Quality grading method for frame paper multi-source production characteristic parameters

Abstract

The present application belongs to the technical field of paper quality detection and control, and relates to a quality grading method for multi-source production characteristic parameters of frame paper, comprising: receiving and spatiotemporally aligning heterogeneous multi-dimensional dynamic characteristic parameter data, separating moisture characteristic change rate variables and quantitative characteristic deviation degree variables to calculate a nonlinear fluctuation dispersion parameter, generating real-time quality grading codes, generating coherent driving data using the product of the two variables during continuous early warning, calculating a coherence coefficient to trigger state truncation, and updating state machine decision time thresholds using reset action physical time difference values. The present application introduces a coherent state machine control path of time sequence waveform polarity evolution, provides in-situ feedforward decoupling capability during signal polarity mutual exclusion mutation, eliminates signal annihilation caused by polarity hedging, blocks false stable state determination, reduces cache overhead, and ensures the output stability of the grading code.

Description

Technical Field

[0001] This invention belongs to the field of papermaking quality inspection and control technology. This invention relates to a quality grading method for multi-source production characteristic parameters of frame paper. Background Technology

[0002] Currently, using online detection systems to continuously monitor the basis weight and moisture content of paper during the papermaking process and implement closed-loop adjustments based on the detection results has become a common quality control method in modern papermaking production. However, existing technologies still have certain limitations in terms of the stability of long-term equipment operation, the impact of mechanical switching under high-speed production conditions, and the dynamic response and coordinated control of multiple production variables.

[0003] First, an electromechanical switch is used to control the scanning carriage for synchronous monitoring. For example, Chinese invention patent CN103194928B discloses a scanning carriage and paper machine synchronization system for paper basis weight and moisture monitoring. This system controls the scanning carriage carriage with an O-frame structure to move along the bottom linear guide rail through a synchronization controller, while the upper and lower sensor probes move synchronously along the crossbeam guide rail to monitor the basis weight and moisture of the paper. This scheme relies heavily on multiple mechanical limit switches set in the bottom guide rail and crossbeam guide rail when controlling the movement, speed change and reset of the scanning carriage. Long-term frequent switching can easily cause wear and fatigue of mechanical contacts due to collision, which not only increases maintenance costs, but also makes it difficult to maintain synchronous measurement accuracy for a long time. In addition, this scheme mainly completes online monitoring and fails to form active feedback adjustment for the production process.

[0004] Secondly, mechanical rotation and repositioning are used to classify and rewind finished products. For example, Chinese invention patent CN105197655A discloses a device and method for classifying and rewinding paper rolls on a slow-speed paper machine. This technology uses a moisture metering device to detect paper density and moisture content, and transmits the detection signals through a wireless transmitting module and a wireless receiving module. After detecting a change in paper density or moisture content, a microcontroller drives a rotary motor to rotate the entire paper roll 180° to complete the repositioning. Then, the paper cutting mechanism cuts the paper, thereby achieving automatic grading and rewinding of the finished paper. This technology mainly performs passive classification of finished products at the end of the production process and is specifically designed for slow-speed paper machines. If it is applied to modern high-speed paper production lines, frequently driving a large-tonnage roll to rotate 180° at high speed will generate significant mechanical inertial impact and compression, which can easily cause paper web breakage or severe equipment vibration, making it difficult to meet the requirements for continuous and stable rewinding under high-speed conditions.

[0005] Third, conventional closed-loop control of paper basis weight is adopted using flow feedback. For example, Chinese invention patent application CN113800459A discloses a control system, control method, and production system for reconstituted tobacco in a thick pulp papermaking process. This system obtains the moisture content and basis weight of the product through an online finished product basis weight scanning rack. The central control unit calculates the oven-dry basis weight. When the oven-dry basis weight exceeds the set threshold, the system calculates the adjusted flow rate and outputs a control signal to the feeding unit to adjust the real-time flow rate. This system is mainly designed for high-concentration, high-viscosity thick pulp papermaking conditions with a concentration of 10wt% to 40wt%. In manual setting mode, it is necessary to extend the data acquisition cycle and remove the maximum, minimum, and related values ​​before calculating the average value. This averaging process reduces the system's ability to respond to instantaneous high-frequency changes, making it difficult to directly apply to traditional low-concentration, high-flow-rate papermaking systems for rapid response to upstream disturbances. In addition, this system focuses on single-variable adjustment of oven-dry basis weight and is difficult to dynamically decouple and coordinate the changes in the mutual influence between basis weight and moisture.

[0006] Therefore, how to improve the response capability of the control system to instantaneous changes and coordinate the coupling relationship between multiple production variables, so that the detection and feedback regulation are kept within the safe range actually allowed by the equipment throughout the entire process of high-speed production, has become the technical problem to be solved by this invention. Summary of the Invention

[0007] The purpose of this invention is to overcome the shortcomings of the prior art and solve the technical problems of signal offsetting and annihilation, false stability determination or computing power lag in the prior art in terms of handling the long-term operation stability of production equipment, high-speed mechanical switching impact and multi-variable time-series response control. The invention provides a quality classification method for multi-source production characteristic parameters of frame paper.

[0008] To achieve the above-mentioned objective, this invention provides a quality grading method for multi-source production characteristic parameters of frame paper, the method comprising the following steps:

[0009] Step S101, Receive heterogeneous multidimensional dynamic characteristic parameters input from the bus: Receive heterogeneous multidimensional dynamic characteristic parameter data representing the multi-source production status of frame paper input from the data bus; Step S102: Extract feature variables, calculate dispersion, and generate grading code: Decouple and spatiotemporally align heterogeneous multidimensional dynamic feature parameter data, separate the water feature change rate variable and the quantitative feature deviation variable, and calculate the nonlinear fluctuation dispersion parameter based on the water feature change rate variable and the quantitative feature deviation variable to generate a real-time quality grading code containing a third-level quality warning code. Step S103, multiplying moisture and quantitative characteristic variables to generate coherent data: when the real-time quality grading code is continuously at the third-level quality warning code, multiply the moisture characteristic change rate variable and the quantitative characteristic deviation variable to generate coherent driving data; Step S104, calculate the coherence coefficient to trigger state truncation and output code: extract the reverse mutation feature from the coherent driving data and calculate the coherence coefficient that characterizes the degree of correlation. When the coherence coefficient is in the range of 0.75 to 0.85, trigger the adaptive state truncation, suspend the calculation of the nonlinear fluctuation dispersion parameter, and output the second-level quality classification code. Step S105: Calculate the physical time difference for compensation to update the judgment time threshold: Calculate the physical time difference of the quality flag command reset action within 10 consecutive sampling periods, and add the physical time difference as an adaptive time compensation parameter to the system's preset benchmark judgment time threshold to update the state machine judgment time threshold, limiting the maximum compensation step size to within 2ms.

[0010] Step S102 of the present invention includes the following sub-steps: Step S1021, constructing a data topology structure in the feature storage area, mapping heterogeneous multidimensional dynamic feature parameter data to the data topology structure to align the sampling time window, and generating spatiotemporally aligned feature data; Step S1022, calculating the first-order difference and variance envelope in the data topology structure, and separating the water feature change rate variable and the quantitative feature deviation variable.

[0011] Step S102 of the present invention includes the following sub-steps: Step S1023, using a rolling calculation window to extract spatiotemporal alignment feature data for 50 sampling periods, calculating the rolling variance and spectral entropy of the spatiotemporal alignment feature data, and generating nonlinear fluctuation dispersion parameters.

[0012] Step S102 of the present invention includes the following sub-steps: Step S1024, comparing the nonlinear fluctuation dispersion parameter with the classification threshold interval, generating a first-level quality classification code when the nonlinear fluctuation dispersion parameter is in the normal interval, and generating a third-level quality warning code when the nonlinear fluctuation dispersion parameter is in the abnormal interval.

[0013] Step S103 of the present invention includes the following sub-steps: Step S1031, normalizing the water characteristic change rate variable and the quantitative characteristic deviation variable to obtain the normalized water characteristic change rate variable and the normalized quantitative characteristic deviation variable; Step S1032, inputting the normalized water characteristic change rate variable and the normalized quantitative characteristic deviation variable into the phase-locked filter model to retrieve the product filter and generate coherent driving data.

[0014] Step S104 of the present invention includes the following sub-steps: Step S1041, extracting phase angle jump value and amplitude envelope reversal feature from coherent driving data, and determining the abrupt change feature where the water characteristic change rate variable and the quantitative characteristic deviation variable simultaneously jump in opposite directions; Step S1042, calculating the coherence coefficient when the water characteristic change rate variable and the quantitative characteristic deviation variable simultaneously jump in opposite directions.

[0015] Step S104 of the present invention includes the following sub-steps: Step S1043, suspend the calculation thread of the nonlinear fluctuation dispersion parameter, clear the data cache queue and jump to the control flow branch, and output the second-level quality classification code.

[0016] Step S105 of the present invention includes the following sub-steps: Step S1051, acquiring physical timestamp data of quality flag instruction reset actions within 10 consecutive sampling periods using a high-resolution timer; Step S1052, calculating the average absolute deviation between adjacent physical timestamp data, and using the average absolute deviation as an adaptive time compensation parameter.

[0017] Step S105 of the present invention includes the following sub-steps: Step S1053, performing an algebraic addition operation on the adaptive time compensation parameter and the benchmark decision time threshold, so that the value of the state machine decision time threshold is updated to the sum of the adaptive time compensation parameter and the benchmark decision time threshold.

[0018] The present invention also includes the following steps: Step S106, converting the output second-level quality grading code or third-level quality warning code into control compensation data, and transmitting the control compensation data to the paper machine adjustment mechanism through the industrial bus interface, so that the multi-physics coupling dissipation is stabilized within the set range.

[0019] Compared with the prior art, the present invention has at least the following beneficial effects: 1. By distinguishing the direction of change of the moisture characteristic change rate variable and the quantitative characteristic deviation variable, and using the difference term to accumulate the absolute values ​​of the changes in opposite directions, it can avoid the mutual cancellation of positive and negative signals in conventional weighted calculation, retain the effective change characteristics corresponding to the actual production fluctuations, and reduce the situation of misjudging abnormal changes as a stable state. This processing does not require extending the filtering window, can reduce data buffering and register occupation, reduce the interference of feature overlap caused by mechanical bounce on the state judgment, and make the real-time quality grading code output when the state machine switches control branches more stable.

[0020] 2. When the third-level quality warning code is continuously output, the coherence judgment is made by using the product relationship between the moisture characteristic change rate variable and the quantitative characteristic deviation variable. This can promptly identify sudden changes in two sets of signals changing in opposite directions at the same time. When the coherence coefficient meets the set conditions, the calculation of the nonlinear fluctuation dispersion parameter is suspended by adaptive state truncation, and the second-level quality classification code is directly output. This can reduce unnecessary repetitive calculations, shorten the time required for state judgment and control flow switching, and reduce the possibility of data processing lag at high sampling frequencies.

[0021] 3. The state machine judgment time threshold is updated based on the actual time difference of the quality flag instruction reset action, and the compensation range of each update is limited. This can automatically compensate for the register processing delay generated during frequent state switching and prevent the delay from accumulating. As a result, the timing drift of state judgment over time can be reduced, and the control loop can remain stable under continuously changing production conditions, thereby improving the reliability of data processing equipment during long-term, high-frequency operation. Attached Figure Description

[0022] Figure 1 This is a schematic diagram illustrating the implementation process of the quality grading method for multi-source production characteristic parameters of frame paper according to the present invention; Figure 2 This is a block diagram showing the connection between the digital data processing frame and the paper machine adjustment mechanism of the present invention. Detailed Implementation

[0023] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments and accompanying drawings.

[0024] Example 1: This embodiment discloses a quality grading method for multi-source production characteristic parameters of frame paper, the method including the following steps: Step S101, Receive heterogeneous multidimensional dynamic characteristic parameters input from the bus: Receive heterogeneous multidimensional dynamic characteristic parameter data representing the multi-source production status of frame paper input from the data bus; Step S102: Extract feature variables, calculate dispersion, and generate grading code: Decouple and spatiotemporally align heterogeneous multidimensional dynamic feature parameter data, separate the water feature change rate variable and the quantitative feature deviation variable, and calculate the nonlinear fluctuation dispersion parameter based on the water feature change rate variable and the quantitative feature deviation variable to generate a real-time quality grading code containing a third-level quality warning code. Step S103, multiplying moisture and quantitative characteristic variables to generate coherent data: when the real-time quality grading code is continuously at the third-level quality warning code, multiply the moisture characteristic change rate variable and the quantitative characteristic deviation variable to generate coherent driving data; Step S104, calculate the coherence coefficient to trigger state truncation and output code: extract the reverse mutation feature from the coherent driving data and calculate the coherence coefficient that characterizes the degree of correlation. When the coherence coefficient is in the range of 0.75 to 0.85, trigger the adaptive state truncation, suspend the calculation of the nonlinear fluctuation dispersion parameter, and output the second-level quality classification code. Step S105: Calculate the physical time difference for compensation to update the judgment time threshold: Calculate the physical time difference of the quality flag command reset action within 10 consecutive sampling periods, and add the physical time difference as an adaptive time compensation parameter to the system's preset benchmark judgment time threshold to update the state machine judgment time threshold, limiting the maximum compensation step size to within 2ms.

[0025] Step S102 in this embodiment includes the following sub-steps: Step S1021, constructing a data topology structure in the feature storage area, mapping heterogeneous multidimensional dynamic feature parameter data to the data topology structure to align the sampling time window, and generating spatiotemporally aligned feature data; Step S1022, calculating the first-order difference and variance envelope in the data topology structure, and separating the water feature change rate variable and the quantitative feature deviation variable.

[0026] Step S102 in this embodiment includes the following sub-steps: Step S1023, using a rolling calculation window to extract spatiotemporal alignment feature data for 50 sampling periods, calculating the rolling variance and spectral entropy of the spatiotemporal alignment feature data, and generating nonlinear fluctuation dispersion parameters.

[0027] Step S102 in this embodiment includes the following sub-steps: Step S1024, compare the nonlinear fluctuation dispersion parameter with the classification threshold interval, generate a first-level quality classification code when the nonlinear fluctuation dispersion parameter is in the normal interval, and generate a third-level quality warning code when the nonlinear fluctuation dispersion parameter is in the abnormal interval.

[0028] Step S103 in this embodiment includes the following sub-steps: Step S1031, normalize the water characteristic change rate variable and the quantitative characteristic deviation variable to obtain the normalized water characteristic change rate variable and the normalized quantitative characteristic deviation variable; Step S1032, input the normalized water characteristic change rate variable and the normalized quantitative characteristic deviation variable into the phase-locked filter model to retrieve the product filter and generate coherent driving data.

[0029] Step S104 in this embodiment includes the following sub-steps: Step S1041, extracting phase angle jump value and amplitude envelope reversal feature from coherent driving data, and determining the abrupt change feature where the water characteristic change rate variable and the quantitative characteristic deviation variable simultaneously jump in opposite directions; Step S1042, calculating the coherence coefficient when the water characteristic change rate variable and the quantitative characteristic deviation variable simultaneously jump in opposite directions.

[0030] Step S104 in this embodiment includes the following sub-steps: Step S1043, suspend the calculation thread of the nonlinear fluctuation dispersion parameter, clear the data cache queue and jump to the control flow branch, and output the second-level quality classification code.

[0031] Step S105 in this embodiment includes the following sub-steps: Step S1051, acquiring physical timestamp data of quality flag command reset actions within 10 consecutive sampling periods using a high-resolution timer; Step S1052, calculating the average absolute deviation between adjacent physical timestamp data, and using the average absolute deviation as an adaptive time compensation parameter.

[0032] Step S105 in this embodiment includes the following sub-steps: Step S1053, performing an algebraic addition operation on the adaptive time compensation parameter and the baseline decision time threshold, so that the value of the state machine decision time threshold is updated to the sum of the adaptive time compensation parameter and the baseline decision time threshold.

[0033] The embodiment also includes the following steps: Step S106, converting the output second-level quality grading code or third-level quality warning code into control compensation data, and transmitting the control compensation data to the paper machine adjustment mechanism through the industrial bus interface, so that the multi-physics coupling dissipation is stabilized within the set range.

[0034] Example 2: In this embodiment, the first and second channel electrical signals are continuously acquired under the forming conditions of frame paper using an industrial-grade multi-bus sensor detection link. These two signals are then transmitted to a high-speed data buffer in a digital data processing rack. The first channel electrical signal corresponds to the moisture data sequence, and the second channel electrical signal corresponds to the quantitative data sequence. When high-frequency transient vibrations occur in the forming wire section or pressure fluctuations occur in the pulp supply system, the two sets of data may experience opposite polarity abrupt changes within a local sampling period. This causes the positive transition and negative deviation in the conventional weighted accumulator to algebraically cancel each other out, leading to a misjudgment of process fluctuations as a stable state. The paper machine's regulating mechanism receives an incorrect control command. The central processing unit (CPU) calls the high-speed data buffer to convert the received moisture and quantitative data sequences into discrete digital data streams. In the feature storage area, a data topology is constructed using sampling channels, sampling positions, and sampling times as related terms. Each sampled data point is mapped to the data topology according to its sampling position and timestamp. Spatiotemporal alignment is achieved through a unified sampling time window, generating spatiotemporally aligned feature data. The CPU calculates the first-order difference and variance envelope of the two data streams within the data topology, and separates the moisture characteristic change rate variable based on the first-order difference of the moisture data sequence. Based on the deviation of the quantitative data sequence from the variance envelope, the quantitative characteristic deviation variable is separated. ,in, This represents the rate of change of moisture characteristics. The variable representing the deviation of quantitative characteristics is continuously monitored by the central processing unit. and The sign polarity in discrete time series; when the fluctuation polarities of the two are consistent, they are sent to the regular weighted path; when the two are positive and negative and transition simultaneously, the current sampling position is marked as a candidate position for reverse mutation for subsequent coherence determination.

[0035] The central processing unit (CPU) uses a rolling computation window to extract spatiotemporally aligned feature data over 50 sampling periods, calculates the rolling variance and spectral entropy of the data within that window to determine the fluctuation state of the current window, and then uses the separated data within that window as a basis for further analysis. and Generate nonlinear fluctuation dispersion parameters In this embodiment, Calculate according to the following formula: ,in, For nonlinear fluctuation dispersion parameters, The moisture weighting coefficient is set to 0.65. The quantitative weighting coefficient is set to 0.35. Let be the discrete correction factor and take the value 0.001. and When a sudden change in polarity occurs due to high-frequency transient vibration in the forming mesh section, the positive and negative components in the numerator of the formula cancel each other out algebraically, and the difference term in the denominator accumulates in absolute value due to the subtraction of the negative value. The calculation converges towards below 0.5. This result is used to distinguish between homogeneous disturbances caused by mechanical bouncing and changes in paper processing conditions, avoiding the generation of a first-level quality grading code due to mutual cancellation after the two signals are directly added; the central processing unit will... Compared with the hierarchical threshold interval, when When the value is within the normal range of 0 to 0.2, a first-level quality grading code is generated; when... When the value is greater than 0.5 and is in the abnormal range, a third-level quality warning code is generated; 0.2 to 0.5 is the transition range corresponding to the second-level quality grading code, which is written into the second-level quality grading code by the subsequent adaptive state truncation branch; the third-level quality warning code is transmitted to the paper machine adjustment mechanism via the industrial bus interface to initiate subsequent coherent determination.

[0036] When the real-time quality grading code remains at the Level 3 quality warning code, the central processing unit... and Normalization is performed, and the normalized rate of change of moisture characteristics and the normalized deviation of quantitative characteristics are input into the phase-locked filter model. The phase-locked filter model generates coherent driving data through orthogonal product operations and low-pass filtering. Their product relationship is Central processing unit Phase and amplitude analysis is performed to extract phase angle jump values ​​and amplitude envelope inversion characteristics; when the two variables transition simultaneously in opposite directions, calculations are performed within a sliding window of 50 data points. The mean, and divide that mean by and The product of the mean squared errors yields the coherence coefficient, which characterizes the degree of correlation between the two variables. ;when When the value is in the range of 0.75 to 0.85, the central processing unit triggers adaptive state truncation and suspends the operation. The computation thread clears the data buffer queue and jumps the state machine control flow to the output branch of the second-level quality grading code. When the second-level quality grading code is written to the quality flag register, the state machine is reset from the third-level quality warning state to the second-level quality grading state and executes the quality flag instruction reset action. Through this process, the mechanical disturbance formed by the reverse transition of the two signals is introduced into the transition adjustment state, avoiding the conventional linear summation from being judged as the first-level quality state due to signal cancellation.

[0037] During frequent state machine transitions, the central processing unit (CPU) acquires the physical timestamp data of quality flag instruction reset actions within 10 consecutive sampling periods using a high-resolution timer, calculates the average absolute deviation between adjacent physical timestamp data, and uses this average absolute deviation as an adaptive time compensation parameter. Central processing unit The system performs algebraic addition with the preset benchmark judgment time threshold to update the state machine judgment time threshold to the sum of the two. The maximum compensation step size for each update is limited to within 2ms to compensate for the delay generated by the register during continuous state switching. When converting the quality grading code into control compensation data, the system uses a preset mapping matrix to convert the second-level quality grading code into a 4mA current compensation benchmark and the third-level quality warning code into a 12mA current compensation benchmark. The control compensation data is transmitted via the industrial bus interface to the steam valve, wire section frequency converter, and take-up actuator of the paper machine regulating mechanism, respectively adjusting the steam inlet, headbox instantaneous pressure, and tension motor drive frequency. Within a 30ms control cycle, the paper machine regulating mechanism compensates for fluctuations caused by kinetic and thermal energy transfer, stabilizing the multi-physics coupling dissipation within a set range, which is within 0.05J / s in terms of dissipation rate. The industrial bus interface continuously outputs discrete frequency adjustment codes, enabling the data from the multi-channel sensor link to be transmitted to the tension motor via the digital data processing rack, maintaining the control state under continuous production conditions.

[0038] Example 3: This embodiment uses a digital-physical hybrid simulation platform to test the processing performance of the digital data processing rack. The platform has a signal sampling rate of 1000Hz and a digital resolution of 0.01%. During the test, the first and second channel electrical signals are acquired at the data bus input terminal, and Gaussian white noise with a signal-to-noise ratio of 20dB and power frequency harmonic interference of 50Hz are superimposed on the two signals. This generates a test input data stream containing random fluctuations, which is used to reproduce the process changes when high-frequency transient jitter occurs in the forming mesh section. In the test loop, the reference sampling period is used as the basis for the number of terms. Control adaptive time compensation parameters The calculation range The timing drift rate of the electrical signal is determined based on the processor register reclamation frequency and is used to balance the tracking sensitivity of the state machine's decision time threshold and the CPU's computational load. As the timing drift rate increases, the sensitivity decreases. To reduce stack latency within the registers; when the timing offset reaches 0.15ms per sampling period, Set to 10, and calculate based on the physical timestamp data obtained from continuous sampling. This is used to update the state machine to determine the time threshold.

[0039] The test procedure included a control group, a partially missing control group, an out-of-range control group, and the present invention sample group. The mechanical bounce amplitude of the mesh was used to characterize the problem intensity. The test input data used three vibration displacement gradients: 2.1%, 5.4%, and 12.8%. Each group input the raw voltage signal collected by the sensor into the feature storage area and operated under the same power frequency interference environment. The control group used a conventional linear weighted accumulation method; the partially missing control group was omitted. and The fluctuation polarity mutually exclusive criterion is determined by directly summing the two variables; the out-of-range control group will... The value is set to 0.45; the sample group of this invention will The value was set to 0.65, and the complete quality grading method was run. This method performs spatiotemporal alignment on heterogeneous multidimensional dynamic feature parameter data and separates them based on first-order difference and variance envelope. and Then, the rolling variance and spectral entropy are calculated on the spatiotemporal alignment feature data to generate... When the third-level quality warning code is continuously output, coherent driving data is generated using the product of the two variables, and adaptive state truncation is performed according to the coherence coefficient. Under a low vibration gradient of 2.1%, the two opposite-polarity signals of the control group undergo algebraic cancellation in the central processing unit register, with a misjudgment rate of 14.3%. After processing the 50Hz power frequency harmonic interference, the interference suppression ratio of the sample group of this invention reaches 32.5dB, and the conformity between the second-level quality classification code and the actual quality state reaches 98.4%. This result shows that under low vibration conditions, branching the two signals according to the wave polarity and coherence state can preserve the change characteristics that are canceled when directly linearly accumulated.

[0040] When the vibration displacement gradient increased from 2.1% to 5.4% and 12.8%, the cancellation ratio of useful signals in the control group rose to 32.6% and 58.2%, respectively, and a third-level quality warning code was missed. The sample group of this invention... The difference term in the denominator handles positive and negative signals, resulting in the algebraic absolute value accumulation when the two variables are in opposite directions; under vibration displacement gradients of 5.4% and 12.8%, the conformity between the quality grading code and the actual quality state was 97.2% and 95.6%, respectively. The partially missing control group did not perform fluctuation polarity mutual exclusion determination, and the conformity dropped to 62.4% under high vibration conditions; in the out-of-range control group, when When it is 0.45, It enters the numerical saturation region and remains at 0.34, unable to follow... Transient jump; will When set to 0.85, control flow overshoot occurs, and the quality classification error rises to 38.7%. This is in conjunction with the sample of this invention. The results when the value is 0.65 When the value is within the range of 0.5 to 0.8, the quality grading error can be reduced when two sets of heterogeneous characteristic parameters overlap and conflict. During the test, the industrial bus interface continuously transmits the discrete frequency adjustment code formed by the conversion of the quality grading code to the paper machine adjustment mechanism. The frequency converter adjusts the drive frequency of the tension motor, and the digital data processing frame completes the transmission of multi-channel sensor link data to the paper machine adjustment mechanism accordingly. Under the condition of mechanical vibration and power frequency harmonic interference, the quality grading code is continuously output.

[0041] Example 4: This embodiment combines Figures 1 to 2 The quality grading method for multi-source production characteristic parameters of framed paper is explained, such as... Figure 1 As shown, the implementation process sequentially covers the following steps: Step S101 receives heterogeneous multidimensional dynamic feature parameters from the bus input, proceeds to Step S102 extracts feature variables, calculates dispersion, and generates hierarchical codes, then executes Step S103 multiplies water content and quantitative feature variables to generate coherent data, then proceeds to Step S104 to calculate the coherence coefficient, triggers state truncation, and outputs codes, and finally implements Step S105 to calculate the physical time difference for compensation to update the judgment time threshold.

[0042] like Figure 2 As shown, the multi-channel bus sensing detection link acquires the first and second channel electrical signals under the paper forming process and inputs the two signals to the high-speed data buffer of the digital data processing rack via the data bus. The high-speed data buffer transmits the received data sequence to the feature storage area to construct the data topology structure, and simultaneously loads the discrete digital data stream into the central processing unit. The phase-locked filter model inside the central processing unit calls the orthogonal phase-sensitive detector to perform orthogonal product modulation on the normalized moisture characteristic change rate variable and the normalized quantitative characteristic deviation variable to generate coherent driving data. The coherent driving data is filtered by a low-pass digital filter to remove high-frequency incoherent noise components and then input to the coherent wave spectrum separator. The coherence spectrum separator calculates the coherence coefficient by combining the coherence threshold stored in the read-only memory to trigger adaptive state truncation; the central processing unit calls the high-resolution timer to obtain the physical timestamp data of the quality flag instruction reset action within the continuous sampling period to calculate the adaptive time compensation parameters and update the state machine decision time threshold; the central processing unit outputs the generated quality grading code and control compensation data to the paper machine adjustment mechanism via the industrial bus interface; the paper machine adjustment mechanism distributes and transmits the control compensation data from the industrial bus interface to the steam valve, the wire section frequency converter, and the winding actuator respectively; the winding actuator sends the drive control signal to the frequency converter to adjust the drive frequency of the tension motor.

[0043] Example 5: In this embodiment, when the real-time quality grading code remains at the third-level quality warning code, the continuous mechanical resonance of the forming mesh and the long-cycle pressure pulsation of the slurry supply system may create a combined disturbance, affecting the high-speed data buffer. and High-frequency unidirectional or reverse fluctuations are generated. Conventional fixed-time-window integrators cannot distinguish between paper process changes and periodic mechanical noise, easily causing register decision delays and data overflows. This prevents the winding control mechanism in the paper machine's adjustment mechanism from receiving control commands in a timely manner. This embodiment utilizes the existing processor computing resources of the digital data processing rack to perform coherent determination on two sets of heterogeneous features in continuous warning states while keeping the external hardware filter and fitting model unchanged. The central processing unit calls the high-speed data cache... and First, the two variables are normalized. Then, the normalized rate of change of moisture characteristics and the normalized deviation of quantitative characteristics are input into the orthogonal phase-sensitive detector of the phase-locked filter model. The orthogonal phase-sensitive detector uses an internal reference oscillation signal to track the frequency and phase of the two input variables, and performs product modulation through an orthogonal product operator to generate coherent driving data. : The modulated composite signal enters a low-pass digital filter with a cutoff frequency of 50Hz to filter out high-frequency incoherent noise components, and outputs the signal for coherence determination. .

[0044] Central processing unit will Input coherent wave spectrum separator, calculate within a discrete sliding sampling window containing 50 data points The mean, and divide that mean by and The product of the mean squared errors yields the dimensionless coherence coefficient. The coherent wave spectrum separator simultaneously processes one-dimensional real-valued forms. Perform a Hilbert transform to construct an analytic signal containing real and imaginary parts. Calculate the instantaneous phase angle series based on the arctangent function of the ratio of the imaginary to real parts of the analytic signal. Calculate the instantaneous amplitude envelope series based on the magnitude of the analytic signal. When the absolute value of the instantaneous phase angle jump within two consecutive sampling periods exceeds 1.2 radians, and the first-order difference sign of the instantaneous amplitude envelope changes from positive to negative, the signal is determined to be... and Simultaneously, it transitions in the opposite direction and enters the coherence coefficient determination; the coherence threshold is stored in read-only memory, with its lower limit set at 0.75. When the value is greater than 0.75 but not exceeding 0.85, the central processing unit (CPU) determines that the fluctuation corresponding to the third consecutive level-three quality warning code is related to the parasitic noise from the same source generated by the bouncing of the mechanical mesh. It then triggers an adaptive state cutoff. After triggering, the CPU suspends... The computation thread clears the data cache queue and jumps the control flow to the output branch of the second-level quality grading code, outputting the second-level quality grading code; the state machine resets from the third-level quality warning state to the second-level quality grading state and executes the quality flag instruction reset action. When the value exceeds 0.85, it indicates that the moisture data and quantitative data are in an extremely strong coupled state. The central processing unit does not perform adaptive state truncation, but instead jumps out of the current control flow branch and issues an emergency shutdown protection command to the main control rack.

[0045] When executing the quality flag command reset action, the high-resolution timer acquires physical timestamp data within 10 consecutive sampling periods, calculates the average absolute deviation between adjacent physical timestamp data, and uses it as an adaptive time compensation parameter. The central processing unit will The time threshold is accumulated to the system's preset baseline judgment time threshold, so that the state machine judgment time threshold is updated to the sum of the two. The maximum compensation step size for each update is limited to within 2ms to compensate for the timing drift caused by long-cycle data flow. The quality grading code after adaptive state truncation correction is transmitted to the winding controlled mechanism in the paper machine adjustment mechanism through the industrial bus interface. The frequency converter adjusts the drive frequency of the tension motor. After the above processing, the response delay of the tension motor is shortened from 15.2ms under conventional control to 4.6ms. Without extending the physical filtering window of the high-speed data buffer, the digital data processing rack completes the continuous transmission of multi-channel sensor link data to the paper machine adjustment mechanism and reduces the control overshoot and phase lag caused by high-frequency transient jitter in the wire section.

[0046] Example 6: This embodiment is used for the initial deployment when converting between different basis weight specifications of frame paper production lines. Before connecting to the field data bus, the digital data processing rack receives a calibration dataset for initializing the stable state. This calibration dataset consists of historical sampled data streams containing known process fluctuations. The central processing unit calls the high-speed data buffer to convert the initial digital signal stream in the calibration dataset into a discrete digital data stream and establishes an initial feature distribution matrix under controlled conditions in the feature storage area. Subsequently, the signal deviation amplitude in each sampling period is normalized to form the reference zero point used for subsequent fluctuation polarity determination.

[0047] The central processing unit determines the baseline working window parameters suitable for the current basis weight specification of the frame paper based on the variance variation in the initial feature distribution matrix, and calculates the moisture weighting coefficient according to the following formula. : ,in, The calibrated variance is the rate of change of the discrete characteristics of moisture content. To quantify the calibrated variance of the deviation of discrete features, when 0.13 When the value is 0.07, the calculated result is The value is 0.65. The central processing unit writes the calibration result as a global parameter into the read-only memory to determine the decoupling boundary of the multi-source production characteristic parameters. After the parameter writing is completed without extending the sampling filter window, the digital data processing rack enters a ready state without time drift.

Claims

1. A method for quality grading of multi-source production characteristic parameters of frame paper, characterized in that, The method includes the following steps: Step S101: Receive heterogeneous multidimensional dynamic characteristic parameter data representing the multi-source production status of frame paper, which is input from the data bus. Step S102: Decouple and spatiotemporally align heterogeneous multidimensional dynamic feature parameter data, separate the water feature change rate variable and the quantitative feature deviation variable, and calculate the nonlinear fluctuation dispersion parameter based on the water feature change rate variable and the quantitative feature deviation variable to generate a real-time quality classification code containing a third-level quality early warning code. Step S103: When the real-time quality grading code remains at the third-level quality warning code, multiply the moisture characteristic change rate variable and the quantitative characteristic deviation variable to generate coherent driving data. Step S104: Extract the reverse mutation feature from the coherent driving data and calculate the coherence coefficient representing the degree of correlation. When the coherence coefficient is in the range of 0.75 to 0.85, trigger the adaptive state truncation, suspend the calculation of the nonlinear fluctuation dispersion parameter, and output the second-level quality classification code. Step S105: Calculate the physical time difference of the quality flag command reset action within 10 consecutive sampling periods, and add the physical time difference as an adaptive time compensation parameter to the preset benchmark judgment time threshold of the system to update the state machine judgment time threshold, limiting the maximum compensation step size to within 2ms.

2. The quality grading method for multi-source production characteristic parameters of frame paper according to claim 1, characterized in that, Step S102 includes the following sub-steps: Step S1021, construct a data topology structure in the feature storage area, map heterogeneous multidimensional dynamic feature parameter data to the data topology structure to align the sampling time window, and generate spatiotemporally aligned feature data; Step S1022: Calculate the first difference and variance envelope in the data topology to separate the water characteristic change rate variable and the quantitative characteristic deviation variable.

3. The quality grading method for multi-source production characteristic parameters of frame paper according to claim 2, characterized in that, Step S102 includes the following sub-steps: Step S1023, using a rolling calculation window to extract spatiotemporal alignment feature data for 50 sampling periods, calculating the rolling variance and spectral entropy of the spatiotemporal alignment feature data, and generating nonlinear fluctuation dispersion parameters.

4. The quality grading method for multi-source production characteristic parameters of frame paper according to claim 1, characterized in that, Step S102 includes the following sub-steps: Step S1024, compare the nonlinear fluctuation dispersion parameter with the classification threshold interval, generate the first-level quality classification code when the nonlinear fluctuation dispersion parameter is in the normal interval, and generate the third-level quality warning code when the nonlinear fluctuation dispersion parameter is in the abnormal interval.

5. The quality grading method for multi-source production characteristic parameters of frame paper according to claim 1, characterized in that, Step S103 includes the following sub-steps: Step S1031, normalize the water characteristic change rate variable and the quantitative characteristic deviation variable to obtain the normalized water characteristic change rate variable and the normalized quantitative characteristic deviation variable; Step S1032, input the normalized water characteristic change rate variable and the normalized quantitative characteristic deviation variable into the phase-locked filter model to retrieve the product filter and generate coherent driving data.

6. The quality grading method for multi-source production characteristic parameters of frame paper according to claim 1, characterized in that, Step S104 includes the following sub-steps: Step S1041, extract the phase angle jump value and amplitude envelope reversal feature from the coherent driving data, and determine the abrupt change feature where the water characteristic change rate variable and the quantitative characteristic deviation variable simultaneously jump in opposite directions; Step S1042, calculate the coherence coefficient when the water characteristic change rate variable and the quantitative characteristic deviation variable simultaneously jump in opposite directions.

7. The quality grading method for multi-source production characteristic parameters of frame paper according to claim 1, characterized in that, Step S104 includes the following sub-steps: Step S1043, suspend the calculation thread of the nonlinear fluctuation dispersion parameter, clear the data cache queue and jump to the control flow branch, and output the second-level quality classification code.

8. The quality grading method for multi-source production characteristic parameters of frame paper according to claim 1, characterized in that, Step S105 includes the following sub-steps: Step S1051, acquiring physical timestamp data of quality flag command reset actions within 10 consecutive sampling periods using a high-resolution timer; Step S1052, calculating the average absolute deviation between adjacent physical timestamp data, and using the average absolute deviation as an adaptive time compensation parameter.

9. The quality grading method for multi-source production characteristic parameters of frame paper according to claim 8, characterized in that, Step S105 includes the following sub-steps: Step S1053, perform algebraic addition to sum the adaptive time compensation parameter and the baseline decision time threshold, so that the value of the state machine decision time threshold is updated to the sum of the adaptive time compensation parameter and the baseline decision time threshold.

10. The quality grading method for multi-source production characteristic parameters of frame paper according to claim 1, characterized in that, It also includes the following steps: Step S106: Convert the output second-level quality grading code or third-level quality warning code into control compensation data, and transmit the control compensation data to the paper machine adjustment mechanism through the industrial bus interface to stabilize the multi-physics coupling dissipation within the set range.

Images