Precision evaluation method for quantitative pumps driven by servo motors

Abstract

The present invention relates to the field of quantitative pump accuracy assessment, and in particular to a quantitative pump accuracy assessment method based on a servo motor-driven quantitative pump. The method constructs flow time series data by continuously sampling the flow output of the servo motor-driven quantitative pump, and performs window division to obtain a quantitative pump flow time series data window; obtains a bubble disturbance assessment factor at a target moment based on the bubble disturbance structural characteristics in the quantitative pump flow time series data window and the output characteristics of the quantitative pump under current working conditions; obtains the bubble disturbance degree of the quantitative pump at the target moment based on the fluctuation difference between the head and tail regions in the quantitative pump flow time series data window and the bubble disturbance assessment factor at the target moment; and optimizes the error index of the quantitative pump flow time series data using the bubble disturbance degree of the quantitative pump at the target moment, thereby obtaining a dynamic precision evaluation result that can reflect the influence of the bubble disturbance, thereby improving the precision of the quantitative pump accuracy assessment.

Description

Technical Field

[0001] The present invention relates to the field of quantitative pump accuracy assessment, and in particular to a quantitative pump accuracy assessment method based on a servo motor driven quantitative pump. Background Art

[0002] With the continuous development of servo control technology and intelligent manufacturing systems, servo motor-driven metering pumps are widely used in high-precision liquid delivery scenarios, such as pharmaceutical liquid filling, high-purity chemical reagent ratio, fine chemical process control and other fields. The collaboration between metering pumps and servo motors can achieve high-precision control of fluid output volume, which is particularly suitable for production lines with extremely low tolerance for unit filling errors. Compared with traditional pneumatic or open-loop motor drive systems, servo systems have high responsiveness and strong robustness, and can dynamically adjust the output flow through closed-loop feedback, greatly improving the control performance of the system. In the filling system, the time it takes for a metering pump to complete a filling cycle is usually hundreds of milliseconds. The system needs to monitor the output flow in real time and use this as the basis for evaluating control accuracy and anomaly identification. To this end, building a reasonable filling accuracy evaluation method has become a key link in achieving high-quality manufacturing.

[0003] A commonly used accuracy assessment method in current filling systems is the sliding window-based instantaneous flow error identification algorithm. This algorithm primarily compares a real-time flow data series with a target filling rate model, evaluating the magnitude and trend of the instantaneous error within a small time window to identify potential output instability. The basic process is as follows: First, the system acquires instantaneous flow data at a high-frequency sampling rate (e.g., 1kHz) during the filling cycle to form a complete flow time series. This sequence is then traversed frame by frame using a sliding window, and within each window, the error between the current flow value and the target flow model is calculated (e.g., using mean squared error or maximum absolute error). The calculated results are used to identify any abnormal fluctuations within the current window that deviate from the target model. If the error exceeds a preset threshold for multiple consecutive windows, a "potential filling cycle with inaccuracy" is identified. This algorithm offers the advantages of simple implementation and fast response in control systems. It can provide a preliminary, structured assessment of flow behavior within a cycle, making it a widely used fast accuracy assessment module in industrial metering pump systems.

[0004] In the practical application of servo motor-driven metering pumps, especially during high-speed, high-frequency liquid metering filling, transient disturbances in the pump's operating state are closely related to liquid flow behavior. Liquid disturbances, residual gas in the pipeline, pressure fluctuations, or reversal lags at the pump's suction end can cause bubbles to enter the fluid circuit. These bubble disturbances can disrupt the original flow stability, manifesting as transient flow collapse, rebound reflux, and micro-periodic oscillations during the filling process.

[0005] However, existing quantitative pump accuracy assessment methods primarily rely on error statistics within a sliding window (such as mean square error and maximum error) for static judgment. These methods lack the ability to identify dynamic disturbances in flow time series, particularly when disturbances are short-lived and the overall filling volume remains within a reasonable range. This makes it impossible to identify hidden structural errors. More seriously, the system may also contain a large number of non-bubble disturbances (such as pressure disturbances caused by suction reversal, rebound between pump control intervals, and transient compression of the liquid). These behaviors may resemble bubble disturbances in terms of flow rate variation, but they do not fundamentally affect the actual accuracy of the pump. Existing methods can easily misjudge these as anomalies, resulting in distorted accuracy assessments.

[0006] Therefore, how to effectively identify the bubble disturbance behavior in the flow time series during the filling process of the metering pump, accurately distinguish between real disturbance and non-disturbance behavior, and reasonably introduce the disturbance characteristics into the accuracy evaluation process, thereby constructing a set of metering pump accuracy evaluation methods with dynamic recognition capabilities, strong scene adaptability, and reliable evaluation results, has become an important problem that needs to be solved urgently in current technology. Summary of the Invention

[0007] In view of this, an embodiment of the present invention provides a method for evaluating the accuracy of a metering pump driven by a servo motor, so as to solve the problem of inaccurate accuracy evaluation caused by bubble disturbance during the accuracy evaluation of the metering pump driven by the servo motor.

[0008] An embodiment of the present invention provides a method for evaluating the accuracy of a metering pump driven by a servo motor, the method comprising the following steps:

[0009] By continuously sampling the flow output of the quantitative pump driven by the servo motor, flow time series data for quantitative pump accuracy evaluation is constructed, and the flow time series data is windowed to obtain the flow time series data window of the quantitative pump;

[0010] Obtaining a bubble disturbance evaluation factor at a target moment based on the bubble disturbance structural characteristics in the flow time series data window of the metering pump and the output characteristics of the metering pump under the current working condition;

[0011] Obtaining the bubble disturbance degree of the dosing pump at the target moment according to the fluctuation difference between the head and tail regions in the dosing pump flow time series data window and the bubble disturbance evaluation factor at the target moment;

[0012] The error index of the flow time series data of the dosing pump is optimized by using the bubble disturbance degree of the dosing pump at the target moment, and a dynamic accuracy evaluation result that can reflect the influence of the bubble disturbance is obtained;

[0013] The step of obtaining a bubble disturbance evaluation factor at a target moment based on the bubble disturbance structural characteristics in the flow time series data window of the metering pump and the output characteristics of the metering pump under the current working condition includes:

[0014] A trend expression equation at the target moment is obtained by performing least squares fitting on a time series data window of the metering pump flow rate at the target moment, and the difference between the prediction result of the trend expression equation at the target moment and the monitoring data of the metering pump flow rate at the target moment is used as a trend deviation assessment at the target moment; and a calculation result of dividing the trend deviation assessment at the target moment by the flow rate mean of the time series data window of the metering pump flow rate at the target moment is used as the trend deviation degree at the target moment;

[0015] The local collapse strength at the target moment is obtained by performing a cumulative difference evaluation on the flow change data in the time series data window of the quantitative pump flow at the target moment;

[0016] By performing a cumulative evaluation of the flow rate change information in the flow time series data window of the quantitative pump at the target moment, a disturbance diffusion inertia evaluation at the target moment is obtained;

[0017] The bubble disturbance assessment factor at the target moment is obtained by performing nonlinear compression fusion analysis on the trend deviation degree, local collapse intensity and disturbance diffusion inertia assessment at the target moment;

[0018] The step of obtaining the bubble disturbance degree of the quantitative pump at the target moment according to the fluctuation difference between the head and tail regions in the quantitative pump flow time series data window and the bubble disturbance evaluation factor at the target moment includes:

[0019] Obtain the second-order difference data of the quantitative pump flow time series data window at the target moment, and use the second-order difference data at any moment in the quantitative pump flow time series data window at the target moment as the disturbance intensity weight at that moment;

[0020] In the quantitative pump flow time series data window at the target moment, the data point with the highest disturbance intensity weight is used as the disturbance center estimation point at the target moment;

[0021] The number of data points of half the window length of the disturbance center estimation point at the target moment is used as the disturbance center area at the target moment; the area to the left of the disturbance center area in the quantitative pump flow time series data window at the target moment is used as the head area at the target moment; the area to the right of the disturbance center area in the quantitative pump flow time series data window at the target moment is used as the tail area at the target moment;

[0022] The bubble disturbance degree of the metering pump at the target moment is obtained by evaluating the fluctuation difference between the head region and the tail region at the target moment and the bubble disturbance evaluation factor at the target moment.

[0023] Preferably, the local collapse strength at the target moment is obtained by performing cumulative difference evaluation on the flow change data in the quantitative pump flow time series data window at the target moment, including:

[0024] For the target moment's metering pump flow timing data window, starting from the flow data at the second moment in the timing data window, the calculation result of subtracting the flow data at each moment from the flow data at the previous moment is used as the flow difference evaluation at that moment; for any moment in the target moment's metering pump flow timing data window, if the flow difference evaluation at that moment is less than 0, the flow difference evaluation at that moment is used as the decline amplitude evaluation at that moment, otherwise the value 0 is used as the decline amplitude evaluation at that moment; the calculation result of adding up the decline amplitude evaluations of all moments in the target moment's metering pump flow timing data window is used as the local collapse strength at the target moment.

[0025] Preferably, the evaluation of the disturbance diffusion inertia at the target moment is obtained by cumulatively evaluating the flow rate change information in the flow rate time series data window of the quantitative pump at the target moment, including:

[0026] For the flow difference evaluation at each moment in the quantitative pump flow timing data window at the target moment, starting from the third moment, the absolute value of the calculation result of subtracting the flow difference evaluation at each moment from the flow difference evaluation at the previous moment is used as the flow change rate difference evaluation at that moment; the average of all the flow change rate difference evaluations in the quantitative pump flow timing data window at the target moment is used as the disturbance diffusion inertia evaluation at the target moment.

[0027] Preferably, the bubble disturbance assessment factor at the target moment is obtained by performing nonlinear compression fusion analysis on the trend deviation degree, local collapse intensity and disturbance diffusion inertia assessment at the target moment, including:

[0028] Obtaining the trend deviation degree, local collapse intensity and disturbance diffusion inertia assessment at the target moment;

[0029] The trend deviation degree, local collapse intensity and disturbance diffusion inertia assessment at the target moment are added to a constant 1 and logarithmically mapped to calculate the results as the first assessment factor, the second assessment factor and the third assessment factor at the target moment;

[0030] The result of multiplying the first evaluation factor, the second evaluation factor, and the third evaluation factor at the target moment is used as the numerator, the square of the result of adding the flow rate mean value of the quantitative pump flow rate time series data window at the target moment to a set minimum positive number is used as the denominator, and the resulting fraction is used as the first bubble disturbance evaluation factor at the target moment;

[0031] A linear normalized mapping result of the first bubble disturbance evaluation factor at the target moment is used as the bubble disturbance evaluation factor at the target moment.

[0032] Preferably, obtaining the bubble disturbance degree of the metering pump at the target moment by evaluating the fluctuation difference between the head region and the tail region at the target moment and the bubble disturbance evaluation factor at the target moment includes:

[0033] Obtaining a set smoothing adjustment factor; subtracting the numerical variance of the head region at the target moment from the numerical variance of the tail region at the target moment as a calculation result of the fluctuation difference evaluation between the head region and the tail region at the target moment;

[0034] The maximum value between the fluctuation difference evaluation of the head region and the tail region at the target moment and a constant of 0 is used as the first confidence evaluation factor of the target moment;

[0035] Multiplying the first confidence assessment factor at the target moment by the set smoothing adjustment factor and taking the negative of the result as the second confidence assessment factor at the target moment;

[0036] Performing exponential mapping on the second confidence evaluation factor at the target moment with a natural constant as the base, and using the corresponding exponential mapping result as the bubble disturbance confidence optimization factor at the target moment;

[0037] The result of multiplying the bubble disturbance confidence optimization factor at the target moment by the bubble disturbance evaluation factor at the target moment is used as the bubble disturbance degree of the metering pump at the target moment.

[0038] Preferably, the calculation result of subtracting the numerical variance of the head area at the target moment from the numerical variance of the tail area at the target moment is used as the fluctuation difference evaluation of the head area and the tail area at the target moment, wherein, if there is no data point in the head area or the tail area at the target moment, the numerical variance of the area is set to 0.

[0039] Preferably, the method of optimizing the error index of the flow time series data of the dosing pump by utilizing the bubble disturbance degree of the dosing pump at the target moment, and obtaining a dynamic accuracy evaluation result that can reflect the influence of the bubble disturbance, includes:

[0040] Obtaining a set error threshold for use in evaluating the accuracy of the metering pump; obtaining a reference target flow value set by the metering pump working system; and using the result of adding the bubble disturbance degree of the metering pump at the target moment to a constant of 1 as the error evaluation optimization weight at the target moment;

[0041] For any moment in the quantitative pump flow rate time series data window at the target moment, the square of the difference between the actual flow rate monitoring value at the moment and the reference target flow rate value at the moment is used as the first error evaluation at the moment, the first error evaluation at the moment is weighted by the error evaluation optimization weight at the moment, and the corresponding calculation result is used as the second error evaluation at the moment; the average of the second error evaluations of all moments in the quantitative pump flow rate time series data window at the target moment is used as the error optimization evaluation at the target moment;

[0042] By comparing the error optimization evaluation at the target moment with the error threshold, a dynamic accuracy evaluation result that can reflect the influence of bubble disturbance is obtained.

[0043] Preferably, the error optimization evaluation at the target moment is compared with the error threshold to obtain a dynamic accuracy evaluation result that can reflect the influence of bubble disturbance, including:

[0044] Obtaining a first error threshold and a second error threshold from the error thresholds; if the error optimization evaluation at the target moment is less than the first error threshold, it is considered that the current working process of the metering pump is stable and in a high-precision state;

[0045] If the error optimization evaluation at the target moment is greater than or equal to the first error threshold and less than the second error threshold, it is considered that there is a medium degree of disturbance in the current working process of the metering pump and it is in a medium accuracy state;

[0046] If the error optimization evaluation at the target moment is greater than or equal to the second error threshold, it is considered that there is a strong disturbance in the current working process of the metering pump and the pump is in a low-precision state.

[0047] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0048] This technical solution constructs a two-layer optimization mechanism for bubble disturbance identification and credibility assessment, and combines the flow time series data characteristics of the servo motor-driven metering pump in the actual filling process to propose a precision dynamic assessment method with the ability to extract disturbance structure and identify pseudo-disturbance. The bubble disturbance assessment factor is used to achieve sensitive identification of bubble disturbance behavior, and the bubble disturbance confidence optimization factor is combined to perform credibility correction on the recovery trend of non-bubble disturbances such as switching. A disturbance score adjustment mechanism is further introduced in the error assessment, so that the system can accurately distinguish between real disturbances and non-disturbance fluctuations, effectively improving the accuracy of the precision assessment. This method has the characteristics of strong real-time performance, simple deployment, and high adaptability. It is particularly suitable for dynamic identification and quality control of abnormal filling behavior in high-frequency, high-precision liquid metering pump control systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0050] Figure 1 This is a flow chart of a method for evaluating the accuracy of a metering pump based on a servo motor driven metering pump provided in the first embodiment of the present invention. DETAILED DESCRIPTION

[0051] The embodiments of the present disclosure are described in detail below, and examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and intended to be used to explain the present disclosure, but should not be understood as limiting the present disclosure.

[0052] It should be noted that the terms "first," "second," and the like in the specification of the present disclosure and the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present disclosure. Instead, they are merely examples of devices and methods consistent with some aspects of the present disclosure.

[0053] In order to illustrate the technical solution of the present invention, specific embodiments are provided below.

[0054] See also Figure 1 , is a flow chart of a method for evaluating the accuracy of a quantitative pump based on a servo motor driven quantitative pump provided in the first embodiment of the present invention, such as Figure 1As shown, the method may include:

[0055] Step S1 : constructing flow time series data for evaluating the accuracy of a quantitative pump by continuously sampling the flow output of a quantitative pump driven by a servo motor, and performing window division on the flow time series data to obtain a quantitative pump flow time series data window.

[0056] To identify and accurately assess the dynamic behavior of a servo motor-driven fixed-displacement pump during operation, it is necessary to first establish a time-series data acquisition mechanism based on flow monitoring. This step samples the real-time flow output from the pump control system to construct a continuous flow monitoring data sequence, providing data support for subsequent disturbance identification and error modeling.

[0057] Specifically, in a quantitative pump control system, the sampling end uses a flow sensor to obtain the instantaneous flow output value of the servo motor-controlled pump head during each filling cycle. This is continuously sampled at a set frequency to generate raw flow time series data. The sampling frequency can be set based on the pump control cycle and fluid characteristics, preferably between 100Hz and 500Hz to ensure the ability to capture micro-perturbations.

[0058] For the obtained quantitative pump flow time series data, set the sliding window length , and obtain the flow time series sliding window according to the sliding window length. The time series data of the quantitative pump flow at the moment, and its corresponding time series data window is:

[0059]

[0060] in, Indicates from time arrive Traffic time series data window; Indicates the The instantaneous flow value at the sampling moment, in units of ; Indicates the sliding window length, the unit is the number of sampling points.

[0061] It should be noted that since this method needs to be deployed in a real-time control system, the judgment at any moment must be based on current and historical data. Therefore, the window adopts a causal structure, that is, the traffic time series data window is formed by the data before the current moment, thereby ensuring that the entire feature extraction process meets the real-time judgment requirements.

[0062] Thus, by continuously sampling the flow output of the servo motor-driven quantitative pump, flow time series data for quantitative pump accuracy evaluation is constructed, and the flow time series data is windowed to obtain the flow time series data window of the quantitative pump.

[0063] Step S2: obtaining a bubble disturbance evaluation factor at a target moment according to the bubble disturbance structural characteristics in the flow time series data window of the metering pump and the output characteristics of the metering pump under the current working condition.

[0064] Traditional sliding window error assessment algorithms generally use static statistics (such as mean square error or maximum absolute error) to evaluate flow errors within the window, without considering the time-dependent evolution and dynamic structure of the error value. This results in an inability to effectively identify disturbances such as the "instantaneous collapse and compensatory rebound" caused by bubbles entering the fluid circuit. This phenomenon is particularly common when using plunger or diaphragm metering pumps to process low-viscosity liquids. If left unaddressed, it will directly affect the metering pump's real-time accuracy assessment and control feedback response.

[0065] In order to achieve targeted modeling and detection of this problem, the present invention extracts three-stage features from the physical evolution process of bubble disturbances: trend deviation, collapse formation, and disturbance diffusion, and constructs three sub-features: trend deviation rate (TDR), local collapse slope (LGF), and disturbance inertia change degree (DIC). Different from the traditional inorganic combination method of simply weighting and summing multiple features, the present invention constructs a structurally dependent functional relationship and logically closes the loop modeling of the above three features in the physical evolution order of "trend deviation trigger → flow collapse unfolds → disturbance diffusion intensifies", forming a disturbance scoring function with semantic consistency and behavioral integrity. The bubble disturbance assessment factor finally obtained can output in real time whether there is bubble disturbance behavior in the current sliding window, as well as the continuity assessment result of the disturbance intensity, with a high degree of scene binding and problem uniqueness.

[0066] In the actual pumping process, when bubbles are introduced into the fluid circuit, the disturbance occurs gradually, not instantaneously. The specific behavior is as follows:

[0067] 1. Instantaneous collapse: Because bubbles are sucked into the pump instead of liquid, the fluid output volume per unit time drops sharply, resulting in a short-term flow drop;

[0068] 2. Recovery overshoot: The system immediately compensates after detecting the flow deviation, but due to inertia or controller overshoot response, a short-term compensatory overshoot often occurs;

[0069] 3. Latency stabilization: Traffic may experience slight fluctuations or short-term rebounds before returning to a stable state.

[0070] This process as a whole forms an asymmetric combination of peaks and troughs on the time axis, first collapsing and then slightly overshooting and rebounding.

[0071] In view of the above scenario, the present invention performs disturbance pattern analysis through the following process to obtain a bubble disturbance evaluation factor for optimizing the error evaluation process of flow time series data.

[0072] After obtaining the metering pump flow time series data window, the flow time series data window can be evaluated for trend deviation and trend anomaly judgment can be made. Specifically, by performing least squares fitting on the metering pump flow time series data window at the target moment, the trend performance equation at the target moment is obtained, and the difference between the predicted result of the trend performance equation at the target moment and the metering pump flow monitoring data at the target moment is used as the trend deviation evaluation at the target moment; the trend deviation evaluation at the target moment is divided by the flow mean of the metering pump flow time series data window at the target moment, and the calculation result is used as the trend deviation degree at the target moment.

[0073] In one embodiment, for The calculation formula for the trend deviation rate of flow monitoring data at a moment is:

[0074]

[0075] in, Indicates the The trend deviation rate of the flow monitoring data at each moment; express Fitting trend forecast value at the moment; express The actual monitoring value of the instantaneous flow of the quantitative pump at the moment; Indicates that in the traffic time series data window The mean flow rate in ; Represents a very small positive number, used to prevent the denominator from being .

[0076] It should be noted that when Significantly greater than When the flow rate begins to deviate from the normal trend, this indicates that the current flow rate is beginning to deviate from the normal trend. This may be a precursor to an abnormal local flow rate drop caused by the entry of bubbles. Bubble disturbances initially manifest as a flow rate drop after the metering pump draws gas. Therefore, this step is used to preemptively determine whether the flow rate has deviated from the expected trend and serves as a precondition for subsequent disturbance determination. If there is no significant deviation, subsequent collapse analysis can be skipped, saving system judgment costs.

[0077] After obtaining the trend deviation rate of the flow monitoring data, the preliminary judgment in the bubble disturbance identification process is completed. After that, the local collapse is further identified. For the quantitative pump flow timing data window at the target moment, starting from the flow data at the second moment in the timing data window, the flow data at each moment is subtracted from the flow data at the previous moment as the flow difference evaluation at that moment; for any moment in the quantitative pump flow timing data window at the target moment, if the flow difference evaluation at that moment is less than 0, the flow difference evaluation at that moment is used as the decline amplitude evaluation at that moment, otherwise the value 0 is used as the decline amplitude evaluation at that moment; the calculation result of adding up the decline amplitude evaluations of all moments in the quantitative pump flow timing data window at the target moment is used as the local collapse intensity at the target moment.

[0078] In one embodiment, the The calculation formula of the local collapse strength at a certain moment is:

[0079]

[0080] in,

[0081]

[0082] in, Indicates the The local collapse intensity of the quantitative pump flow time series data at each moment; Indicates the magnitude of the decline when a decline occurs; Indicates the The instantaneous flow value of the metering pump at a certain moment; Indicates the The instantaneous flow value of the metering pump at a certain moment; Indicates absolute value calculation.

[0083] It's important to note that when analyzing bubble disturbance patterns in metering pump flow time series data, trend deviations can be caused by random fluctuations and are insufficient to prove the presence of a disturbance. Therefore, only when a continuous decline occurs after a trend deviation can a bubble disturbance be considered to be forming, thus preventing normal fluctuations from being misinterpreted as abnormal behavior. The second key determination is to assess the local collapse strength of metering pump flow time series data, confirming the actual formation of a "collapse pattern."

[0084] Because when bubble disturbance occurs during the operation of the metering pump, the bubble disturbance is often not static, but further develops under pressure and pump chamber movement, that is, structural diffusion, so it is necessary to further evaluate the disturbance diffusion inertia in the metering pump flow monitoring time series data to determine whether the disturbance continues to intensify. Specifically, for the flow difference evaluation at each moment in the metering pump flow time series data window at the target moment, starting from the third moment, the absolute value of the calculation result of subtracting the flow difference evaluation at each moment from the flow difference evaluation at the previous moment is used as the flow change rate difference evaluation at that moment; the average of all flow change rate difference evaluations in the metering pump flow time series data window at the target moment is used as the disturbance diffusion inertia evaluation at the target moment.

[0085] In one embodiment, the The calculation formula for the disturbance diffusion inertia evaluation at a moment is:

[0086]

[0087] in, Indicates the The inertia of the diffusion of disturbance at a certain moment; Indicates the set traffic time series data window length; Indicates the The local instantaneous flow rate change rate at a moment; Indicates the The local instantaneous flow rate change at time .

[0088] After obtaining the disturbance diffusion inertia in the quantitative pump flow monitoring time series data, a comprehensive analysis can be performed through the trend deviation rate, local collapse intensity and disturbance diffusion inertia in the quantitative pump flow monitoring time series data to obtain a bubble disturbance evaluation factor for bubble disturbance pattern recognition. Specifically, the trend deviation degree, local collapse intensity and disturbance diffusion inertia evaluation at the target moment are obtained; the trend deviation degree, local collapse intensity and disturbance diffusion inertia evaluation at the target moment are added to a constant 1 and the results of logarithmic mapping are used as the first evaluation factor, second evaluation factor and third evaluation factor at the target moment; the result of multiplying the first evaluation factor, the second evaluation factor and the third evaluation factor at the target moment is used as the numerator, the square of the result of adding the flow mean of the quantitative pump flow time series data window at the target moment to a set minimum positive number is used as the denominator, and the resulting fraction is used as the first bubble disturbance evaluation factor at the target moment; the linear normalized mapping result of the first bubble disturbance evaluation factor at the target moment is used as the bubble disturbance evaluation factor at the target moment.

[0089] In one embodiment, the The calculation formula of the bubble disturbance evaluation factor at a moment is:

[0090]

[0091] in, Indicates The bubble disturbance evaluation factor used for bubble disturbance pattern recognition at each moment; Indicates the The trend deviation rate of the flow monitoring data at each moment; Indicates the The local collapse intensity of the quantitative pump flow time series data at each moment; Indicates the The inertia of the diffusion of disturbance at a certain moment; Indicates Logarithmic function with base ; Indicates The mean of the instantaneous flow monitoring values in the local window of the momentary flow time series data; Represents a very small positive number, used to prevent the denominator from being .

[0092] It should be noted that in the calculation formula of the bubble disturbance assessment factor, the molecular part reacts to the stage-by-stage coupling logic of the disturbance through the product structure, and the molecular part is composed of three disturbance behavior indicators, namely, trend deviation rate , local collapse strength and disturbance diffusion inertia , the numerical stability and disturbance response sensitivity of these three sub-items are enhanced by logarithmic transformation. Specifically, each sub-item is The nonlinear compression of the bubble disturbance assessment factor (BDI) ensures that it remains sensitive to small disturbances while suppressing large anomalies. This prevents the factor from exploding due to drastic flow rate fluctuations and enhances the stability of the accuracy assessment process. This approach ensures that the BDI output approaches zero under normal filling conditions, while gradually increasing during successive phases of typical bubble disturbance behavior. A multiplication of the three sub-items after nonlinear compression is used to evaluate the common bubble disturbance pattern. This structure embodies the continuous evolution of disturbance behavior from onset to accumulation and then to nonlinear enhancement, effectively highlighting structural disturbances rather than localized noise. The mean of the instantaneous flow monitoring values in the denominator represents the base output level of the metering pump under current operating conditions and is used to translate disturbance behavior into a relative intensity expression relative to the system's operating state. During high flow rates, the system is less sensitive to minor disturbances, while at low flow rates, it is more sensitive to the same absolute disturbance. Finally, linear normalization is used to map the BDI to the desired response range, making the output of the BDI more adaptable to different application scenarios, including different pump types, different liquids, and different control accuracy levels.

[0093] In this step, the bubble disturbance evaluation factor is introduced , significantly enhanced the quantitative pump accuracy assessment system's ability to perceive the bubble disturbance behavior in the fluid, and solved the "structural blind spot" problem of the existing sliding window error assessment method when facing short-term nonlinear dynamic disturbances. First, compared with the static error statistics method, the optimization factor It can capture the sudden flow collapse caused by bubbles mixing into the pump cavity through real-time structural analysis, and effectively avoid misjudging normal fluctuations as abnormalities through combined judgment of trend deviation, collapse continuity and disturbance inertia. The output of can not only be used as an independent disturbance identification result, but can also be further used in the weighted judgment module in the error evaluation process to assign dynamic confidence to the error results within the disturbance window, thereby improving the robustness of the entire accuracy evaluation model under non-ideal working conditions.

[0094] In addition, the optimization factor construction process has clear engineering feasibility. All feature calculations are based on the collected flow data, relying on the physical interpretability of quantitative features, without introducing additional sensors or external model dependencies, and has good system integration capabilities. The introduction of not only realizes the quantitative identification of bubble disturbance behavior and error interference suppression, but also provides a credible disturbance prior for the subsequent construction of multi-step disturbance judgment and control feedback mechanism, thus laying a key foundation for the precision evaluation and robust control of servo motor driven quantitative pumps.

[0095] At this point, the bubble disturbance evaluation factor at the target moment is obtained based on the bubble disturbance structural characteristics in the flow time series data window of the metering pump and the output characteristics of the metering pump under the current working condition.

[0096] Step S3, obtaining the bubble disturbance degree of the dosing pump at the target moment according to the fluctuation difference between the head and tail areas in the dosing pump flow time series data window and the bubble disturbance evaluation factor at the target moment.

[0097] In step S2, the bubble disturbance assessment factor at the target moment has been obtained and used for dynamic detection and structural scoring of typical bubble disturbance patterns in the flow time series data of the metering pump. However, in actual scenarios, the metering pump still has non-bubble disturbance behaviors during operation, especially short-term nonlinear flow rate mutations caused by suction reversal delays, pipeline pressure differential rebound, etc., which manifest themselves in the flow time series monitoring data of the metering pump in a similar way to bubble disturbances and will also be judged as abnormal by the bubble disturbance assessment factor, thereby causing incorrect judgment of the accuracy assessment. In order to further improve the disturbance credibility assessment capability of bubble disturbance identification, it is necessary to further optimize the accuracy of bubble disturbance identification to ensure the accuracy of the metering pump precision assessment.

[0098] After analyzing a large amount of real canning data, it was found that the most common type of pseudo-bubble disturbance comes from the instantaneous pressure difference disturbance caused by pump-controlled reversing operation. This type of disturbance usually manifests as flow collapse and recovery rising behavior within a period of time. However, the main difference from the real bubble disturbance is that the pseudo-bubble disturbance usually quickly returns to a stable state after the disturbance occurs, while the real bubble disturbance will cause the disturbance tail to be accompanied by a longer oscillation recovery period, which is manifested as a significantly higher system fluctuation amplitude in the time period after the disturbance.

[0099] Therefore, the present invention uses the change in system fluctuation stability after disturbance as the core feature of pseudo-disturbance identification to obtain the bubble disturbance confidence optimization factor for evaluating the credibility of the disturbance scoring result output by the bubble disturbance evaluation factor.

[0100] In the process of evaluating the credibility of the disturbance scoring results output by the bubble disturbance assessment factor, it is necessary to analyze the fluctuation changes in the window in the quantitative pump flow time series data to determine the difference between the fluctuations caused by bubble disturbances and the fluctuations caused by pseudo-bubble disturbances. For the difference analysis of the fluctuation pattern, it is necessary to divide the flow time series data window and perform a difference assessment between the window tail area and the window head area to determine the disturbance change.

[0101] However, in the process of dividing the head area and the tail area, the disturbance may not always appear in the front part of the window. Sometimes the disturbance may also occur in the middle part or even in the back part of the window. If the division is only performed through fixed window sub-areas, the disturbance part may be regarded as the tail area at some moments, resulting in misjudgment. Even the head area and the tail area may be in the middle and the front part of the disturbance respectively, which will completely lose the meaning of the evaluation. Therefore, it is necessary to first evaluate the disturbance position in the window, and then divide the head area and the tail area relative to this position.

[0102] For the evaluation of the disturbance position, the second-order difference data of the quantitative pump flow time series data window at the target moment is obtained, and the second-order difference data at any moment in the quantitative pump flow time series data window at the target moment is used as the disturbance intensity weight of the moment; in the quantitative pump flow time series data window at the target moment, the data point with the highest disturbance intensity weight is used as the disturbance center estimation point at the target moment; the number of data points of half the window length of the disturbance center estimation point at the target moment is used as the disturbance center area at the target moment; the area to the left of the disturbance center area in the quantitative pump flow time series data window at the target moment is used as the head area at the target moment; the area to the right of the disturbance center area in the quantitative pump flow time series data window at the target moment is used as the tail area at the target moment; if there is no data point in the head area or tail area at the target moment, the numerical variance of the area is set to 0.

[0103] After obtaining the head area and the tail area in the quantitative pump flow time series data window at the target moment, the bubble disturbance confidence optimization factor can be evaluated by the fluctuation difference between the head area and the tail area. Specifically, the set smoothing adjustment factor is obtained; the calculation result of subtracting the numerical variance of the head area at the target moment from the numerical variance of the tail area at the target moment is used as the fluctuation difference evaluation of the head area and the tail area at the target moment; the maximum value between the fluctuation difference evaluation of the head area and the tail area at the target moment and the constant 0 is used as the first confidence evaluation factor at the target moment; the calculation result of multiplying the first confidence evaluation factor at the target moment by the set smoothing adjustment factor and taking the negative number is used as the second confidence evaluation factor at the target moment; the second confidence evaluation factor at the target moment is subjected to exponential mapping with a natural constant as the base, and the corresponding exponential mapping result is used as the bubble disturbance confidence optimization factor at the target moment.

[0104] In one embodiment, the The calculation formula of the bubble disturbance confidence optimization factor at a moment is:

[0105]

[0106] in, Indicates the Bubble disturbance confidence optimization factor at the moment; Indicates the The instantaneous flow variance in the head area of the quantitative pump flow time series data window at each moment; Indicates the The instantaneous flow variance in the tail area of the quantitative pump flow time series data window at each moment; Represents the smoothing adjustment factor, which is used to control the change of the disturbance recovery characteristic The impact speed is set in this embodiment. The value range is In this embodiment, it is set ,The smoothing adjustment factor can be adjusted according to the actual scenario and is not required; Represents a very small positive number, used to prevent the denominator from being .

[0107] It should be noted that in the quantitative pump flow time series data, if the disturbance is caused by reversing, it usually returns to stability quickly after the disturbance, that is, Obviously positive, at this time If the disturbance credibility is reduced, it should be suppressed during the bubble disturbance identification process; ,at this time for , it means that the tail fluctuation has not recovered or even strengthened, and the disturbance trend continues to exist, which is more consistent with the real bubble disturbance. At this time, the identification of the bubble disturbance should be retained.

[0108] After obtaining the bubble disturbance confidence optimization factor at the target moment, the bubble disturbance assessment factor at the target moment can be confidence optimized using the bubble disturbance confidence optimization factor at the target moment. Specifically, the result of multiplying the bubble disturbance confidence optimization factor at the target moment by the bubble disturbance assessment factor at the target moment is used as the bubble disturbance degree of the metering pump at the target moment.

[0109] In one embodiment, the The calculation formula for the bubble disturbance degree of the quantitative pump at a certain moment is:

[0110]

[0111] in, Indicates the The degree of bubble disturbance in the metering pump at each moment; Indicates The bubble disturbance evaluation factor used for bubble disturbance pattern recognition at each moment; Indicates the The bubble disturbance confidence optimization factor at each moment is used to evaluate the disturbance assessment credibility of the bubble disturbance assessment factor.

[0112] It should be noted that the evaluation of the bubble disturbance degree of the quantitative pump adopts a multiplicative coupling method to integrate the two dimensions of whether the disturbance exists and whether it is credible. is a structural score used to determine whether the disturbance exhibits typical evolutionary characteristics of bubble disturbances in the time series structure. It is a confidence score, which is mainly used to evaluate whether the current disturbance behavior is real and to judge the credibility of the behavior based on the disturbance tail response characteristics. The multiplication structure ensures that the degree of bubble disturbance in the final metering pump is significantly amplified only when the disturbance intensity is high and the disturbance credibility is high, thereby effectively reducing misjudgment.

[0113] Thus, the bubble disturbance degree of the metering pump at the target moment is obtained according to the fluctuation difference between the head and tail regions in the metering pump flow time series data window and the bubble disturbance evaluation factor at the target moment.

[0114] Step S4, optimizing the error index of the flow time series data of the dosing pump using the bubble disturbance degree of the dosing pump at the target moment, and obtaining a dynamic accuracy evaluation result that can reflect the influence of the bubble disturbance.

[0115] In step S3, the degree of bubble disturbance of the metering pump is obtained, and the degree of bubble disturbance of the metering pump can reflect the structural strength and credibility of the flow anomaly in the current window in real time. This step is intended to optimize the actual canning accuracy assessment through the degree of bubble disturbance of the metering pump, thereby improving the accuracy of the precision judgment of the metering pump under the condition of bubble disturbance. Specifically, the set error threshold is obtained for the quantitative pump accuracy assessment; the reference target flow value set by the metering pump working system is obtained; the calculation result of adding the bubble disturbance degree of the metering pump at the target moment to the constant 1 is used as the error assessment optimization weight of the target moment; for any moment in the metering pump flow time series data window at the target moment, the square of the difference between the actual flow monitoring value at the moment and the reference target flow value at the moment is used as the first error assessment at the moment, the first error assessment at the moment is weighted by the error assessment optimization weight at the moment, and the corresponding calculation result is used as the second error assessment at the moment; the average of the second error assessments of all moments in the metering pump flow time series data window at the target moment is used as the error optimization assessment of the target moment.

[0116] In one embodiment, the The calculation formula for the error optimization evaluation at each moment is:

[0117]

[0118] in, Indicates the Error optimization evaluation at each moment; Indicates the set constant flow rate pump time series data window length; Indicates the The degree of bubble disturbance in the metering pump at each moment; Indicates the The actual flow value at a moment; Indicates the system settings in The reference target flow value at a certain moment.

[0119] After obtaining the quantitative pump optimization evaluation error for quantitative pump accuracy optimization evaluation, the accuracy evaluation can be performed using the error threshold to obtain the first error threshold and the second error threshold in the error threshold;

[0120] If the error optimization evaluation at the target moment is less than the first error threshold, it is considered that the current working process of the metering pump is stable and in a high-precision state;

[0121] If the error optimization evaluation at the target moment is greater than or equal to the first error threshold and less than the second error threshold, it is considered that there is a medium degree of disturbance in the current working process of the metering pump and it is in a medium accuracy state;

[0122] If the error optimization evaluation at the target moment is greater than or equal to the second error threshold, it is considered that there is a strong disturbance in the current working process of the metering pump and the pump is in a low-precision state.

[0123] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A method for evaluating the accuracy of a quantitative pump driven by a servo motor, characterized in that: The method for evaluating the accuracy of a quantitative pump driven by a servo motor includes: By continuously sampling the flow output of a quantitative pump driven by a servo motor, flow time series data for evaluating the accuracy of the quantitative pump is constructed, and the flow time series data is divided into windows to obtain a flow time series data window of the quantitative pump; based on the structural characteristics of the bubble disturbance in the flow time series data window of the quantitative pump and the output characteristics of the quantitative pump under the current working conditions, a bubble disturbance evaluation factor at a target moment is obtained; based on the fluctuation difference between the head and tail areas in the flow time series data window of the quantitative pump and the bubble disturbance evaluation factor at the target moment, the bubble disturbance degree of the quantitative pump at the target moment is obtained; using the bubble disturbance degree of the quantitative pump at the target moment, the error index of the flow time series data of the quantitative pump is optimized, and a dynamic accuracy evaluation result that can reflect the influence of the bubble disturbance is obtained; The method of obtaining a bubble disturbance assessment factor at a target moment based on the bubble disturbance structural characteristics in the flow time series data window of the quantitative pump and the output characteristics of the quantitative pump under the current working conditions includes: obtaining a trend performance equation at the target moment by performing least square fitting on the flow time series data window of the quantitative pump at the target moment, and taking the difference between the prediction result of the trend performance equation at the target moment and the flow monitoring data of the quantitative pump at the target moment as the trend deviation assessment at the target moment; dividing the trend deviation assessment at the target moment by the flow mean of the flow time series data window of the quantitative pump at the target moment as the trend deviation degree at the target moment; obtaining the local collapse strength at the target moment by performing cumulative difference assessment on the flow change data in the flow time series data window of the quantitative pump at the target moment; obtaining the disturbance diffusion inertia assessment at the target moment by performing cumulative change speed assessment on the flow change information in the flow time series data window of the quantitative pump at the target moment; obtaining the bubble disturbance assessment factor at the target moment by performing nonlinear compression fusion analysis on the trend deviation degree, local collapse strength and disturbance diffusion inertia assessment at the target moment; The bubble disturbance assessment factor at the target moment is obtained by performing nonlinear compression fusion analysis on the trend deviation degree, local collapse intensity and disturbance diffusion inertia assessment at the target moment, including: obtaining the trend deviation degree, local collapse intensity and disturbance diffusion inertia assessment at the target moment; adding the trend deviation degree, local collapse intensity and disturbance diffusion inertia assessment at the target moment to a constant 1 and performing logarithmic mapping to calculate the results as the first assessment factor, the second assessment factor and the third assessment factor at the target moment; using the result of multiplying the first assessment factor, the second assessment factor and the third assessment factor at the target moment as the numerator, using the square of the result of adding the flow rate mean of the quantitative pump flow time series data window at the target moment to a set minimum positive number as the denominator, and using the formed fraction as the first bubble disturbance assessment factor at the target moment; using the linear normalized mapping result of the first bubble disturbance assessment factor at the target moment as the bubble disturbance assessment factor at the target moment; The method of obtaining the bubble disturbance degree of the quantitative pump at the target moment according to the fluctuation difference between the head and tail areas in the quantitative pump flow time series data window and the bubble disturbance evaluation factor at the target moment includes: obtaining second-order difference data of the quantitative pump flow time series data window at the target moment, and using the second-order difference data at any moment in the quantitative pump flow time series data window at the target moment as the disturbance intensity weight of the moment; using the data point with the highest disturbance intensity weight in the quantitative pump flow time series data window at the target moment as the disturbance center estimation point at the target moment; using the number of data points of half the window length of the disturbance center estimation point at the target moment as the disturbance center area at the target moment; using the area to the left of the disturbance center area in the quantitative pump flow time series data window at the target moment as the head area at the target moment; and using the area to the right of the disturbance center area in the quantitative pump flow time series data window at the target moment as the tail area at the target moment; and obtaining the bubble disturbance degree of the quantitative pump at the target moment by evaluating the fluctuation difference between the head area and the tail area at the target moment and the bubble disturbance evaluation factor at the target moment; The method of obtaining the bubble disturbance degree of the quantitative pump at the target moment by evaluating the fluctuation difference between the head region and the tail region at the target moment and the bubble disturbance evaluation factor at the target moment includes: obtaining a set smoothing adjustment factor; subtracting the numerical variance of the head region at the target moment from the numerical variance of the tail region at the target moment as the fluctuation difference evaluation between the head region and the tail region at the target moment; taking the maximum value between the fluctuation difference evaluation between the head region and the tail region at the target moment and a constant 0 as the first confidence evaluation factor at the target moment; multiplying the first confidence evaluation factor at the target moment by the set smoothing adjustment factor and taking the negative value as the second confidence evaluation factor at the target moment; performing exponential mapping on the second confidence evaluation factor at the target moment with a natural constant as the base, and taking the corresponding exponential mapping result as the bubble disturbance confidence optimization factor at the target moment; and multiplying the bubble disturbance confidence optimization factor at the target moment by the bubble disturbance evaluation factor at the target moment as the bubble disturbance degree of the quantitative pump at the target moment.

2. The method for evaluating the accuracy of a quantitative pump driven by a servo motor according to claim 1, wherein: The method of obtaining the local collapse strength at the target moment by performing cumulative difference evaluation on the flow change data in the flow time series data window of the quantitative pump at the target moment comprises: For the target moment's metering pump flow timing data window, starting from the flow data at the second moment in the timing data window, the calculation result of subtracting the flow data at each moment from the flow data at the previous moment is used as the flow difference evaluation at that moment; for any moment in the target moment's metering pump flow timing data window, if the flow difference evaluation at that moment is less than 0, the flow difference evaluation at that moment is used as the decline amplitude evaluation at that moment, otherwise the value 0 is used as the decline amplitude evaluation at that moment; the calculation result of adding up the decline amplitude evaluations of all moments in the target moment's metering pump flow timing data window is used as the local collapse strength at the target moment.

3. The method for evaluating the accuracy of a quantitative pump driven by a servo motor according to claim 1, wherein: The method of performing a cumulative evaluation of the flow rate change information in the flow rate time series data window of the quantitative pump at the target moment to obtain the disturbance diffusion inertia evaluation at the target moment includes: For the flow difference evaluation at each moment in the quantitative pump flow timing data window at the target moment, starting from the third moment, the absolute value of the calculation result of subtracting the flow difference evaluation at each moment from the flow difference evaluation at the previous moment is used as the flow change rate difference evaluation at that moment; the average of all the flow change rate difference evaluations in the quantitative pump flow timing data window at the target moment is used as the disturbance diffusion inertia evaluation at the target moment.

4. The method for evaluating the accuracy of a quantitative pump driven by a servo motor according to claim 1, wherein: The calculation result of subtracting the numerical variance of the head area at the target moment from the numerical variance of the tail area at the target moment is used as the fluctuation difference evaluation of the head area and the tail area at the target moment, wherein if there is no data point in the head area or the tail area at the target moment, the numerical variance of the area is set to 0.

5. The method for evaluating the accuracy of a quantitative pump driven by a servo motor according to claim 1, wherein: The error index of the flow time series data of the dosing pump is optimized by utilizing the bubble disturbance degree of the dosing pump at the target moment, and a dynamic accuracy evaluation result reflecting the influence of the bubble disturbance is obtained, including: Obtaining a set error threshold for use in evaluating the accuracy of the metering pump; obtaining a reference target flow value set by the metering pump working system; and using the result of adding the bubble disturbance degree of the metering pump at the target moment to a constant of 1 as the error evaluation optimization weight at the target moment; For any moment in the quantitative pump flow rate time series data window at the target moment, the square of the difference between the actual flow rate monitoring value at the moment and the reference target flow rate value at the moment is used as the first error evaluation at the moment, the first error evaluation at the moment is weighted by the error evaluation optimization weight at the moment, and the corresponding calculation result is used as the second error evaluation at the moment; the average of the second error evaluations of all moments in the quantitative pump flow rate time series data window at the target moment is used as the error optimization evaluation at the target moment; By comparing the error optimization evaluation at the target moment with the error threshold, a dynamic accuracy evaluation result that can reflect the influence of bubble disturbance is obtained.

6. The method for evaluating the accuracy of a quantitative pump driven by a servo motor according to claim 5, characterized in that: The error optimization evaluation at the target moment is compared with the error threshold to obtain a dynamic accuracy evaluation result that can reflect the influence of bubble disturbance, including: Obtaining a first error threshold and a second error threshold from the error thresholds; if the error optimization evaluation at the target moment is less than the first error threshold, it is considered that the current working process of the metering pump is stable and in a high-precision state; If the error optimization evaluation at the target moment is greater than or equal to the first error threshold and less than the second error threshold, it is considered that there is a medium degree of disturbance in the current working process of the metering pump and it is in a medium accuracy state; If the error optimization evaluation at the target moment is greater than or equal to the second error threshold, it is considered that there is a strong disturbance in the current working process of the metering pump and the pump is in a low-precision state.

Images