Water scale generation trend prediction and flow regulation and control system of flow meter

Abstract

The invention discloses a scale generation trend prediction and flow regulation and control system of a flowmeter, particularly relates to the technical field of scale prevention control of an industrial circulating water system, and is used for solving the problems of prediction deviation and regulation and control failure caused by neglecting a synergistic effect of microbial pollution and mineral deposition in the prior art. Water hardness, microbial pollution activeness and flow velocity parameters are acquired in real time through a multi-source parameter acquisition module; the metabolic influence analysis module quantifies influence factors of microbial metabolism on the ion binding rate; the dynamic threshold identification module locates a low-flow-velocity deposition area; the composite prediction modeling module is used for establishing a partition prediction model of the scale generation rate and the microorganism activity degree; the pulse regulation and control execution module drives the valve to execute periodic flow velocity sudden change washing operation; the model iterative optimization module iteratively updates the prediction parameters based on the feedback data; and through multi-factor dynamic correlation analysis and a closed-loop feedback mechanism, high-precision scale trend prediction and targeted flushing regulation and control are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of scale prevention and control of industrial circulating water systems, and more particularly to a scale generation trend prediction and flow control system for a flow meter. Background Art

[0002] In industrial circulating water systems and fluid conveying equipment, flow meters, as core monitoring devices, have long been used to detect the real-time flow rate and flow data of fluids in pipelines, providing basic parameter support for scale prevention and control. In existing technologies, scale deposition is usually suppressed by adjusting valves or pump speeds based on the flow rate data collected by the flow meter, combined with static indicators such as water hardness. However, this type of method only uses the flow meter as a flow rate feedback tool, and does not deeply explore the correlation between its dynamic data and multi-source pollution factors in the pipeline, resulting in insufficient adaptability of the scale prediction model to complex working conditions.

[0003] However, during long-term operation, the flow meter is exposed to fluids contaminated with microorganisms, and its internal sensor unit and pipe connection area are prone to form a composite deposition layer of microbial film and mineral scale. This not only changes the measurement accuracy of the flow meter, but also accelerates the abnormal accumulation of scale due to the metabolic activity of the biofilm, making the traditional control strategy based on flow data ineffective. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a scale formation trend prediction and flow control system for a flow meter to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A scale generation trend prediction and flow control system for a flow meter, comprising:

[0007] Multi-source parameter acquisition module: The built-in sensor group of the flow meter is used to obtain the water hardness parameters, microbial contamination activity parameters and flow rate parameters of the fluid in the pipeline in real time;

[0008] Metabolic impact analysis module: Based on the correlation between microbial contamination activity parameters and water hardness parameters, the module analyzes the influencing factors of microbial membrane metabolic activity on the ion binding rate of scale components;

[0009] Dynamic threshold identification module: Based on the influencing factors and flow rate parameters, the dynamic threshold method is used to identify low-velocity areas in the pipeline where the flow rate is lower than the preset flushing threshold;

[0010] Composite prediction modeling module: Based on the location distribution and ion binding rate of low flow rate areas, a composite prediction model of scale formation rate and microbial contamination activity is established;

[0011] Pulse control execution module: generates pulse flow control parameters based on the output results of the composite prediction model, and drives the valve to perform periodic flow rate mutation operations to flush low flow rate areas;

[0012] Model iterative optimization module: After the periodic flow rate mutation operation is completed, the impact factor is recalculated based on the updated sensor group data, and the composite prediction model is iteratively optimized.

[0013] In a preferred embodiment, the flow meter's built-in sensor group is used to obtain real-time water hardness parameters, microbial contamination activity parameters, and flow rate parameters of the fluid in the pipeline, including:

[0014] The conductivity sensor collects the fluid conductivity data and converts it into water hardness parameters;

[0015] The adenosine triphosphate concentration data of the fluid is detected by a biofilm activity sensor and converted into a microbial contamination activity parameter;

[0016] The flow velocity parameters of the fluid are measured in real time by a flow velocity sensor;

[0017] The water hardness parameters, microbial contamination activity parameters and flow rate parameters are synchronously transmitted through parallel threads and stored in a ring buffer after being marked with timestamps.

[0018] In a preferred embodiment, based on the correlation between the microbial contamination activity parameter and the water hardness parameter, the factors affecting the ion binding rate of scale components by microbial membrane metabolic activity are analyzed, including:

[0019] The correlation coefficient between the microbial contamination activity parameter and the water hardness parameter was calculated through dynamic correlation;

[0020] The dynamic weight factor of the ion binding rate is calculated based on the correlation coefficient and the real-time flow rate parameter;

[0021] The influence factor of microbial membrane metabolic activity on the ion binding rate of scale components is quantified by the ratio of the dynamic weight factor of the ion binding rate to the preset benchmark rate;

[0022] Normalize the impact factors and flow rate parameters to generate input feature vectors;

[0023] The dynamic weight distribution ratio of the influencing factors is modified according to the matching degree between the input feature vector and the historical sedimentary dataset;

[0024] The calculation rule of the dynamic weight factor of the ion binding rate is updated based on the revised dynamic weight distribution ratio.

[0025] In a preferred embodiment, the dynamic correlation relationship is established based on the fluctuation range of the water hardness parameter when the microbial contamination activity parameter is highly active in historical data.

[0026] In a preferred embodiment, based on the influencing factors and the flow rate parameters, a low flow rate area in the pipeline where the flow rate is lower than a preset flushing threshold is identified by a dynamic threshold method, including:

[0027] Dynamically adjust the preset flushing threshold according to the value range of the influencing factor;

[0028] Determine the adaptive adjustment interval of the dynamic threshold method based on the relationship between the fluctuation standard deviation of the real-time flow rate parameter and the preset fluctuation threshold;

[0029] Compare the normalized impact factors and flow rate parameters with the adaptive adjustment interval to generate a coordinate set of the low flow rate area in the pipeline;

[0030] Continuous low velocity areas are screened out according to the spacing density of adjacent coordinates in the coordinate set, and their spatial coverage area is calculated;

[0031] The coordinates and coverage area of ​​the continuous low flow velocity area are stored in the historical scour record database;

[0032] Based on the updated data of the historical scour record database, the adaptive adjustment interval of the dynamic threshold method is recalibrated.

[0033] In a preferred embodiment, the value range of the influencing factor is divided by the correlation between the microbial contamination activity parameter and the water hardness parameter in the historical data.

[0034] In a preferred embodiment, a composite prediction model of scale formation rate and microbial contamination activity is established based on the location distribution and ion binding rate of the low flow rate area, including:

[0035] Integrate the position distribution data of the low flow rate area with the ion binding rate data to generate a multidimensional feature matrix;

[0036] Based on the distribution pattern of microbial contamination activity parameters in the multidimensional feature matrix, high, medium and low activity sub-areas are divided;

[0037] In the high-activity sub-region, a local prediction equation for scale formation rate is constructed through the dynamic correlation between ion binding rate and flow rate parameters.

[0038] In the medium and low activity sub-regions, a global prediction equation is constructed based on the linear relationship between the mean ion binding rate and the flow rate parameter;

[0039] The local prediction equation and the global prediction equation are weightedly fused to generate a composite prediction model, with the weight distribution ratio determined by the correlation coefficient between the spatial coverage area of ​​the sub-region and the historical sedimentation.

[0040] By matching the sub-region division rules with the coordinates of the real-time low-flow-rate area, the corresponding prediction equation is dynamically called to calculate the scale formation rate.

[0041] In a preferred embodiment, the position distribution data is labeled by a pipeline three-dimensional coordinate system, and the ion binding rate data is calibrated by historical deposition experiments.

[0042] In a preferred embodiment, a pulse flow control parameter is generated based on the output of the composite prediction model, and a valve is driven to perform a periodic flow rate mutation operation to flush the low flow rate area, including:

[0043] The scale formation rate output by the composite prediction model is converted into a pulse amplitude parameter, and the conversion rule is based on the linear mapping relationship between the scale formation rate and the preset scouring intensity;

[0044] According to the spatial coverage area of ​​the low velocity area and the historical scour record data, the time interval of the periodic velocity mutation operation is set;

[0045] Generate a valve control signal through pulse amplitude parameters and time interval, the control signal including pulse peak flow rate and duration;

[0046] Drive the valve to execute the control signal, so that the flow rate in the pipeline increases from the baseline value to the pulse peak flow rate within the set time interval and maintains the set duration;

[0047] Adjust the pulse amplitude parameters and time interval based on real-time feedback data on flushing effects. The feedback data is calculated by the rate of change of coverage area in the low velocity area after flushing.

[0048] The adjusted pulse parameters and corresponding flushing effect data are stored in the historical optimization database.

[0049] In a preferred embodiment, after the periodic flow rate mutation operation is completed, the impact factor is recalculated based on the updated sensor group data, and the composite prediction model is iteratively optimized, including:

[0050] The sensor group collects water hardness parameters, microbial contamination activity parameters and flow rate parameters of the fluid in the pipeline after flushing;

[0051] Substitute the updated water hardness parameter and microbial pollution activity parameter into the dynamic correlation equation to recalculate the impact factor;

[0052] Adjust the weight distribution ratio of the composite prediction model based on the recalculated impact factors and the coverage area change rate of the low-velocity area after scouring;

[0053] The adjusted weight distribution ratio is fused with the parameters in the historical optimization database through normalization to generate new model parameters;

[0054] Update the regression coefficients and linear relationship terms in the composite prediction equation according to the new model parameters;

[0055] The updated composite forecast model parameters are stored in the historical optimization database and timestamped for subsequent retrieval.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] 1. Through dynamic correlation analysis of multiple source parameters and a closed-loop feedback optimization mechanism, the accuracy of scale formation trend prediction and the real-time performance of flow control are significantly improved. Based on the dynamic correlation between microbial contamination activity parameters and water hardness, the catalytic effect of biofilm metabolic activity on the binding of scale component ions is quantified as an influencing factor. This overcomes the limitations of traditional methods that rely on static water quality indicators and accurately identifies the synergistic effects of microbial contamination and mineral deposition. A dynamic threshold method combined with adaptive identification of the location distribution of low-flow velocity areas effectively distinguishes between normal and abnormal accumulation areas, avoiding misjudgments caused by localized flow velocity fluctuations. A composite prediction model, based on partitioned modeling (local equations for high-activity sub-regions and global equations for medium and low-activity sub-regions) and a weighted fusion mechanism, dynamically predicts scale formation rates under complex operating conditions, addressing the limited generalization capabilities of a single model. Pulsed flow control and periodic flushing operations directly target low-flow velocity areas, removing complex deposits through sudden changes in flow velocity. Simultaneously, the model iterative optimization module continuously adjusts prediction parameters based on real-time feedback data, forming a closed loop of "perception-prediction-control-optimization" to ensure the long-term stability and reliability of the system.

[0058] 2. Incorporating microbial metabolic activity into the analysis of scale formation mechanisms, a predictive model combining biological and chemical factors is constructed. Through the collaborative design of dynamic thresholds and adaptive control, this approach balances deposition prevention and flow stability, avoiding the surge in energy consumption associated with traditional wide-range valve regulation. The system's modular architecture seamlessly integrates multi-source data, dynamic decision-making, and optimized execution, eliminating the need for complex algorithms or costly hardware modifications. This makes it highly practical and scalable for industrial field deployments. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 This is a structural schematic diagram of a scale generation trend prediction and flow control system for a flow meter of the present invention. DETAILED DESCRIPTION

[0060] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0061] Example: Figure 1 The present invention provides a structural schematic diagram of a flow meter scale generation trend prediction and flow control system, which includes the following modules:

[0062] Multi-source parameter acquisition module: The built-in sensor group of the flow meter is used to obtain the water hardness parameters, microbial contamination activity parameters and flow rate parameters of the fluid in the pipeline in real time;

[0063] Metabolic impact analysis module: Based on the correlation between microbial contamination activity parameters and water hardness parameters, the module analyzes the influencing factors of microbial membrane metabolic activity on the ion binding rate of scale components;

[0064] Dynamic threshold identification module: Based on the influencing factors and flow rate parameters, the dynamic threshold method is used to identify low-velocity areas in the pipeline where the flow rate is lower than the preset flushing threshold;

[0065] Composite prediction modeling module: Based on the location distribution and ion binding rate of low flow rate areas, a composite prediction model of scale formation rate and microbial contamination activity is established;

[0066] Pulse control execution module: generates pulse flow control parameters based on the output results of the composite prediction model, and drives the valve to perform periodic flow rate mutation operations to flush low flow rate areas;

[0067] Model iterative optimization module: After the periodic flow rate mutation operation is completed, the impact factor is recalculated based on the updated sensor group data, and the composite prediction model is iteratively optimized.

[0068] The system obtains water hardness parameters, microbial contamination activity parameters and flow rate parameters in real time through the multi-source parameter acquisition module, and transmits them to the metabolic impact analysis module. The influencing factor of the ion binding rate is calculated based on the dynamic correlation between the microbial contamination activity parameter and the water hardness parameter (such as the Pearson correlation coefficient); the dynamic threshold recognition module combines the influencing factor and flow rate parameter, and identifies the coordinates of the low flow rate area through the dynamic threshold method (adaptively adjusting the preset flushing threshold based on historical data); the composite prediction modeling module integrates the location distribution and ion binding rate data of the low flow rate area to construct a composite prediction model of scale formation rate and microbial activity (such as local linear equation and global weighted fusion); the pulse control execution module generates pulse flow control parameters (such as peak flow rate, duration) according to the model output, and drives the valve to perform periodic flow rate mutation operations; the model iteration optimization module recalculates the influencing factor based on the sensor data updated after flushing (such as the coverage area change rate), and updates the regression coefficient and weight distribution ratio of the composite prediction model by normalizing and fusing the historical optimization database parameters to form a closed-loop feedback.

[0069] The multi-source parameter acquisition module provides the data foundation for the entire system, achieving multi-dimensional parameter synchronization through a sensor group (conductivity, ATP detection, and flow rate sensors). The metabolic impact analysis module quantifies the dynamic impact of microbial metabolic activity on scale formation into calculable influencing factors, addressing the problem of traditional methods ignoring biochemical synergistic effects. The dynamic threshold identification module achieves adaptive identification of low-flow rate areas through the dynamic correlation between flow rate and influencing factors, overcoming misjudgments caused by fixed thresholds. The composite prediction modeling module improves prediction accuracy under complex working conditions through partitioned prediction equations (local equations for high-activity sub-regions and global equations for medium and low activity) and a weighted fusion mechanism. The pulse control execution module converts prediction results into actionable flow rate mutation control signals, achieving precise flushing through valve actuation. The model iterative optimization module continuously corrects model parameters based on real-time flushing effect feedback data to ensure dynamic adaptability of prediction and control. Each module is connected through a closed-loop data flow and control flow, achieving full process automation of "perception-analysis-decision-execution-optimization".

[0070] The specific implementation method of obtaining the water hardness parameters, microbial contamination activity parameters and flow rate parameters of the fluid in the pipeline in real time through the built-in sensor group of the flow meter is as follows:

[0071] The conductivity sensor is installed at the center of the straight section of the flowmeter's inlet pipe to measure the conductivity of the fluid within the pipe. The conductivity sensor has a range of 0 to 2000 microsiemens per centimeter (μS / cm) with an accuracy of ±2%. The method for converting water hardness parameters includes the following steps: During the laboratory calibration phase, calcium carbonate (CaCO3) standard solutions with concentrations of 100 mg / L, 300 mg / L, and 500 mg / L are prepared. The water hardness values ​​are determined using a standard titration method, and the conductivity values ​​are simultaneously measured to establish a linear relationship table between conductivity and water hardness. This linear relationship table is stored in the flowmeter's embedded memory unit. During real-time conversion, the table is queried based on the current conductivity measurement value to obtain the corresponding water hardness parameter (in mg / L). For example, if the conductivity sensor measures 800 μS / cm, the water hardness parameter obtained by querying the table is 320 mg / L. The calibration process is repeated every six months to correct for deviations caused by sensor aging.

[0072] The biofilm activity sensor uses fluorescence detection to measure adenosine triphosphate (ATP) concentration in fluids. The specific steps include: mixing the fluid sample with a freeze-dried fluorescent reagent. The luciferase in the reagent reacts with ATP to produce luciferin, and the fluorescence intensity (unit: relative light units, RLU) is measured using a photomultiplier tube. A standard curve comparing fluorescence intensity and ATP concentration is generated by calibrating ATP solutions of known concentrations over a range of 0 to 20 nanomoles per liter (nmol / L). During conversion, the standard curve is queried based on the current fluorescence intensity value to obtain the ATP concentration. The microbial activity parameter is then assigned according to the following criteria: ATP concentrations below 2 nmol / L are considered low activity, between 2 nmol / L and 10 nmol / L are considered medium activity, and above 10 nmol / L are considered high activity. For example, a fluorescence intensity of 5000 RLU corresponds to an ATP concentration of 5 nmol / L, resulting in a medium activity parameter. The fluorescent reagent is automatically replaced by a robotic arm after every 1000 tests, and the replacement process takes no more than 30 seconds.

[0073] The flow velocity sensor is a turbine-type sensor, installed at the center of the straight section of the flowmeter's outlet pipe, to measure fluid velocity. The rotational frequency of the turbine blades is linearly related to the flow velocity, with a measurement range of 0 to 5 meters per second (m / s) and a measurement frequency of 10 times per second. The raw flow velocity data is processed using a sliding average filter with a filter window size of 10 data points. The calculation method is to take the flow velocity measurements at the current moment and the previous nine moments and calculate the arithmetic mean as the filtered real-time flow velocity parameter (in m / s). For example, if the raw flow velocity sequence is [1.2, 1.3, 1.1, 1.4, 1.2, 1.3, 1.0, 1.5, 1.1, 1.2], the filtered output value is 1.21 m / s. If the standard deviation of 10 consecutive flow velocity data points exceeds 0.1 m / s, the filter window is automatically expanded to 20 data points. For example, when the standard deviation is 0.15 m / s, the window is expanded to 20 data points, and the output value fluctuation range is reduced from ±0.12 m / s to ±0.05 m / s.

[0074] Water hardness parameters, microbial contamination activity parameters, and flow rate parameters are transmitted synchronously through parallel threads. Parallel threads are implemented by creating three independent threads in the flow meter's microcontroller to process the data from the conductivity sensor, biofilm activity sensor, and flow rate sensor, respectively. A mutex mechanism is used between threads to ensure data synchronization. Each parameter is appended with a millisecond timestamp of the system clock when transmitted (for example, the timestamp 1625097600000 represents 00:00:00 on July 1, 2021). Data storage uses a circular buffer structure with a buffer capacity of 1,000 sets of data. Each set of data contains a water hardness parameter (unit: mg / L), a microbial contamination activity parameter (low / medium / high), a flow rate parameter (unit: m / s), and a timestamp. When the buffer is full, new data overwrites the oldest stored data according to the first-in-first-out rule. For example, the 1001th set of data will overwrite the first set of data with a timestamp of 1625097600000.

[0075] Water hardness parameters, microbial contamination activity parameters, and flow rate parameters at the same timestamp are associated via a hash table. The hash table's key is the timestamp, and the value is the storage address of the corresponding parameter. For example, the timestamp 1625097600000 corresponds to the storage address 0x0012FF34. Reading the data at this address yields a water hardness parameter of 320 mg / L, a microbial contamination activity parameter of high activity, and a flow rate parameter of 1.24 m / s. If data from a sensor is lost due to a communication interruption, the system automatically marks the timestamp as invalid and skips it in subsequent processing.

[0076] The conductivity sensor is recalibrated every 6 months using a standard solution. The calibration method is: use calcium carbonate standard solutions with concentrations of 100 mg / L, 300 mg / L, and 500 mg / L, measure their conductivity values ​​respectively, and update the corresponding relationship table between conductivity and water hardness. For example, if the measurement value of the aged conductivity sensor in a 100 mg / L standard solution drifts from the initial 200 μS / cm to 210 μS / cm, the entry for 100 mg / L in the corresponding relationship table is adjusted to 210 μS / cm. The fluorescent reagent of the biofilm activity sensor is replaced after every 1,000 tests. During replacement, the robotic arm automatically removes the old reagent kit and installs the pre-packaged new reagent kit. The replacement process takes no more than 30 seconds.

[0077] The specific implementation method of analyzing the influencing factors of microbial membrane metabolic activity on the ion binding rate of scale components based on the correlation between microbial pollution activity parameters and water hardness parameters is as follows:

[0078] The correlation coefficient between the microbial contamination activity parameter and the water hardness parameter is calculated through a dynamic correlation relationship. The dynamic correlation relationship is established based on the fluctuation range of the water hardness parameter when the microbial contamination activity parameter is highly active in the historical data. The specific method is: screen all time points where the microbial contamination activity parameter is highly active in the historical data, extract the water hardness parameter values ​​at the corresponding moments, calculate the standard deviation and mean of these water hardness parameter values, and generate a fluctuation range. For example, when the microbial contamination activity parameter is highly active, the corresponding mean of the water hardness parameter is 300 mg / L, the standard deviation is 50 mg / L, and the fluctuation range is 250 mg / L to 350 mg / L. The correlation coefficient is calculated using the Pearson correlation coefficient method. The input parameters are the historical data sequence of the microbial contamination activity parameter and the water hardness parameter. The output result is a value between -1 and 1, indicating the degree of linear correlation between the two. For example, when the correlation coefficient is 0.8, it indicates that the microbial contamination activity parameter is strongly positively correlated with the water hardness parameter.

[0079] The correlation coefficient is calculated using the Pearson correlation coefficient method, and the specific formula is:

[0080]

[0081] Among them, A i represents the microbial contamination activity parameter at the i-th sampling moment, in dimensionless grade values ​​(low activity = 0.5, medium activity = 1.0, high activity = 1.5); B i represents the water hardness parameter at the i-th sampling moment, in milligrams per liter (mg / L); The arithmetic mean of the sequence of parameters representing the activity of microbial contamination; represents the arithmetic mean of the water hardness parameter sequence; N represents the number of samples in the historical data sequence, which must satisfy N ≥ 30 to ensure statistical significance; R αβ It represents the linear correlation coefficient between the microbial contamination activity parameter and the water hardness parameter, with a value range of -1 to 1.

[0082] The dynamic weighting factor for the ion binding rate is calculated based on the correlation coefficient and the real-time flow velocity parameter. The dynamic weighting factor is used to characterize the degree of influence of flow velocity on the ion binding rate. The calculation method is as follows: when the real-time flow velocity parameter is less than 1.5 m / s, the dynamic weighting factor is the product of the correlation coefficient and the flow velocity parameter; when the real-time flow velocity parameter is greater than or equal to 1.5 m / s, the dynamic weighting factor is the product of the correlation coefficient and a fixed coefficient of 0.8. For example, when the real-time flow velocity parameter is 1.2 m / s and the correlation coefficient is 0.8, the dynamic weighting factor is 0.8 × 1.2 = 0.96; when the real-time flow velocity parameter is 2.0 m / s, the dynamic weighting factor is 0.8 × 0.8 = 0.64.

[0083] The impact factor of microbial membrane metabolic activity on the ion binding rate of scale components is quantified by the ratio of the dynamic weighting factor of the ion binding rate to a preset baseline rate. The preset baseline rate is the ion binding rate measured in the laboratory under conditions without microbial contamination, for example, a baseline rate of 0.5 mg / (L·h). The impact factor is calculated as: Impact Factor = Dynamic Weighting Factor / Baseline Rate. For example, when the dynamic weighting factor is 0.96, the impact factor is 0.96 / 0.5 = 1.92, indicating that the microbial membrane metabolic activity increases the ion binding rate to 1.92 times the baseline rate.

[0084] The impact factor and flow velocity parameter are normalized to generate the input feature vector. Normalization uses the maximum-minimum method, mapping the impact factor and flow velocity parameter to a range of 0 to 1. The specific method is: Based on historical data, the maximum value of the impact factor is determined to be 3.0 and the minimum value is 0.2; the maximum value of the flow velocity parameter is 5.0 m / s and the minimum value is 0 m / s. The normalization formula is: Normalized value = (original value - minimum value) / (maximum value - minimum value). For example, when the impact factor is 1.92, the normalized value is (1.92-0.2) / (3.0-0.2) = 0.614; when the flow velocity parameter is 1.2 m / s, the normalized value is (1.2-0) / (5.0-0) = 0.24. The input feature vector is a two-dimensional array containing the normalized impact factor and normalized flow velocity parameter, for example, [0.614, 0.24].

[0085] The dynamic weight distribution ratio of the influencing factors is modified based on the degree of match between the input feature vector and the historical sedimentation dataset. The historical sedimentation dataset contains multiple input feature vectors and the corresponding actual scale deposition data. The matching degree is calculated by calculating the Euclidean distance between the current input feature vector and all feature vectors in the dataset, selecting the top 10% with the smallest distance as high-matching samples, and calculating the average dynamic weight factor of these samples as the modified distribution ratio. For example, if the current input feature vector has the smallest distance with 5% of the feature vectors in the dataset, and the average dynamic weight factor of these samples is 0.85, the modified distribution ratio is 0.85.

[0086] The calculation rules for the dynamic weight factor of the ion binding rate are updated based on the revised dynamic weight allocation ratio. This update is performed by replacing the fixed coefficient 0.8 in the dynamic weight factor calculation formula with the revised allocation ratio. For example, when the revised allocation ratio is 0.85, the dynamic weight factor calculation rule for real-time flow rate parameters greater than or equal to 1.5 m / s is updated to: Dynamic weight factor = Correlation coefficient × 0.85. This update is performed every 24 hours to ensure that the weight allocation ratio is adapted to real-time operating conditions.

[0087] Based on the influencing factors and flow rate parameters, the specific implementation method of identifying the low flow rate area in the pipeline where the flow rate is lower than the preset flushing threshold by using the dynamic threshold method is as follows:

[0088] The preset flushing threshold is dynamically adjusted according to the value range of the influencing factor. The value range of the influencing factor is divided by the correlation between the microbial contamination activity parameter and the water hardness parameter in the historical data. The specific method is: in the historical data, all data points with high microbial contamination activity parameters and water hardness parameters higher than 300 mg / L are screened, and the distribution range of the influencing factors of these data points is counted. The lower limit and upper limit of the distribution range are used as the adjustment boundaries of the preset flushing threshold. For example, if the minimum value of the influencing factor in the historical data is 0.5 and the maximum value is 3.0, then the preset flushing threshold is adjusted to 0.5 to 3.0 times the real-time flow rate parameter. When the real-time flow rate parameter is 1.2 m / s, the dynamic range of the preset flushing threshold is 0.6 m / s to 3.6 m / s.

[0089] The adaptive adjustment range of the dynamic threshold method is determined based on the relationship between the standard deviation of the real-time velocity parameter fluctuations and the preset fluctuation threshold. The standard deviation of the real-time velocity parameter fluctuations is calculated using a sliding window method, with the window size consisting of the velocity data from the past hour. The standard deviation is calculated as the square root of the average of the squared deviations of all data points from the mean. The preset fluctuation threshold is set at 20% of the pipeline's design flow rate. For example, for a design flow rate of 5 m / s, the preset fluctuation threshold is 1 m / s. When the standard deviation of the real-time velocity parameter fluctuations exceeds the preset fluctuation threshold, the lower and upper limits of the adaptive adjustment range are each expanded by 20%. When the standard deviation falls below the preset fluctuation threshold, the adaptive adjustment range remains unchanged. For example, when the standard deviation is 1.2 m / s and the preset fluctuation threshold is 1 m / s, the adaptive adjustment range expands from 0.6-3.6 m / s to 0.72-4.32 m / s.

[0090] The normalized influencing factors and flow rate parameters are compared with the adaptive adjustment interval to generate a coordinate set of the low flow rate area in the pipeline. The normalized influencing factors and flow rate parameters are mapped to the range of 0 to 1 by the maximum and minimum method. The comparison method is: if the normalized flow rate parameter is lower than the lower limit of the adaptive adjustment interval, the coordinates of the position are marked as a low flow rate area. For example, when the normalized flow rate parameter is 0.2 (corresponding to the original value 1.0m / s) and the lower limit of the adaptive adjustment interval is 0.3 (corresponding to the original value 1.5m / s), the position is determined to be a low flow rate area. The coordinate set is generated based on the three-dimensional model of the pipeline. The origin of the coordinate system is the center point of the flowmeter inlet, the X-axis is the extension direction of the pipeline, and the Y-axis and Z-axis are radial coordinates.

[0091] Continuous low velocity areas are screened out based on the spacing density of adjacent coordinates in the coordinate set, and their spatial coverage area is calculated. Spacing density is defined as the number of coordinate points per square meter, and the spacing between adjacent coordinates is calculated using the Euclidean distance. When the spacing density is higher than the preset density threshold (e.g. 5 points / m 2 ), it is determined to be a continuous low flow rate area. The spatial coverage area is determined by the convex polygon area calculation method, specifically, the boundary coordinates of the continuous area are connected into a polygon and the polygon area is calculated. For example, when the boundary coordinates of the continuous area are (0,0,0), (2,0,0), (2,1,0), (0,1,0), the coverage area is 2m 2 .

[0092] The coordinates and coverage area of ​​the continuous low flow area are stored in the historical scour record database. The database adopts a relational structure, and each record contains a coordinate set (format: JSON array), coverage area (unit: m 2), timestamp (in milliseconds), and the corresponding adaptive adjustment range. For example, a record might be: {"coordinates":[[0,0,0],[2,0,0],[2,1,0],[0,1,0]],"area":2,"timestamp":1625097600000,"threshold_range":[0.72,4.32]}. The database is stored in the flow meter's embedded memory chip, with a read / write speed of 10MB / s and a storage capacity of 64GB.

[0093] The adaptive adjustment interval of the dynamic threshold method is recalibrated based on updated data from the historical flushing record database. This recalibration method involves extracting the records from the database for the last seven days and calculating the average of the upper and lower limits of the adaptive adjustment interval, which serves as the new adjustment interval baseline. For example, if the average lower limit of the interval for the last seven days is 0.8 and the average upper limit is 4.0, the recalibrated adaptive adjustment interval will be 0.8-4.0. This recalibration is automatically performed every seven days to ensure that the threshold dynamically adapts to the operating conditions.

[0094] The specific implementation method of establishing a composite prediction model of scale formation rate and microbial contamination activity based on the location distribution and ion binding rate of low flow rate areas is as follows:

[0095] The position distribution data of the low flow rate area and the ion binding rate data are integrated to generate a multidimensional feature matrix. The position distribution data is marked by the three-dimensional coordinate system of the pipeline. The origin of the coordinate system is the center point of the flowmeter inlet, the X axis is along the extension direction of the pipeline, and the Y axis and Z axis are radial coordinates. The coordinates of each low flow rate area are recorded in meters (m). For example, the coordinates (2.5, 0.3, 0.3) represent a position 2.5 meters away from the inlet and 0.3 meters radially offset. The ion binding rate data is calibrated through historical deposition experiments. The experimental method is as follows: water of different hardness is injected into the laboratory simulation pipeline, and the microbial contamination activity parameter is controlled to high, medium and low. The scale deposition within 24 hours is measured and the average ion binding rate is calculated in milligrams per liter per hour (mg / (L·h)). For example, when the microbial contamination activity parameter is high activity, the ion binding rate is 1.2mg / (L·h).

[0096] Based on the distribution pattern of the microbial contamination activity parameters in the multidimensional feature matrix, high, medium and low activity sub-areas are divided. The distribution pattern is implemented by the K-means clustering algorithm, the number of clusters is 3, and the input feature is the historical data sequence of the microbial contamination activity parameters. For example, the cluster centers are high activity (mean 1.8), medium activity (mean 1.0), and low activity (mean 0.5), respectively. The sub-area division boundaries are: high activity ≥ 1.5, medium activity 0.8-1.5, and low activity ≤ 0.8. The spatial range of the sub-area is defined by the minimum circumscribed cube in the three-dimensional coordinate system of the pipeline. For example, the high activity sub-area range is 0-5 meters on the X axis, -0.5 to 0.5 meters on the Y axis, and -0.5 to 0.5 meters on the Z axis.

[0097] A local prediction equation for the scale formation rate was constructed within the high-activity subregion using the dynamic correlation between the ion binding rate and the flow velocity parameter. This dynamic correlation was established through multiple linear regression analysis, with the ion binding rate and flow velocity parameter as the independent variables and the scale formation rate as the dependent variable. The regression coefficients were determined by fitting historical data using the least squares method. For example, when the ion binding rate was 1.2 mg / (L·h) and the flow velocity parameter was 0.8 m / s, the local prediction equation for the scale formation rate was: scale formation rate = 0.8 × ion binding rate + 0.2 × flow velocity parameter, resulting in a calculated value of 1.12 mg / (L·h).

[0098] The specific formula for constructing the local prediction equation of scale formation rate in the high activity sub-region through the dynamic correlation between ion binding rate and flow rate parameters is:

[0099] S k =β0+β1C k +β2F k +∈ k

[0100] Among them, S k represents the scale formation rate of the kth sample, in milligrams per liter per hour (mg / (L·h)); C k F represents the ion binding rate of the kth sample, in milligrams per liter per hour (mg / (L·h)); k represents the flow rate parameter of the kth sample, in meters per second (m / s); β0 represents the intercept term of the regression equation, in milligrams per liter per hour (mg / (L·h)); β1 represents the linear influence coefficient of the ion binding rate on the scale formation rate, dimensionless; β2 represents the linear influence coefficient of the flow rate parameter on the scale formation rate, dimensionless; ∈ k The random error term of the k-th sample is expressed in milligrams per liter per hour (mg / (L·h)). The subscript k represents the k-th sample in the historical dataset. The total number of samples must be ≥ 100 to ensure regression significance.

[0101] A global prediction equation is constructed within the medium and low activity sub-regions based on the linear relationship between the mean ion binding rate and the flow velocity parameter. The mean ion binding rate is calculated using all samples in the corresponding sub-region from the historical data. For example, the mean ion binding rate for the medium activity sub-region is 0.9 mg / (L·h), and for the low activity sub-region it is 0.4 mg / (L·h). The linear relationship is: scale formation rate = mean ion binding rate × flow velocity parameter. For example, when the flow velocity parameter is 1.0 m / s, the predicted value for the medium activity sub-region is 0.9 mg / (L·h), and for the low activity sub-region it is 0.4 mg / (L·h).

[0102] In the medium and low activity sub-regions, a global prediction equation is constructed based on the linear relationship between the mean ion binding rate and the flow rate parameter. The specific formula is:

[0103] G j =M j ×V j

[0104] Among them, G j M represents the predicted value of scale formation rate in the j-th sub-area (medium activity or low activity), in milligrams per liter per hour (mg / (L·h)); j represents the historical mean ion binding rate of the j-th subregion, in milligrams per liter per hour (mg / (L·h)), and is calculated as:

[0105]

[0106] Where T j represents the total number of historical samples in the j-th sub-region, C t represents the measured value of the ion binding rate of the tth sample, and the subscript t represents the sample number;

[0107] V j represents the real-time flow velocity parameter of the j-th sub-area, in meters per second (m / s); the subscript j represents the sub-area type (medium activity j=1, low activity j=2).

[0108] A composite prediction model is generated by weighted fusion of the local and global prediction equations. The weighting ratio is determined by the correlation coefficient between the spatial coverage area of ​​the subregion and the historical sedimentation. The correlation coefficient is calculated using the Pearson correlation coefficient method, with the input being the subregion area sequence and the historical sedimentation sequence. For example, if the correlation coefficient between the area and sedimentation of a high-activity subregion is 0.7, its weight ratio is 0.7; if the correlation coefficients of the medium and low-activity subregions are 0.3 and 0.1, respectively, their weight ratios are 0.3 and 0.1, respectively. The weighted fusion formula is: composite prediction value = 0.7 × local prediction value + 0.3 × global prediction value.

[0109] The scale formation rate is calculated by dynamically calling the corresponding prediction equation based on the coordinates of the real-time low-flow area and the sub-region division rules. The matching method is as follows: the coordinates of the real-time low-flow area are compared with the spatial extent of the sub-region. If the coordinates are within the high-activity sub-region, the local prediction equation is called; if they are within the medium or low-activity sub-region, the global prediction equation is called. For example, if the real-time coordinates are (3.0, 0.2, 0.1) and are within the high-activity sub-region, the local prediction equation is used; if the coordinates are (6.0, 0.6, 0.6) and are within the medium-activity sub-region, the global prediction equation is used.

[0110] The specific implementation method of generating pulse flow control parameters based on the output results of the composite prediction model and driving the valve to perform periodic flow rate mutation operation to flush the low flow rate area is as follows:

[0111] The scale generation rate output by the composite prediction model is converted into a pulse amplitude parameter. The conversion rule is based on a linear mapping relationship between the scale generation rate and the preset scouring intensity. Specifically, the preset scouring intensity is 1.5 times the scale generation rate, and the pulse amplitude parameter is calculated by multiplying the scale generation rate by a proportional coefficient of 0.8. For example, when the scale generation rate output by the composite prediction model is 1.0 mg / (L·h), the pulse amplitude parameter is 1.0×0.8=0.8 m / s. The proportional coefficient is calibrated using historical scouring effect data. The calibration method is: the scouring efficiency under different proportional coefficients is tested in the laboratory, and the coefficient value that reduces the deposition by 50% is selected as the optimal value.

[0112] The time interval of periodic velocity mutation operation is set according to the spatial coverage area of ​​the low velocity area and the historical flushing record data. The historical flushing record data includes the coverage area of ​​the low velocity area and the corresponding time interval after each flushing operation. The setting rule is: when the coverage area of ​​the low velocity area is greater than 1.2 times the historical average, the time interval is shortened by 20%; when the coverage area is less than 0.8 times the historical average, the time interval is extended by 20%. For example, the historical average coverage area is 2.0m 2 , the current coverage area is 2.5m 2 , the time interval is shortened from 60 minutes to 48 minutes.

[0113] The valve control signal is generated using the pulse amplitude parameter and the time interval. The control signal consists of a pulse peak velocity and duration. The peak velocity is the sum of the baseline velocity and the pulse amplitude parameter, and the duration is set to 50% of the time interval. For example, if the baseline velocity is 1.0 m / s and the pulse amplitude parameter is 0.8 m / s, the peak velocity is 1.8 m / s; if the time interval is 48 minutes, the duration is 24 minutes. The control signal is output as an analog voltage signal with a voltage range of 0-10V, corresponding to a flow rate range of 0-5 m / s.

[0114] The valve drives the control signal, increasing the flow rate in the pipeline from the baseline value to the pulse peak flow rate within a set time interval and maintaining it for the set duration. The valve is an electric control valve with a response time of 5 seconds and a control accuracy of ±0.1 m / s. For example, if the baseline flow rate is 1.0 m / s, the valve will increase the flow rate to 1.8 m / s within 5 seconds, maintain it for 24 minutes, and then return it to the baseline value within 5 seconds.

[0115] Adjust the pulse amplitude parameters and time interval based on the real-time feedback data of the flushing effect. The feedback data is calculated by the coverage area change rate of the low flow area after flushing. The change rate formula is (original coverage area - coverage area after flushing) / original coverage area × 100%. The adjustment rule is: when the change rate is less than 20%, the pulse amplitude parameter is increased by 10%; when the change rate is higher than 40%, the time interval is extended by 15%. For example, the original coverage area is 2.5m 2 , after scouring, it is 1.8m 2 , the change rate is 28%, and the pulse amplitude parameter increases from 0.8m / s to 0.88m / s.

[0116] The adjusted pulse parameters and corresponding flushing effect data are stored in a historical optimization database. The database uses a time-series data storage structure. Each record contains the pulse amplitude parameter (unit: m / s), time interval (unit: minutes), coverage area change rate (unit: %), and timestamp (unit: milliseconds). For example, a record is: {"pulse_amp":0.88,"interval":48,"change_rate":28,"timestamp":1625097600000}. The database is implemented using MySQL 8.0, with the timestamp index field, and the query response time is less than 100 milliseconds.

[0117] After the periodic flow rate mutation operation is completed, the impact factor is recalculated based on the updated sensor group data, and the iterative optimization composite prediction model is specifically implemented as follows:

[0118] The sensor group collects water hardness, microbial activity, and flow rate parameters for the fluid in the flushed pipe. The sensor group includes a conductivity sensor, a biofilm activity sensor, and a flow rate sensor. The sampling frequency is once per second, with 10 seconds of data collected and the average value taken as the effective value. For example, the conductivity sensor collects a value of 810 μS / cm, corresponding to a water hardness parameter of 324 mg / L; the biofilm activity sensor detects an ATP concentration of 6.2 nmol / L, corresponding to a medium microbial activity parameter; and the flow rate sensor measures a value of 1.3 m / s.

[0119] Substitute the updated water hardness parameter and microbial contamination activity parameter into the dynamic correlation equation to recalculate the impact factor. The dynamic correlation equation is a linear equation with a slope coefficient determined by fitting historical data. The input is the water hardness parameter (unit: mg / L) and the microbial contamination activity parameter (dimensionless level), and the output is the impact factor (dimensionless). For example, when the water hardness parameter is 324 mg / L and the microbial contamination activity parameter is medium activity (level value 1.0), the calculation formula is: Impact factor = 0.003 × water hardness parameter + 0.5 × activity level value, and the calculated result is 1.47.

[0120] The weight distribution ratio of the composite prediction model is adjusted based on the recalculated impact factors and the coverage area change rate of the low velocity area after scouring. The coverage area change rate is calculated by the difference in the area of ​​the low velocity area before and after scouring. For example, the area before scouring is 2.5m 2 , after scouring, it is 1.8m 2 The change rate is 28%. The adjustment rule is: for every 0.1 increase in the impact factor, the weight allocation ratio increases by 2%; for every 5% increase in the coverage area change rate, the weight allocation ratio increases by 3%. For example, if the impact factor is 1.47 and the change rate is 28%, the weight allocation ratio is adjusted to the original value of 1.47 × 2% + 28 / 5 × 3% = 5.4% + 16.8% = 22.2%.

[0121] The adjusted weight distribution ratio is fused with the parameters in the historical optimization database through normalization to generate new model parameters. Normalization uses the maximum-minimum method to map the weight distribution ratio and historical parameters to the range of 0 to 1. For example, if the adjusted weight is 22.2% and the historical average is 18%, the normalized value is (22.2-10) / (30-10)=0.61 (assuming the historical minimum is 10% and the maximum is 30%). The fusion method is the weighted average method, and the new parameter = 0.7×normalized value + 0.3×historical average. For example, if the historical average is 0.6, the new parameter = 0.7×0.61+0.3×0.6=0.607.

[0122] The regression coefficients and linear relationship terms in the composite forecast equation are updated based on the new model parameters. The regression coefficients are refitted to historical data using the least squares method, with the past seven days of historical data used as the training set for each update. For example, if the original coefficient of the local forecast equation is 0.8 and the new parameter is 0.607, the coefficient is adjusted to 0.8 × 0.607 / 0.7 ≈ 0.7 (to two decimal places) after fitting the historical data for that month. The slope of the linear relationship term is scaled proportionally to the new parameter value. For example, if the original slope of the global forecast equation is 1.0, the adjusted slope becomes 1.0 × 0.607 = 0.607.

[0123] The updated composite forecast model parameters are stored in the historical optimization database and timestamped for subsequent access. The database uses a time series structure for storage, with each record containing the parameter name, parameter value (unit: dimensionless), timestamp (accurate to milliseconds), and operation type label. For example, a record might be: {"param":"local_coeff","value":0.70,"timestamp":1625097600000,"type":"update"}. The database indexes the timestamp and parameter name, supporting fast searches by time period or parameter type.

[0124] This embodiment integrates the multi-source sensor data (water hardness, microbial activity and flow rate parameters) built into the flow meter to construct a dynamic correlation model of the combination of microbial metabolic activity and scale component ions, and incorporates the chemical impact of biofilm metabolites on local water quality into the quantitative analysis of scale formation rate, rather than simply superimposing independent parameters. In traditional methods, flow rate regulation is only based on setting a fixed threshold based on historical experience. However, this embodiment uses a dynamic threshold method combined with real-time flushing effect feedback to achieve adaptive optimization of low flow rate area identification, breaking through the limitations of human experience. In addition, the closed-loop feedback mechanism introduced in the iterative optimization process of the model (such as the dynamic mapping of the coverage area change rate after flushing and the weight distribution) enables the prediction model to continuously track changes in the deposition state in the pipeline, rather than relying on initial calibration data. This technical path of multi-dimensional data fusion and dynamic adaptation overcomes the lack of applicability of traditional methods under complex working conditions, and all technical features are implemented based on sensors and control units that can be deployed on industrial sites, ensuring the engineering feasibility of the solution.

[0125] The calculations involved in the embodiments are all dimensionless numerical calculations, and the preset parameters and thresholds in the calculations are set by those skilled in the art according to actual conditions.

[0126] The above embodiments may be implemented in whole or in part through software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments may be implemented in whole or in part in the form of a computer program product.

[0127] Those skilled in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application of the technical solution and the invention constraints. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0128] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0129] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the modules is only a logical function division. In actual implementation, there may be other division methods, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0130] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0131] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A flow meter scale generation trend prediction and flow control system, characterized in that: include: Multi-source parameter acquisition module: The built-in sensor group of the flow meter is used to obtain the water hardness parameters, microbial contamination activity parameters and flow rate parameters of the fluid in the pipeline in real time; Metabolic impact analysis module: Based on the correlation between microbial contamination activity parameters and water hardness parameters, the module analyzes the influencing factors of microbial membrane metabolic activity on the ion binding rate of scale components; Dynamic threshold identification module: Based on the influencing factors and flow rate parameters, the dynamic threshold method is used to identify low-velocity areas in the pipeline where the flow rate is lower than the preset flushing threshold; Composite prediction modeling module: Based on the location distribution and ion binding rate of low flow rate areas, a composite prediction model of scale formation rate and microbial contamination activity is established; Pulse control execution module: generates pulse flow control parameters based on the output results of the composite prediction model, and drives the valve to perform periodic flow rate mutation operations to flush low flow rate areas; Model iterative optimization module: After the periodic flow rate mutation operation is completed, the impact factor is recalculated based on the updated sensor group data, and the composite prediction model is iteratively optimized.

2. A flow meter scale generation trend prediction and flow control system according to claim 1, characterized in that: The built-in sensor group of the flow meter can obtain the water hardness parameters, microbial contamination activity parameters and flow rate parameters of the fluid in the pipeline in real time, including: The conductivity sensor collects the fluid conductivity data and converts it into water hardness parameters; The adenosine triphosphate concentration data of the fluid is detected by a biofilm activity sensor and converted into a microbial contamination activity parameter; The flow velocity parameters of the fluid are measured in real time by a flow velocity sensor; The water hardness parameters, microbial contamination activity parameters and flow rate parameters are synchronously transmitted through parallel threads and stored in a ring buffer after being marked with timestamps.

3. The scale generation trend prediction and flow control system of a flow meter according to claim 1, characterized in that: Based on the correlation between microbial contamination activity parameters and water hardness parameters, the factors affecting the ion binding rate of scale components by microbial membrane metabolic activity are analyzed, including: The correlation coefficient between the microbial contamination activity parameter and the water hardness parameter was calculated through dynamic correlation; The dynamic weight factor of the ion binding rate is calculated based on the correlation coefficient and the real-time flow rate parameter; The influence factor of microbial membrane metabolic activity on the ion binding rate of scale components is quantified by the ratio of the dynamic weight factor of the ion binding rate to the preset benchmark rate; Normalize the impact factors and flow rate parameters to generate input feature vectors; The dynamic weight distribution ratio of the influencing factors is modified according to the matching degree between the input feature vector and the historical sedimentary dataset; The calculation rule of the dynamic weight factor of the ion binding rate is updated based on the revised dynamic weight distribution ratio.

4. A flow meter scale generation trend prediction and flow control system according to claim 3, characterized in that: The dynamic correlation relationship is established based on the fluctuation range of water hardness parameters when the microbial contamination activity parameters are highly active in historical data.

5. The scale generation trend prediction and flow control system of a flow meter according to claim 1, characterized in that: Based on the influencing factors and flow rate parameters, the dynamic threshold method is used to identify low-velocity areas in the pipeline where the flow rate is lower than the preset flushing threshold, including: Dynamically adjust the preset flushing threshold according to the value range of the influencing factor; Determine the adaptive adjustment interval of the dynamic threshold method based on the relationship between the fluctuation standard deviation of the real-time flow rate parameter and the preset fluctuation threshold; Compare the normalized impact factors and flow rate parameters with the adaptive adjustment interval to generate a coordinate set of the low flow rate area in the pipeline; Continuous low velocity areas are screened out according to the spacing density of adjacent coordinates in the coordinate set, and their spatial coverage area is calculated; The coordinates and coverage area of ​​the continuous low flow velocity area are stored in the historical scour record database; Based on the updated data of the historical scour record database, the adaptive adjustment interval of the dynamic threshold method is recalibrated.

6. A scale generation trend prediction and flow control system for a flow meter according to claim 5, characterized in that: The value range of the influencing factor is divided by the correlation between the microbial contamination activity parameter and the water hardness parameter in historical data.

7. The scale generation trend prediction and flow control system of a flow meter according to claim 1, characterized in that: Based on the location distribution and ion binding rate of low flow rate areas, a composite prediction model of scale formation rate and microbial contamination activity was established, including: Integrate the position distribution data of the low flow rate area with the ion binding rate data to generate a multidimensional feature matrix; Based on the distribution pattern of microbial contamination activity parameters in the multidimensional feature matrix, high, medium and low activity sub-areas are divided; In the high-activity sub-region, a local prediction equation for scale formation rate is constructed through the dynamic correlation between ion binding rate and flow rate parameters. In the medium and low activity sub-regions, a global prediction equation is constructed based on the linear relationship between the mean ion binding rate and the flow rate parameter; The local prediction equation and the global prediction equation are weightedly fused to generate a composite prediction model, with the weight distribution ratio determined by the correlation coefficient between the spatial coverage area of ​​the sub-region and the historical sedimentation. By matching the sub-region division rules with the coordinates of the real-time low-flow-rate area, the corresponding prediction equation is dynamically called to calculate the scale formation rate.

8. The scale generation trend prediction and flow control system of a flow meter according to claim 7, characterized in that: The position distribution data are marked by the pipeline three-dimensional coordinate system, and the ion binding rate data are calibrated by historical deposition experiments.

9. The scale generation trend prediction and flow control system of a flow meter according to claim 1, characterized in that: Based on the output of the composite prediction model, pulse flow control parameters are generated to drive the valve to perform periodic flow rate mutation operations to flush low flow rate areas, including: The scale formation rate output by the composite prediction model is converted into a pulse amplitude parameter, and the conversion rule is based on the linear mapping relationship between the scale formation rate and the preset scouring intensity; According to the spatial coverage area of ​​the low velocity area and the historical scour record data, the time interval of the periodic velocity mutation operation is set; Generate a valve control signal through pulse amplitude parameters and time interval, the control signal including pulse peak flow rate and duration; Drive the valve to execute the control signal, so that the flow rate in the pipeline increases from the baseline value to the pulse peak flow rate within the set time interval and maintains the set duration; Adjust the pulse amplitude parameters and time interval based on real-time feedback data on flushing effects. The feedback data is calculated by the rate of change of coverage area in the low velocity area after flushing. The adjusted pulse parameters and corresponding flushing effect data are stored in the historical optimization database.

10. The scale generation trend prediction and flow control system of a flow meter according to claim 1, characterized in that: After the periodic flow rate mutation operation is completed, the impact factors are recalculated based on the updated sensor group data, and the composite prediction model is iteratively optimized, including: The sensor group collects water hardness parameters, microbial contamination activity parameters and flow rate parameters of the fluid in the pipeline after flushing; Substitute the updated water hardness parameter and microbial pollution activity parameter into the dynamic correlation equation to recalculate the impact factor; Adjust the weight distribution ratio of the composite prediction model based on the recalculated impact factors and the coverage area change rate of the low flow velocity area after scouring; The adjusted weight distribution ratio is fused with the parameters in the historical optimization database through normalization to generate new model parameters; Update the regression coefficients and linear relationship terms in the composite prediction equation according to the new model parameters; The updated composite forecast model parameters are stored in the historical optimization database and timestamped for subsequent retrieval.

Images