Color master batch filtering pressure constant adjusting method based on proportional integral derivative control

Abstract

The invention belongs to the technical field of industrial process control, and particularly relates to a color master batch filtering pressure constant adjusting method based on proportional integral derivative control, and the method comprises the steps: obtaining the current pressure, the current flow and the target pressure, and calculating the pressure deviation and the deviation change rate; based on the current pressure and the current flow, stage parameters representing the filtering resistance state are obtained; extracting low-frequency drift characteristics of the pressure deviation on a time sequence and calculating a deviation degree according to the low-frequency drift characteristics; performing fuzzy reasoning on the pressure deviation and the deviation change rate to obtain an original increment; adjusting the original increment according to the stage parameters and the deviation degree to obtain a target increment; and superposing the target increment to the historical instruction to obtain a flow instruction, and outputting the flow instruction to the pumping device for execution. According to the invention, stage parameters and low-frequency drift characteristics cooperate to drive the regulatory factors, and constant control of the filter pressure of the color master batch in the whole period is realized.

Description

Technical Field

[0001] This invention relates to the field of industrial process control technology. More specifically, this invention relates to a method for constant pressure regulation of masterbatch filtration based on proportional-integral-derivative control. Background Technology

[0002] Color masterbatch is a commonly used colorant carrier in the plastics processing industry. Its production process requires filtration of the slurry to achieve solid-liquid separation. In industrial filtration stages such as plate and frame filters, the slurry is pumped into the filter, where solid particles are trapped by the filter screen, and the liquid phase is discharged through the screen. The filtration pressure at the inlet of the filter is a core process parameter for measuring the stability of the filtration process. Too low a filtration pressure will lead to decreased filtration efficiency, while too high a pressure may damage the filter screen or cause fluctuations in product quality. Therefore, maintaining the filtration pressure at the inlet of the filter consistently near the target process value throughout the entire filtration cycle is a fundamental requirement for ensuring both the quality of color masterbatch products and production efficiency.

[0003] Currently, the industry commonly uses a fixed-parameter proportional-integral-derivative (PID) controller to regulate the filtration pressure of masterbatch. Traditional PID controllers calculate the adjustment amount of the pump flow rate based on the current pressure deviation and its integral and derivative terms, and then send the adjustment command to the frequency converter to drive the pump motor. In practical applications, operators typically set a fixed proportional coefficient, integral time, and derivative time before filtration begins based on experience. The controller operates continuously with this set of fixed parameters throughout the filtration cycle, adjusting the pump flow rate in real time to bring the filtration pressure closer to the target value. This solution is simple in structure and low in implementation cost, and can basically meet the pressure control requirements under ideal operating conditions where the filtration resistance changes relatively smoothly.

[0004] However, in the actual process of masterbatch filtration, the filter cake continuously accumulates on the filter screen surface as filtration continues, leading to a non-linear increase in filtration resistance. This, in turn, causes significant changes in the dynamic characteristics of the filtration system throughout the entire filtration cycle. Existing fixed-parameter control schemes have the following obvious drawbacks when dealing with such complex conditions. First, traditional controllers cannot sense the actual physical stage of the current filtration process. Using the same control parameters at the beginning and end of filtration makes it difficult to adapt to the drastically different system gains at different stages. Second, facing the rapid increase in resistance in the later stages of filtration, the system lacks detection of systematic pressure drift, resulting in a slow adjustment response and an inability to provide sufficient adjustment force to compensate for pressure drift in a timely manner. Finally, the resolution of fixed-parameter controllers is limited, making it impossible to simultaneously accommodate the large pressure fluctuations in the early stages of filtration and the fine-tuning for minor pressure deviations in the later stages. These inherent defects cause the filtration pressure at the inlet of the filter device to easily deviate from the target value continuously in the middle and later stages of filtration, seriously affecting the filtration quality and production stability of the final masterbatch product. Summary of the Invention

[0005] To address the technical problem of fixed-parameter controllers being unable to adapt to dynamic characteristic changes, leading to filtration pressure drift in the later stages of filtration, this invention proposes a method for constant filtration pressure regulation of masterbatch based on proportional-integral-derivative control, which can achieve constant control of filtration pressure throughout the entire filtration cycle.

[0006] This invention provides a method for constant pressure regulation of masterbatch filtration based on proportional-integral-derivative control, comprising: acquiring the current pressure, current flow rate, and target pressure at the inlet of the filtration device; calculating the difference between the target pressure and the current pressure to obtain a pressure deviation, and calculating the variation value of the pressure deviation to obtain a deviation variation rate; acquiring stage parameters characterizing the current degree of filter cake accumulation based on the current pressure and the current flow rate; extracting the low-frequency drift characteristics of the pressure deviation in the time series, and calculating the deviation degree to characterize the degree of pressure drift in the later stage of filtration; performing fuzzy inference based on the pressure deviation and the deviation variation rate to obtain the original increment; calculating an adjustment factor based on the stage parameters and the deviation degree, and adjusting the original increment using the adjustment factor to obtain the target increment; superimposing the target increment onto historical commands to obtain a flow command and outputting it to the pumping device for execution.

[0007] This invention effectively transforms the current stage of the filtration process into control parameters by acquiring stage parameters characterizing the current degree of filter cake accumulation. Simultaneously, it extracts low-frequency drift features of pressure deviation over time to identify systematic pressure drift during filtration. By synergistically driving the adjustment factor using stage parameters and low-frequency drift features, it achieves adaptive adjustment of the original increment of the fuzzy inference output, enabling the controller to better match the adjustment intensity required by the current operating condition throughout the entire filtration cycle. This adjustment mechanism alleviates pressure deviation caused by the nonlinear increase in filter cake resistance, allowing the masterbatch filtration pressure to approach the target setpoint more effectively and ensuring the stability of the production process.

[0008] Preferably, calculating the resistance coefficient based on the current pressure and the current flow rate includes: obtaining a preset lower flow rate limit; determining whether the current flow rate is lower than the lower flow rate limit; if the current flow rate is lower than the lower flow rate limit, then the lower flow rate limit is taken as the effective flow rate; if the current flow rate is not lower than the lower flow rate limit, then the current flow rate is taken as the effective flow rate; calculating the ratio of the current pressure to the effective flow rate to obtain the resistance coefficient.

[0009] This invention introduces a preset lower flow limit to protect the current flow rate. The current flow rate below the lower flow limit is replaced with the lower flow limit in the calculation of the resistance coefficient. This prevents the resistance coefficient from approaching infinity when the pumping flow rate is extremely low. It ensures that the resistance coefficient can still stably and accurately reflect the current filter cake accumulation state during the start-up and shutdown transition of the pumping device, and avoids incorrect stage judgments caused by calculation abnormalities.

[0010] Preferably, obtaining the stage parameter characterizing the current filter cake accumulation level includes: normalizing the resistance coefficient to obtain a stage factor as the stage parameter; the normalization includes: the stage factor is the ratio of the difference between the resistance coefficient and the initial resistance to the difference between the maximum resistance and the initial resistance, and the calculation result is limited to between 0 and 1; wherein, the initial resistance is the arithmetic mean of the resistance coefficients in the initial filtration state, and the maximum resistance is the arithmetic mean of the resistance coefficients in the historical filtration saturation state.

[0011] This invention normalizes the resistance coefficient with reference to the initial resistance and the maximum resistance to obtain a stage factor with a fixed value range of 0 to 1. This eliminates the difference in absolute resistance caused by different batches of slurry concentration and different specifications of filter screens, so that the calculation of subsequent adjustment factors is not affected by specific working conditions, and enhances the universality of the adjustment method among different production batches.

[0012] Preferably, the step of extracting the low-frequency drift feature of the pressure deviation over a time series and calculating the deviation based thereon includes: obtaining the trend mean of the pressure deviation over a preset time period as the low-frequency drift feature; obtaining a preset trend reference value and calculating the deviation based on the trend mean and the trend reference value; obtaining the trend mean of the pressure deviation over the preset time period includes: obtaining the incrementing number of the current sampling time and comparing the incrementing number with a preset constant, selecting the smaller value as the effective length; obtaining multiple pressure deviations within a time window equal to the effective length from the current sampling time, and calculating the arithmetic mean of the multiple pressure deviations to obtain the trend mean.

[0013] Preferably, before performing fuzzy inference, the method further includes: dynamically adjusting the mapping coefficients of the fuzzy controller based on the fluctuation amplitude of the pressure deviation to achieve equivalent universe of discourse scaling, including: obtaining the maximum and minimum deviations of multiple pressure deviations within the time window, and subtracting the minimum deviation from the maximum deviation as the fluctuation amplitude; calculating the ratio of the fluctuation amplitude to the total width of the standard universe of discourse to obtain the utilization ratio, wherein the total width of the standard universe of discourse is twice half the width of the standard universe of discourse; obtaining a preset utilization reference value, calculating the ratio of the utilization ratio to the utilization reference value to obtain the scaling factor; and after limiting the scaling factor, dividing it by the initially set mapping coefficient to obtain the target mapping coefficient at the current sampling time.

[0014] This invention uses the difference between the maximum and minimum pressure deviations within a statistical time window as the fluctuation amplitude. By using the amplitude limiting calculation of ratio and scaling factor, a new domain of discourse half-amplitude is obtained, which makes the domain of discourse of the membership function of the fuzzy controller dynamically shrink or expand with the actual distribution range of the pressure deviation. When the pressure deviation fluctuation narrows in the later stage of filtering, the domain of discourse is automatically refined, which improves the ability to distinguish small pressure deviations and avoids insufficient control accuracy in the later stage caused by a fixed domain of discourse.

[0015] Preferably, fuzzy reasoning is performed on the pressure deviation and the deviation variation rate to obtain the original increment, including: obtaining a preset standard membership function with a fixed shape and vertex position, wherein the preset standard membership function contains multiple fuzzy subsets distributed in a triangular pattern within a standard domain; standardizing and mapping the pressure deviation and the deviation variation rate using mapping coefficients to obtain fuzzy variables; inputting the fuzzy variables into a fuzzy rule table for minimization and maximization operations to obtain a comprehensive fuzzy set; defuzzifying the comprehensive fuzzy set using the centroid method to obtain a preliminary output, and multiplying the preliminary output by a preset scaling factor to obtain the original increment.

[0016] Preferably, the fuzzy variables are input into a fuzzy rule table and subjected to minimum and maximum operations to obtain a comprehensive fuzzy set, including: if the pressure deviation is positive and the deviation change rate is positive, then a rule output for increasing the pumping flow rate is matched; if the pressure deviation is negative and the deviation change rate is negative, then a rule output for decreasing the pumping flow rate is matched; if both the pressure deviation and the deviation change rate are close to zero, then a rule output for maintaining the current pumping flow rate is matched; the minimum operation is performed on all activated rule outputs to obtain the activation intensity, and the maximum operation is performed on each rule output to obtain the comprehensive fuzzy set.

[0017] Preferably, the adjustment factor is calculated based on the stage parameter and the deviation, and the original increment is adjusted using the adjustment factor to obtain the target increment, including: the target increment is equal to the product of the adjustment factor and the original increment; wherein, the adjustment factor is equal to 1 plus the product of the value corresponding to the stage parameter and the deviation, and the deviation is the ratio of the absolute value of the trend mean to the trend reference value; if the adjustment factor is greater than a preset adjustment upper limit, then the adjustment upper limit is taken as the adjustment factor; if the adjustment factor is less than a preset adjustment lower limit, then the adjustment lower limit is taken as the adjustment factor.

[0018] This invention constructs an adjustment factor using a stage factor and a deviation degree, subjecting the adjustment factor to both the stage position of the filtration process and the degree of filtration pressure drift. In the initial stage of filtration, when the stage factor is small, the adjustment factor approaches 1, maintaining stable control. In the later stage of filtration, when the stage factor is large, the adjustment factor significantly amplifies with the increase of the deviation degree, effectively addressing the filtration pressure drift during the rapid increase of filter cake resistance.

[0019] Preferably, superimposing the target increment onto historical instructions to obtain a flow instruction and outputting it for execution includes: adding the target increment to the historical instructions to obtain a preliminary instruction; comparing the preliminary instruction with a preset upper working limit and a preset lower working limit; if the preliminary instruction is greater than the upper working limit, then using the upper working limit as the flow instruction; if the preliminary instruction is less than the lower working limit, then using the lower working limit as the flow instruction; and sending the flow instruction to the drive device for its output execution.

[0020] Preferably, after the flow command is sent to the driving device for its output execution, the method further includes: obtaining a preset factor threshold, determining whether the stage parameter is greater than the factor threshold, and determining whether the flow command is equal to the working upper limit; if the stage parameter is greater than the factor threshold and the flow command is equal to the working upper limit, then a filter replacement signal is generated and the filter replacement signal is output.

[0021] This invention combines the filter cake resistance saturation state and the pumping capacity limit as triggering conditions by simultaneously judging whether the stage factor exceeds the preset factor threshold and whether the flow command has reached the preset working limit. It generates and outputs a filter replacement signal, effectively avoiding false alarms caused by judging a single condition. This allows operators to replace the filter at the appropriate time and prevents the quality of the masterbatch product from being affected by the continuous loss of control over the filtration pressure.

[0022] The beneficial effects of this invention are as follows: This invention transforms the current stage position of the filtration process into control parameters by calculating the resistance coefficient and extracting stage parameters in real time. Simultaneously, it extracts low-frequency drift characteristics and calculates the deviation degree using the arithmetic mean of pressure deviations within a time window to accurately identify systematic pressure drift. Based on these two factors, it calculates the adjustment factor to adaptively amplify the original increment of the fuzzy inference output. Furthermore, it employs an industry-standard mapping coefficient adjustment mechanism to dynamically scale the equivalent domain of discourse, ensuring that the controller operates with a resolution and adjustment level matching the current operating conditions throughout the entire filtration cycle, thus stabilizing the masterbatch filtration pressure near the target pressure.

[0023] Furthermore, this invention coordinates the stage parameters and deviation to adjust the output increment of fuzzy inference, and adaptively adjusts the mapping coefficients within the framework of industrial common sense. This enables the controller to operate robustly with a wider domain of discourse in the early stage of filtration and to make fine adjustments with a narrower domain of discourse and enhanced strength in the later stage of filtration. This fundamentally overcomes the nonlinear changes in the dynamic characteristics of the system and achieves constant control of the filtration pressure throughout the entire filtration cycle of the masterbatch. Attached Figure Description

[0024] Figure 1 This is a flowchart of the method for constant adjustment of masterbatch filtration pressure based on proportional-integral-derivative control in this invention; Figure 2 This is a degenerate curve of the trend mean changing with the stage factor in this invention; Figure 3 This is a comparison curve of the current pressure change of the color masterbatch during a complete filtration cycle in this invention. Detailed Implementation

[0025] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0026] This invention discloses a method for constant filtration pressure adjustment of color masterbatch based on proportional-integral-derivative control, referring to... Figure 1 This includes steps S1-S6: S1. Obtain the current pressure, current flow rate, and target pressure at the inlet of the filter device.

[0027] In one optional embodiment, the slurry in the masterbatch production line is pumped through a pipeline into a filtration device, where solid-liquid separation is achieved via a filter screen. However, as filtration continues, filter cake accumulates on the filter screen surface, leading to a continuous increase in filtration resistance and consequently, a continuous drift in filtration pressure. To stabilize the filtration pressure near the set value that meets process requirements, it is necessary to collect the current pressure and flow rate at the slurry inlet of the filtration device in real time as feedback signals and obtain a pre-set target pressure.

[0028] Furthermore, a pressure sensor and a flow meter are installed on the slurry inlet pipe of the filtration device to achieve sampling at a fixed period. Get the current pressure in real time and current traffic In this process, variables are set. This is an incrementing number representing the current sampling time, with values ​​increasing sequentially from 0 starting from the beginning of filtering. Sampling period. The specific value is determined by the response speed of the filtration system, and is usually taken as a fraction of the system response time constant. to The sampling period in this invention The value is 1 second. At the same time, the target pressure, pre-set according to the product manufacturing requirements, also needs to be obtained. The unit of measurement for this parameter is Pa. For example, in the plate and frame filter section of a masterbatch production line, the target pressure... The pressure is set to 0.4 MPa, and the duration of a single filtration cycle is usually between 2 and 4 hours. During this period, the thickness of the filter cake will gradually increase from 0 to about 30 mm.

[0029] In this way, by collecting real-time data through pressure sensors and flow meters, the current pressure and current flow rate at the inlet of the filter device are obtained synchronously in each sampling cycle. Combined with the preset target pressure, accurate raw data input is provided for subsequent deviation calculation and resistance sensing.

[0030] S2. Calculate the difference between the target pressure and the current pressure to obtain the pressure deviation, and calculate the change value of the pressure deviation to obtain the deviation change rate; calculate the resistance coefficient based on the current pressure and current flow rate, and obtain the stage parameters characterizing the filter resistance state.

[0031] In an optional embodiment, the current pressure at the inlet of the filter device is obtained. Current traffic and target pressure Next, it is necessary to simultaneously calculate control quantities such as pressure deviation and deviation variation rate, which reflect the current degree of pressure deviation, and stage parameters, which reflect the degree of filter cake accumulation on the filter screen surface. These control quantities characterize the current operating status of the filtration device from different dimensions, and together provide a decision basis for the subsequent adaptive adjustment of the fuzzy controller.

[0032] Furthermore, when calculating the pressure deviation and the rate of change of deviation, the target pressure is... With current pressure By taking the difference, the pressure deviation can be obtained:

[0033] in, For the first The pressure deviation at each sampling time is expressed in Pa. The target pressure is measured in Pa. For the first The current pressure at each sampling time is expressed in Pa; when A positive value indicates that the pressure at the inlet of the filter device is lower than the target pressure at the current moment. A negative value indicates that the current pressure is higher than the target pressure.

[0034] It should be noted that pressure deviation directly reflects the degree of deviation of the current pressure during filtration from the target pressure set in the process. In the initial stage of filtration when the filter cake is thin, the pressure deviation is usually small and does not fluctuate much; as the filter cake continues to accumulate, the filtration resistance increases non-linearly, and if the pump flow rate is not adjusted in time, the pressure deviation will continue to increase.

[0035] When calculating the rate of change of deviation, the rate of change of deviation can be obtained by subtracting the pressure deviation at the current moment from the pressure deviation at the previous moment:

[0036] in, For the first The rate of change of deviation at each sampling time, in Pa; For the first The pressure deviation at each sampling time is expressed in Pa. For the first The pressure deviation at each sampling time is expressed in Pa.

[0037] It should be noted that at the initial moment, i.e. At that time, the pressure deviation of the previous moment... Setting it to 0 results in a deviation rate of 0 at the initial time. equal This allows the system to reflect the change of the first sampling point relative to the 0-deviation state, and this setting will not cause abnormal output from the controller.

[0038] To measure the current level of filter cake accumulation, the resistance coefficient needs to be calculated based on the current pressure and flow rate at the inlet of the filter device. Specifically, first, a preset lower flow rate limit is obtained, and it is determined whether the current flow rate is lower than this limit. If the current flow rate is lower than the lower flow rate limit, the lower flow rate limit is taken as the effective flow rate; if the current flow rate is not lower than the lower flow rate limit, the current flow rate is directly taken as the effective flow rate. Finally, the ratio of the current pressure to the effective flow rate is calculated to obtain the resistance coefficient.

[0039] in, For the first The drag coefficient at each sampling time is expressed in Pa·s / m³. The current pressure is expressed in Pa. The current flow rate is expressed in m³ / s. The lower limit of flow rate is measured in m³ / s. The lower limit of flow rate is taken from the theoretical flow rate value corresponding to the pump at the lowest operating frequency. It can be found through the pump's performance curve. The purpose of introducing the lower limit of flow rate is to prevent the resistance coefficient from tending to infinity when the current flow rate is extremely low, which would lead to calculation abnormalities. This is the effective traffic.

[0040] It should be noted that this formula is based on the fundamental principle of Darcy's Law. Since the pressure drop during filtration is directly proportional to the slurry flow rate and the filter cake resistance, the ratio of the current pressure to the effective flow rate directly reflects the overall flow resistance characteristics of the filter screen and filter cake. In actual operation, the filter screen is clean and the filter cake has not yet formed in the initial stage of filtration, resulting in a relatively small resistance coefficient. As filtration progresses, the filter cake thickness continuously increases, and the resistance coefficient exhibits a monotonically increasing trend until it reaches its maximum value when the filter cake is close to saturation at the end of filtration. The resistance coefficient accurately reflects the current stage of the filtration process.

[0041] Next, to accurately characterize the filter cake accumulation state, stage parameters are calculated and obtained. In a preferred embodiment, the resistance coefficient is normalized to obtain a stage factor, which is then used as the stage parameter.

[0042] in, For the first The stage parameters at each sampling time are dimensionless and range from [value missing]. ; The drag coefficient has the dimension of Pa·s / m³. The initial resistance is expressed in Pa·s / m³. The maximum resistance is expressed in Pa·s / m³.

[0043] It should be noted that the initial resistance The method for obtaining the resistance coefficient is to take the arithmetic mean of the resistance coefficients over the first 5 sampling periods after each filter replacement. At this time, the filter cake has not yet formed, and the resistance coefficient only reflects the resistance of the clean filter itself. Maximum resistance It is obtained based on historical production data, specifically the arithmetic mean of the resistance coefficient at the last sampling moment before filter replacement in the past 10 filtration cycles, with the maximum resistance being... This represents the resistance level when the filter cake is close to saturation.

[0044] During the calculation process, if the calculated stage factor is less than 0, it is set to 0; if the stage factor is greater than 1, it is set to 1, ensuring that the stage factor always remains within the range of 0 to 1. If insufficient historical data or abnormal parameter configuration results in the maximum resistance being less than or equal to the initial resistance, the stage factor is set to 0, causing the controller to run with the initial default parameters and issue an error message. This normalization process eliminates the influence of differences in slurry concentration between different batches and differences in the absolute value of resistance between different filter screen specifications, allowing the stage factor to accurately map the current filtration process's relative position from start to finish.

[0045] In this way, the calculation of pressure deviation, deviation variation rate, drag coefficient, and stage parameters is completed simultaneously within a sampling period. By comprehensively measuring the current filtering state from two dimensions—pressure deviation characteristics and filtering stage characteristics—complete decision inputs are provided for subsequent steps such as trend mean acquisition, deviation calculation, domain scaling, and rule strength adjustment.

[0046] S3. Extract the low-frequency drift characteristics of the pressure deviation in the time series, and calculate the deviation degree used to characterize the degree of pressure drift in the later stage of filtration.

[0047] In an optional embodiment, to identify whether there is pressure drift in the filter, it is necessary to obtain the low-frequency drift characteristics of the pressure deviation within a preset time period. Specifically, a sliding window is used to calculate the trend mean as the low-frequency drift characteristic, that is, the incrementing number of the current sampling time is obtained. and compare it with a preset constant. The smaller value is selected as the valid length after comparison.

[0048] in, The effective length is dimensionless. An incrementing number representing the current sampling time; This is a preset constant, dimensionless.

[0049] It should be noted that the preset constant During the commissioning phase, a step response test was conducted to determine the filter pressure. This involved applying a step change to the pump flow rate and observing the time it took for the filter pressure to stabilize at the new value. This time was then divided by the sampling period. The number of sampling points corresponding to the response time can be obtained, and then multiplied by 3 to 5 times to obtain the value of the preset constant. In this invention, the preset constant... Take 20.

[0050] Next, obtain the first From the sampling time to the... Multiple pressure deviations within a time window of effective length for each sampling time are considered, and the arithmetic mean of these multiple pressure deviations within the time window is calculated to obtain the trend mean. The trend mean satisfies the following relationship:

[0051] in, For the first The trend mean at each sampling time, with the dimension Pa; For the first The pressure deviation at each sampling time is expressed in Pa. For effective length, dimensionless. When hour, Take 1 and define the trend mean as 0; when hour, The actual number of existing sampling points is used, and the effect gradually takes effect as the number of sampling points accumulates; when back, The sliding window has ample data, and the adaptive mechanism is fully effective.

[0052] It should be noted that the arithmetic mean can smooth out short-term disturbances caused by pumping pulsations or slurry air bubbles. The trend mean, which is a low-frequency drift characteristic, will only significantly deviate from zero when the filtration pressure continuously deviates from the target pressure over multiple consecutive sampling periods. A positive trend mean indicates that the filtration pressure is consistently lower than the target pressure within the aforementioned time window, while a negative trend mean indicates that the filtration pressure is consistently higher than the target pressure.

[0053] Furthermore, after obtaining the trend mean as a characteristic of low-frequency drift, the deviation is calculated to characterize the degree of drift in the filter pressure. First, a preset trend reference value needs to be obtained. Trend reference value Pressure for the target 5% of It can be adjusted within the range of 3% to 8% of the target pressure according to the actual control accuracy requirements. Subsequently, the ratio of the absolute value of the trend mean to the trend reference value is calculated as the deviation, i.e. Deviation, as a normalized indicator, can accurately reflect the proportion of the systematic deviation of the current filtration pressure relative to the allowable reference range.

[0054] Thus, by extracting the low-frequency drift characteristics of the filtration pressure within the time window and calculating the deviation, the short-term high-frequency disturbances caused by pumping pulsation were successfully filtered out. The actual systemic pressure drift degree of the filtration system at the current stage was accurately identified, providing a reliable decision input for the calculation of adjustment factors and adaptive compensation in subsequent steps.

[0055] S4. Perform fuzzy reasoning based on the pressure deviation and the deviation variation rate to obtain the original increment.

[0056] In an optional embodiment, before performing fuzzy inference, in order to improve the ability to distinguish small deviations in the later stages of filtering, the mapping coefficients of the fuzzy controller need to be dynamically adjusted according to the actual fluctuation range of the pressure deviation to achieve industry-standard equivalent adaptive scaling of the domain of discourse.

[0057] Specifically, the effective length is counted. Within the corresponding time window, obtain the maximum pressure deviation among multiple deviations. With minimum deviation The difference between the maximum and minimum deviations is taken as the fluctuation range of the pressure deviation at the current moment. Calculate the ratio of this fluctuation amplitude to the total width of the standard universe of discourse to obtain the utilization ratio. The calculation method for the utilization ratio is as follows:

[0058] in, For the first The utilization rate at each sampling time, dimensionless; The maximum deviation of pressure within the time window is expressed in Pa. The minimum deviation of pressure within the time window, in Pa; The half-amplitude of the defined standard domain, with dimensions in Pa, for example, the target pressure. 10%; the total width of the standard domain is twice half the width of the standard domain, that is ;when At that time, utilization ratio Defined as 1 to ensure no abnormal output during initialization. Then, a preset utilization reference value is obtained. The scaling factor is obtained by calculating the ratio of the utilization rate to the reference utilization value.

[0059] in, For the first The scaling factor for each sampling time, dimensionless; To utilize a reference value that is dimensionless, 0.6 is used in this invention; The lower bound of the scaling factor is set to 0.3; The upper limit of the scaling factor is set to 1.5. The lower and upper limits of the scaling factor ensure that the scaling process always operates within a safe range.

[0060] Next, the preset initial mapping coefficients are adjusted by a scaling factor to achieve scaling of the equivalent universe of discourse, without directly modifying the membership function vertices of the underlying fuzzy matrix. The target mapping coefficients are calculated as follows:

[0061] in, For the first The target mapping coefficients at each sampling time; The mapping coefficients initially set for the controller. When filtering the scaling factor later. When the target mapping coefficient is reduced, such as to 0.5, the target mapping coefficient... This will increase to twice the initial value. This allows minute physical pressure deviations, after being multiplied by the amplified mapping coefficient, to fill the entire standard universe of discourse, thus doubling the ability to resolve minute pressure deviations without changing the standard membership function. This is an efficient design that aligns with the underlying logic of industry.

[0062] After obtaining the current target mapping coefficients, the preset standard membership function is retrieved. This preset standard membership function remains fixed and contains seven fuzzy subsets distributed in a triangular pattern, defined in order from negative to positive as negative large, negative medium, negative small, zero, positive small, positive medium, and positive large subsets. Each fuzzy subset is distributed within the standard normalized universe of discourse, and its vertex position and base width remain statically fixed.

[0063] Subsequently, using the target mapping coefficient Pressure deviation at the current moment and deviation change rate Standardization mapping is performed to transform the variables into fuzzy variables. The transformed fuzzy variables are then input into a fuzzy rule table for minimization and maximization operations to obtain a comprehensive fuzzy set. The fuzzy rule table uses pressure deviation and deviation change rate as input dimensions, and the original increment of pump flow rate adjustment as the output dimension. If the pressure deviation is positive and the deviation change rate is positive, the rule will match an output that increases the pump flow rate; if the pressure deviation is negative and the deviation change rate is negative, the rule will match an output that decreases the pump flow rate; if both are close to zero, the current flow rate remains unchanged. The activation intensity is obtained by minimizing the output of all activated rules, and the comprehensive fuzzy set is obtained by maximization.

[0064] Finally, the centroid method is used to defuzzify the comprehensive fuzzy set to obtain a preliminary dimensionless output value. This output value is then multiplied by a preset scaling factor (Ku) to obtain the precise physical output variable, i.e., the original increment. The dimension is Hz.

[0065] Thus, by dynamically adjusting the mapping coefficient of the input signal according to the actual fluctuation amplitude of the pressure deviation, the equivalent adaptive scaling of the universe of discourse is cleverly achieved without changing the physical structure of the standard membership function of the underlying control system at all. This enables the fuzzy controller to effectively capture and amplify small pressure deviations in the later stage of filtration, greatly enhancing the refined resolution and adjustment ability for small deviations. At the same time, through a standardized fuzzy inference process, the current pressure state is successfully transformed into a preliminary raw increment output.

[0066] S5. Calculate the adjustment factor according to the stage parameter and the deviation degree, and use the adjustment factor to adjust the raw increment to obtain the target increment.

[0067] In an optional embodiment, the raw increment is the inference result obtained under the instantaneous deviation information at the current sampling moment, without considering the stage position of the current cake accumulation and the drift degree of the filtration pressure. In the later stage of filtration, the cake resistance accelerates to increase. If the adjustment intensity is not strengthened for the same pressure deviation, the adjustment amplitude of the pumping flow rate will not be sufficient to catch up with the rapid change of the cake resistance, resulting in the continuous deviation of the filtration pressure from the target pressure. Therefore, it is necessary to dynamically adjust the raw increment according to the stage parameter reflecting the cake accumulation and the deviation degree to obtain the target increment.

[0068] Furthermore, the target increment is equal to the product of the adjustment factor and the raw increment, where the adjustment factor is equal to 1 plus the product of the value corresponding to the stage parameter and the deviation degree. To prevent the driving device from acting violently due to over-adjustment, it is necessary to obtain the preset adjustment upper limit and the preset adjustment lower limit, and perform amplitude limiting processing on the adjustment factor. The adjustment factor satisfies the following relational expression:

[0069] where, is the adjustment factor at the th sampling moment, dimensionless; is the stage parameter adopted in this embodiment (i.e., the stage factor obtained by normalizing the resistance coefficient), and its value range is [0, 1]; is the trend mean value, with the dimension of Pa; is the trend reference value, with the dimension of Pa; is the adjustment lower limit, which is taken as 1 in this invention to ensure that the target increment is always not lower than the raw increment; is the adjustment upper limit, which is taken as 2 in this invention to ensure that the target increment is at most amplified to twice the raw increment.

[0070] It should be noted that the adjustment logic of the adjustment factor reflects the synergistic effect of stage perception and drift detection. That is, when the trend mean is 0, it indicates that there is no drift in the filtering pressure within the time window. At this time, the deviation is 0, the adjustment factor is 1, and the target increment is equal to the original increment. When the absolute value of the trend mean is large and the stage parameter is large, it indicates that it is in the middle and late stages of filtering and there is an obvious drift in the filtering pressure. At this time, the adjustment factor is greater than 1, and the target increment is amplified to strengthen the compensation intensity, driving the filtering pressure to return to the target pressure more quickly.

[0071] However, the significance of the stage parameter as a multiplier is that in the initial stage of filtering, the stage parameter is close to 0. At this time, even if there is a certain trend mean, the adjustment factor is close to 1, and there is no rush to make large adjustments. In the late stage of filtering, the stage parameter is close to 1. At this time, the accumulation of the filter cake accelerates, and the same trend mean will be fully amplified, thus driving a more active adjustment response. The direction information of the pressure deviation is carried by the positive and negative signs of the original increment. The adjustment factor only responsible for adjusting the amplitude and does not perform the function of direction judgment.

[0072] The calculation method of the target increment is as follows:

[0073] where is the target increment at the th sampling moment, with the dimension of Hz; is the adjustment factor, dimensionless; is the original increment obtained by fuzzy inference, with the dimension of Hz.

[0074] Exemplarily, assume that the target pressure is 0.4 MPa and the trend reference value is 0.02 MPa. In the initial stage of filtering, if the stage factor is 0.1 and the trend mean is 0.01 MPa, then the adjustment factor is , which means that the target increment is only amplified by 5%, and the adjustment intensity is relatively mild. In the late stage of filtering, if the stage factor is 0.8 and the trend mean is 0.015 MPa, the adjustment factor is , which means that the target increment is amplified by 60%, and the adjustment intensity is significantly enhanced. In the final stage of filtering, if the stage factor is 0.9 and the trend mean is 0.03 MPa, the adjustment factor is Since the value exceeded the adjustment limit, the adjustment limit protection was triggered, and the final adjustment factor was set to 2, limiting the target increment to twice the original increment.

[0075] Reference Figure 2 This study demonstrates that under conditions of nonlinear deterioration in filter cake resistance, the stage factor, through normalization, characterizes the severity of filter cake accumulation on the filter surface. When the stage factor exceeds 0.7, entering the stage of rapid increase in filtration resistance, the filtration pressure drift phenomenon under existing control technologies rapidly worsens, and the system's regulatory capability severely degrades. In contrast, this invention effectively suppresses the degradation trend of control performance by real-time detection of the trend mean and dynamically amplifying the target increment accordingly. Even when the filter cake is close to saturation, it can still limit the deterioration of filtration pressure drift to a level far lower than that of existing technologies.

[0076] By using the synergistic drive of stage parameters and deviations characterizing filter cake accumulation, the adjustment factor can automatically match the adjustment intensity to adapt to the current physical conditions at different filtration stages. This achieves adaptive enhancement of the control rule output compensation intensity as the filtration process progresses, thereby fundamentally and effectively addressing the severe pressure drift problem caused by the rapid increase in filter cake resistance in the later stages of filtration.

[0077] S6. Add the target increment to the historical instructions to obtain the flow instruction and output it to the pumping device for execution.

[0078] In an optional embodiment, after obtaining the target increment, it needs to be superimposed on the historical instructions to generate the current flow instruction. After the flow instruction is processed by the limiting protection, it is sent to the output of the drive equipment, i.e., the frequency converter of the pumping device, etc., to complete a complete closed-loop control cycle.

[0079] Furthermore, by adding the target increment to the historical instructions, the preliminary instructions can be obtained, which satisfy the following relationship:

[0080] in, For the first The initial command for each sampling time, in Hz; This is a historical instruction saved from the previous sampling time, with the dimension Hz; The target increment is expressed in Hz.

[0081] It should be noted that the initial historical instructions The preset operating limit is 50% to ensure that the pump motor can quickly establish the initial filtration pressure at a medium flow rate during the initial startup phase.

[0082] The above-described superposition structure employs an incremental output method. This means that each sampling cycle only outputs the change in the control quantity and superimposes it onto the historical command, rather than directly outputting the absolute value of the control quantity. This incremental structure has a significant advantage: when the controller is affected by interference or abnormal conditions, the negative impact is limited to the target increment of the current sampling cycle, preventing a step-like change in the flow command. This significantly reduces the physical impact on the pumping motor and pipeline system, and substantially improves the safety and stability of equipment operation.

[0083] To prevent the generated flow command from exceeding the device's capacity, it is necessary to compare the initial command with the preset upper and lower operating limits. That is, the flow command is obtained after the initial command is limited. The flow command satisfies the following relationship:

[0084] in, For the first The flow rate command at each sampling time, in Hz; The preset lower limit of operation is measured in Hz. In this invention, the value is 10Hz. This value is determined by the hardware specifications of the frequency converter. When the frequency is lower than this lower limit, the motor torque may be insufficient, resulting in a low pumping flow rate, which in turn causes the slurry to deposit in the pipeline, causing a risk of blockage. The preset operating upper limit is set to 50Hz in this invention. This value is also determined by the hardware specifications of the frequency converter. When the operating upper limit frequency is exceeded, the motor will face overload, which may cause equipment damage.

[0085] The logic for limiting the flow rate is as follows: if the initial command is greater than the preset upper limit, then the upper limit is used as the flow rate command; if the initial command is less than the preset lower limit, then the lower limit is used as the flow rate command. Finally, the determined flow rate command is... The data is transmitted to the drive unit via a communication interface, which then adjusts the pump motor speed accordingly, thereby changing the actual flow rate of the pumping device. As the pumping flow rate changes, the filtration pressure at the inlet of the filter device also changes, and the pressure sensor and flow meter will detect this at the next sampling time. Once new current pressure and flow data are collected, the next round of closed-loop calculation begins.

[0086] After issuing the flow command to the drive device for execution, the adjustment method also includes actively monitoring the filter's status. First, a pre-set factor threshold based on production experience is obtained; in this invention, it is set to 0.9. This factor threshold indicates that when the stage parameter characterizing the filter cake accumulation exceeds 0.9, the filter cake is considered close to saturation. Then, it is determined whether the current stage parameter is greater than the factor threshold, and simultaneously whether the currently issued flow command is equal to the preset working upper limit. If the stage parameter is greater than the factor threshold and the flow command is equal to the working upper limit, it indicates that the filter cake resistance has reached its limit. At this point, even if the pumping capacity is at full output, it is still insufficient to maintain the set target pressure. Under this extreme condition, the system generates a filter replacement signal and outputs it to the control terminal to prompt the operator to replace the filter and start a new filtration cycle. By jointly judging the factor threshold and the flow command upper limit as the triggering basis, false alarms that may occur due to single-condition judgment are effectively avoided, ensuring that operators can perform maintenance at the appropriate time.

[0087] Reference Figure 3 In the initial stage of filtration, both methods can effectively follow the target pressure. However, in the later stage of filtration, as the filter cake accumulates and the filtration resistance increases non-linearly, existing technologies, unable to detect the stage changes in the filtration process, exhibit significant filtration pressure drift. This invention, through the coordinated adaptive adjustment of stage parameters and low-frequency drift characteristics, successfully overcomes the changes in the dynamic characteristics of the filtration system, reducing the pressure deviation by approximately 15% to 20% in the later stage of filtration and achieving constant control of the filtration pressure throughout the entire filtration cycle.

[0088] Thus, through incremental superposition calculation and a bilateral limiting protection mechanism, the target increment is successfully and safely converted into an executable flow command and sent to the pumping device for execution. Simultaneously, when the filter cake approaches saturation and pumping capacity is limited, the system proactively prompts for filter replacement. This constructs a complete closed-loop control system encompassing data acquisition, state parameter sensing, mapping adjustment and fuzzy inference, force compensation adjustment, physical command output, and state monitoring, thereby ensuring that the filtration pressure of the masterbatch remains stable near the target pressure throughout the entire filtration cycle.

[0089] It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept, and these all fall within the scope of protection of this invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A method for constant filtration pressure regulation of masterbatch based on proportional-integral-derivative control, characterized in that, include: Obtain the current pressure, current flow rate, and target pressure at the inlet of the filter device; Calculate the difference between the target pressure and the current pressure to obtain the pressure deviation, and calculate the change in the pressure deviation to obtain the deviation change rate; obtain stage parameters characterizing the current degree of filter cake accumulation based on the current pressure and the current flow rate; Extract the low-frequency drift characteristics of the pressure deviation in the time series, and calculate the deviation degree to characterize the degree of pressure drift in the later stage of filtration. The original increment is obtained by performing fuzzy reasoning based on the pressure deviation and the rate of change of the deviation. Calculate the adjustment factor based on the stage parameters and the deviation, and adjust the original increment using the adjustment factor to obtain the target increment; The target increment is superimposed on the historical instructions to obtain a flow instruction, which is then output to the pumping device for execution to control the current flow rate.

2. The method for constant pressure adjustment of masterbatch filtration based on proportional-integral-derivative control according to claim 1, characterized in that, Calculating the resistance coefficient based on the current pressure and the current flow rate includes: obtaining a preset lower flow rate limit; determining whether the current flow rate is lower than the lower flow rate limit; if the current flow rate is lower than the lower flow rate limit, then the lower flow rate limit is taken as the effective flow rate; if the current flow rate is not lower than the lower flow rate limit, then the current flow rate is taken as the effective flow rate; calculating the ratio of the current pressure to the effective flow rate to obtain the resistance coefficient.

3. The method for constant pressure adjustment of masterbatch filtration based on proportional-integral-derivative control according to claim 2, characterized in that, The step of obtaining the stage parameter characterizing the current filter cake accumulation level includes: normalizing the resistance coefficient to obtain a stage factor as the stage parameter; the normalization includes: the stage factor is the ratio of the difference between the resistance coefficient and the initial resistance to the difference between the maximum resistance and the initial resistance, and the calculation result is limited to between 0 and 1; wherein, the initial resistance is the arithmetic mean of the resistance coefficients in the initial filtration state, and the maximum resistance is the arithmetic mean of the resistance coefficients in the historical filtration saturation state.

4. The method for constant pressure adjustment of masterbatch filtration based on proportional-integral-derivative control according to claim 1, characterized in that, The step of extracting the low-frequency drift feature of the pressure deviation over a time series and calculating the deviation based thereon includes: obtaining the trend mean of the pressure deviation over a preset time period as the low-frequency drift feature; obtaining a preset trend reference value and calculating the deviation based on the trend mean and the trend reference value; obtaining the trend mean of the pressure deviation over the preset time period includes: obtaining the incrementing number of the current sampling time and comparing the incrementing number with a preset constant, selecting the smaller value as the effective length; obtaining multiple pressure deviations within a time window equal to the effective length from the current sampling time, and calculating the arithmetic mean of the multiple pressure deviations to obtain the trend mean.

5. The method for constant pressure adjustment of masterbatch filtration based on proportional-integral-derivative control according to claim 4, characterized in that, Before performing fuzzy inference, the method further includes: dynamically adjusting the mapping coefficients of the fuzzy controller based on the fluctuation amplitude of the pressure deviation to achieve equivalent universe of discourse scaling, including: obtaining the maximum and minimum deviations of multiple pressure deviations within the time window, and subtracting the minimum deviation from the maximum deviation as the fluctuation amplitude; calculating the ratio of the fluctuation amplitude to the total width of the standard universe of discourse to obtain the utilization ratio, wherein the total width of the standard universe of discourse is twice half the width of the standard universe of discourse; obtaining a preset utilization reference value, calculating the ratio of the utilization ratio to the utilization reference value to obtain the scaling factor; and after limiting the scaling factor, dividing it by the initially set mapping coefficient to obtain the target mapping coefficient at the current sampling time.

6. The method for constant pressure adjustment of masterbatch filtration based on proportional-integral-derivative control according to claim 1, characterized in that, The process of performing fuzzy inference on the pressure deviation and the deviation variation rate to obtain the original increment includes: obtaining a preset standard membership function with a fixed shape and vertex position, wherein the preset standard membership function contains multiple fuzzy subsets distributed in a triangular pattern within a standard domain; standardizing and mapping the pressure deviation and the deviation variation rate using mapping coefficients to obtain fuzzy variables; inputting the fuzzy variables into a fuzzy rule table for minimization and maximization operations to obtain a comprehensive fuzzy set; defuzzifying the comprehensive fuzzy set using the centroid method to obtain a preliminary output, and multiplying the preliminary output by a preset scaling factor to obtain the original increment.

7. The method for constant pressure adjustment of masterbatch filtration based on proportional-integral-derivative control according to claim 6, characterized in that, The fuzzy variables are input into a fuzzy rule table and subjected to minimum and maximum operations to obtain a comprehensive fuzzy set. This includes: if the pressure deviation is positive and the deviation change rate is positive, then a rule output for increasing the pumping flow rate is matched; if the pressure deviation is negative and the deviation change rate is negative, then a rule output for decreasing the pumping flow rate is matched; if both the pressure deviation and the deviation change rate are close to zero, then a rule output for maintaining the current pumping flow rate is matched; the minimum operation is performed on all activated rule outputs to obtain the activation intensity, and the maximum operation is performed on each rule output to obtain the comprehensive fuzzy set.

8. The method for constant pressure adjustment of masterbatch filtration based on proportional-integral-derivative control according to claim 4, characterized in that, The adjustment factor is calculated based on the stage parameters and the deviation, and the original increment is adjusted using the adjustment factor to obtain the target increment. This includes: the target increment is equal to the product of the adjustment factor and the original increment; wherein the adjustment factor is equal to 1 plus the product of the value corresponding to the stage parameter and the deviation, and the deviation is the ratio of the absolute value of the trend mean to the trend reference value; if the adjustment factor is greater than a preset adjustment upper limit, then the adjustment upper limit is taken as the adjustment factor; if the adjustment factor is less than a preset adjustment lower limit, then the adjustment lower limit is taken as the adjustment factor.

9. The method for constant pressure adjustment of masterbatch filtration based on proportional-integral-derivative control according to claim 1, characterized in that, The process of adding the target increment to historical instructions to obtain a flow instruction and then outputting it for execution includes: adding the target increment to the historical instructions to obtain a preliminary instruction; comparing the preliminary instruction with a preset upper working limit and a preset lower working limit; if the preliminary instruction is greater than the upper working limit, then using the upper working limit as the flow instruction; if the preliminary instruction is less than the lower working limit, then using the lower working limit as the flow instruction; and sending the flow instruction to the drive device for its output execution.

10. The method for constant pressure adjustment of masterbatch filtration based on proportional-integral-derivative control according to claim 9, characterized in that, After the flow command is sent to the driving device for its output execution, the method further includes: obtaining a preset factor threshold, determining whether the stage parameter is greater than the factor threshold, and determining whether the flow command is equal to the working upper limit; if the stage parameter is greater than the factor threshold and the flow command is equal to the working upper limit, then a filter replacement signal is generated and the filter replacement signal is output.

Images