A method for producing recycled granules based on waste plastic fiber bags

Abstract

The present application relates to the technical field of plastic regeneration, in particular to a regeneration granulation production method based on waste plastic fiber bag, comprising: obtaining multi-source detection data of the waste plastic fiber bag; wherein the multi-source detection data comprises transmission spectrum data, surface texture image and thickness measurement data; performing feature analysis based on the multi-source detection data to obtain raw material composite defect characteristic values; wherein the raw material composite defect characteristic values comprise fiber residue characteristic values, pollutant distribution characteristic values and mechanical strength attenuation characteristic values. Through multi-source data fusion of transmission spectrum, surface texture and thickness measurement, the present application extracts fiber residue, pollutant distribution and mechanical strength attenuation characteristic values, and introduces the concept of plasticization collapse point and dynamic coupling boundary line. By calculating the vertical distance from the current raw material coordinate point to the dynamic boundary, the synergistic damage effect of fiber and pollutant can be quantified, so that the plasticization collapse risk can be predicted before production, and the product qualification rate is improved.

Description

Technical Field

[0001] This invention relates to the field of plastic recycling technology, specifically to a method for producing recycled granules based on waste plastic fiber bags. Background Technology

[0002] With the rapid development of the plastics industry and the increasing awareness of environmental protection, the recycling and reuse of waste plastics has become an important part of the resource circular economy. In the traditional waste plastic recycling and granulation process, a standardized process of crushing-washing-melting-extrusion-granulation is usually adopted. In order to ensure the quality of recycled granules, operators often rely on experience to set process parameters such as melting temperature, screw speed and die head pressure.

[0003] However, existing recycled granulation production methods typically focus only on a single physical indicator of the raw material, such as only testing thickness or impurity content, or relying solely on manual visual inspection. This one-sided testing approach leads to poor production stability and often uses fixed process parameters for control. Due to the extreme heterogeneity of waste plastic raw materials, fixed parameters are difficult to adapt to the mechanical strength decay characteristics of different batches of raw materials. When the raw material strength is low, if production is still carried out according to conventional parameters, it is easy to cause melt fracture. If the parameters are set too conservatively, energy consumption will surge, and the overall process parameter settings are relatively rigid. Summary of the Invention

[0004] To achieve the above objectives, the present invention provides the following technical solution: a method for producing recycled granules based on waste plastic fiber bags, comprising: Acquire multi-source detection data of waste plastic fiber bags; the multi-source detection data includes transmission spectral data, surface texture images, and thickness measurement data; Feature analysis was performed based on multi-source detection data to obtain the composite defect characteristic values ​​of raw materials; among them, the composite defect characteristic values ​​of raw materials include fiber residue characteristic values, contaminant distribution characteristic values, and mechanical strength attenuation characteristic values; Using pre-calibrated fiber residue weights and contaminant influence weights, the fiber residue characteristic values ​​and contaminant distribution characteristic values ​​are weighted and summed to generate a fiber interference index; based on the mechanical strength attenuation characteristic values, the pre-built strength attenuation-compensation coefficient comparison table is consulted to obtain the melt strength compensation coefficient. The fiber interference index and melt strength compensation coefficient are input into a pre-built process response surface database, and the initial production parameters are obtained through multidimensional interpolation. The initial production parameters include the initial melt temperature, screw speed and die head pressure. Historical energy consumption records of similar granulation production lines are obtained, and an energy consumption regression equation is established with melt temperature, screw speed and output as variables. The initial production parameters are input into the energy consumption regression equation to calculate the predicted unit energy consumption. When the predicted unit energy consumption exceeds the preset energy consumption threshold, the screw speed and melt temperature are adjusted in the direction of reducing unit energy consumption to obtain the corrected production parameters. Granulation is initiated by correcting production parameters, and the melt pressure signal at the die head is collected in real time and the root mean square of pressure fluctuation is calculated. When the root mean square of pressure fluctuation is greater than the fluctuation threshold, the die head pressure is compensated in a closed loop through a proportional-integral controller to complete the production management of regenerated granulation.

[0005] Preferably, feature analysis is performed based on multi-source detection data to obtain feature values ​​of composite defects in the raw materials, including: Baseline correction and characteristic absorption peak integration were performed on the transmission spectral data in the multi-source detection data to extract fiber residual characteristic values; The surface texture image in the multi-source detection data is divided into grids, the short run advantage of the gray-level run matrix of each grid is calculated, and the pollutant distribution characteristic value is obtained by weighted summation. The thickness variation coefficient is calculated from the thickness measurement data in the multi-source detection data and used as a characteristic value of mechanical strength attenuation. The characteristic values ​​of fiber residue, pollutant distribution, and mechanical strength attenuation are combined to form the characteristic values ​​of composite defects in raw materials.

[0006] Preferably, feature analysis is performed based on multi-source detection data to obtain feature values ​​of composite defects in the raw materials, including: Baseline correction and characteristic absorption peak integration were performed on the transmission spectral data in the multi-source detection data to extract fiber residual characteristic values; The surface texture image in the multi-source detection data is divided into grids, the short run advantage of the gray-level run matrix of each grid is calculated, and the pollutant distribution characteristic value is obtained by weighted summation. The thickness variation coefficient is calculated from the thickness measurement data in the multi-source detection data and used as a characteristic value of mechanical strength attenuation. The characteristic values ​​of fiber residue, pollutant distribution, and mechanical strength attenuation are combined to form the characteristic values ​​of composite defects in raw materials.

[0007] Preferably, the fiber interference index and melt strength compensation coefficient are input into a pre-constructed process response surface database, and initial production parameters are obtained through multidimensional interpolation, including: Using fiber interference index and melt strength compensation coefficient as two-dimensional input variables, four neighboring grid points on the corresponding response surface grid are located in the process response surface database. The process response surface database pre-stores the optimal parameter combinations under different combinations of fiber interference index and melt strength compensation coefficient, and the optimal parameter combinations include initial melting temperature, screw speed and die head pressure. Using bilinear interpolation, the initial melting temperature, screw speed, and die head pressure are calculated based on the current fiber interference index, melt strength compensation coefficient, and distance weights of four neighboring grid points. These parameters are then combined as the initial production parameters. If the current fiber interference index or melt strength compensation coefficient exceeds the calibration boundary of the database, the parameter value corresponding to the nearest boundary grid point will be used as the initial production parameter, and a boundary extrapolation warning will be issued.

[0008] Preferably, historical energy consumption records of similar granulation production lines are obtained, and an energy consumption regression equation is established with melt temperature, screw speed, and output as variables, including: Multiple sets of operating condition data were collected from the historical operation database of similar granulation production lines. Each set of operating condition data included at least the melting temperature, screw speed, output per unit time, and the corresponding measured unit energy consumption. A multiple regression analysis method was adopted, with melting temperature, screw speed and output as independent variables and unit energy consumption as dependent variable, to construct an initial regression model containing linear terms, quadratic terms and interaction terms between independent variables. The historical energy consumption records were fitted using the least squares method, the regression coefficients of each term were calculated, and insignificant terms were removed through stepwise regression or analysis of variance to obtain the final energy consumption regression equation.

[0009] Preferably, the initial production parameters are input into the energy consumption regression equation to calculate the predicted unit energy consumption. When the predicted unit energy consumption exceeds the preset energy consumption threshold, the screw speed and melting temperature are adjusted to reduce the unit energy consumption, resulting in corrected production parameters, including: The initial melting temperature and initial screw speed in the initial production parameters, along with the preset target output, are substituted into the energy consumption regression equation to calculate the predicted unit energy consumption. The predicted unit energy consumption is compared with the preset energy consumption threshold. If the predicted unit energy consumption is less than or equal to the preset energy consumption threshold, the initial production parameters are directly used as the corrected production parameters. If the predicted unit energy consumption is greater than the preset energy consumption threshold, the step-by-step joint adjustment process will be initiated.

[0010] Preferably, the process proceeds to a step-by-step joint adjustment process, including: Under the condition that the initial value of the melting temperature remains unchanged, the screw speed is gradually reduced in a preset step size. After each reduction, the screw speed is recalculated by substituting it into the energy consumption regression equation until the predicted unit energy consumption drops below the threshold or the screw speed reaches the lower limit. If the screw speed has been adjusted to the lower limit but the predicted unit energy consumption is still higher than the threshold, the screw speed is restored to the initial value. While keeping the screw speed constant, the melting temperature is gradually reduced in a preset step size. The predicted unit energy consumption is recalculated after each reduction until the predicted unit energy consumption drops below the threshold or the melting temperature reaches the lower limit. If the above steps fail to reduce the predicted unit energy consumption below the threshold, then the screw speed and melting temperature will be gradually reduced in small increments, and the predicted unit energy consumption will be recalculated after each synchronous adjustment until the threshold requirement is met or the preset maximum number of adjustments is reached. The final melt temperature and screw speed that meet the predicted unit energy consumption threshold will be combined with the die pressure in the initial production parameters to form the corrected production parameters; if the threshold requirement still cannot be met after all adjustments are tried, the set of parameters with the lowest energy consumption will be used as the corrected production parameters and a warning will be output.

[0011] Preferably, granulation is initiated by correcting production parameters, and the melt pressure signal at the die head is collected in real time and the root mean square of pressure fluctuation is calculated, including: The corrected production parameters are called up, and the corrected melt temperature, corrected screw speed and corrected die head pressure are written into the granulator control system. The extruder and die head heating are started, and the screw is driven to run at the corrected screw speed to start regeneration granulation. After the granulation process is running stably, the pressure sensor installed at the die head is used to continuously collect the melt pressure signal at the die head at a preset high-frequency sampling period to form a real-time pressure time series. Extract the corrected head pressure included in the corrected production parameters as the benchmark target pressure, calculate the deviation value between each sampling point in the real-time pressure time series and the benchmark target pressure, and generate a pressure deviation series; Based on a preset sliding time window, the pressure deviation sequence within the window is squared and the arithmetic mean is calculated. Then, the square root of the average value is taken to obtain the root mean square of the pressure fluctuation at the current moment.

[0012] Preferably, when the root mean square of the pressure fluctuation is greater than the fluctuation threshold, a proportional-integral controller performs closed-loop compensation of the die head pressure to complete the regenerated granulation production management, including: The root mean square of the pressure fluctuation is compared with a set fluctuation threshold. If the root mean square of the pressure fluctuation is greater than the fluctuation threshold, a trigger signal is generated. The trigger signal and the root mean square pressure fluctuation are input to the proportional-integral controller. At the same time, the corrected head pressure included in the corrected production parameters is read as the target setpoint, and the real-time head melt pressure at the current moment is read as the feedback value. The current pressure deviation is calculated based on the target setpoint and real-time feedback value. Combined with the historical pressure deviation sequence, the proportional gain coefficient and integral gain coefficient are used for weighted calculation to generate the proportional control component and integral control component. The proportional and integral control components are superimposed to obtain the total control amount; the total control amount is used to reverse the screw speed to generate a new target screw speed; the speed is reduced when the pressure is too high and increased when the pressure is too low. If the absolute value of the total adjustment exceeds the preset speed adjustment limit, the remaining adjustment will be used to assist in correcting the melting temperature and generate a new target melting temperature. The new target screw speed and new target melting temperature are sent to the granulator control system to replace the original modified production parameters, and the melt pressure signal at the die head is continuously monitored. Recalculate the root mean square of pressure fluctuation. If it drops below the fluctuation threshold, maintain the current parameters and clear the integral term of the proportional-integral controller. If the root mean square of pressure fluctuation continues to exceed the limit and the adjustment has reached its limit, trigger an alarm and shut down for inspection.

[0013] Compared with the prior art, the beneficial effects of the present invention are: This invention uses multi-source data fusion, including transmission spectroscopy, surface texture, and thickness measurement, to extract characteristic values ​​of fiber residue, contaminant distribution, and mechanical strength attenuation. It also introduces the concepts of plasticization collapse point and dynamic coupling boundary line. By calculating the vertical distance from the current raw material coordinate point to the dynamic boundary, the synergistic destructive effect of fibers and contaminants can be quantified, enabling the prediction of plasticization collapse risk before production. This significantly reduces impurities and mechanical performance defects in recycled granules, thereby improving product qualification rate. This invention establishes a process response surface database with fiber interference index and melt strength compensation coefficient as dual inputs. Through multidimensional interpolation algorithm, it can automatically match the best historical working conditions and generate personalized initial melting temperature, screw speed and die head pressure, avoiding melt fracture or poor plasticization caused by improper parameters, and ensuring the process adaptability of different batches of raw materials. This invention establishes a multivariate energy consumption regression equation containing quadratic and cross terms, which can predict unit energy consumption before production starts. By adjusting the process step by step, it can automatically find the combination of process parameters with the lowest energy consumption while ensuring output and quality. This not only avoids energy waste caused by blind production, but also replaces manual trial and error and improves the level of precision in production management. Attached Figure Description

[0014] Figure 1 This is a schematic flowchart of the overall method in one embodiment of the present invention. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] Example 1, please refer to Figure 1This invention provides a technical solution: a method for producing recycled granules based on waste plastic fiber bags, comprising: S1. Obtain multi-source detection data of waste plastic fiber bags; wherein, the multi-source detection data includes transmission spectral data, surface texture images and thickness measurement data; S2. Based on multi-source detection data, feature analysis is performed to obtain the composite defect feature values ​​of raw materials; among which, the composite defect feature values ​​of raw materials include fiber residue feature values, contaminant distribution feature values, and mechanical strength attenuation feature values; S3. Using the pre-calibrated fiber residue weight and contaminant influence weight, the fiber residue characteristic value and contaminant distribution characteristic value are weighted and summed to generate the fiber interference index; based on the mechanical strength attenuation characteristic value, the pre-built strength attenuation-compensation coefficient comparison table is consulted to obtain the melt strength compensation coefficient. S4. Input the fiber interference index and melt strength compensation coefficient into the pre-constructed process response surface database, and obtain the initial production parameters through multidimensional interpolation; among which, the initial production parameters include the initial melt temperature, screw speed and die head pressure; S5. Obtain historical energy consumption records of similar granulation production lines and establish an energy consumption regression equation with melt temperature, screw speed and output as variables; input the initial production parameters into the energy consumption regression equation to calculate the predicted unit energy consumption; when the predicted unit energy consumption exceeds the preset energy consumption threshold, adjust the screw speed and melt temperature in the direction of reducing unit energy consumption to obtain the corrected production parameters. S6. Start granulation by correcting production parameters, collect the melt pressure signal of the die head in real time and calculate the root mean square of pressure fluctuation; when the root mean square of pressure fluctuation is greater than the fluctuation threshold, perform closed-loop compensation of the die head pressure through a proportional-integral controller to complete the production management of regenerated granulation.

[0017] It should be noted that transmission spectral data refers to optical characteristic data obtained by scanning waste plastic fiber bags using a near-infrared spectrometer. For example, for polypropylene woven bags, characteristic absorption peaks appear near wavelengths of 1200 nm and 1650 nm. The spectrometer determines the degree of fiber residue by detecting the light absorption intensity in these two wavelength bands. Surface texture images refer to grayscale or color images obtained by capturing the surface of raw materials using an industrial CCD camera. For example, if there are many irregular white bright spots or dark patches in the image captured by the camera, it indicates the presence of mud or ink contaminants. The system divides the image into a 10x10 grid and counts the number of abnormal pixels in each grid to quantify the contamination situation. Thickness measurement data refers to a numerical sequence obtained by continuously measuring the thickness of raw materials on a conveyor belt using a laser thickness gauge. For example, the laser thickness gauge collects 100 data points per second, and the ratio of the standard deviation to the mean of these 100 points is calculated, i.e., the thickness variation coefficient. The larger this value, the more severe the aging of the raw material and the lower its mechanical strength. The characteristic values ​​of composite defects in raw materials include fiber residue characteristic values, contaminant distribution characteristic values, and mechanical strength attenuation characteristic values. The fiber residue characteristic value is calculated based on the integral area of ​​the characteristic absorption peaks in the transmission spectrum. For example, if the area of ​​the absorption peak at 1200 nm in the spectrum is greater than 5000 units, the fiber residue characteristic value is recorded as 0.8; the smaller the area, the smaller the characteristic value. The contaminant distribution characteristic value is calculated based on the short run dominance value of the grayscale run matrix of the surface texture image. For example, by calculating the frequency of short grayscale line segments in the image, a higher frequency indicates more micro-cracks or oil contamination on the surface. If the calculation result is 0.65, the contaminant distribution characteristic value is 0.65. The mechanical strength attenuation characteristic value is a value directly converted from the thickness variation coefficient. For example, when the thickness variation coefficient exceeds 0.25, the mechanical strength attenuation characteristic value is set to 0.6; the larger the variation coefficient, the closer the characteristic value is to 1. The generation process of the fiber interference index involves the calculation of a dynamic coupling boundary line. For example, the system pre-stores 100 sets of data on historical plasticization collapses. These data form a critical curve in a coordinate system with fiber residue as the horizontal axis and contaminant distribution as the vertical axis. If the current raw material's coordinate points are 0.7 and 0.6, the system calculates the vertical distance from that point to the critical curve. Assuming the distance is 0.15, this distance serves as a dynamic coupling factor. The weight compensation function adjusts the weights of fiber residue and contaminants based on this factor. If the distance is very small, it indicates proximity to the collapse zone, and the weights will increase. Finally, the adjusted weights are multiplied by the corresponding characteristic values ​​and summed to obtain the fiber interference index. The melt strength compensation coefficient is obtained by consulting a strength attenuation-compensation coefficient lookup table. For example, the lookup table specifies that when the mechanical strength attenuation characteristic value is 0.3, the compensation coefficient is 1.0, indicating no additional compensation is needed; when the attenuation characteristic value is 0.8, the compensation coefficient is 1.2, indicating that the melt strength needs to be increased. The process response surface database is a three-dimensional lookup table storing historical optimal process parameters. For example, the database records that when the fiber interference index is 0.6 and the melt strength compensation coefficient is 1.1, the corresponding optimal parameters are a melt temperature of 170 degrees Celsius, a screw speed of 280 rpm, and a die pressure of 12 MPa. Multidimensional interpolation uses bilinear interpolation. Assuming the currently calculated fiber interference index is 0.65 and the melt strength compensation coefficient is 1.15, this point is located between four grid points in the database: 0.6 and 0.7, and 1.1 and 1.2. Based on the positional relationship of 0.65 between 0.6 and 0.7, the system assigns a weight of 0.5 to both sides. Similarly, it assigns a weight of 0.5 to 1.15. The weighted average values ​​of temperature, screw speed, and pressure are calculated to obtain the initial production parameters. The energy consumption regression equation is a mathematical relationship established based on multiple regression analysis. For example, 500 sets of data were collected from a historical database, each set including melting temperature, screw speed, output, and measured energy consumption. Through fitting, it was found that energy consumption is not only related to the first power of these three factors, but also to the square of temperature, the square of speed, and the product of temperature and speed. The final equation is: energy consumption equals 0.05 times temperature plus 0.02 times speed minus 0.01 times output plus quadratic terms and cross terms. Substituting the initial production parameters of 180 degrees Celsius, 300 rpm, and preset output of 500 kg / h into the equation, the predicted unit energy consumption is calculated to be 105 kWh / ton. If the preset energy consumption threshold is 100 kWh / ton, it is judged to exceed the standard. The root mean square (RMS) calculation of pressure fluctuation is based on a sliding time window. For example, setting the sliding window length to 10 seconds and the sampling frequency to 100 times per second, the system extracts the head pressure of 12 MPa from the corrected production parameters as the baseline value. It calculates the difference between each sampling point within the window and the baseline value, squares these differences, sums them, and divides the sum by the number of sampling points to obtain the average value. Finally, it takes the square root of the average value. Assuming the calculated result is 0.6 MPa, and the set fluctuation threshold is 0.5 MPa, the fluctuation is considered too large. The closed-loop compensation process of the proportional-integral (PI) controller includes proportional and integral regulation. For example, if the current real-time head pressure is 12.6 MPa, which is higher than the target value of 12.0 MPa... With a deviation of 0.6 MPa, the proportional term is directly multiplied by the proportional gain coefficient to generate the adjustment amount, assuming a reduction of 20 rpm in rotational speed. The integral term accumulates the pressure deviation over a period of time. If the deviation persists, the integral term will further increase the adjustment amount. If the total adjustment amount exceeds the upper limit of rotational speed adjustment of -30 rpm, for example, if the calculation requires a reduction of 40 rpm, the rotational speed is first reduced by 30 rpm, and the remaining 10 rpm adjustment amount is converted into an adjustment of the melting temperature, for example, a reduction of 3 degrees Celsius. The system sends the new rotational speed and temperature to the granulator and continuously monitors the pressure, recalculating the root mean square. If the pressure drops below 0.5 MPa, the parameters are maintained and the integral term is cleared.

[0018] In an optional embodiment, feature analysis is performed based on multi-source detection data to obtain feature values ​​of composite defects in the raw materials, including: Baseline correction and characteristic absorption peak integration were performed on the transmission spectral data in the multi-source detection data to extract fiber residual characteristic values; It should be noted that baseline correction refers to a preprocessing step to eliminate background noise and instrument drift. For example, it involves fitting a smooth baseline using an algorithm and raising or flattening the spectral data to highlight the true absorption characteristics. Characteristic absorption peak integration refers to calculating the area covered by the absorption peak within a specific wavelength range. This area value directly reflects the content of fiber components. For example, when the integrated area of ​​the absorption peak at 1200 nm is greater than 5000 units, the system determines that there is a lot of fiber residue and sets the fiber residue characteristic value to 0.8. The smaller the area value, the lower the characteristic value. The surface texture image in the multi-source detection data is divided into grids, the short run advantage of the gray-level run matrix of each grid is calculated, and the pollutant distribution characteristic value is obtained by weighted summation. It should be noted that grid partitioning refers to dividing the entire image into several small regions for independent analysis. For example, a 1000-pixel by 1000-pixel image can be divided into 10 rows and 10 columns, totaling 100 grid cells. The grayscale run-length matrix is ​​a method for statistically analyzing image texture features. It records the length distribution of pixel segments with the same grayscale value appearing consecutively in a certain direction. Short run-length advantage is a statistic in this matrix, specifically used to measure the frequency of short-length pixel segments. The larger the value, the more fragmented the surface microstructure or the more impurities there are. For example, if a large number of irregular white bright spots or dark patches are detected in a certain grid, the calculated short run-length advantage value is 0.65. The system uses this value as a pollutant distribution feature value. The higher the value, the denser the pollutant distribution. The thickness variation coefficient is calculated from the thickness measurement data in the multi-source detection data and used as a characteristic value of mechanical strength attenuation. It should be noted that the thickness variation coefficient is a statistical indicator describing the degree of thickness fluctuation. It is obtained by calculating the ratio of the standard deviation to the average value of thickness data over a period of time. For example, if a laser thickness gauge collects 100 data points per second for 10 seconds, totaling 1000 points, and the average value of these 1000 points is calculated to be 0.1 mm with a standard deviation of 0.025 mm, then the thickness variation coefficient is 0.25. This coefficient directly reflects the aging uniformity and mechanical property stability of the raw material; the larger the coefficient, the more severe the decrease in mechanical strength. For example, when the thickness variation coefficient exceeds 0.25, the system directly sets the mechanical strength decrease characteristic value to 0.6; the smaller the coefficient, the closer the characteristic value is to 0. The characteristic values ​​of fiber residue, contaminant distribution, and mechanical strength attenuation are combined to form the characteristic values ​​of raw material composite defects; It should be noted that the composite defect feature value of raw materials is a comprehensive index formed by combining the above three independent feature values ​​in vector form. For example, the system packages the extracted fiber residue feature value (0.8), contaminant distribution feature value (0.65), and mechanical strength attenuation feature value (0.6) into a three-dimensional array, which serves as the basic input for subsequent calculation of the fiber interference index.

[0019] In an optional embodiment, a fiber interference index is generated by weighting and summing the fiber residue characteristic values ​​and the contaminant distribution characteristic values ​​using pre-calibrated fiber residue weights and contaminant influence weights, including: Based on a pre-built historical production database, plasticization collapse point data of waste plastic fiber bags during the melting and plasticization stage are extracted; among them, the plasticization collapse point data are the fiber residue characteristic value and contaminant distribution characteristic value corresponding to the critical point of sudden increase in melt pressure and sudden drop in tensile strength. It should be noted that plasticization collapse point data refers to specific critical state data extracted from historical production records. Specifically, the system backtracks through the historical database to identify moments when waste plastic fiber bags experience a sudden surge in melt pressure and a sharp drop in tensile strength during the melting and plasticizing stage, and records the corresponding raw material test values ​​at that moment. For example, if the historical database contains 100 such abnormal records, one record shows that when the fiber residue characteristic value is 0.8 and the contaminant distribution characteristic value is 0.6, the melt pressure suddenly rises from 10MPa to 25MPa, while the tensile strength drops from 20MPa to 5MPa. This pair of data points with 0.8 and 0.6 constitutes a plasticization collapse point. The system collects all such critical point data as the basis for constructing the boundary line. A feature space is constructed with fiber residue characteristic value as the abscissa and pollutant distribution characteristic value as the ordinate. Plasticization collapse point data is mapped into the feature space, and a dynamic coupling boundary line is generated using a nonlinear fitting algorithm. The vertical distance from the coordinate point where the fiber residue characteristic value and pollutant distribution characteristic value of the current raw material are located to the dynamic coupling boundary line is calculated as the dynamic coupling factor. It should be noted that the feature space refers to a two-dimensional planar coordinate system established with fiber residue characteristic values ​​on the horizontal axis and contaminant distribution characteristic values ​​on the vertical axis; the dynamic coupling boundary line is a critical curve generated after processing the plasticization collapse point data using a nonlinear fitting algorithm; for example, projecting the above 100 plasticization collapse points onto the coordinate system and using a polynomial fitting algorithm to draw a smooth curve, the equation of which may be expressed as the vertical axis value equal to 0.5 times the square of the horizontal axis value plus 0.2. This curve divides the coordinate system into a safe zone and a risk zone; the dynamic coupling factor refers to the vertical distance from the current raw material's coordinate point to this dynamic coupling boundary line; for example, if the current raw material's fiber residue characteristic value is 0.7 and the contaminant distribution characteristic value is 0.6, forming a point (0.7, 0.6) in the coordinate system, the vertical distance from this point to the fitted curve is calculated, assuming the measured distance is 0.15, this 0.15 is the dynamic coupling factor, and the smaller the value, the closer to the collapse boundary; The dynamic coupling factor is input into the pre-constructed weight compensation function to perform nonlinear correction on the fiber residue weight and pollutant influence weight, so as to obtain the corrected fiber residue influence coefficient and pollutant distribution influence coefficient. It should be noted that the weight compensation function is a pre-defined mathematical relationship used to adjust the base weights according to the magnitude of the dynamic coupling factor. For example, in the pre-calibrated base weights, the weight of fiber residue is 0.6 and the weight of pollutant influence is 0.4. When the dynamic coupling factor is 0.15, the weight compensation function outputs a correction coefficient of 1.2, so the corrected fiber residue influence coefficient becomes 0.72 and the corrected pollutant distribution influence coefficient becomes 0.48. If the dynamic coupling factor increases to 0.3, it indicates an increased risk, and the weight compensation function may output a larger correction coefficient, such as 1.5, increasing the fiber residue influence coefficient to 0.9, thus giving higher attention to fiber residue in the calculation. The first weighted component is obtained by multiplying the corrected fiber residue influence coefficient by the current fiber residue characteristic value, and the second weighted component is obtained by multiplying the corrected pollutant distribution influence coefficient by the current pollutant distribution characteristic value. It should be noted that the first weighted component is the result of multiplying the corrected fiber residue influence coefficient by the current fiber residue characteristic value; for example, if the corrected fiber residue influence coefficient is 0.72 and the currently detected fiber residue characteristic value is 0.7, the product is 0.504, which is the first weighted component. The second weighted component is the result of multiplying the corrected contaminant distribution influence coefficient by the current contaminant distribution characteristic value; for example, if the corrected contaminant distribution influence coefficient is 0.48 and the current contaminant distribution characteristic value is 0.6, the product is 0.288, which is the second weighted component. These two components represent the degree of contribution of fibers and contaminants to the final interference index after risk correction. The first weighted component and the second weighted component are vector-superimposed to generate the fiber interference index, which characterizes the synergistic destructive effect of fibers and pollutants. It should be noted that vector superposition refers to the operation of directly adding the first weighted component and the second weighted component arithmetically. For example, adding the first weighted component 0.504 and the second weighted component 0.288 obtained above yields a sum of 0.792. This sum is the fiber interference index, which not only includes the influence of a single defect but also introduces the nonlinear influence of their interaction through dynamic coupling factors and weight compensation functions. When the index value is large, it indicates a strong synergistic destructive effect between the fiber and the contaminant. Based on this, the system determines that more conservative process parameters should be adopted to avoid plasticization collapse.

[0020] In an optional embodiment, the fiber interference index and melt strength compensation coefficient are input into a pre-built process response surface database, and initial production parameters are obtained through multidimensional interpolation, including: Using fiber interference index and melt strength compensation coefficient as two-dimensional input variables, four neighboring grid points on the corresponding response surface grid are located in the process response surface database. The process response surface database pre-stores the optimal parameter combinations under different combinations of fiber interference index and melt strength compensation coefficient, and the optimal parameter combinations include initial melting temperature, screw speed and die head pressure. It should be noted that the four nearest grid points refer to the four closest known data points surrounding the coordinates of the current input variable in the process response surface database. For example, if the currently calculated fiber interference index is 0.65 and the melt strength compensation coefficient is 1.15, in the database grid, 0.65 is located between 0.6 and 0.7, and 1.15 is located between 1.1 and 1.2. The system will locate the grid cell consisting of the four coordinate points (0.6, 1.1), (0.6, 1.2), (0.7, 1.1), and (0.7, 1.2). These four points are the nearest grid points of the current input variable, and the system will extract the corresponding melt temperature, screw speed, and die pressure values ​​for subsequent calculations. Using bilinear interpolation, the initial melting temperature, screw speed, and die head pressure are calculated based on the current fiber interference index, melt strength compensation coefficient, and distance weights of four neighboring grid points. These parameters are then combined as the initial production parameters. It should be noted that bilinear interpolation is a calculation method that uses distance weights for weighted averaging. Taking 0.65 and 1.15 as examples, this point has a horizontal weight of 0.05 at distances of 0.6 and 0.05 at distances of 0.7, each accounting for half the weight; vertically, it has a vertical weight of 0.05 at distances of 1.1 and 1.2, each accounting for half the weight. The system calculates the contribution of the four neighboring grid points to the current point. For example, for the melting temperature, if the temperatures of the four neighboring points are 170, 175, 172, and 177 degrees Celsius, the system will assign different weights to 0.65 and 1.15 according to their relative positions in the grid, and the calculated weighted average is approximately 173.5 degrees Celsius. Similarly, the screw speed is calculated to be approximately 287.5 revolutions per minute, and the die head pressure is approximately 12.25 MPa. Combining these three calculation results yields the initial production parameters. If the current fiber interference index or melt strength compensation coefficient exceeds the calibration boundary of the database, the parameter value corresponding to the nearest boundary grid point will be used as the initial production parameter, and a boundary extrapolation warning will be issued. It should be noted that the boundary extrapolation warning refers to the handling mechanism when the input variable exceeds the coverage range of the database; for example, the effective range of the fiber interference index in the process response surface database is 0 to 0.9. If the fiber interference index calculated for the current raw material is 0.95, exceeding the upper limit of 0.9, the system cannot perform interpolation calculation; at this time, the system will select the nearest boundary grid point, that is, the column of grid points corresponding to the fiber interference index of 0.9, as a reference, and directly read its corresponding parameters as the initial production parameters; at the same time, the system will trigger an alarm prompt, informing the operator that the defect level of the current raw material has exceeded the range of historical experience, belongs to extrapolation prediction, and may have a large risk, requiring manual intervention or careful monitoring; if the calculated value of the melt strength compensation coefficient is 1.6, exceeding the upper limit of 1.5, the handling logic is the same, the system will select the parameters corresponding to the 1.5 boundary and issue a warning.

[0021] In an optional embodiment, historical energy consumption records of similar granulation production lines are obtained, and an energy consumption regression equation is established with melt temperature, screw speed, and output as variables, including: Multiple sets of operating condition data were collected from the historical operation database of similar granulation production lines. Each set of operating condition data included at least the melting temperature, screw speed, output per unit time, and the corresponding measured unit energy consumption. It should be noted that the historical operation database stores a large number of operational records from actual production processes. Each set of operating data corresponds to a stable production period, such as continuous operation for more than 30 minutes, during which the melt temperature, screw speed, and output remain basically unchanged, while the measured unit energy consumption is recorded. For example, the system extracted 500 sets of data from the database. One set of data showed a melt temperature of 200℃, a screw speed of 300rpm, an output of 150kg / h per unit time, and a measured unit energy consumption of 105kWh / t; another set of data showed a melt temperature of 190℃, a screw speed of 280rpm, an output of 160kg / h, and a measured unit energy consumption of 102kWh / t. These data together constitute the basic sample for establishing the regression equation. A multiple regression analysis method was adopted, with melting temperature, screw speed and output as independent variables and unit energy consumption as dependent variable, to construct an initial regression model containing linear terms, quadratic terms and interaction terms between independent variables. It should be noted that multiple regression analysis is a statistical modeling method used to reveal the quantitative relationship between multiple independent variables and a dependent variable. The initial regression model considers not only the linear impact of individual changes in each independent variable on energy consumption, but also the impact of nonlinear changes in the independent variables themselves. For example, energy consumption does not increase uniformly but accelerates when the temperature rises, as well as the mutual influence between independent variables. For example, when temperature and screw speed increase simultaneously, the combined effect on energy consumption exceeds the sum of their individual effects. For instance, the system sets the initial regression model as follows: energy consumption equals a constant term plus a linear term of melting temperature multiplied by regression coefficient one, plus a quadratic term of melting temperature multiplied by regression coefficient two, plus a linear term of screw speed multiplied by regression coefficient three, plus a quadratic term of screw speed multiplied by regression coefficient four, plus a linear term of output multiplied by regression coefficient five, plus a quadratic term of output multiplied by regression coefficient six, plus the product of melting temperature and screw speed multiplied by regression coefficient seven, plus the product of melting temperature and output multiplied by regression coefficient eight, plus the product of screw speed and output multiplied by regression coefficient nine. This model can more accurately describe the energy consumption variation pattern in actual production. The historical energy consumption records were fitted using the least squares method, the regression coefficients of each term were calculated, and insignificant terms were removed through stepwise regression or analysis of variance to obtain the final energy consumption regression equation. It should be noted that the least squares method is a parameter estimation method that finds the most suitable regression coefficient value by minimizing the sum of squares of the differences between the predicted energy consumption and the actual energy consumption of all sample points. Stepwise regression refers to gradually introducing or removing a term from all candidate independent variables, introducing the term with the most significant impact on the dependent variable each time and removing the term with an insignificant impact, until all terms in the model meet the significance requirement. Analysis of variance, on the other hand, determines whether a term should be retained by calculating the contribution of each term to the change in energy consumption and the statistical test value. For example, the system substitutes 500 sets of historical data into the initial regression model, calculates the specific values ​​of each regression coefficient using the least squares method, and then finds through stepwise regression that the quadratic term of output has a very weak impact on energy consumption, with a statistical test value below the significance level of 0.05, so it is removed from the model. At the same time, it is found that the interaction term between melting temperature and output is also insignificant, so it is also removed. The final energy consumption regression equation only retains the first term of melting temperature, the first term of screw speed, the first term of output, the quadratic term of melting temperature, the quadratic term of screw speed, and the interaction term between melting temperature and screw speed.

[0022] In an optional embodiment, initial production parameters are input into an energy consumption regression equation to calculate predicted unit energy consumption. When the predicted unit energy consumption exceeds a preset energy consumption threshold, the screw speed and melting temperature are adjusted to reduce unit energy consumption, resulting in corrected production parameters, including: The initial melting temperature and initial screw speed in the initial production parameters, along with the preset target output, are substituted into the energy consumption regression equation to calculate the predicted unit energy consumption. It should be noted that the target output is a fixed value preset based on the production plan and equipment capacity, and it usually remains unchanged during adjustments. The predicted unit energy consumption is an estimate of the energy consumption that may be generated under the current combination of process parameters, based on a regression equation; for example, assuming that the initial melting temperature in the initial production parameters is 215℃, the initial screw speed is 320rpm, and the preset target output is 150kg / h, the system substitutes these values ​​into the previously established energy consumption regression equation, and after calculation, the predicted unit energy consumption is 136kWh / t. The predicted unit energy consumption is compared with the preset energy consumption threshold. If the predicted unit energy consumption is less than or equal to the preset energy consumption threshold, the initial production parameters are directly used as the corrected production parameters. It should be noted that the preset energy consumption threshold is a maximum allowable unit energy consumption value set by the enterprise based on energy-saving goals and equipment performance. If the predicted unit energy consumption does not exceed this threshold, it means that the initial production parameters have met the energy-saving requirements and do not need to be adjusted; they can be used directly for production. For example, assuming the preset energy consumption threshold is 140 kWh / t, and the calculated predicted unit energy consumption is 136 kWh / t, 136 is less than 140, which meets the requirements. In this case, the system will directly use the initial melting temperature of 215℃, the initial screw speed of 320 rpm, and the initial die head pressure of 2.8 MPa as the corrected production parameters without any further adjustments. If the predicted unit energy consumption is greater than the preset energy consumption threshold, then the step-by-step joint adjustment process will be initiated. It should be noted that the step-by-step joint adjustment process is a successive approximation optimization method. It is activated when the initial parameter energy consumption exceeds the limit. The process prioritizes a single parameter adjustment strategy, first trying to reduce only the screw speed. If this is ineffective, it then tries to reduce only the melt temperature, and only then considers reducing both parameters simultaneously. This sequence is based on field experience: adjusting the screw speed has a more direct impact on energy consumption and a relatively smaller impact on product quality, while reducing the melt temperature requires caution to avoid poor plasticization. Therefore, adjusting the speed is prioritized.

[0023] In an optional embodiment, the step-by-step joint adjustment process includes: Under the condition that the initial value of the melting temperature remains unchanged, the screw speed is gradually reduced in a preset step size. After each reduction, the screw speed is recalculated by substituting it into the energy consumption regression equation until the predicted unit energy consumption drops below the threshold or the screw speed reaches the lower limit. It should be noted that the preset step size is a fixed change in screw speed each time, for example, a decrease of 10 rpm each time; the lower limit of screw speed is the minimum allowable speed to ensure stable plasticizing and extrusion processes. Below this value, it may lead to uneven melt mixing or reduced output; the system recalculates energy consumption after each speed reduction and determines whether it meets the standard; for example: the initial melt temperature is kept constant at 215℃, and the initial screw speed is 320 rpm; the system first reduces the speed to 310 rpm, and the predicted unit energy consumption is calculated to be 134 kWh / t by substituting into the regression equation, which is still higher than the threshold of 130 kWh / t; the second time it is reduced to 300 rpm, and the calculated energy consumption is 132 kWh / t, which still exceeds the standard; the third time it is reduced to 290 rpm, and the calculated energy consumption is 130 kWh / t, which is exactly equal to the threshold; at this point, the adjustment stops, and the current parameters are recorded. If the energy consumption still exceeds the standard after reducing to the lower limit, for example, 250 rpm, then proceed to the next step; If the screw speed has been adjusted to the lower limit but the predicted unit energy consumption is still higher than the threshold, the screw speed is restored to the initial value. While keeping the screw speed constant, the melting temperature is gradually reduced in a preset step size. The predicted unit energy consumption is recalculated after each reduction until the predicted unit energy consumption drops below the threshold or the melting temperature reaches the lower limit. It should be noted that the lower limit of the melting temperature is the minimum temperature at which the plastic can be fully melted without thermal degradation. Temperatures below this value may lead to poor plasticization and affect granulation quality. This step size usually has a more significant energy consumption effect than the speed adjustment step size; therefore, the temperature reduction step size is small, for example, 5℃ each time. For example, assuming the screw speed has been reduced to the lower limit of 250 rpm, but the calculated predicted unit energy consumption is still 132 kWh / t, higher than the threshold of 130 kWh / t, the system restores the screw speed to the initial 320 rpm, then keeps the speed constant and begins to reduce the melting temperature, from 215℃ to 210℃, with a calculated energy consumption of 133 kWh / t; reducing it to 205℃, the calculated energy consumption is 130 kWh / t, which meets the standard. If reducing it to the lower limit, for example, 190℃, still does not meet the standard, then proceed to the next step. If the above steps fail to reduce the predicted unit energy consumption below the threshold, then the screw speed and melting temperature will be gradually reduced in small increments, and the predicted unit energy consumption will be recalculated after each synchronous adjustment until the threshold requirement is met or the preset maximum number of adjustments is reached. It should be noted that when adjusting a single parameter fails to achieve the energy-saving target, it indicates that two parameters need to be reduced synergistically. In this case, the step size is smaller than that used when adjusting a single parameter. For example, each adjustment reduces the screw speed by 5 rpm while simultaneously reducing the temperature by 2°C to avoid over-adjustment leading to deterioration of plasticizing quality. The maximum number of adjustments is used to prevent infinite loops; for example, a maximum of 10 adjustments are set. For instance, after the first two adjustments, if the screw speed drops to the lower limit but energy consumption still exceeds the limit, and the melt temperature drops to the lower limit but still exceeds the limit, the system begins synchronous adjustment: First, the screw speed is reduced from 320 rpm to 315 rpm, and the temperature is reduced from 215°C to 213°C, with a calculated energy consumption of 132 kWh / t, still higher than 130 kWh / t; Second, the screw speed is reduced to 310 rpm, and the temperature is reduced to 211°C, with a calculated energy consumption of 129 kWh / t, below the threshold, so the adjustment stops, and the current parameters are recorded. The final melt temperature and screw speed that meet the predicted unit energy consumption threshold will be combined with the die head pressure in the initial production parameters to form the corrected production parameters; if the threshold requirement still cannot be met after all adjustments are tried, the set of parameters with the lowest energy consumption will be used as the corrected production parameters and a warning will be output. It should be noted that the corrected production parameters are the final process parameters actually used to start granulation. These include the adjusted melt temperature and screw speed, as well as the constant die head pressure. If all possible adjustments fail to reach the energy consumption threshold, it indicates that the expected energy-saving target cannot be achieved under the current raw material and equipment conditions. In this case, the system will select the combination of parameters with the lowest predicted energy consumption as a compromise. Simultaneously, a warning signal will be issued to remind operators to pay attention to the high energy consumption issue. For example, after 10 simultaneous adjustments, it is found that even if the speed is reduced to 250 rpm and the temperature to 190℃, the predicted unit energy consumption is still 131 kWh / t, higher than the threshold of 130 kWh / t. The system reviews all tried parameter combinations and finds that the lowest predicted energy consumption is 130.5 kWh / t at a speed of 260 rpm and a temperature of 195℃. Therefore, this set of parameters is combined with the initial die head pressure of 2.8 MPa as the corrected production parameters, and a warning message is output: "Energy consumption exceeds the standard; the lowest energy consumption parameter has been used. Please check the raw materials or equipment."

[0024] In an optional embodiment, granulation is initiated by correcting production parameters, and the melt pressure signal at the die head is collected in real time and the root mean square of the pressure fluctuation is calculated, including: The corrected production parameters are called up, and the corrected melt temperature, corrected screw speed and corrected die head pressure are written into the granulator control system. The extruder and die head heating are started, and the screw is driven to run at the corrected screw speed to start regeneration granulation. It should be noted that writing the corrected production parameters into the control system means sending the values ​​to the programmable logic controller or direct digital controller via an industrial communication protocol. The controller then drives the heater and motor to perform corresponding actions based on these set values. For example, the system writes the corrected melt temperature of 205°C into the temperature controller of each section of the extruder, the corrected screw speed of 280 rpm into the motor frequency converter, and the corrected die head pressure of 2.8 MPa as the target value for die head opening adjustment. Then the system issues a start command, the extruder starts heating up, and after the temperature reaches the set value, the motor drives the screw to rotate at a speed of 280 rpm. The raw material enters the barrel from the feed port, is melted and plasticized, and then extruded from the die head, completing the start process of regenerated granulation. After the granulation process is running stably, the pressure sensor installed at the die head is used to continuously collect the melt pressure signal at the die head at a preset high-frequency sampling period to form a real-time pressure time series. It should be noted that stable operation of the granulation process typically refers to the period from startup to 3 to 5 minutes, during which the temperature and pressure of each section reach the set values ​​with minimal fluctuations. The high-frequency sampling period is generally set to 50 to 200 times per second to capture the details of pressure fluctuations. The real-time pressure time series is a series of pressure values ​​arranged in chronological order, reflecting the dynamic changes in the pressure at the die head. For example, after the granulator has been running for 5 minutes, the temperature of each section has stabilized at around 205°C. The system then begins to collect pressure signals, with a sampling period set to 100 times per second, meaning that a pressure value is collected every 0.01 seconds. After 10 seconds of continuous collection, 1000 pressure data points are obtained. These data points are arranged in chronological order to form a pressure time series. For example, the first data point is 2.81 MPa, the second point is 2.83 MPa, and so on. Extract the corrected head pressure included in the corrected production parameters as the benchmark target pressure, calculate the deviation value between each sampling point in the real-time pressure time series and the benchmark target pressure, and generate a pressure deviation series; It should be noted that the baseline target pressure is the ideal pressure value to be achieved, usually the corrected head pressure. The deviation value indicates the degree to which the actual pressure deviates from the target pressure. It can be a positive value indicating that the actual pressure is higher than the target, or a negative value indicating that the actual pressure is lower than the target. The pressure deviation sequence is obtained by arranging the deviations of each sampling point in chronological order. For example, if the corrected head pressure is 2.80 MPa, and this is used as the baseline target pressure, the first sampling point in the pressure time series is 2.81 MPa with a deviation of +0.01 MPa; the second sampling point is 2.83 MPa with a deviation of +0.03 MPa; the third sampling point is 2.78 MPa with a deviation of -0.02 MPa. The deviation values ​​of all 1000 sampling points are calculated sequentially to obtain the pressure deviation sequence. Based on a preset sliding time window, the pressure deviation sequence within the window is squared and the arithmetic mean is calculated. Then, the square root of the average value is taken to obtain the root mean square of the pressure fluctuation at the current moment. It should be noted that the sliding time window is a fixed-length period, such as 10 seconds. The system calculates the sum of squares of all deviation values ​​within the window at shorter time intervals, such as once per second. This sum is divided by the average number of sampling points within the window, and the square root is taken to obtain the root mean square (RMS) value. This RMS value comprehensively reflects the magnitude of pressure fluctuations within the window period; the greater the fluctuation, the larger the RMS value. For example, if the sliding window length is set to 10 seconds and the sampling frequency is 100 times / second, each window contains 1000 deviation values. At the end of the 10th second after startup and stabilization, the system calculates the sum of squares of the 1000 deviations in the first window, divides it by 1000 to obtain the average value, and then takes the square root, assuming the result is 0.32 MPa. At the end of the 11th second, the window slides forward 1 second, incorporating the new data from the 11th second and removing the old data from the 1st second, recalculating to obtain 0.35 MPa. This yields one RMS value of pressure fluctuation per second, used for real-time monitoring of pressure stability.

[0025] In an optional embodiment, when the root mean square of the pressure fluctuation is greater than the fluctuation threshold, a proportional-integral controller performs closed-loop compensation of the die head pressure to complete the regenerated granulation production management, including: The root mean square of the pressure fluctuation is compared with a set fluctuation threshold. If the root mean square of the pressure fluctuation is greater than the fluctuation threshold, a trigger signal is generated. It should be noted that the fluctuation threshold is a maximum allowable root mean square value preset according to product quality requirements and process stability requirements; the trigger signal is a Boolean flag used to initiate the compensation action of the proportional-integral controller; for example, if the fluctuation threshold is set to 0.30 MPa, and the currently calculated root mean square pressure fluctuation is 0.32 MPa, 0.32 is greater than 0.30, which meets the trigger condition, and the system immediately generates a trigger signal, which is a logical true value, indicating that closed-loop compensation needs to be initiated; The trigger signal and the root mean square pressure fluctuation are input to the proportional-integral controller. At the same time, the corrected head pressure included in the corrected production parameters is read as the target setpoint, and the real-time head melt pressure at the current moment is read as the feedback value. It should be noted that the proportional-integral controller is a closed-loop control algorithm. It starts working after receiving a trigger signal and uses the deviation between the target setpoint and the feedback value to generate the control output. The magnitude of the root mean square of the pressure fluctuation can be used to dynamically adjust the controller's gain parameter. The larger the fluctuation, the stronger the controller response. For example, if the trigger signal is true, the system will use the corrected head pressure of 2.80 MPa as the target setpoint, the real-time head melt pressure of 2.95 MPa measured by the pressure sensor at the current moment as the feedback value, and input the root mean square pressure fluctuation of 0.32 MPa into the controller's adaptive module. This module will appropriately increase the proportional gain coefficient to enhance the response speed. The current pressure deviation is calculated based on the target setpoint and real-time feedback value. Combined with the historical pressure deviation sequence, the proportional gain coefficient and integral gain coefficient are used for weighted calculation to generate the proportional control component and integral control component. It should be noted that the current pressure deviation is the target setpoint minus the real-time feedback value. A positive deviation indicates that the actual pressure is lower than the target, and a negative deviation indicates that the actual pressure is higher than the target. The proportional control component is the deviation value multiplied by the proportional gain coefficient, used to quickly respond to the current deviation. The integral control component is the sum of historical pressure deviations multiplied by the integral gain coefficient, used to eliminate long-term steady-state errors. For example, if the target setpoint is 2.80 MPa, the real-time feedback value is 2.95 MPa, the current pressure deviation is -0.15 MPa, the proportional gain coefficient is set to 1.5, and the proportional control component is -0.225, indicating that the speed needs to be reduced by 0.225 multiplied by a certain conversion coefficient. The historical pressure deviation includes the deviation values ​​recorded every 0.1 seconds over the past 10 seconds, and their sum is -2.4 MPa·s. The integral gain coefficient is set to 0.1, and the integral control component is -0.24. Adding the two parts together gives the total control component of -0.465. The proportional and integral control components are superimposed to obtain the total control amount; the total control amount is used to reverse the screw speed to generate a new target screw speed; the speed is reduced when the pressure is too high and increased when the pressure is too low. It should be noted that the total adjustment is a dimensionless relative value and needs to be multiplied by the speed conversion factor to obtain the specific speed change value. The meaning of reverse correction is: when the pressure is higher than the target, the speed needs to be reduced to reduce the melt delivery and thus reduce the pressure; when the pressure is lower than the target, the speed needs to be increased to increase the melt delivery and thus increase the pressure. For example, a total adjustment of -0.465 indicates that the pressure is too high and the speed needs to be reduced. The speed conversion factor is set to 40 rpm per unit adjustment, so the speed that needs to be reduced is -0.465 multiplied by 40, which equals -18.6 rpm. The current corrected screw speed is 280 rpm, and the new target screw speed is 280 minus 18.6, which is 261.4 rpm, rounded to 261 rpm. If the absolute value of the total adjustment exceeds the preset speed adjustment limit, the remaining adjustment will be used to assist in correcting the melting temperature and generate a new target melting temperature. It should be noted that the upper limit of the speed adjustment is to ensure that the screw does not reduce speed excessively, which could lead to poor melt mixing or a significant drop in output. When the required speed change exceeds the upper limit, the system first adjusts the speed to the upper limit value, and then converts the remaining adjustment amount into a change in the melting temperature. Lowering the temperature can increase the melt viscosity and increase the die head pressure. Therefore, for cases where the pressure is too high, in addition to reducing the speed, the temperature can also be appropriately reduced; for cases where the pressure is too low, the temperature is increased. For example, if the preset upper limit of the speed adjustment is a maximum reduction of 30 rpm in a single operation, and the current required speed reduction is 50 rpm, which exceeds the upper limit by 20 rpm, the system first reduces the speed by 30 rpm, from 280 rpm to 250 rpm. The remaining 20 rpm needs to be converted into a temperature adjustment, with a conversion coefficient of 2℃ per 10 rpm. The remaining 20 rpm corresponds to a reduction of 4℃. The current corrected melting temperature is 205℃, and the new target melting temperature is 201℃. The new target screw speed and new target melting temperature are sent to the granulator control system to replace the original modified production parameters, and the melt pressure signal at the die head is continuously monitored. It should be noted that the distribution operation writes the new setpoints into the corresponding register of the controller via the industrial communication network. The controller then adjusts the motor frequency and heating power, replacing the original corrected production parameters. This means that subsequent pressure monitoring and fluctuation calculations will be based on the new parameters. Continuous monitoring refers to maintaining high-frequency sampling and constantly calculating the root mean square of pressure fluctuations to prepare for the next possible compensation. Recalculate the root mean square of pressure fluctuation. If it drops below the fluctuation threshold, maintain the current parameters and clear the integral term of the proportional-integral controller. If the root mean square of pressure fluctuation continues to exceed the limit and the adjustment has reached its limit, trigger an alarm and shut down for inspection. It should be noted that the recalculation of the root mean square of pressure fluctuation is performed after the new parameters are issued and a stabilization period is waited, usually 30 to 60 seconds. If the root mean square of fluctuation is already below the threshold, it indicates that the compensation is effective, and the system continues to operate with the current parameters. At the same time, the accumulated historical deviations in the integral controller are cleared to avoid integral saturation during subsequent adjustments. If the adjustment has reached its limit, for example, the speed has dropped to the lower limit and the temperature has dropped to the lower limit, but the pressure fluctuation is still excessive, it indicates that there is a serious problem that cannot be solved by parameter adjustment, and an alarm should be triggered and the machine should be stopped for inspection.

[0026] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. A method for producing recycled granules based on waste plastic fiber bags, characterized in that, include: Acquire multi-source detection data of waste plastic fiber bags; the multi-source detection data includes transmission spectral data, surface texture images, and thickness measurement data; Feature analysis was performed based on multi-source detection data to obtain the composite defect characteristic values ​​of raw materials; among them, the composite defect characteristic values ​​of raw materials include fiber residue characteristic values, contaminant distribution characteristic values, and mechanical strength attenuation characteristic values; Using pre-calibrated fiber residue weights and contaminant influence weights, the fiber residue characteristic values ​​and contaminant distribution characteristic values ​​are weighted and summed to generate a fiber interference index; based on the mechanical strength attenuation characteristic values, the pre-built strength attenuation-compensation coefficient comparison table is consulted to obtain the melt strength compensation coefficient. The fiber interference index and melt strength compensation coefficient are input into a pre-built process response surface database, and the initial production parameters are obtained through multidimensional interpolation. The initial production parameters include the initial melt temperature, screw speed and die head pressure. Historical energy consumption records of similar granulation production lines are obtained, and an energy consumption regression equation is established with melt temperature, screw speed and output as variables. The initial production parameters are input into the energy consumption regression equation to calculate the predicted unit energy consumption. When the predicted unit energy consumption exceeds the preset energy consumption threshold, the screw speed and melt temperature are adjusted in the direction of reducing unit energy consumption to obtain the corrected production parameters. Granulation is initiated by correcting production parameters, and the melt pressure signal at the die head is collected in real time and the root mean square of pressure fluctuation is calculated. When the root mean square of pressure fluctuation is greater than the fluctuation threshold, the die head pressure is compensated in a closed loop through a proportional-integral controller to complete the production management of regenerated granulation.

2. The method for producing recycled granules based on waste plastic fiber bags according to claim 1, characterized in that, Feature analysis based on multi-source detection data yields characteristic values ​​of composite defects in raw materials, including: Baseline correction and characteristic absorption peak integration were performed on the transmission spectral data in the multi-source detection data to extract fiber residual characteristic values; The surface texture image in the multi-source detection data is divided into grids, the short run advantage of the gray-level run matrix of each grid is calculated, and the pollutant distribution characteristic value is obtained by weighted summation. The thickness variation coefficient is calculated from the thickness measurement data in the multi-source detection data and used as a characteristic value of mechanical strength attenuation. The characteristic values ​​of fiber residue, pollutant distribution, and mechanical strength attenuation are combined to form the characteristic values ​​of composite defects in raw materials.

3. The method for producing recycled granules based on waste plastic fiber bags according to claim 2, characterized in that, Using pre-calibrated fiber residue weights and contaminant influence weights, the fiber residue characteristic values ​​and contaminant distribution characteristic values ​​are weighted and summed to generate a fiber interference index, including: Based on a pre-built historical production database, plasticization collapse point data of waste plastic fiber bags during the melting and plasticization stage are extracted; among them, the plasticization collapse point data are the fiber residue characteristic value and contaminant distribution characteristic value corresponding to the critical point of sudden increase in melt pressure and sudden drop in tensile strength. A feature space is constructed with fiber residue characteristic value as the abscissa and pollutant distribution characteristic value as the ordinate. Plasticization collapse point data is mapped into the feature space, and a dynamic coupling boundary line is generated using a nonlinear fitting algorithm. The vertical distance from the coordinate point where the fiber residue characteristic value and pollutant distribution characteristic value of the current raw material are located to the dynamic coupling boundary line is calculated as the dynamic coupling factor. The dynamic coupling factor is input into the pre-constructed weight compensation function to perform nonlinear correction on the fiber residue weight and pollutant influence weight, so as to obtain the corrected fiber residue influence coefficient and pollutant distribution influence coefficient. The first weighted component is obtained by multiplying the corrected fiber residue influence coefficient by the current fiber residue characteristic value, and the second weighted component is obtained by multiplying the corrected pollutant distribution influence coefficient by the current pollutant distribution characteristic value. The first and second weighted components are vector-superimposed to generate the fiber interference index, which characterizes the synergistic destructive effect of fibers and pollutants.

4. The method for producing recycled granules based on waste plastic fiber bags according to claim 3, characterized in that, The fiber interference index and melt strength compensation coefficient are input into a pre-built process response surface database, and initial production parameters are obtained through multidimensional interpolation, including: Using fiber interference index and melt strength compensation coefficient as two-dimensional input variables, four neighboring grid points on the corresponding response surface grid are located in the process response surface database. The process response surface database pre-stores the optimal parameter combinations under different combinations of fiber interference index and melt strength compensation coefficient, and the optimal parameter combinations include initial melting temperature, screw speed and die head pressure. Using bilinear interpolation, the initial melting temperature, screw speed, and die head pressure are calculated based on the current fiber interference index, melt strength compensation coefficient, and distance weights of four neighboring grid points. These parameters are then combined as the initial production parameters. If the current fiber interference index or melt strength compensation coefficient exceeds the calibration boundary of the database, the parameter value corresponding to the nearest boundary grid point will be used as the initial production parameter, and a boundary extrapolation warning will be issued.

5. The method for producing recycled granules based on waste plastic fiber bags according to claim 4, characterized in that, Obtain historical energy consumption records for similar granulation production lines and establish an energy consumption regression equation with melt temperature, screw speed, and output as variables, including: Multiple sets of operating condition data were collected from the historical operation database of similar granulation production lines. Each set of operating condition data included at least the melting temperature, screw speed, output per unit time, and the corresponding measured unit energy consumption. A multiple regression analysis method was adopted, with melting temperature, screw speed and output as independent variables and unit energy consumption as dependent variable, to construct an initial regression model containing linear terms, quadratic terms and interaction terms between independent variables. The historical energy consumption records were fitted using the least squares method, the regression coefficients of each term were calculated, and insignificant terms were removed through stepwise regression or analysis of variance to obtain the final energy consumption regression equation.

6. The method for producing recycled granules based on waste plastic fiber bags according to claim 5, characterized in that, The initial production parameters are input into the energy consumption regression equation to calculate the predicted unit energy consumption. When the predicted unit energy consumption exceeds the preset energy consumption threshold, the screw speed and melting temperature are adjusted to reduce the unit energy consumption, resulting in corrected production parameters, including: The initial melting temperature and initial screw speed in the initial production parameters, along with the preset target output, are substituted into the energy consumption regression equation to calculate the predicted unit energy consumption. The predicted unit energy consumption is compared with the preset energy consumption threshold. If the predicted unit energy consumption is less than or equal to the preset energy consumption threshold, the initial production parameters are directly used as the corrected production parameters. If the predicted unit energy consumption is greater than the preset energy consumption threshold, the step-by-step joint adjustment process will be initiated.

7. The method for producing recycled granules based on waste plastic fiber bags according to claim 6, characterized in that, Enter the step-by-step joint adjustment process, including: Under the condition that the initial value of the melting temperature remains unchanged, the screw speed is gradually reduced in a preset step size. After each reduction, the screw speed is recalculated by substituting it into the energy consumption regression equation until the predicted unit energy consumption drops below the threshold or the screw speed reaches the lower limit. If the screw speed has been adjusted to the lower limit but the predicted unit energy consumption is still higher than the threshold, the screw speed is restored to the initial value. While keeping the screw speed constant, the melting temperature is gradually reduced in a preset step size. The predicted unit energy consumption is recalculated after each reduction until the predicted unit energy consumption drops below the threshold or the melting temperature reaches the lower limit. If the above steps fail to reduce the predicted unit energy consumption below the threshold, then the screw speed and melting temperature will be gradually reduced in small increments, and the predicted unit energy consumption will be recalculated after each synchronous adjustment until the threshold requirement is met or the preset maximum number of adjustments is reached. The final melt temperature and screw speed that meet the predicted unit energy consumption threshold will be combined with the die pressure in the initial production parameters to form the corrected production parameters; if the threshold requirement still cannot be met after all adjustments are tried, the set of parameters with the lowest energy consumption will be used as the corrected production parameters and a warning will be output.

8. The method for producing recycled granules based on waste plastic fiber bags according to claim 7, characterized in that, Granulation is initiated by adjusting production parameters, and the melt pressure signal at the die head is collected in real time and the root mean square of pressure fluctuation is calculated, including: The corrected production parameters are called up, and the corrected melt temperature, corrected screw speed and corrected die head pressure are written into the granulator control system. The extruder and die head heating are started, and the screw is driven to run at the corrected screw speed to start regeneration granulation. After the granulation process is running stably, the pressure sensor installed at the die head is used to continuously collect the melt pressure signal at the die head at a preset high-frequency sampling period to form a real-time pressure time series. Extract the corrected head pressure included in the corrected production parameters as the benchmark target pressure, calculate the deviation value between each sampling point in the real-time pressure time series and the benchmark target pressure, and generate a pressure deviation series; Based on a preset sliding time window, the pressure deviation sequence within the window is squared and the arithmetic mean is calculated. Then, the square root of the average value is taken to obtain the root mean square of the pressure fluctuation at the current moment.

9. A method for producing recycled granules based on waste plastic fiber bags according to claim 8, characterized in that, When the root mean square of the pressure fluctuation exceeds the fluctuation threshold, a proportional-integral controller performs closed-loop compensation for the die head pressure to complete the regenerated granulation production management, including: The root mean square of the pressure fluctuation is compared with a set fluctuation threshold. If the root mean square of the pressure fluctuation is greater than the fluctuation threshold, a trigger signal is generated. The trigger signal and the root mean square pressure fluctuation are input to the proportional-integral controller. At the same time, the corrected head pressure included in the corrected production parameters is read as the target setpoint, and the real-time head melt pressure at the current moment is read as the feedback value. The current pressure deviation is calculated based on the target setpoint and real-time feedback value. Combined with the historical pressure deviation sequence, the proportional gain coefficient and integral gain coefficient are used for weighted calculation to generate the proportional control component and integral control component. The proportional and integral control components are superimposed to obtain the total control amount; the total control amount is used to reverse the screw speed to generate a new target screw speed; the speed is reduced when the pressure is too high and increased when the pressure is too low. If the absolute value of the total adjustment exceeds the preset speed adjustment limit, the remaining adjustment will be used to assist in correcting the melting temperature and generate a new target melting temperature. The new target screw speed and new target melting temperature are sent to the granulator control system to replace the original modified production parameters, and the melt pressure signal at the die head is continuously monitored. Recalculate the root mean square of pressure fluctuation. If it drops below the fluctuation threshold, maintain the current parameters and clear the integral term of the proportional-integral controller. If the root mean square of pressure fluctuation continues to exceed the limit and the adjustment has reached its limit, trigger an alarm and shut down for inspection.

Images