the preset quality fluctuation threshold, the system will analyze the posterior feedback data within the abnormal time period. Through Bayesian updates, the standard deviation of the distribution parameter is adjusted. If the actual process effect fluctuation assessed by the quality feedback data is much greater than the preset threshold, the standard deviation of the distribution parameter is increased to reduce the reliance on the current equipment's execution accuracy. The updated operating condition density distribution model more actively adjusts the setpoints in subsequent optimizations to stabilize quality, enhancing the resilience of the control system to sudden disturbances and its ability to maintain quality stability.

[0165] The posterior constraint must include at least one of the following: the quality feedback data exceeds a preset quality fluctuation threshold or the continuous running time reaches a preset time period, before the prior point estimate of the standard deviation of the distribution parameters of the subsequent working condition density distribution model can be updated. If neither condition is met, it indicates that the quality deviation of the target self-combing machine's adjustment parameters is small, and no adjustment or calibration is required.

[0166] In this embodiment, the distribution center setpoint is extracted from the working condition density distribution model that meets the optimization objective, the optimal apparent needle density setpoint is obtained, and control commands are generated and executed to control the needle density, ensuring that intelligent decisions can be accurately and reliably translated into actual changes in the equipment's state. Simultaneously, based on a quality feedback-based posterior learning and Bayesian update mechanism, posterior feedback data is filtered, and the distribution parameters in the working condition density distribution model are updated using Bayesian methods. This enables the model to cope with instantaneous changes in working conditions and, by analyzing long-term production effects, automatically correct internal equipment execution uncertainties. This allows the model to track the slow degradation or improvement of equipment performance, achieving continuous stability and optimization of combing quality under long-term complex and changing conditions.

[0167] The embodiments of this application, through the above specific implementation methods, achieve the following technical effects:

[0168] In this embodiment, by synchronously monitoring the core properties of raw materials and the core operating conditions of equipment online, the discrete and lagging offline detection and experience judgment in the traditional combing process are transformed into continuous and real-time data acquisition, thereby obtaining the operating characteristic data of the combing zone and providing accurate data input for subsequent intelligent decision-making.

[0169] Secondly, by acquiring high-quality historical production data of a specific carding machine that processes specific raw material types, and selecting data from the same source as the target carding machine and the real-time raw material type as sample data based on the historical production data, a first relational model is constructed through mapping functions or rule bases. This achieves a high degree of personalization, conforms to the unique mechanical characteristics of the equipment and the unique processing requirements of the raw materials, and makes the output target working condition needle density more accurate and reliable.

[0170] Furthermore, through historical data analysis, the random deviation between the apparent needle tooth density distribution under the target operating condition and the needle tooth density distribution under the sample operating condition was quantified. Using historical production data, the apparent needle tooth density distribution under the target operating condition and the needle tooth density distribution under the sample operating condition were extracted, and the distribution basis model and parameters were determined. The distribution basis model describes the inherent fluctuation level of a specific piece of equipment when executing density commands, quantifying the execution accuracy. Subsequently, the needle tooth density under the target operating condition, based on the fineness of the raw material, was initialized as the distribution center. The distribution basis model of the deterministic apparent needle tooth density under the target operating condition and the needle tooth density under the sample operating condition were combined. Through the distribution basis model, the optimal probability distribution was obtained, thereby reducing fluctuations caused by various factors during the production process and enhancing the stability and adaptability of the system.

[0171] Simultaneously, a first mechanism model and a second mechanism model are constructed respectively. Then, the centrifugal force and frictional temperature rise mechanisms are integrated and coupled in parallel to obtain a second relational model. This second relational model can calculate comprehensive adjustment parameters based on the first and second mechanism models, combining the predictive power and dynamic compensation capabilities of physical principles to quantify the impact of operational state changes on the combing effect in real time. Subsequently, the distribution center of the working condition density distribution model is adjusted according to the comprehensive adjustment coefficient using the second relational model. The difference between the desired working condition needle density and the target working condition needle density is then calculated, with the goal of minimizing the difference until it falls below a preset convergence threshold, at which point the working condition density distribution model is updated.

[0172] Ultimately, from the working condition density distribution model that meets the optimization objective, the distribution center setpoint is extracted to obtain the optimal apparent needle density setpoint. Control commands are then generated and executed to control the needle density, ensuring that intelligent decisions are accurately and reliably translated into changes in the actual state of the equipment. Simultaneously, based on a posterior learning and Bayesian update mechanism using quality feedback, posterior feedback data is filtered, and the distribution parameters in the working condition density distribution model are updated using Bayesian methods. This allows the model to cope with instantaneous changes in working conditions and automatically correct internal equipment execution uncertainties by analyzing long-term production effects. The model can then track the slow degradation or improvement of equipment performance, achieving continuous stability and optimization of combing quality under long-term complex and changing conditions.

Claims

1. An adaptive control method for needle density of a wool spinning carding machine based on fiber fineness monitoring, characterized in that, include: Real-time acquisition of fiber fineness monitoring data and combing zone operation characteristic data of the target scene; The fiber fineness monitoring data is input into a pre-constructed first relationship model to obtain the matching target working condition needle density; Construct a working condition density distribution model of the needle tooth density of the target combing machine, and initialize the working condition density distribution model based on the target working condition needle tooth density; Combining the working condition density distribution model, the combing zone operation characteristic data and the pre-constructed second relationship model, the desired working condition needle tooth density is calculated, and the working condition density distribution model is iteratively solved and adjusted with the desired working condition needle tooth density approaching the target working condition needle tooth density as the optimization objective. Based on the working condition density distribution model that satisfies the optimization objective, the optimal apparent needle density setpoint is output, control commands are generated, and the needle density actuator of the combing machine is driven to perform actions.

2. The adaptive control method for needle density of a wool spinning carding machine based on fiber fineness monitoring as described in claim 1, characterized in that, The operating characteristic data of the combing zone includes at least the operating speed data of the combing rollers and the characteristic temperature data of the combing zone.

3. The adaptive control method for needle density of a wool spinning carding machine based on fiber fineness monitoring as described in claim 1, characterized in that, Real-time acquisition of fiber fineness monitoring data and combing zone operation characteristic data for the target scene, including: The fiber fineness monitoring data is collected in real time by an online fiber fineness monitoring component configured at the feeding mechanism of the target carding machine, wherein the fiber fineness monitoring data includes at least the average diameter of the fed fiber; By using an encoder assembly and a non-contact temperature measurement assembly configured at the combing roller of the target combing machine, the running speed data of the combing roller and the characteristic temperature data of the combing zone are obtained respectively, and the output is the running characteristic data of the combing zone.

4. The adaptive control method for needle density of a wool spinning carding machine based on fiber fineness monitoring as described in claim 1, characterized in that, The construction of the first relational model includes: Obtain historical production data of the target combing machine, wherein the historical production data includes historical fiber fineness data and corresponding needle density records of high-quality output slivers; Data from the historical production data that are identical to the target combing machine and the real-time raw material type are selected as sample data. Based on the sample data, analysis and training are performed to establish the first relational model with fiber fineness as input, wherein the first relational model is constructed based on either a mapping function or a rule base.

5. The adaptive control method for needle density of a wool spinning carding machine based on fiber fineness monitoring as described in claim 1, characterized in that, Constructing a working condition density distribution model of the needle density of the target combing machine, and initializing the working condition density distribution model based on the target working condition needle density, including: Based on the historical production data of the target combing machine, multiple sets of sample apparent needle density and sample working condition needle density distribution are extracted, and the distribution basis model of the working condition density distribution model is determined by matching accordingly. By combining the aforementioned distribution base model, the apparent needle density of multiple sets of samples, and the needle density distribution under operating conditions of the samples, the distribution parameters of the operating condition density distribution model are analyzed and determined. The target working condition needle density is used as the distribution center of the working condition density distribution model, and initialization is performed in combination with the distribution base model and the distribution parameters.

6. The adaptive control method for needle density of a wool spinning carding machine based on fiber fineness monitoring as described in claim 1, characterized in that, The construction of the second relational model includes: Based on a prior knowledge graph, a first mechanism model of the target combing machine is constructed, wherein the first mechanism model takes the combing operation characteristic data as input and outputs a first equivalent density adjustment coefficient to characterize the influence of centrifugal force. Based on a prior knowledge graph, a second mechanism model of the target combing machine is constructed. The second mechanism model takes the combing operation characteristic data and the working condition density distribution model as inputs and outputs a second equivalent density adjustment coefficient to characterize the influence of frictional temperature rise. The second relationship model is constructed by combining the first mechanism model and the second mechanism model.

7. The adaptive control method for needle density of a wool spinning carding machine based on fiber fineness monitoring as described in claim 1, characterized in that, Combining the operating condition density distribution model, the combing operation characteristic data, and the pre-constructed second relationship model, the desired operating condition needle density is calculated. Using the goal of approximating the target operating condition needle density with the desired operating condition needle density as the optimization objective, the operating condition density distribution model is iteratively solved and adjusted, including: Input the working condition density distribution model and the combing operation characteristic data of the current iteration step into the second relationship model; The first equivalent density adjustment coefficient and the second equivalent density adjustment coefficient under the current iteration step are calculated and output through the second relationship model, and the distribution center of the working condition density distribution model is adjusted accordingly. Based on the adjusted working condition density distribution model, the mathematical expectation is calculated as the expected working condition needle tooth density, and the difference between the expected working condition needle tooth density and the target working condition needle tooth density is calculated. With the goal of minimizing the difference, the distribution center of the working condition density distribution model is iteratively adjusted using an optimization algorithm until the difference is less than a preset convergence threshold.

8. The adaptive control method for needle density of a wool spinning carding machine based on fiber fineness monitoring as described in claim 1, characterized in that, Based on the working condition density distribution model that satisfies the optimization objective, the optimal apparent needle density setpoint is output, control commands are generated, and the needle density actuator of the combing machine is driven to perform actions, including: From the working condition density distribution model that satisfies the optimization objective, extract the distribution center setting value and output the distribution center setting value as the optimal apparent needle density setting value. Based on the optimal apparent needle density setting value, the control command based on machine language is generated and sent to the needle density actuator to perform needle density control. The needle density actuator includes either a needle cloth mechanical mechanism with dynamically adjustable needle density or a needle cloth unit group with differentiated preset needle density.

9. The adaptive control method for needle density of a wool spinning carding machine based on fiber fineness monitoring as described in claim 1, characterized in that, After generating control commands and driving the needle density actuator of the combing machine to perform actions, the process also includes: Record the operation log of the target carding machine, and combine the operation log with preset posterior constraints to obtain the quality feedback data of the output sliver. Based on the quality feedback data and the corresponding optimal apparent needle density setting value, posterior feedback data is analyzed and screened to obtain the posterior feedback data. Based on the posterior feedback data, the distribution parameters in the working condition density distribution model are updated using Bayesian methods.

10. The adaptive control method for needle density of a wool spinning carding machine based on fiber fineness monitoring as described in claim 9, characterized in that, The posterior constraints include at least one of the following: Since the last Bayes update, the target comber has been running continuously for the preset time period. The quality feedback data of the output yarn, which is monitored in real time, exceeds the preset quality fluctuation threshold.

Images