Method for predicting production quality of textile dyeing based on data recognition

Abstract

The present application relates to the technical field of industrial data analysis, in particular to a textile dyeing production quality prediction method based on data recognition, comprising the following steps: extracting dye molecule affinity constant, salt concentration parameter and temperature parameter, solving according to a thermodynamic adsorption model to generate equilibrium adsorption amount distribution data; extracting dye vat replenishment amount, roller pressure and cloth running speed in the continuous pad dyeing color changing production link. In the present application, space for adjusting liquid discharge can be reserved before the pollution peak approaches the processing limit, and single quality prediction can be expanded to comprehensive judgment considering product quality stability, degraded cloth length control, workshop color changing rhythm coordination and back-end processing load balancing, so that it is more beneficial to compress the quality fluctuation range, reduce the impact of invalid liquid discharge, reduce the mutual interference caused by continuous color changing, and improve the response accuracy of production organization to complex working condition changes in actual production.

Description

Technical Field

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

[0002] Industrial data analytics technology is based on multi-source data generated during industrial production. It involves collecting, integrating, processing, and analyzing equipment operating parameters, process parameters, environmental variables, and product quality testing results to achieve quantitative expression and pattern discovery of the production process status.

[0003] While existing technologies can centralize multi-source industrial data and perform general correlation analysis in practical operation, they are prone to problems when dealing with scenarios like textile dyeing, which involve continuous and significant changes, tight coupling, and quality formation processes spanning multiple time periods. These problems include insufficient correlation depth between parameters, discontinuous quality change chains, and difficulty in connecting the effects of processes sequentially. For example, when there are slight fluctuations in the dye liquor state, existing technologies often only detect color difference shifts in subsequent test results, but struggle to sequentially connect changes in adsorption equilibrium, dyeing operation status, the impact of residual discharge, and color tolerance exceedances. As a result, it is difficult to pinpoint the source of quality anomalies, accurately define the range of anomaly diffusion, and rely on empirical estimations for the length of degraded fabric. This further leads to a mismatch between the color change and discharge schedule and the actual pollution load. If multiple machines are in similar color change periods, there may be excessive concentration of discharge times, short-term accumulation of biochemical treatment pressure, and disordered queuing relationships in the workshop. Therefore, improvements are needed. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a data-based method for predicting the production quality of textile dyeing.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a method for predicting the production quality of textile dyeing based on data recognition, comprising the following steps:

[0006] The dye molecule affinity constant, salt concentration parameter, and temperature parameter are extracted and solved according to the thermodynamic adsorption model to generate equilibrium adsorption distribution data; the dye bath replenishment amount, rolling pressure, and fabric running speed in the continuous pad dyeing and color changing production process are extracted and the dyeing fluid dynamic state matrix is ​​solved.

[0007] Based on the dyeing fluid dynamics state matrix, the discharge flow rate and residual liquid concentration parameters are extracted to generate the time-series distribution curve of pollutants; the photoelectric colorimeter measurement parameters are extracted to construct the mapping relationship between transmittance and dye concentration; the photoelectric colorimeter measurement parameters are substituted into the mapping relationship for solution to generate the dye concentration distribution sequence; the dye concentration distribution sequence is mapped to the fabric apparent color prediction sequence according to the dyeing rate prediction model.

[0008] Extract the target color tolerance range parameter, compare each element in the fabric apparent colorimetric prediction sequence with the target color tolerance range parameter, and filter to obtain a colorimetric deviation element set; extract the time series corresponding to the colorimetric deviation element set, and calculate and generate the downgraded fabric length prediction value.

[0009] Extract the degradation limit parameters of the biochemical pool, compare the peak values ​​of the time-series distribution curves of the pollutants with the degradation limit parameters of the biochemical pool, and generate a workshop-level staggered production scheduling sequence by combining the predicted value of the degraded fabric length.

[0010] Preferably, the step of obtaining the dyeing fluid dynamics state matrix is ​​as follows:

[0011] The affinity constant of dye molecules, salt concentration parameters, and temperature parameters were extracted. The affinity constant of dye molecules, salt concentration parameters, and temperature parameters were matched according to a uniform time. The adsorption driving force, salt inhibition correction amount, and temperature compensation amount were calculated for each group. The adsorption increment and adsorption decrement of each group were recorded according to the numerical range of the calculation results to form the equilibrium adsorption amount distribution results.

[0012] Based on the equilibrium adsorption distribution results, the dyeing tank replenishment amount, rolling press pressure, and fabric running speed in the continuous dyeing and color changing production process are extracted. The fabric compression degree is calculated according to the change range of rolling press pressure. Then, the fabric porosity change parameter is estimated segment by segment according to the fabric compression degree. The transfer distance and residence time are allocated point by point in the running direction according to the fabric running speed. The dyeing tank replenishment amount is written into the corresponding position in time sequence to form the pore mass transfer coupling distribution results.

[0013] Based on the pore mass transfer coupling distribution results, the interval adsorption amount in the equilibrium adsorption amount distribution results is mapped segment by segment to the spatial position corresponding to the fabric porosity change parameter. The amount of dye liquor entering, the amount of dye liquor remaining, and the amount of dye liquor migrating at each position are calculated according to the residence time corresponding to the fabric running speed. Then, the concentration change caused by the amount of dye bath replenishment is superimposed time by time. The calculation results of each position are arranged in time order and spatial order to form the dyeing fluid dynamic state matrix.

[0014] Preferably, the step of obtaining the time-series distribution curve of the polluting elements is as follows:

[0015] Based on the dyeing fluid dynamics state matrix, the liquid exchange record, discharge path record, and residual liquid record corresponding to each time node are read one by one. The discharge flow rate and residual liquid concentration parameters corresponding to each time node are extracted. The discharge flow rate is multiplied into the discharge duration of the corresponding time period in chronological order. The residual liquid concentration parameters are written into the residual liquid interval of the corresponding time period. Then, the discharge pollution amount and residual pollution amount of continuous time nodes are accumulated and sorted to generate the time-series distribution curve of pollution elements.

[0016] Preferably, the step of obtaining the fabric apparent color prediction sequence is as follows:

[0017] Based on the time-series distribution curve of the pollutant elements, the incident light intensity record, transmitted light intensity record, detection band identifier, and detection time identifier are extracted from the photoelectric colorimeter measurement parameters. The incident light intensity record and transmitted light intensity record are paired one by one according to the detection time identifier. The transmittance is converted one by one according to the pairing results. The dye concentration conversion level corresponding to the transmittance is established according to the detection band identifier. Then, the photoelectric colorimeter measurement parameters are substituted one by one into the mapping relationship between transmittance and dye concentration. The dye concentration corresponding to each detection time identifier is sequentially analyzed to generate a dye concentration distribution sequence.

[0018] Based on the dye concentration distribution sequence, the dye concentration value, concentration change direction, and concentration change range corresponding to each detection time marker are read one by one. According to the correspondence between the detection time marker and the fabric running position, the dye concentration value corresponding to each detection time marker is written into the corresponding fabric running position. The cumulative surface coloring result and continuous coloring offset result after the dye enters the fabric are calculated segment by segment. Then, the coloring results of each fabric running position are arranged according to the order of the detection time markers to obtain the fabric apparent color prediction sequence.

[0019] Preferably, the step of obtaining the chromaticity deviation element set is as follows:

[0020] Extract the target color tolerance range parameter, analyze the upper color boundary, lower color boundary, and the corresponding chromaticity component position in the target color tolerance range parameter, read the time identifier, position identifier, and chromaticity component value in the fabric apparent chromaticity prediction sequence one by one, map each chromaticity component value to the upper color boundary and lower color boundary respectively, mark the sequence elements that fall outside the upper color boundary or the lower color boundary, and then summarize the marking results according to the time identifier and position identifier to obtain the chromaticity deviation element set.

[0021] Preferably, the step of obtaining the predicted value of the downgraded fabric length is as follows:

[0022] Extract the time identifier corresponding to each deviation element in the set of color deviation elements, arrange the time identifiers in chronological order, and determine whether adjacent time identifiers are continuously progressive. When a continuous progressive interruption is encountered, the current statistical segment ends and the next statistical segment is restarted. The duration of the first and last time identifiers in each statistical segment is converted to generate abnormal time spans.

[0023] Extract the start time and end time identifiers corresponding to the abnormal time spans, call the speed records corresponding to the start time and end time identifiers in the fabric running speed, match the fabric running speed in the corresponding time period according to each abnormal time span, perform multiplication calculation on each abnormal time span and the corresponding fabric running speed segment by segment, and then summarize all the segment calculation results in chronological order to obtain the predicted value of the downgraded fabric length.

[0024] Preferably, the steps for obtaining the workshop-level staggered production scheduling sequence are as follows:

[0025] Extract the degradation limit parameter of the biochemical tank, read the pollution value at each time position in the time-series distribution curve of the pollutant element one by one, select the maximum pollution value in the time-series distribution curve of the pollutant element as the peak value of the time-series distribution curve of the pollutant element, map the peak values ​​of the time-series distribution curve of the pollutant element to the degradation limit parameter of the biochemical tank, determine whether the peak value of the time-series distribution curve of the pollutant element is greater than the degradation limit parameter of the biochemical tank, mark the determination result that is greater than the degradation limit parameter of the biochemical tank as needing intervention and scheduling, mark the determination result that is not greater than the degradation limit parameter of the biochemical tank as maintaining the original discharge arrangement, and obtain the discharge scheduling determination result;

[0026] Based on the drainage scheduling determination result, extract the machine queuing order, planned color change time of each machine, drainage occupation time period of each machine, and pending execution status identifier of each machine from the determination record that requires intervention in scheduling. Map the time nodes corresponding to the predicted value of the downgraded fabric length to the planned color change time of the current machine. Determine whether the time node corresponding to the predicted value of the downgraded fabric length falls within the drainage occupation time period of each machine. If the time node corresponding to the predicted value of the downgraded fabric length falls within the drainage occupation time period of each machine, then postpone the execution time of the color change drainage action of the current machine to the subsequent idle time period. If the time node corresponding to the predicted value of the downgraded fabric length does not fall within the drainage occupation time period of each machine, then retain the execution time of the color change drainage action of the current machine to obtain the color change drainage action time shift result.

[0027] Preferably, the step of obtaining the workshop-level staggered production scheduling sequence further includes:

[0028] Based on the time shift results of the color change and draining actions, the execution times of the color change and draining actions of each machine are rearranged in chronological order. Machine records with the same time are re-layered according to the machine queue order. The adjusted execution times of the color change and draining actions are written into the color change task record of the corresponding machine. Then, the color change task records of all machines are expanded according to the unified timeline of the workshop to form a workshop-level staggered production scheduling sequence.

[0029] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0030] In this invention, the dye adsorption behavior is first linked to the dye bath replenishment amount, padding pressure, and fabric running speed in the continuous padding and color changing process. This allows for the continuous expression of the dye liquor's entry, migration, and retention processes in the fabric. Then, the drain flow rate and residual liquor concentration parameters are further connected with the aforementioned state results to extrapolate the temporal changes of pollutant elements. Simultaneously, photoelectric colorimeter measurement parameters are substituted into the transmittance-dye liquor concentration mapping relationship to further transfer the liquid phase concentration change to the fabric apparent color prediction sequence. Finally, the target color tolerance range parameters, color deviation element set, predicted degraded fabric length, biochemical tank degradation limit parameters, and workshop multi-machine color changing queue status parameters are connected to the same processing chain, enabling quality judgment, degraded length estimation, and drain load estimation. By establishing a closed-loop, continuous processing relationship between the production schedule and workshop scheduling, the gains are not limited to the final interpretation of dyeing results. Instead, dyeing quality, contamination load, drainage timing, and color change queue status are integrated into a traceable data chain. This allows for the identification of abnormal sustained intervals before color deviations spread to the entire fabric, the provision of drainage adjustment space before contamination peaks approach treatment limits, and the expansion of single quality prediction into a comprehensive judgment that considers product quality stability, control of downgraded fabric length, coordination of workshop color change rhythm, and balance of back-end processing capacity. Therefore, in actual production, this is more conducive to compressing the range of quality fluctuations, reducing ineffective drainage impacts, minimizing mutual interference caused by continuous color changes, and improving the responsiveness of production organization to complex working conditions. Attached Figure Description

[0031] Figure 1 This is a schematic diagram of the steps of the present invention. Detailed Implementation

[0032] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0033] Please see Figure 1 This invention provides a technical solution: a method for predicting the production quality of textile dyeing based on data recognition, comprising the following steps:

[0034] The dye molecule affinity constant, salt concentration parameter, and temperature parameter are extracted and solved according to the thermodynamic adsorption model to generate equilibrium adsorption distribution data; the dye bath replenishment amount, rolling pressure, and fabric running speed in the continuous pad dyeing and color changing production process are extracted and the dyeing fluid dynamic state matrix is ​​solved.

[0035] Based on the dyeing fluid dynamics state matrix, the discharge flow rate and residual liquid concentration parameters are extracted to generate the time-series distribution curve of pollutants; the photoelectric colorimeter measurement parameters are extracted to construct the mapping relationship between transmittance and dye concentration, and the photoelectric colorimeter measurement parameters are substituted into the mapping relationship to solve for the dye concentration distribution sequence; the dye concentration distribution sequence is mapped to the fabric apparent color prediction sequence based on the dyeing rate prediction model.

[0036] Extract the target color tolerance range parameter, compare each element in the fabric apparent colorimetric prediction sequence with the target color tolerance range parameter, and filter to obtain the colorimetric deviation element set; extract the time series corresponding to the colorimetric deviation element set, and calculate and generate the downgraded fabric length prediction value.

[0037] The degradation limit parameters of the biochemical pool are extracted, and the peak values ​​of the time-series distribution curves of pollutants are compared with the degradation limit parameters of the biochemical pool. Combined with the predicted value of the degraded fabric length, a workshop-level staggered production scheduling sequence is generated.

[0038] The steps for obtaining the state matrix of the dyed fluid dynamics are as follows:

[0039] The affinity constant of dye molecules, salt concentration parameters, and temperature parameters were extracted. The affinity constant of dye molecules, salt concentration parameters, and temperature parameters were matched according to a uniform time. The adsorption driving force, salt inhibition correction amount, and temperature compensation amount were calculated for each group. The adsorption increment and adsorption decrement of each group were recorded according to the numerical range of the calculation results to form the equilibrium adsorption amount distribution results.

[0040] Based on the equilibrium adsorption distribution results, the dyeing tank replenishment amount, rolling press pressure, and fabric running speed in the continuous dyeing and color changing production process are extracted. The fabric compression degree is calculated according to the change range of rolling press pressure. Then, the fabric porosity change parameter is estimated segment by segment according to the fabric compression degree. The transfer distance and residence time are allocated point by point in the running direction according to the fabric running speed. The dyeing tank replenishment amount is written into the corresponding position in time sequence to form the pore mass transfer coupling distribution results.

[0041] Based on the pore mass transfer coupling distribution results, the interval adsorption amount in the equilibrium adsorption amount distribution results is mapped segment by segment to the spatial position corresponding to the fabric porosity change parameter. The amount of dye liquor entering, the amount of dye liquor remaining, and the amount of dye liquor migrating at each position are calculated according to the residence time corresponding to the fabric running speed. Then, the concentration change caused by the amount of dye bath replenishment is superimposed time by time. The calculation results of each position are arranged in time order and spatial order to form the dyeing fluid dynamic state matrix.

[0042] Specifically, the extracted dye molecule affinity constant, salt concentration parameters, and temperature parameters, after being aligned according to a unified sampling time, are used to calculate the theoretical equilibrium adsorption capacity under the current operating conditions using a commonly used adsorption isotherm model in chemical engineering, such as the Langmuir adsorption isotherm model. The specific parameters of this model include the maximum adsorption capacity. and Langmuir adsorption constant The data is not calculated in real time, but rather obtained by querying a pre-built parameter database. This database is established by systematically changing the temperature (from 40℃ to 100℃ in 10℃ increments) and salt concentration (from 10g / L to 80g / L in 10g / L increments) under laboratory conditions for specific dye-fiber systems (such as Reactive Red 195 and pure cotton fabric). The equilibrium adsorption data for each combination of conditions is measured spectrophotometrically, and then nonlinear fitting is performed on these experimental data to obtain the values ​​at different temperatures and salt concentrations. and The values ​​are stored in a multidimensional lookup table. In actual production, based on the real-time collected temperature and salt concentration parameters, multidimensional interpolation can be performed in this lookup table to obtain the Langmuir model parameters corresponding to the current operating conditions, and then the theoretical equilibrium adsorption capacity can be calculated. Then, the calculated Compared to a fixed process baseline, which is the equilibrium adsorption capacity measured under a standard dyeing process (e.g., temperature 90°C, salt concentration 50 g / L), if the current... If the value is higher than the baseline value, the difference between the two is recorded as the adsorption increment; otherwise, it is recorded as the adsorption decrement. This process is continuously executed over the entire time series to form a balanced adsorption capacity distribution result that reflects the adsorption potential as the process parameters fluctuate.

[0043] Based on the equilibrium adsorption distribution, the dye bath replenishment amount, roller pressure, and fabric running speed in the continuous dyeing and color-changing production process are extracted. First, a pre-calibrated fabric mechanics model is used to calculate the fabric compression degree corresponding to the roller pressure. Specifically, a physical compression test is conducted on the target fabric (e.g., 200 g / m² pure cotton knitted fabric). Pressures from 0 to 3.5 MPa are applied on a material testing machine, while an optical non-contact thickness gauge records the changes in fabric thickness. The obtained "pressure-thickness" data points are fitted into a high-order polynomial function, such as a third-order polynomial. Where P is the pressure, and a, b, c, and d are fitting coefficients, this functional relationship is embedded in the program. During production, the measured rolling mill pressure is substituted into this function to obtain the fabric thickness, and then the degree of compression is calculated. Next, based on the compressed fabric thickness, the fabric porosity change parameter is calculated. The calculation is based on the principle that pore volume equals the total fabric volume minus the fiber volume, i.e. The fabric surface density and fiber density are known physical constants. Then, at the software level, the running fabric is virtualized into a series of continuous computing units with a length of 0.1 meters along its length. Based on the real-time running speed of the fabric, the dwell time required for each unit to pass through each process zone (such as the padding zone) is calculated. Finally, the collected dye bath replenishment data is accurately assigned to the virtual computing unit that is currently at the physical location of the dye bath replenishment port according to its timestamp, and recorded as an instantaneous input event. By integrating the porosity, dwell time and replenishment event of all computing units, the pore mass transfer coupling distribution result is formed.

[0044] Based on the pore mass transfer coupling distribution results, a discretization calculation method based on the finite difference method is adopted to simulate the dye transport and adsorption process in the fabric. The adsorption amount is mapped piecewise to form a state matrix. This method divides the fabric in the running direction into several discrete control volumes (i.e., the aforementioned virtual calculation units) and also discretizes time into small time steps. (For example, 0.1 seconds), within each time step, the following calculations are performed sequentially for each control volume. First, the convection process is simulated, calculating the amount of dye flowing from the previous control volume to the next control volume due to the overall movement of the fabric, based on the fabric's running speed. Second, the diffusion process is simulated, calculating the net diffusion flux of the dye based on the discrete form of Fick's first law, according to the dye concentration difference between the current control volume and its adjacent control volumes. The effective diffusion coefficient used is... This is an empirical constant, obtained from a preset parameter table based on the dye type and temperature. Next, the adsorption reaction process is simulated, and a reaction kinetic model is used to calculate the amount of dye transferred from the liquid phase to the fiber within that time step. The reaction rate is proportional to the difference between the current liquid phase concentration and the theoretical equilibrium adsorption amount at that location (provided by the equilibrium adsorption amount distribution). The reaction rate constant used is... Similarly, the dye bath replenishment amount is treated as a boundary condition of the system, based on empirical values ​​calibrated through experiments. When a replenishment event occurs, the dye concentration values ​​of all control bodies within the dye bath area are directly updated. After completing the convection, diffusion, and adsorption calculations of all control bodies within a time step, the state of each control body (including liquid phase concentration and solid phase adsorption amount) is updated, and then the calculation for the next time step begins. This process is repeated iteratively until the complete state data of all control bodies at all time points are obtained. These data are then organized into a two-dimensional array according to spatial location (control body number) and time order to form the dyeing fluid dynamics state matrix.

[0045] The steps for obtaining the time-series distribution curves of pollutants are as follows:

[0046] Based on the state matrix of the dyeing fluid dynamics, the liquid exchange record, discharge path record, and residual liquid record corresponding to each time node are read one by one. The discharge flow rate and residual liquid concentration parameters corresponding to each time node are extracted. The discharge flow rate is multiplied into the discharge duration of the corresponding time period in chronological order. The residual liquid concentration parameters are written into the residual liquid interval of the corresponding time period. Then, the discharge pollution amount and residual pollution amount of continuous time nodes are accumulated and sorted to generate the time series distribution curve of pollution elements.

[0047] Specifically, based on the dyeing fluid dynamics state matrix, data for each time node arranged chronologically in the matrix is ​​read one by one. At each time node, liquid exchange records, discharge path records, and residual liquid records related to all discharge ports (e.g., main discharge valve, cleaning water discharge valve) are parsed out. From these records, the instantaneous discharge flow rate of each discharge port (e.g., obtained by reading the flow meter reading corresponding to the opening degree of the discharge valve) and the corresponding dye concentration parameter of the discharged liquid (this concentration value comes directly from the concentration state of the dyeing tank or cleaning tank in the state matrix at the current moment) are extracted. Taking a time step of 1 second as the discharge duration, the instantaneous discharge flow rate is multiplied by the time step to obtain the discharge volume within that second. Then, multiply this volume by the corresponding discharge concentration to calculate the mass of pollutant elements (i.e., dyes) discharged through the discharge port in that second. At the same time, extract the volume and concentration of the residual liquid inside the fabric after it has been squeezed by the rolling mill from the residual liquid record. Multiply the two to obtain the mass of pollutant elements currently remaining on the fabric. Then, sum the mass of pollutant elements discharged from all discharge ports and the mass of residual pollutant elements on all fabric segments in the same time node to obtain the total amount of pollutant elements generated in that time node. Finally, arrange the total amount of pollutant elements generated in each time node in chronological order to form a curve with time as the horizontal axis and pollutant element mass as the vertical axis, generating a pollutant element time-series distribution curve.

[0048] The steps for obtaining the fabric apparent color prediction sequence are as follows:

[0049] Based on the time-series distribution curves of pollutants, the incident light intensity record, transmitted light intensity record, detection band identifier, and detection time identifier are extracted from the photoelectric colorimeter measurement parameters. The incident light intensity record and transmitted light intensity record are paired one by one according to the detection time identifier. The transmittance is calculated one by one according to the pairing results. The dye concentration conversion level corresponding to the transmittance is established according to the detection band identifier. Then, the photoelectric colorimeter measurement parameters are substituted one by one into the mapping relationship between transmittance and dye concentration. The dye concentration corresponding to each detection time identifier is analyzed sequentially to generate a dye concentration distribution sequence.

[0050] Based on the dye concentration distribution sequence, the dye concentration value, concentration change direction, and concentration change range corresponding to each detection time marker are read one by one. According to the correspondence between the detection time marker and the fabric running position, the dye concentration value corresponding to each detection time marker is written into the corresponding fabric running position. The cumulative surface coloring result and continuous coloring offset result after the dye enters the fabric are calculated segment by segment. Then, the coloring results of each fabric running position are arranged according to the order of the detection time markers to obtain the fabric apparent color prediction sequence.

[0051] Specifically, based on the measurement parameters collected in real time by the photoelectric colorimeter on the production line, including incident light intensity records, transmitted light intensity records, detection band markers, and detection time markers, firstly, according to each detection time marker, the incident light intensity collected at the same time is... and transmitted light intensity Pairing is performed, and the transmittance at each time point is calculated based on the pairing results. Next, based on the detection band identifier (e.g., 530nm for red dye and 620nm for blue dye), a pre-established mapping relationship between transmittance and dye concentration is invoked. This mapping relationship is established as follows: First, a series of standard dye samples with known concentrations are prepared, for example, from 0.005 g / L to 0.1 g / L, a total of 20 gradients. The transmittance of each standard sample is measured using a photoelectric colorimeter of the same model as the production line at the same detection band, and then its absorbance is calculated. According to Beer-Lambert's law, absorbance is directly proportional to concentration. By performing linear regression analysis on the "concentration-absorbance" data points, a calibration curve is obtained (e.g., concentration = 1.25 * absorbance - 0.001). This regression equation represents the mapping relationship between transmittance and dye concentration. In actual conversion, the transmittance at each time point calculated in real time is converted into absorbance, and then the absorbance value is substituted into the above regression equation for solution. This allows the dye concentration value corresponding to each detection time marker to be analyzed. Finally, all the calculated concentration values ​​are arranged in chronological order of detection time to generate a dye concentration distribution sequence.

[0052] Based on the dye concentration distribution sequence, the dye concentration value corresponding to each detection time marker in the sequence is first read sequentially. The direction (positive or negative) and magnitude of concentration change are obtained by calculating the difference between concentration values ​​at adjacent time points. Then, using a preset time-space transformation relationship, each detection time marker is mapped to a specific running position on the fabric. This transformation relationship is calculated based on the real-time running speed of the fabric and the fixed installation position of the photoelectric colorimeter. For example, the concentration detected at time t corresponds to the fabric position at time (t - L / v) after passing through the dye bath, where L is the distance from the dye bath to the colorimeter, and v is the fabric running speed. In this way, the dye concentration values ​​of the time series are written to the corresponding fabric running positions. Next, the Kubelka-Munk theoretical model is used to calculate the coloring effect of the dye concentration on the fabric. Specifically, the dye concentration at the fabric position is converted into the dye concentration adsorbed on the fabric fibers through a pre-established "dye concentration - dyeing amount" relationship. This relationship is obtained through laboratory immersion dyeing experiments. Then, the dye concentration on the fibers is substituted into the Kubelka-Munk equation. The theoretical reflectance of the fabric surface is calculated, where K / S is the ratio of absorption to scattering coefficients, c is the dye concentration on the fiber, and k is the dye absorption constant. Finally, the calculated theoretical reflectance is converted into the L*, a*, b chromaticity values ​​of the fabric appearance using the CIELAB color space conversion model (which is also established by measuring and regressing standard color samples). This chromaticity value is the cumulative surface coloring result at that location, while the continuous coloring shift result is obtained by calculating the Euclidean distance between the chromaticity values ​​of adjacent fabric locations. Finally, the L*, a*, b* chromaticity values ​​of all fabric locations are arranged in spatial order to obtain the fabric appearance chromaticity prediction sequence.

[0053] The steps to obtain the set of chromaticity deviation elements are as follows:

[0054] Extract the target color tolerance range parameter, analyze the upper color boundary, lower color boundary, and the corresponding chromaticity component position in the target color tolerance range parameter, read the time identifier, position identifier, and chromaticity component value in the fabric apparent chromaticity prediction sequence one by one, map each chromaticity component value to the upper color boundary and lower color boundary respectively, mark the sequence elements that fall outside the upper color boundary or the lower color boundary, and then summarize the marking results according to the time identifier and position identifier to obtain the chromaticity deviation element set.

[0055] Specifically, the extracted target color tolerance range parameter first analyzes the upper and lower color limits contained within this parameter. These limits are set for the L*, a*, and b chromaticity components of the CIELAB color space. For example, the tolerance range for a target red might be defined as L between (50, 55), a between (40, 48), and b between (15, 20). These boundary values ​​are determined by measuring the standard color sample provided by the customer using a high-precision spectrophotometer and combining this with the allowable fluctuation range of the production process. For example, if the measured L of the standard color sample is 52.5, then the upper limit boundary can be set as 52.5 + 2.5 = 55, and the lower limit boundary as 52.5 - 2.5 = 50. Then, each data point in the fabric apparent color prediction sequence is read sequentially. Each data point contains a time marker, a location marker, and the corresponding L, a*, and b chromaticity component values. The L, a*, and b values ​​of each data point are then analyzed. The b-value is compared with the preset upper and lower color boundaries. For example, the predicted L-value is compared with the (50, 55) range, the a-value with the (40, 48) range, and the b-value with the (15, 20) range. If any of the component values ​​falls outside its corresponding boundary range, for example, if the L* value is 49, which is lower than the lower limit of 50, the sequence element is marked as a deviation element. The marking content includes its time identifier and position identifier. Finally, all sequence elements marked as deviations are summarized and uniquely integrated according to their time identifier and position identifier to obtain the color deviation element set.

[0056] The steps for obtaining the predicted length of the downgraded fabric are as follows:

[0057] Extract the time stamp corresponding to each deviation element in the color deviation element set, arrange the time stamps in chronological order, and check whether adjacent time stamps are continuously progressive. When a continuous progressive interruption is encountered, the current statistical segment ends and the next statistical segment is started. The duration of the first and last time stamps in each statistical segment is converted to generate abnormal time spans.

[0058] Extract the start and end time identifiers corresponding to the abnormal time spans, call the speed records corresponding to the start and end time identifiers in the fabric running speed, match the fabric running speed in the corresponding time period according to each abnormal time span, perform multiplication calculation on each abnormal time span and the corresponding fabric running speed segment by segment, and then summarize all the segment calculation results in chronological order to obtain the predicted value of the downgraded fabric length.

[0059] Specifically, each deviation element recorded in the color deviation element set is extracted, and its corresponding time stamp is extracted. These time stamps are then arranged in ascending order of value to form an ordered time sequence. Next, starting from the first time stamp in the sequence, each adjacent time stamp is checked against a preset continuity criterion. This criterion is set as a fixed time threshold, such as 1.5 seconds. This threshold is based on the minimum time interval for data acquisition on the production line (e.g., 1 second) plus a fault tolerance buffer time (0.5 seconds) to allow for minor network latency or data processing fluctuations. If the difference between two adjacent time stamps is less than or equal to 1.5 seconds, they are determined to belong to the same continuous abnormal time period, and the process continues until a pair of adjacent time stamps is encountered. If the difference between time markers is greater than 1.5 seconds (for example, if the previous time marker is 10:05:30 and the next is 10:05:32, the difference is 2 seconds, which is greater than 1.5 seconds), then the continuous progression is considered to be interrupted at this point. In this case, all consecutive time markers from the beginning of the sequence to the point of interruption are considered as a complete statistical segment. The first time marker (start time) and the last time marker (end time) of this statistical segment are recorded. Then, starting from the time marker at the interruption point, a new statistical segment is started, and the above judgment process is repeated until all time markers have been traversed. For each statistical segment divided in this way, the duration of the color anomaly in that segment is obtained by calculating the difference between its end time and start time. The duration of all statistical segments is recorded to generate the abnormal time span.

[0060] Extract the start and end time identifiers corresponding to each abnormal time span record. Using these two time identifiers, query and retrieve all speed records within that time period from pre-stored historical fabric running speed data. Since fabric running speed may change during production, it is necessary to obtain the complete speed change sequence within that time period, rather than a single speed value. Then, for each abnormal time span, match it with the fabric running speed sequence within the corresponding queried time period. Specifically, divide the abnormal time span into multiple smaller sub-time periods according to the time points in the speed sequence where speed changes occur. Within each sub-time period, the fabric running speed can be considered constant. Then... For each sub-period, its duration (the end time of the sub-period minus the start time) is multiplied by the constant fabric running speed within that period to calculate the length of the degraded fabric produced in that sub-period. For example, if an abnormal time span is 10 seconds, with the speed at 1.2 m / s for the first 4 seconds and 1.0 m / s for the last 6 seconds, then the lengths are calculated as follows: 4 seconds 1.2 m / s = 4.8 m and 6 seconds 1.0 m / s = 6.0 m. Finally, all the segmented degraded fabric lengths calculated for all abnormal time spans are summed up in chronological order to obtain a total length value. This total length value is the predicted value of the total length of fabric that needs to be degraded due to color deviation in the entire production batch, i.e., the predicted length of degraded fabric.

[0061] The steps for obtaining the workshop-level staggered production scheduling sequence are as follows:

[0062] Extract the degradation limit parameters of the biochemical tank, read the pollution values ​​at each time position in the time-series distribution curve of each pollutant element, select the maximum pollution value in the time-series distribution curve of the pollutant element as the peak value of the time-series distribution curve of the pollutant element, map the peak values ​​of the time-series distribution curve of the pollutant element to the degradation limit parameters of the biochemical tank, and determine whether the peak value of the time-series distribution curve of the pollutant element is greater than the degradation limit parameters of the biochemical tank. The judgment result that is greater than the degradation limit parameters of the biochemical tank is marked as needing intervention and scheduling, and the judgment result that is not greater than the degradation limit parameters of the biochemical tank is marked as maintaining the original discharge arrangement, thus obtaining the discharge scheduling judgment result;

[0063] Based on the drainage scheduling judgment results, extract the machine queuing order, planned color change time of each machine, drainage occupation time of each machine, and pending execution status identifier of each machine from the judgment record that requires intervention in scheduling. Map the time nodes corresponding to the predicted value of the downgraded fabric length to the planned color change time of the current machine. Determine whether the time node corresponding to the predicted value of the downgraded fabric length falls within the drainage occupation time of each machine. If the time node corresponding to the predicted value of the downgraded fabric length falls within the drainage occupation time of each machine, then postpone the execution time of the color change drainage action of the current machine to the subsequent idle time. If the time node corresponding to the predicted value of the downgraded fabric length does not fall within the drainage occupation time of each machine, then retain the execution time of the color change drainage action of the current machine to obtain the color change drainage action time shift result.

[0064] Based on the time shift results of the color change and draining actions, the execution times of the color change and draining actions of each machine are rearranged in chronological order. Machine records with the same time are re-layered according to the machine queue order. The adjusted execution times of the color change and draining actions are written into the color change task record of the corresponding machine. Then, the color change task records of all machines are expanded according to the unified timeline of the workshop to form a workshop-level staggered production scheduling sequence.

[0065] Specifically, the degradation limit parameter of the biological treatment tank is extracted. This parameter is not a single value, but a set of graded thresholds set according to different types of dyes (e.g., reactive, disperse, and acid dyes) and the chemical oxygen demand (COD) concentration in the wastewater. For example, for reactive dye wastewater with a COD concentration below 500 mg / L, the instantaneous degradation limit load of the biological treatment tank is set at 1.2 kg COD / hour, while for wastewater with a COD concentration between 500-1000 mg / L, the limit is set at 1.0 kg COD / hour. These limit values ​​are determined based on statistical analysis of long-term operating data of the biological treatment tank and combined with the results of microbial activity tests. Then, the pollution value corresponding to each time point in the time-series distribution curve of the pollutant element (i.e., the pollution level) is read one by one. The peak value of the pollutant element time distribution curve is found by traversing the entire curve. This peak value is then compared with the specific degradation limit value that matches the current dye type and the estimated wastewater COD concentration, obtained from the degradation limit parameters of the biological treatment tank. If the peak value of the calculated pollutant element time distribution curve is greater than the corresponding degradation limit parameter of the biological treatment tank (for example, if the predicted peak value is 1.3 kg COD / hour, which is greater than the set 1.2 kg COD / hour), the discharge scheduling judgment result of this color change event is marked as "intervention is required". If the peak value is not greater than the degradation limit parameter, it is marked as "maintain the original discharge arrangement". Finally, the judgment results of all color change events are summarized to obtain the discharge scheduling judgment result.

[0066] Based on the drainage scheduling results, all records marked as "requiring intervention" are selected. For each such record, the corresponding multi-machine color-changing queue status parameters are extracted from the Workshop Production Execution System (MES). These parameters include the current queue order of all dyeing vats in the workshop (e.g., dyeing vat 3 is before dyeing vat 5), the planned start time of color changing for each machine (e.g., dyeing vat 3 is scheduled to start color changing at 14:30), the expected drainage time period for each machine's color changing process (e.g., dyeing vat 3 is expected to drain from 14:45 to 15:00), and the pending execution status flag for each machine (indicating whether the color changing task has started). Then, the time node corresponding to the predicted length of the downgraded fabric obtained in the previous steps (i.e., the production period for producing downgraded fabric) is matched with the original planned color changing time of the machine currently requiring intervention. The corresponding process is not a direct comparison, but rather a determination of whether the production process of the downgraded fabric overlaps with the originally planned drainage period. For example, if the downgraded fabric is produced from 14:50 to 14:55, while the originally planned drainage period is from 14:45 to 15:00, it is determined to fall within the drainage period. If it is determined to fall within the period, a scheduling adjustment is performed, postponing the execution time of the color-changing drainage action of the current machine. The target period for the postponement is the first empty period on the workshop drainage main pipeline timetable. This empty period is found by querying the drainage periods of all other machines to find a sufficiently long (e.g., greater than 15 minutes) time window that is not occupied by any machine. If it is determined not to fall within the period, the original execution time of the color-changing drainage action of the current machine is retained unchanged. After completing this determination and adjustment for all machines that need to be intervened in the scheduling, the result of the color-changing drainage action time shift is obtained.

[0067] Based on the time shift results of the color-changing and draining actions, the execution times of the color-changing and draining actions for all machines (including those that have been adjusted and those that haven't) are first collected. These times are then rearranged in order from morning to night to form a preliminary timeline. If, during the rearrangement process, it is found that the execution times of the color-changing and draining actions for multiple machines have been adjusted to the same time point—for example, dyeing vats 3 and 5 are both adjusted to start draining at 15:30—then stratification is required based on their machine queuing order in the original workshop multi-machine color-changing queuing status parameters. For example, if the queuing order of machine 3 is higher than that of machine 5, then the draining time of machine 5 will be within the draining time slot occupied by machine 3. After the process ends, it is postponed to the next available idle period. In this way, it is ensured that only one machine is discharging high-concentration liquid into the main pipeline at any given time. After all conflict resolution is completed, the finalized and adjusted color change and liquid discharge execution time is written back to the color change task record of the corresponding machine in the production execution system, and its task plan is updated. Finally, the color change task records of all machines (including the updated liquid discharge start time and liquid discharge occupation period) are integrated and visualized according to a unified workshop-level timeline. For example, on a Gantt chart, the production and liquid discharge arrangements of each machine in the future period are clearly marked, forming a workshop-level staggered production scheduling sequence.

[0068] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for predicting the production quality of textile dyeing based on data recognition, characterized in that, Includes the following steps: The dye molecule affinity constant, salt concentration parameter, and temperature parameter are extracted and solved according to the thermodynamic adsorption model to generate equilibrium adsorption distribution data; the dye bath replenishment amount, rolling pressure, and fabric running speed in the continuous pad dyeing and color changing production process are extracted and the dyeing fluid dynamic state matrix is ​​solved. Based on the dyeing fluid dynamics state matrix, the discharge flow rate and residual liquid concentration parameters are extracted to generate the time-series distribution curve of pollutants; the photoelectric colorimeter measurement parameters are extracted to construct the mapping relationship between transmittance and dye concentration; the photoelectric colorimeter measurement parameters are substituted into the mapping relationship for solution to generate the dye concentration distribution sequence; the dye concentration distribution sequence is mapped to the fabric apparent color prediction sequence according to the dyeing rate prediction model. Extract the target color tolerance range parameter, compare each element in the fabric apparent colorimetric prediction sequence with the target color tolerance range parameter, and filter to obtain a colorimetric deviation element set; extract the time series corresponding to the colorimetric deviation element set, and calculate and generate the downgraded fabric length prediction value. Extract the degradation limit parameters of the biochemical pool, compare the peak values ​​of the time-series distribution curves of the pollutants with the degradation limit parameters of the biochemical pool, and generate a workshop-level staggered production scheduling sequence by combining the predicted value of the degraded fabric length.

2. The method for predicting the production quality of textile dyeing based on data recognition according to claim 1, characterized in that, The steps for obtaining the dyeing fluid dynamics state matrix are as follows: The affinity constant of dye molecules, salt concentration parameters, and temperature parameters were extracted. The affinity constant of dye molecules, salt concentration parameters, and temperature parameters were matched according to a uniform time. The adsorption driving force, salt inhibition correction amount, and temperature compensation amount were calculated for each group. The adsorption increment and adsorption decrement of each group were recorded according to the numerical range of the calculation results to form the equilibrium adsorption amount distribution results. Based on the equilibrium adsorption distribution results, the dyeing tank replenishment amount, rolling press pressure, and fabric running speed in the continuous dyeing and color changing production process are extracted. The fabric compression degree is calculated according to the change range of rolling press pressure. Then, the fabric porosity change parameter is estimated segment by segment according to the fabric compression degree. The transfer distance and residence time are allocated point by point in the running direction according to the fabric running speed. The dyeing tank replenishment amount is written into the corresponding position in time sequence to form the pore mass transfer coupling distribution results. Based on the pore mass transfer coupling distribution results, the interval adsorption amount in the equilibrium adsorption amount distribution results is mapped segment by segment to the spatial position corresponding to the fabric porosity change parameter. The amount of dye liquor entering, the amount of dye liquor remaining, and the amount of dye liquor migrating at each position are calculated according to the residence time corresponding to the fabric running speed. Then, the concentration change caused by the amount of dye bath replenishment is superimposed time by time. The calculation results of each position are arranged in time order and spatial order to form the dyeing fluid dynamic state matrix.

3. The method for predicting the production quality of textile dyeing based on data recognition according to claim 1, characterized in that, The steps for obtaining the time-series distribution curves of the polluting elements are as follows: Based on the dyeing fluid dynamics state matrix, the liquid exchange record, discharge path record, and residual liquid record corresponding to each time node are read one by one. The discharge flow rate and residual liquid concentration parameters corresponding to each time node are extracted. The discharge flow rate is multiplied into the discharge duration of the corresponding time period in chronological order. The residual liquid concentration parameters are written into the residual liquid interval of the corresponding time period. Then, the discharge pollution amount and residual pollution amount of continuous time nodes are accumulated and sorted to generate the time-series distribution curve of pollution elements.

4. The method for predicting the production quality of textile dyeing based on data recognition according to claim 1, characterized in that, The steps for obtaining the fabric apparent color prediction sequence are as follows: Based on the time-series distribution curve of the pollutant elements, the incident light intensity record, transmitted light intensity record, detection band identifier, and detection time identifier are extracted from the photoelectric colorimeter measurement parameters. The incident light intensity record and transmitted light intensity record are paired one by one according to the detection time identifier. The transmittance is converted one by one according to the pairing results. The dye concentration conversion level corresponding to the transmittance is established according to the detection band identifier. Then, the photoelectric colorimeter measurement parameters are substituted one by one into the mapping relationship between transmittance and dye concentration. The dye concentration corresponding to each detection time identifier is sequentially analyzed to generate a dye concentration distribution sequence. Based on the dye concentration distribution sequence, the dye concentration value, concentration change direction, and concentration change range corresponding to each detection time marker are read one by one. According to the correspondence between the detection time marker and the fabric running position, the dye concentration value corresponding to each detection time marker is written into the corresponding fabric running position. The cumulative surface coloring result and continuous coloring offset result after the dye enters the fabric are calculated segment by segment. Then, the coloring results of each fabric running position are arranged according to the order of the detection time markers to obtain the fabric apparent color prediction sequence.

5. The method for predicting the production quality of textile dyeing based on data recognition according to claim 1, characterized in that, The steps for obtaining the set of color deviation elements are as follows: Extract the target color tolerance range parameter, analyze the upper color boundary, lower color boundary, and the corresponding chromaticity component position in the target color tolerance range parameter, read the time identifier, position identifier, and chromaticity component value in the fabric apparent chromaticity prediction sequence one by one, map each chromaticity component value to the upper color boundary and lower color boundary respectively, mark the sequence elements that fall outside the upper color boundary or the lower color boundary, and then summarize the marking results according to the time identifier and position identifier to obtain the chromaticity deviation element set.

6. The method for predicting the production quality of textile dyeing based on data recognition according to claim 1, characterized in that, The steps for obtaining the predicted length of the downgraded fabric are as follows: Extract the time identifier corresponding to each deviation element in the set of color deviation elements, arrange the time identifiers in chronological order, and determine whether adjacent time identifiers are continuously progressive. When a continuous progressive interruption is encountered, the current statistical segment ends and the next statistical segment is restarted. The duration of the first and last time identifiers in each statistical segment is converted to generate abnormal time spans. Extract the start time and end time identifiers corresponding to the abnormal time spans, call the speed records corresponding to the start time and end time identifiers in the fabric running speed, match the fabric running speed in the corresponding time period according to each abnormal time span, perform multiplication calculation on each abnormal time span and the corresponding fabric running speed segment by segment, and then summarize all the segment calculation results in chronological order to obtain the predicted value of the downgraded fabric length.

7. The method for predicting the production quality of textile dyeing based on data recognition according to claim 1, characterized in that, The steps for obtaining the workshop-level staggered production scheduling sequence are as follows: Extract the degradation limit parameter of the biochemical tank, read the pollution value at each time position in the time-series distribution curve of the pollutant element one by one, select the maximum pollution value in the time-series distribution curve of the pollutant element as the peak value of the time-series distribution curve of the pollutant element, map the peak values ​​of the time-series distribution curve of the pollutant element to the degradation limit parameter of the biochemical tank, determine whether the peak value of the time-series distribution curve of the pollutant element is greater than the degradation limit parameter of the biochemical tank, mark the determination result that is greater than the degradation limit parameter of the biochemical tank as needing intervention and scheduling, mark the determination result that is not greater than the degradation limit parameter of the biochemical tank as maintaining the original discharge arrangement, and obtain the discharge scheduling determination result; Based on the drainage scheduling determination result, extract the machine queuing order, planned color change time of each machine, drainage occupation time period of each machine, and pending execution status identifier of each machine from the determination record that requires intervention in scheduling. Map the time nodes corresponding to the predicted value of the downgraded fabric length to the planned color change time of the current machine. Determine whether the time node corresponding to the predicted value of the downgraded fabric length falls within the drainage occupation time period of each machine. If the time node corresponding to the predicted value of the downgraded fabric length falls within the drainage occupation time period of each machine, then postpone the execution time of the color change drainage action of the current machine to the subsequent idle time period. If the time node corresponding to the predicted value of the downgraded fabric length does not fall within the drainage occupation time period of each machine, then retain the execution time of the color change drainage action of the current machine to obtain the color change drainage action time shift result.

8. The method for predicting the production quality of textile dyeing based on data recognition according to claim 7, characterized in that, The steps for obtaining the workshop-level staggered production scheduling sequence also include: Based on the time shift results of the color change and draining actions, the execution times of the color change and draining actions of each machine are rearranged in chronological order. Machine records with the same time are re-layered according to the machine queue order. The adjusted execution times of the color change and draining actions are written into the color change task record of the corresponding machine. Then, the color change task records of all machines are expanded according to the unified timeline of the workshop to form a workshop-level staggered production scheduling sequence.

Images