High-speed wire rotational flow pool liquid level correction control method and system based on curve model filtering

Abstract

The invention discloses a curve model filtering-based high-speed wire rotational flow pool liquid level correction control method and system, and relates to the technical field of steel production, and the method comprises the following steps: collecting a liquid level signal, and constructing a fixed curve model according to a process stage; a staged deviation zone is set according to the model, deviation is compared and calculated in real time, and abnormity judgment is carried out; performing previous normal value substitution, model prediction value substitution or triggering process alarm on the abnormal section according to the duration to obtain a substitution sequence; and performing composite filtering on the alternative sequence, and outputting a corrected liquid level for closed-loop control and visualization. Through four-stage fixed curve modeling and staged deviation band judgment, graded substitution is triggered by a pull-back amount, a phase drift amount and a logic consistency rate, and compound filtering closed-loop control is performed on a substitution sequence, so that accurate anomaly detection, low false alarm, continuous curve and stable control are realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of steel production, in particular to a high-speed wire vortex pool liquid level correction control method and system based on curve model filtering. BACKGROUND

[0002] In actual production rolling site, analog quantity sensors are widely used in various production links, real-time acquisition of device state in closed-loop control system. The automatic control of heating furnace step beam, vortex pool lifting pump, rolling mill cascade, collection area core rod and other devices cannot be separated from analog quantity sensors. The field environment is relatively complex, electromagnetic interference is everywhere, in order to improve the stability of analog quantity sensors in feedback link, signal isolation module is often added before the sensor is connected to the central program control system. This can handle most of the interference generated between the sensor and the control cabinet, but it is powerless to the interference between the isolation module output and the central program control system input, and such interference has a great impact on the signal and data, which will directly affect the normal production rolling. Therefore, the existing technology has great drawbacks and defects, which are specifically shown in the following aspects. First, the interference between the isolation module output and the central program control system input will cause abnormality of key rolling process tracking signal, and then cause steel stacking in production rolling process, and then cause reduction of yield in production rolling process. Then, the interference between the isolation module output and the central program control system input will cause failure of key closed-loop control in production rolling process, because the collected basic data are problematic, and then cause out-of-control state of closed-loop regulation in the whole rolling process, and then cause abnormality of finished product quality in production rolling process, and then also cause great reduction of yield and great reduction of output. Finally, the interference between the isolation module output and the central program control system input will cause various control soft faults in production rolling process, and then greatly prolong the fault handling time, and then cause great bottleneck problem to the whole production stability. SUMMARY

[0003] Based on the above-mentioned shortcomings of the prior art, the purpose of the present application is to provide a high-speed wire vortex pool liquid level correction control method and system based on curve model filtering, to solve the above technical problems.

[0004] To achieve the above-mentioned purpose, the present application provides the following technical scheme: a high-speed wire vortex pool liquid level correction control method based on curve model filtering, comprising: S1: collecting vortex pool liquid level sensing signal, generating trend curve of liquid level change with time, and dividing the trend curve into four stages of rising, peak value keeping, falling and valley value keeping according to working conditions; S2: a fixed curve model is established based on the normal production condition, the actual liquid level change trend is matched in time interval segmentation and amplitude change, and segmentation and positioning are performed on the time axis and amplitude axis to form a model reference; S3: a phased deviation band is set according to the fixed curve model, the real-time curve and the model curve are compared, the deviation is calculated, and different threshold values are determined in different stages; S4: when the deviation exceeds the corresponding stage deviation band, or the liquid level change exceeds the mutation threshold value within the preset scanning period window, the section is determined as abnormal; S5: the abnormal section is replaced by the previous normal value, the model prediction value or the process alarm replacement strategy, and a replacement sequence is obtained; S6: the replaced liquid level sequence is subjected to composite filtering processing, and a corrected liquid level sequence is output.

[0005] The application further sets that the rising stage corresponds to the liquid level rising after the water pump is started, the peak value maintaining stage corresponds to the stable drainage state, the falling stage corresponds to the liquid level falling after the water pump is stopped, and the valley value maintaining stage corresponds to the minimum working liquid level.

[0006] The application further sets that the reference condition of the fixed curve model is a sawtooth periodic curve in which the liquid level rises from the lower limit of the working liquid level to the upper limit of the working liquid level and then falls, which is used to limit the reference range of the time interval and the amplitude interval of the model.

[0007] The application further sets that in step S3, the phased deviation band is set for each of the rising, peak value maintaining, falling and valley value maintaining stages according to the fixed curve model, the real-time liquid level curve and the fixed curve model are compared at each time point during the operation to obtain the deviation, and the corresponding dynamic threshold value is applied for judgment according to the stage.

[0008] The application further sets that in step S4, the mutation threshold criterion is that the absolute value of the difference between adjacent samples of the liquid level signal exceeds the upper limit of the allowed change rate given by the fixed curve model in the preset scanning period window; wherein the preset scanning period window is determined by an integral multiple of the scanning period of the control task, and the upper limit of the allowed change rate is determined according to the upper limit of the first derivative of the fixed curve model in the corresponding stage; The criterion for the deviation exceeding the corresponding stage deviation band is that the absolute value of the residual of the real-time liquid level relative to the fixed curve model exceeds the outer boundary of the stage deviation band and continuously meets the gating condition within the corresponding stage, which is determined as abnormal; wherein the phased deviation band is determined according to the model residual distribution in the stage, and the continuous gating condition is determined according to the device lag and the process propagation time delay.

[0009] The application is further configured to calculate parameters for strategy decision for the abnormal section before performing hierarchical replacement on the abnormal section, including: pullback amount, phase shift amount, and logic consistency rate; The pullback amount is used to represent a comprehensive dimensionless indication of the regression trend and regression speed of the real-time liquid level relative to the fixed curve model after the abnormality occurs, and is obtained according to the degree of monotonous approach and the approach rate of the real-time liquid level to the model curve in the abnormal section; The phase shift amount is used to represent a dimensionless indication of the instantaneous phase misalignment degree of the real-time liquid level change rhythm relative to the fixed curve model in the corresponding stage, and is determined according to the relative order and alignment relationship of the feature event sequence of the real-time curve and the model curve in the stage; The logic consistency rate is used to represent a dimensionless indication of the consistent matching degree of the pump valve working condition switching event and the liquid level derivative direction change in the specified reaction window, and is determined according to the time corresponding relationship and trigger causal relationship of the working condition switching signal and the liquid level derivative sign.

[0010] The application is further configured that the trigger of the previous normal value replacement takes the pullback amount as the decision parameter, when the pullback amount meets the time gating and the shape gating conditions linked with the stage information of the fixed curve model, it is determined that the previous normal value is replaced in the abnormal section and written into the replacement sequence; wherein, the time gating is used to limit the short-time interference scene, and the shape gating is used to ensure that the curve after replacement is monotonous and consistent with the model curve in the stage and boundary continuous; The trigger of the model prediction replacement takes the phase shift amount as the decision parameter, when the phase shift amount meets the time gating and the rhythm gating conditions linked with the stage information of the fixed curve model, it is determined that the predicted value of the fixed curve model is replaced in the abnormal section and written into the replacement sequence; wherein, the rhythm gating is used to maintain the traceable continuity of the event sequence alignment relationship and the stage boundary in the stage; The trigger of the process alarm takes the logic consistency rate as the decision parameter, when the logic consistency rate does not meet the predetermined consistency criterion and the criterion is linked with the stage information of the fixed curve model, it is determined to trigger the process alarm and execute the corresponding protection strategy; wherein, the consistency criterion is used to verify the matching and causal consistency of the pump valve working condition switching event and the liquid level derivative direction change in the specified reaction window.

[0011] The application is further configured that the step S5 is further configured to: For the judged abnormal section, hierarchical replacement is performed according to the duration and stage working condition, wherein, the short-time interference is replaced by the previous normal value, the medium-time interference is replaced by the predicted value of the fixed curve model, and the long-term interference triggers the process alarm; and a replacement sequence is obtained.

[0012] The application is further configured to implement composite filtering processing on the replaced liquid level sequence in step S6 and output a corrected liquid level sequence, wherein the composite filtering processing at least includes inertia type smoothing, median suppression and multi-scale denoising.

[0013] The application also provides a high line cyclone pool liquid level correction control system based on curve model filtering, which is used to implement the high line cyclone pool liquid level correction control method based on curve model filtering. The signal acquisition module acquires the cyclone pool liquid level sensing signal, generates a trend curve of the liquid level change over time, and divides the trend curve into four stages of rising, peak value maintaining, falling and valley value maintaining according to the working condition. The model establishment module establishes a fixed curve model that is matched with the actual liquid level change trend in time interval segmentation and amplitude change, and segments and positions the model according to the time axis and the amplitude axis to form a model reference. The curve comparison module compares the real-time curve with the model curve according to the stage-based deviation band of the fixed curve model, calculates the deviation and executes different threshold judgments according to the stages. The abnormality judgment module judges that the segment is abnormal when the deviation exceeds the corresponding stage deviation band or the liquid level change exceeds the mutation threshold within two program scanning periods. The sequence replacement module adopts a hierarchical replacement strategy of previous normal value replacement, model prediction value replacement or process alarm triggering according to the duration of the abnormal segment to obtain a replacement sequence. The correction output module implements composite filtering processing on the replaced liquid level sequence and outputs a corrected liquid level sequence.

[0014] The application provides a high line cyclone pool liquid level correction control method and system based on curve model filtering. The method acquires a cyclone pool liquid level sensing signal, generates a trend curve of the liquid level change over time, and divides the trend curve into four stages of rising, peak value maintaining, falling and valley value maintaining according to the working condition. A fixed curve model that is matched with the actual liquid level change trend in time interval segmentation and amplitude change is established based on the normal production condition, and the model is segmented and positioned according to the time axis and the amplitude axis to form a model reference. The real-time curve is compared with the model curve according to the stage-based deviation band of the fixed curve model, the deviation is calculated, and different threshold judgments are executed according to the stages. The segment is judged to be abnormal when the deviation exceeds the corresponding stage deviation band or the liquid level change exceeds the mutation threshold within a preset scanning period window. A hierarchical replacement strategy of previous normal value replacement, model prediction value replacement or process alarm triggering is adopted according to the duration of the abnormal segment to obtain a replacement sequence. Composite filtering processing is implemented on the replaced liquid level sequence, and a corrected liquid level sequence is output. The beneficial effects include: More accurate judgment, less false positives: phase modeling combined with phase bias band mutation criterion, real-time curve comparison at each moment, process rhythm, reduce misjudgment and missed judgment and can quickly capture abnormalities; More stable decision-making, more controllable: three types of treatment are triggered by pullback amount, phase drift amount, and logical consistency rate, combined with phase gating and morphology / rhythm gating, to achieve short-term suppression, rhythm synchronization, and execution chain consistency guardianship, maintaining curve continuity and closed-loop stability. Stronger noise resistance, easier landing: replace the implementation of composite filtering to improve signal quality; clear process and module boundaries, fewer but key parameters, facilitating integration and maintenance on existing control platforms, and adapting to multiple working conditions to improve system robustness.

[0015] The above description is only a summary of the technical solutions of the present application. In order to more clearly understand the technical means of the present application, it can be implemented according to the content of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the following specific embodiments of the present application are described. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained without creative labor. In the drawings: Figure 1 The flowchart of the high-line vortex pool liquid level correction control method based on curve model filtering according to an exemplary embodiment of the present application is shown. DETAILED DESCRIPTION

[0017] The embodiments of the present application will be described below with reference to the drawings and preferred embodiments, and those skilled in the art can easily understand other advantages and effects of the present application from the content disclosed in the specification. The present application can also be implemented or applied by different specific embodiments, and the details in the specification can be modified or changed based on different views and applications without departing from the spirit of the present application. It should be understood that the preferred embodiments are only for illustration of the present application, not for limitation of the protection scope of the present application.

[0018] It should be noted that the diagrams provided in the following embodiments only illustrate the basic concept of the present application in a schematic manner, and the diagrams only show the components related to the present application, not the number, shape and size of the components during actual implementation. The actual implementation of each component may be a random change, and the component layout pattern may be more complex.

[0019] In the following description, numerous specific details are discussed in order to provide a thorough understanding of embodiments of the present application. However, those skilled in the art will recognize that the embodiments of the present application can be practiced without these specific details. In other instances, well-known structures and devices are not described in detail in order to avoid obscuring embodiments of the present application. Embodiments

[0020] A high line cyclone pool liquid level correction control method based on curve model filtering, as shown in Figure 1 , includes: S1: Collecting cyclone pool liquid level sensing signals, generating a trend curve of liquid level change over time, and dividing the trend curve into four stages of rising, peak value maintaining, falling, and valley value maintaining according to working conditions; S2: Based on normal production conditions, establishing a fixed curve model that completely overlaps and matches the actual liquid level change trend in time interval segmentation and amplitude change, and segmenting and positioning according to the time axis and the amplitude axis to form a model reference; S3: According to the fixed curve model, set a phased deviation band, compare the real-time curve with the model curve, calculate the deviation, and execute different threshold judgments according to the stage; S4: When the deviation exceeds the corresponding stage deviation band, or the liquid level change exceeds the mutation threshold within the preset scanning period window, the section is determined as abnormal; S5: For the abnormal section, use the previous normal value replacement, model prediction value replacement, or trigger process alarm replacement strategy according to the duration to obtain a replacement sequence; S6: Implementing composite filtering processing on the replaced liquid level sequence, and outputting a corrected liquid level sequence.

[0021] Specifically, the high line cyclone pool liquid level curve trend is obtained through the signal acquisition software of the industrial control terminal. The feedback curve generated during operation has the following characteristics: taking time as the X-axis and variable as the Y-axis, a curve can be divided into four stages of rising / falling, maintaining peak value / valley value, falling / rising, and maintaining valley value / peak value. Different stages have their own stability, and the transition between stages generally has a gentle characteristic; the signal acquisition of the industrial control terminal refers to the control process completed by the liquid level sensor in the cyclone pool field, signal transmission cable, signal receiver, data recognition, and trend curve generation unit; Combined with the characteristics of the analog quantity sensor and the program control system, a fixed curve model is established, which completely overlaps and matches the high line cyclone pool liquid level change trend in time interval segmentation and amplitude change during normal rolling process; According to the key node in the model, a threshold is set, when the real-time curve deviates from the threshold to a certain extent, a set alternative value is triggered to filter the abnormal signal, and then the intelligent closed-loop correction of the abnormal signal is realized; The key node in the model refers to the quantized amplitude point corresponding to the time interval and time coordinate based on the characteristics in the full dimension curve of the whole model. The threshold value is set based on the upper and lower limit amplitude interval value of the curve reference value, and then the intelligent identification and intelligent correction of abnormal data and signal can be realized. Triggering a set alternative value means that intelligent triggering and intelligent assignment are realized, and then the intelligent correction closed-loop control of the curve is realized; The process optimization test driven control system refers to the effect test and characteristic test based on the curve model and control model through the quantized signal in the production process, and then the double superposition optimization of the model curve and dynamic process is realized. The high line cyclone pool liquid level signal refers to the dynamic fluctuation value of the cyclone pool liquid level in the normal production rolling process. The Y axis is 4.5, which corresponds to a characteristic pointing point, including time point positioning and amplitude point positioning; Design the process optimization test driven control system, such as high line cyclone pool liquid level signal, in normal production process, the Y axis is 4.5 (pump full stop) around gently rising to 6.0 (pump full open), and then quickly falling to 4.5, the whole curve is a sawtooth curve with stable peaks and valleys. In the non-peak and valley interval, the curve changes are gentle. If the Y axis data mutation exceeds 1.0 in 2 program scanning periods (50ms), it can be judged as an error signal, and the value before 50ms can be used to replace the value; Design and develop filter software, and then run, optimize, debug on the experimental platform to ensure the precision of integrated output; The filter software refers to the intelligent control program that can realize the function of "according to the key node in the model, a threshold is set, when the real-time curve deviates from the threshold to a certain extent, a set alternative value is triggered to filter the abnormal signal". Running, optimizing and debugging on the experimental platform means testing and optimizing again on the non-production platform before actual application, and then realizing the optimization of integrated output control; The actual application in the high line cyclone pool control system means that the control program with abnormal signal intelligent correction and related control components are integrated and applied in the actual production of the high line cyclone pool control process. The key variables in the running process are the liquid level signal, flow signal and pump running signal of the cyclone pool in the normal rolling process; In the actual application of the high line cyclone pool control system, the key variables in the running process are collected and identified, and the deviation early warning control is designed in the program.

[0022] The actual running state and the optimal model state are designed as visual dynamic curves on the program and man-machine operation picture, which is convenient for visual operation and control; The actual running state is the integrated state output of all liquid level signals, flow signals and pump running signals of the cyclone tank in the normal production rolling process and the production gap of the parking inspection. The optimal model state is the integrated state of all liquid level signals, flow signals and pump running signals when the production is smooth and each production process and link is optimally matched. The visual dynamic curve is the continuous identification collection and continuous process control of the variable. The different time interval segmentation parameters are the process control parameter variables matched based on the interval segmentation and time point positioning of the time axis. The auxiliary calibration parameters are the corresponding related process control parameters matched based on the characteristic amplitude of the amplitude axis. The parameter optimization is the matching parameter optimization based on the actual running condition and the dynamic change in the production process.

[0023] The application is further provided with the rising stage corresponding to the liquid level rising after the water pump starts, the peak holding stage corresponding to the stable drainage state, the falling stage corresponding to the liquid level falling after the water pump stops, and the valley holding stage corresponding to the minimum working liquid level. Specifically, the liquid level sensing signal, the pump start-stop state signal, the related valve position and flow signal are collected; deburring and robust smoothing are performed on the liquid level signal to obtain three types of quantitative indications of the liquid level trend, the liquid level change rate and the change rate trend for judgment, so as to align the multi-source signals on the unified time axis of the control system and mark the time at the pump start-stop edge; the inflection point and the platform segment start-stop point of the liquid level signal are automatically marked as stage boundary candidates; the rising stage is entered when the conditions of "pump start signal valid" and "liquid level change rate positive and continuously in the stable window" are met; the termination of the stage is determined by "change rate falling into the stable band" or "entering the platform candidate segment"; the peak holding stage is entered when the conditions of "liquid level change rate absolute value in the stable band and change rate trend limited" are met; the stage is exited when "pump stop signal valid" or "change rate turns negative and continuously in the stable window"; the falling stage is entered when the conditions of "pump stop signal valid" and "liquid level change rate negative and continuously in the stable window" are met; the stage is exited when "change rate absolute value falls into the stable band"; the valley holding stage is entered when the conditions of "liquid level change rate absolute value in the stable band and liquid level in the lower limit neighborhood feature" are met; the stage is exited when "pump starts again" or "change rate turns positive and continuously in the stable window".

[0024] The application is further configured that the reference working condition of the fixed curve model is a sawtooth periodic curve of the liquid level rising from the lower limit of the working liquid level to the upper limit of the working liquid level and then falling, which is used to limit the reference range of the time interval and the amplitude interval of the model. Specifically, the time period when the production line is in stable production is selected, and the liquid level, pump start-stop state, main valve position and flow signals are collected synchronously; the alarm, self-checking, irregular switching and obvious missing segments are removed, and the candidate period with regular start-stop is retained; in a single candidate period, according to the pump start-stop event and the liquid level form, the valley steady state section and the peak steady state section are located; the representative liquid level is extracted from the two types of steady state sections as the lower limit and the upper limit of the working liquid level, and the corresponding entering / exiting time and steady state duration interval are recorded; the key events of “pump start-up—rise completion—pump stop—fall completion” are used for time alignment of multiple periods; the normalization of time scale and amplitude scale is performed on each period to obtain the standardized sawtooth trajectory; the multiple standardized trajectories are robustly aggregated to form the reference sawtooth template; the reference sawtooth template is divided into four stages of rising, peak holding, falling and valley holding; the entering / exiting event, monotonicity requirement, curve smoothness requirement and traceable continuity requirement of the stage boundary are labeled for each stage; according to the reference sawtooth template, the time interval reference range and the amplitude interval reference range are given for each stage in the original dimension: the time interval reference range is used to constrain the shortest / longest duration and the sequence of key events; the amplitude interval reference range is used to limit the amplitude activity area allowed in the stage with the lower limit and the upper limit of the working liquid level as the boundary; accordingly, the fixed curve model is generated, which includes: stage sequence, key event, reference curve, time / amplitude reference interval and form constraint.

[0025] The application is further configured to set a phased deviation band for each stage of the rising, peak holding, falling and valley holding stages according to the fixed curve model in step S3, compare the real-time liquid level curve with the fixed curve model at each moment during operation to obtain the deviation, and apply the corresponding dynamic threshold for judgment according to the stage. Specifically, after completing the fixed curve model and four-stage division, a "relative stage progress" timeline is established for each stage. In operation, the real-time liquid level curve is aligned to the relative progress of the corresponding stage with a unified time scale for subsequent comparison at each moment; for the four stages of rising, peak holding, falling and valley holding, upper and lower boundary curves varying with the relative stage progress are constructed for use as the phased deviation band. The formation of the boundary integrates the form constraint of the reference working condition curve, the stable interval profile of the historical residual, the protection band of the sensor noise and the lag feature of the execution chain, and imposes continuity constraints on boundary smoothing and adjacent stage splicing to ensure no abrupt jump at the stage switching point; within the control sampling period, the liquid level value is read in real time and mapped to the same moment of the fixed curve model, and the difference between the two is calculated as the deviation; at the same time, the direction information and change trend of the deviation are extracted for auxiliary judgment and gating; a dynamic threshold gate is set for each stage to match its stability. The gate is adjusted in conjunction with the current relative stage progress, the deviation change trend, the frequency and persistence of crossing the boundary, and the sensor health indicators: the more stable the stage, the stricter the gate; when in the transition or reversing section, the gate is moderately relaxed to accommodate the expected form transition.

[0026] The application is further configured that, in step S4, the mutation threshold criterion is that, in the preset scanning period window, the absolute value of the difference between adjacent samples of the liquid level signal exceeds the upper limit of the allowed change rate given by the fixed curve model of the current stage; wherein the preset scanning period window is determined by an integral multiple of the scanning period of the control task, and the upper limit of the allowed change rate is determined according to the upper limit of the first derivative of the fixed curve model at the corresponding stage; specifically, the scanning period of the control task is determined, and the preset scanning period window is set by an integral multiple thereof. The first derivatives of the four stages are enveloped and estimated by the fixed curve model, and the upper limit of the allowed change rate of each stage is obtained. The residual distribution of the real-time curve relative to the model curve is calculated according to the historical data to form the outer boundary of the stage deviation band of each stage. Combined with the device lag and process propagation delay, the continuous gate and the entering / leaving hysteresis rule are set to establish the state machine and priority of "attention-warning-abnormality"; the liquid level value and the pump valve state are read at each sampling period, and are mapped to the corresponding stage and relative progress of the fixed curve model; the difference between adjacent samples and its direction are calculated; the residual of the real-time curve relative to the model curve is calculated synchronously, and the sliding window buffer is updated; in the preset scanning period window, the representative measure of the absolute value of the difference between adjacent samples is obtained, and is compared with the upper limit of the allowed change rate of the current stage: if it exceeds and is outside the noise protection band, a "mutation candidate" is generated; the candidate is checked for consistency (continuous at multiple points, stable derivative direction, normal sensor health), and if the checking is passed, a "mutation anomaly" is confirmed, and the abnormal start and end time marks and stage label are output; The criterion for deviation exceeding the deviation band of the corresponding stage is that, in the corresponding stage, the absolute value of the residual of the real-time liquid level relative to the fixed curve model crosses the outer boundary of the deviation band of the stage and continuously meets the gate condition, which is determined as an abnormality; wherein the stage deviation band is determined according to the model residual distribution of the stage, and the continuous gate condition is determined according to the device lag and process propagation delay; specifically, when the absolute value of the residual crosses the outer boundary of the deviation band of the current stage, the continuous gate timing is started; if it falls back to the deviation band before the continuous gate condition is met, it is cleared and recorded as a short-time out-of-bound; if it continuously meets the condition to reach the gate condition, it is confirmed as a "deviation out-of-limit anomaly", and the abnormal start and end time marks and stage label are output; the two types of criteria are executed as "any one meets the criterion is an abnormality". After entering the abnormality, the exit gate of "not mutating and not out-of-limit" is required to be continuously met and the hysteresis release condition is reached, so that the abnormality can be recovered to normal or warning state, avoiding repeated jumping near the boundary.

[0027] The application is further configured that, before performing the hierarchical replacement on the abnormal section, parameters for strategy judgment are calculated for the abnormal section, including: the pullback amount, the phase drift amount, and the logical consistency rate; The pull-back amount is used to represent the comprehensive dimensionless indication of the regression trend and regression speed of the real-time liquid level relative to the fixed curve model after the occurrence of the anomaly, and is obtained according to the degree of monotonous closing-in and the closing-in rate of the real-time liquid level to the model curve in the abnormal section; the trigger of the previous normal value replacement takes the pull-back amount as the decision parameter, and when the pull-back amount meets the time gating and the shape gating conditions linked with the stage information of the fixed curve model, it is determined that the abnormal section uses the previous normal value replacement and writes into the replacement sequence; wherein, the time gating is used to limit the short-time interference scene, and the shape gating is used to ensure that the curve after replacement is monotonous and consistent with the model curve in the stage and boundary continuous. The phase shift amount is used to represent the dimensionless indication of the instantaneous phase misalignment degree of the real-time liquid level change rhythm relative to the fixed curve model in the corresponding stage, and is determined according to the relative sequence and alignment relationship of the feature event sequence of the real-time curve and the model curve in the stage; the trigger of the model prediction replacement takes the phase shift amount as the decision parameter, and when the phase shift amount meets the time gating and the rhythm gating conditions linked with the stage information of the fixed curve model, it is determined that the abnormal section uses the predicted value of the fixed curve model to replace and writes into the replacement sequence; wherein, the rhythm gating is used to maintain the alignment relationship of the event sequence and the traceable continuity of the stage boundary in the stage. The logic consistency rate is used to represent the dimensionless indication of the consistent matching degree of the pump valve working condition switching event and the liquid level derivative direction change in the specified reaction window, and is determined according to the time corresponding relationship and trigger causal relationship of the working condition switching signal and the liquid level derivative sign; the trigger of the process alarm takes the logic consistency rate as the decision parameter, and when the logic consistency rate does not meet the predetermined consistency criterion and the criterion is linked with the stage information of the fixed curve model, it is determined to trigger the process alarm and execute the corresponding protection strategy; wherein, the consistency criterion is used to verify the matching and causal consistency of the pump valve working condition switching event and the liquid level derivative direction change in the specified reaction window.

[0028] Specifically, the liquid level signal, pump start-stop and valve opening degree and other execution chain signals are synchronously acquired from the control system, and are aligned with the stage information of the fixed curve model on the unified time scale; the liquid level signal is deburred and robustly smoothed to obtain the real-time curve, derivative and derivative change trend for judgment. The abnormal candidate section and its stage label and relative stage progress output by step S4 are received; a sliding analysis window is established for the candidate section, and parameter calculation and gating determination are prepared.

[0029] The residual sequence of the real-time curve relative to the fixed curve model is constructed in the abnormal candidate section, and the monotonous closing-in and closing-in rate of the residual are evaluated: when the residual continuously converges to zero and the convergence direction is consistent with the stage monotonicity, the pull-back amount increases; when there is repeated deviation, stagnation or inconsistent direction, the pull-back amount decreases. The result is mapped as a dimensionless indication, and continuity constraints are applied at the stage boundary to avoid shape mutations. In the current stage, the feature event sequence (inflection point, extreme value, platform segment start and end point, etc.) is extracted from the real-time curve and the fixed curve model respectively, and the sequence and alignment relationship of the corresponding events are compared; when there is a stable relative misalignment and the stage monotonicity and boundary continuity requirements are met, the phase drift amount is increased; when the event alignment is good or only short-term disturbance occurs, the phase drift amount is reduced. Missing or ambiguous events are corrected by nearest matching in the local neighborhood of the relative stage progress.

[0030] In the specified reaction window, the pump valve working condition switching event and the liquid level derivative direction change are paired and causally checked: matching and correct sequence are consistent, and opposite direction, no response or no action are inconsistent; the logic consistency rate is determined by the consistent event proportion and the causal sequence consistency. When the execution chain action and the liquid level response are long-term mismatched, the logic consistency rate is reduced.

[0031] Three types of gates are established in conjunction with the stage information of the fixed curve model: time gate: limit the application domain of short-term interference, medium-term deviation and long-term anomaly; shape gate: require the curve change direction to be consistent with the model in the monotonic section, and maintain continuity at the stage boundary; rhythm gate: requires the feature event sequence to maintain traceable alignment relationship within the stage, and suppresses false triggering in the reversing and transition section.

[0032] Previous normal value replacement: only the pullback amount is used as the decision parameter, when the pullback amount meets the time gate and shape gate conditions, it is determined as short-term interference, and the previous normal value replacement is written; model prediction replacement: only the phase drift amount is used as the decision parameter, when the phase drift amount meets the time gate and rhythm gate conditions, it is determined as a phase rhythm misalignment, and the predicted value of the fixed curve model is written; process alarm: only the logic consistency rate is used as the decision parameter, when the logic consistency rate does not meet the consistency criterion and is linked with the stage information, the process alarm is triggered and the protection strategy is executed.

[0033] When the same time window appears and meets the parallel, the arbitration is performed according to the process alarm priority, then the model prediction replacement, and then the previous normal value replacement.

[0034] The application further provides that step S5 is further provided as: For the determined abnormal section, hierarchical replacement is performed according to the duration and stage working condition, wherein the short-time interference is replaced by the previous normal value, the medium-time interference is replaced by the predicted value of the fixed curve model, and the long-term interference triggers a process alarm; and a replacement sequence is obtained. Specifically, when the abnormality is determined as short-time interference by time gating, the latest normal sample before the abnormality is read as a replacement reference, and piecewise previous normal value replacement is performed under the morphology constraint of the current stage; if the abnormality crosses the stage boundary, splicing smoothing is performed at the new stage boundary to ensure sequence continuity; when the abnormality is determined as medium-time interference by time gating, the predicted value of the fixed curve model at the corresponding stage and relative progress is called as the replacement output; in the vicinity of stage reversal or platform turning, the model transition segment of the adjacent progress segment is used for splicing to maintain the alignment relationship of the event sequence and the traceable continuity of the stage boundary; when the abnormality is determined as long-term interference by time gating, a process alarm is triggered, and the trigger reason, stage label and start and end time mark are recorded; during the alarm processing procedure, the value replacement of the section is suspended, and the state information is transmitted to the upper interlocking and safety policy module. The disposal results of each section are written into the replacement sequence in time sequence.

[0035] The application is further provided that, in step S6, composite filtering processing is performed on the replaced liquid level sequence to output a corrected liquid level sequence, wherein the composite filtering processing at least includes inertial smoothing, median suppression and multi-scale denoising. Specifically, the replaced liquid level sequence obtained in step S5 is taken as a filtering input, time alignment is completed according to the stage information and relative stage progress of the fixed curve model, monotonic section and platform section are labeled, and morphology constraint is provided for differential filtering. Three parallel branches of inertial smoothing, median suppression and multi-scale denoising are established: inertial smoothing is used to suppress low-frequency slow fluctuations and maintain overall trend continuity; median suppression is used to eliminate impulse outliers and short spikes; multi-scale denoising is used to reduce multi-frequency high-frequency disturbances and retain event boundaries and inflection points.

[0036] First-order inertial smoothing is performed in the monotonic section, and steady-state maintaining smoothing is performed in the platform section; the time constant is adaptively adjusted according to the stage stability to ensure the consistency of the rising / descending section and the flatness of the platform section. The stage boundary adopts continuous transition to avoid phase mutation; a position-independent robust representative quantity is calculated within a sliding window to replace outliers; the window span is linked with the stage progress, a protection strategy is used in the neighborhood of event inflection points, only significant outliers are suppressed, and the true reversal and key nodes are prevented from being weakened; the input sequence is decomposed in the scale domain, soft suppression and boundary protection are performed on high-frequency components, and the structure information of low-frequency and medium-low-frequency is retained; the suppression strength is reduced in the stage reversal section to ensure the traceability of feature events (inflection points, extreme values, platform start and end).

[0037] Adaptive fusion of three branches output according to phase gating, morphology gating, and rhythm gating: prefer to use median suppression result when pulse is dominant; prefer to use inertia smoothing result when trend slowly drifts; prefer to use multi-scale denoising result when high-frequency noise is dominant. Arbitrate in parallel situation according to the order of pulse priority-scale priority-trend priority, ensure curve continuity and consistent with phase boundary.

[0038] Perform phase delay compensation and amplitude consistency check on the fusion result: use the event sequence of the fixed curve model as an anchor point to align the inflection point and platform boundary; smooth transition at the stage splicing to prevent boundary step and secondary ringing. Output the modified liquid level sequence for closed-loop control and visualization, while recording the branch selection, gating state and fusion identification of each sampling point as the basis for online audit and subsequent parameter gentle update. Embodiment

[0039] The exemplary curve model filtering-based high-line cyclone pool liquid level correction control system is used to implement the curve model filtering-based high-line cyclone pool liquid level correction control method described above, and includes: Signal acquisition module: acquires cyclone pool liquid level sensing signals, generates a trend curve of liquid level change over time, and divides the trend curve into four stages of rising, peak holding, falling, and valley holding according to working conditions; Model establishment module: based on normal production conditions, a fixed curve model that overlaps and matches the actual liquid level change trend in time interval segmentation and amplitude change is established, and segmented and positioned according to the time axis and amplitude axis to form a model reference; Curve comparison module: compare the real-time curve with the model curve according to the stage deviation band set by the fixed curve model, calculate the deviation and perform different threshold judgments according to the stage; Abnormality determination module: when the deviation exceeds the corresponding stage deviation band, or the liquid level change exceeds the mutation threshold within two program scanning periods, the segment is determined as abnormal; Sequence replacement module: replace the abnormal segment with the previous normal value, model prediction value, or trigger a hierarchical replacement strategy for process alarm according to the duration to obtain a replacement sequence; Correction output module: perform composite filtering on the replaced liquid level sequence to output a modified liquid level sequence.

[0040] It should be noted that the high line cyclone pool liquid level correction control system based on curve model filtering provided by the above embodiment and the high line cyclone pool liquid level correction control method based on curve model filtering provided by the above embodiment belong to the same concept, wherein the specific manner in which each module and unit performs operations has been described in detail in the method embodiments, and will not be described here. The high line cyclone pool liquid level correction control system based on curve model filtering provided by the above embodiment can complete the above functions by different functional modules according to the needs in actual application, that is, the internal structure of the system is divided into different functional modules to complete all or part of the functions described above, and this is not limited herein.

[0041] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A high-level vortex pool liquid level correction control method based on curve model filtering, characterized in that, include: S1: Collect the liquid level sensor signal of the vortex pool, generate the trend curve of the liquid level changing over time, and divide the trend curve into four stages according to the working conditions: rising, peak holding, falling, and valley holding. S2: Based on normal production conditions, establish a fixed curve model that overlaps and matches the actual liquid level change trend in terms of time interval segmentation and amplitude change, and segment and position it according to the time axis and amplitude axis to form a model benchmark; S3: Based on the fixed curve model, set a staged deviation band, compare the real-time curve with the model curve, calculate the deviation, and perform different threshold judgments according to the stage. S4: When the deviation exceeds the corresponding stage deviation band, or when the liquid level change exceeds the sudden change threshold within the preset scanning cycle window, the segment is judged as abnormal. S5: For abnormal sections, a graded replacement strategy is adopted based on the duration, using the previous normal value, the model prediction value, or the triggering of a process alarm to obtain a replacement sequence; S6: Perform composite filtering on the replaced liquid level sequence and output the corrected liquid level sequence.

2. The high-level vortex pool liquid level correction control method based on curve model filtering according to claim 1, characterized in that, The rising phase corresponds to the rise in liquid level after the water pump starts, the peak holding phase corresponds to the stable drainage state, the falling phase corresponds to the drop in liquid level after the water pump stops, and the valley holding phase corresponds to the lowest working liquid level.

3. The high-level vortex pool liquid level correction control method based on curve model filtering according to claim 1, characterized in that, The baseline operating condition for the fixed curve model is a sawtooth-shaped periodic curve in which the liquid level rises from the lower limit of the working liquid level to the upper limit of the working liquid level and then falls back. This curve is used to define the reference range of the model's time interval and amplitude interval.

4. The high-level vortex pool liquid level correction control method based on curve model filtering according to claim 3, characterized in that, In step S3, a staged deviation band is set for each stage of rising, peak holding, falling, and valley holding based on the fixed curve model. During operation, the real-time liquid level curve is compared with the fixed curve model at each time step to obtain the deviation, and the corresponding dynamic threshold is applied to determine the deviation according to the stage.

5. The high-level vortex pool liquid level correction control method based on curve model filtering according to claim 1, characterized in that, In step S4, the mutation threshold criterion is: within the preset scanning period window, the absolute value of the difference between adjacent samplings of the liquid level signal exceeds the upper limit of the allowable rate of change given by the fixed curve model in the current stage; wherein, the preset scanning period window is determined by an integer multiple of the scanning period of the control task, and the upper limit of the allowable rate of change is determined based on the upper limit of the first derivative of the fixed curve model in the corresponding stage. The criterion for deviations exceeding the corresponding stage deviation band is as follows: within the corresponding stage, if the absolute value of the residual of the real-time liquid level relative to the fixed curve model crosses the outer boundary of the stage deviation band and continues to reach the gating condition, it is judged as abnormal; whereby the staged deviation band is determined based on the model residual distribution of the stage, and the continuous gating condition is determined based on the equipment lag and process propagation delay.

6. The high-level vortex pool liquid level correction control method based on curve model filtering according to claim 1, characterized in that, Before performing graded replacement on abnormal segments, parameters for strategy determination are calculated for the abnormal segments, including: pullback amount, phase drift amount, and logic consistency rate. The pullback amount is used to characterize the regression trend and regression speed of the real-time liquid level relatively fixed curve model in the corresponding stage after the anomaly occurs. It is obtained by mapping the monotonic convergence degree and convergence rate of the real-time liquid level to the model curve in the abnormal section. Phase drift is a dimensionless indicator used to characterize the degree of instantaneous phase misalignment of a real-time liquid level change rhythm relative to a fixed curve model at a corresponding stage. It is determined based on the relative sequence and alignment of characteristic events between the real-time curve and the model curve at that stage. Logical consistency rate is a dimensionless indicator used to characterize the degree of consistency between pump / valve operating condition switching events and changes in the direction of liquid level derivative within a specified reaction window. It is determined based on the temporal correspondence and triggering causal relationship between the operating condition switching signal and the liquid level derivative sign.

7. The high-level vortex pool liquid level correction control method based on curve model filtering according to claim 6, characterized in that, The trigger for replacing the previous normal value is based on the pullback amount as the decision parameter. When the pullback amount meets the time gating and morphological gating conditions that are linked with the fixed curve model stage information, the abnormal segment is determined to be replaced by the previous normal value and written into the replacement sequence. Among them, the time gating is used to limit short-term interference scenarios, and the morphological gating is used to ensure that the curve after replacement remains monotonically consistent with the model curve and the boundary is continuous within this stage. The model predicts substitution is triggered by the phase drift amount as the decision parameter. When the phase drift amount meets the time gating and rhythm gating conditions that are linked with the stage information of the fixed curve model, the abnormal segment is determined to be replaced by the predicted value of the fixed curve model and written into the substitution sequence. Among them, rhythm gating is used to maintain the alignment relationship of the event sequence and the traceability continuity of the stage boundary within the stage. The process alarm is triggered using the logical consistency rate as the decision parameter. When the logical consistency rate does not meet the predetermined consistency criterion and the criterion is linked with the stage information of the fixed curve model, the process alarm is triggered and the corresponding protection strategy is executed. The consistency criterion is used to verify the matching and causal consistency between the pump valve operating condition switching event and the change in the direction of the liquid level derivative within the specified reaction window.

8. The high-level vortex pool liquid level correction control method based on curve model filtering according to claim 1, characterized in that, Step S5 is also set as follows: For the identified abnormal sections, graded replacements are performed based on the duration and stage of operation. Short-term disturbances are replaced by the previous normal value, medium-term disturbances are replaced by the predicted value of the fixed curve model, and long-term disturbances trigger process alarms; thus, a replacement sequence is obtained.

9. The high-level vortex pool liquid level correction control method based on curve model filtering according to claim 1, characterized in that, In step S6, the replaced liquid level sequence is subjected to composite filtering and a corrected liquid level sequence is output. The composite filtering process includes at least inertial smoothing, median suppression and multi-scale denoising.

10. A high-level vortex pool level correction control system based on curve model filtering, used to implement the high-level vortex pool level correction control method based on curve model filtering as described in any one of claims 1-9, characterized in that, include: Signal acquisition module: Acquires liquid level sensor signals from the cyclone pool, generates a trend curve of liquid level change over time, and divides the trend curve into four stages according to the operating conditions: rising, peak holding, falling, and valley holding. Model building module: Based on normal production conditions, a fixed curve model is built that overlaps and matches the actual liquid level change trend in terms of time interval segmentation and amplitude change. The model is segmented and positioned according to the time axis and amplitude axis to form a model benchmark. Curve comparison module: Based on a fixed curve model, a phased deviation band is set, the real-time curve is compared with the model curve, the deviation is calculated, and different threshold judgments are performed according to the phase. Anomaly detection module: When the deviation exceeds the corresponding stage deviation band, or the liquid level change exceeds the sudden change threshold within two program scan cycles, the segment is judged as abnormal; Sequence Substitution Module: This module employs a tiered substitution strategy based on duration, using the previous normal value, model prediction value, or triggered process alarm to replace abnormal sections, thereby obtaining a substitution sequence. Correction output module: Performs composite filtering on the replaced liquid level sequence and outputs the corrected liquid level sequence.

Images