CN116882079A - Water pump characteristic curve self-adaptive calibration and prediction method - Google Patents

Abstract

The invention discloses an adaptive calibration and prediction method for water pump characteristic curves, which includes the following steps: collecting data on various characteristics of the water pump in real time through external equipment, uploading the collected data to a designated platform, and storing current and historical data; establishing Based on the classification-improved ARMA model, each feature of the water pump is predicted; the predicted values of each feature of the water pump in step S2 and the collected values of each feature of the water pump at the current moment are substituted into the operating rule formula of the water pump at the rated frequency to predict the water pump at the current moment. Correct and predict the characteristic curve; fit the collected pipeline parameters in a least squares manner through formulas; draw the pipeline characteristic curve and the water pump characteristic curve. The intersection of the two curves is to predict the actual working conditions of the water pump. point. The invention has the characteristics of effectively improving accuracy and expanding the scope of application.

Description

An adaptive calibration and prediction method for water pump characteristic curves

Technical field

The invention relates to a water pump characteristic processing method, and in particular to a water pump characteristic curve adaptive calibration and prediction method.

Background technique

The current traditional water pump modeling methods are usually based on mechanism modeling. The changes in the characteristic curves of the same type of water pump in different working conditions and with the increase of use time are not considered in the model. Therefore, the traditional model will There is a large error. In addition, due to the particularity of different industrial scenarios, the collected data will be subject to certain restrictions (the equipment cannot be started and stopped arbitrarily in special industrial scenarios), and only some data points at fixed frequencies or specific frequencies can be collected. For example, in the existing technology patent titled "Modeling Method and System for Mathematical Model of Variable Frequency Adjustable Speed Water Pumps", it is based on a large amount of historical water pump operating data, grouping them according to the same frequency band, and fitting the corresponding frequency band The flow-head and flow-efficiency curves.

However, there are still several shortcomings:

1. This model family is only suitable for fitting the flow-head and flow-efficiency curves of water pumps to data in the same frequency band. It will be used when the collected data cannot meet the requirements of the same frequency band or there are few data points in the same frequency band. restricted.

2. The model's fitting of the water pump characteristic curve based on historical data can, to a certain extent, deal with the problem of characteristic curve deviation caused by long-term use of the water pump. However, when the water pump fails and the collected data deviates greatly, it cannot be discovered in time.

Therefore, the existing technology has problems of poor accuracy and narrow scope of application.

Contents of the invention

The purpose of the present invention is to provide an adaptive calibration and prediction method for water pump characteristic curves in view of the shortcomings of the existing technology. The invention has the characteristics of effectively improving accuracy and expanding the scope of application.

The technical solution of the present invention: an adaptive calibration and prediction method for water pump characteristic curves, including the following steps:

S1. Collect data on various characteristics of the water pump and pipeline in real time through external equipment, upload the collected data to the designated platform, and store current and historical data;

S2. Establish an ARMA model based on classification improvement and predict various characteristics of the water pump;

S3. Substitute the predicted values of each characteristic of the water pump in step S2 and the collected values of each characteristic of the water pump at the current moment into the operating law formula of the water pump at the rated frequency, including the flow-head curve equation, the flow-power curve equation and the efficiency curve equation. The characteristic curve of the water pump at the current moment is corrected, and the predicted flow-head and flow-efficiency curves when the water pump is running in the next time unit are obtained;

S4. Based on the collected pipeline parameters, the pipeline characteristic curve is obtained by fitting the least squares method. The fitting formula is: Among them/> is the head of the pipeline, Q is the flow rate of the pipeline, D and S are the parameters that need to be fitted; draw the pipeline characteristic curve and the water pump characteristic curve, and the intersection of the two curves is the actual working condition point for predicting the work of the water pump.

Further, in step S1, data on various characteristics of the water pump and pipeline are collected once a day, and each piece of data corresponds to the flow rate and head of the water pump and pipeline;

For each collected data, the average value is taken through multiple collections as the measured value of this piece of data.

Further, in step S2, the ARMA model based on classification improvement is implemented through the following steps:

1) Classify the data of each characteristic of the water pump collected in step S1 into data at the same frequency, that is, fixed frequency data, and data at different frequencies, that is, variable frequency data;

2) Classify fixed-frequency data and variable-frequency data by quarter, and extract each feature data information in the same category, including the maximum value, minimum value, mean, censored mean and median of each feature, and select one of the features Data information is used to describe the characteristics of this class, recorded as Represents the median of the j-th feature point collection data under the i-th class;

3) Construct the data information of each feature in each category into a vector, and according to the Legendre best approximation criterion, find the collection data vector closest to the feature data information vector, and record the corresponding date d _i,t in this category;

4) Take m features from the first N categories in the classified data set, and predict the m feature points of the current class, that is, the N+1th class, based on the ARMA model. The specific formula is as follows:

Among them, p is the order of the autoregressive term AR, α _i (i=0,1,K,p) autoregressive coefficient; q is the order of the moving average term MA, β _j (j=1,2,K, q) is the moving average coefficient, ξ _t is white noise, Defines the boundary of noise; E represents mathematical expectation, Var represents variance,/> Represents any collected water pump data of the same type.

5) Substitute the collected values of each feature variable into the above steps to obtain the predicted data information of each feature of the next class. The corresponding date is

6) Based on the measured characteristic values of the last kt day in the Nth class and the predicted future characteristic data information vector, fit the change curve of each characteristic, expressed by the function f _i , and predict with a period of 15 days, Realize the update and iteration of the water pump characteristic curve every 15 days in the next quarter; the specific formula is as follows:

Use the least squares method to solve the formula to obtain the function f _i , where the function f _i represents the prediction function of the i-th feature fitting, and the predicted value at any time in the next class is obtained based on the prediction function.

Further, in step S3, the flow-head curve equation, flow-power curve equation and efficiency curve equation are as follows:

H＝aQ ² +bQ+c

P＝jQ ² +kQ+l

Among them, H is the water pump head, Q is the water pump flow rate, eta is the water pump efficiency, P is the water pump power, a, b, c, j, k, l are the parameters of fitting the quadratic curve.

Furthermore, for the data collected at the same frequency, the flow-head curve equation and the flow-power curve equation at a specific frequency can be obtained through the flow-head curve equation and the flow-power curve equation;

The data collected at different frequencies are fitted by combining the least squares processing method with the mechanism model, as shown below:

H ₀ =aQ ₀ ² +bQ ₀ i+ci ²

P ₀ ＝jQ ₀ ² i+kQ ₀ i ² +li ³

Among them, H ₀ , Q ₀ , P ₀ are the head, flow rate and power when the pump operating frequency is f ₀ , f is the rated frequency, and i is the proportional coefficient between the operating frequency and the rated frequency.

Beneficial effects of the present invention:

Compared with the existing technology, in order to improve the accuracy of the mathematical model of the water pump, the present invention adopts two aspects of modeling for data fusion for the characteristic curve of the water pump at a specific frequency: First, the fitting of the water pump characteristic curve under variable frequency speed regulation , collect the flow, head, and power parameters of the water pump at different frequencies, convert these data into data at the same frequency through the mechanism model, and further give the characteristic curve of the corresponding frequency; the second is the characteristic curve of the water pump at fixed frequency Fitting, by collecting the characteristic parameters of the water pump at the same frequency, and fitting the characteristic curve at that frequency; finally, the fitting of the pipeline characteristic curve, by measuring the pipeline head at different flow rates, Fit the pipeline characteristic curve in this scenario. This invention integrates and automates data collection, data processing, data inspection, and real-time updating of water pump characteristic curves, and no longer requires R&D personnel to perform additional work; it can predict one or more future events through collected data at the same frequency and different frequencies. The flow-head and flow efficiency curve of the water pump within a time unit. This method reduces the work complexity while maintaining the accuracy of the characteristic curve. Moreover, the application scenarios can be generalized and are not only applicable to a specific industrial scenario.

This invention collects data in real time and calibrates the characteristic curves of water pumps and pipelines based on the real-time collected data, updates the model adaptively, analyzes the collected historical data, and determines the current moment based on the improved time series ARMA model. Whether the collected data is abnormal, further predict the future flow-head and flow-efficiency curves, and can be used as a judgment condition for abnormal detection of water pumps. It is suitable for different industrial environments and can simultaneously analyze data collected at fixed frequency and variable frequency. Fit the water pump characteristic curve at the specified frequency. To sum up, the present invention has the characteristics of effectively improving accuracy and expanding the scope of application.

Description of the drawings

Figure 1 is a flow chart of an adaptive calibration and prediction method for a water pump characteristic curve provided by an embodiment of the present invention.

Detailed ways

The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

The invention provides a method for adaptive calibration and prediction of water pump characteristic curves.

What the present invention implements is a method for adaptive testing, calibration and prediction of water pump characteristic curves, which includes four aspects: data collection, data processing and rationality analysis, water pump model calibration, and water pump characteristic curve prediction. It is implemented through the following steps, as shown in the figure 1 shows the specific flow chart:

1. External equipment collects data in real time, and collects data once a day. Each piece of data corresponds to the flow rate, head, etc. of the water pump/pipeline. The data collected in real time by the external equipment is uploaded to the designated platform, and the current and historical data are stored.

For each collected data point, the average value is taken through multiple collections as the measured value of the data.

2. Establishment of ARMA model based on classification improvement. The traditional ARMA model can predict multi-day data in the future based on daily collected data. However, for data points collected from the water pump characteristic curve, which have small changes in the short term, predictions need to be made on a longer time scale. , when predicting data trends, the trend will be unclear and a lot of computing resources will be wasted. Therefore, the traditional model should be improved to fully mine the information of the data and improve the overall accuracy of data prediction. The improved model is implemented through the following steps:

The data collected in 1 is classified into two parts: data at the same frequency (fixed frequency) and data at different frequencies (variable frequency).

Classify the two parts of data by quarter, and extract the information of each feature (flow rate, head, etc.) in the same category, including the maximum value, minimum value, mean value, and censored mean value of each feature (removing the maximum value in the same category and the mean after the minimum value), median, select the median of the feature to describe the feature of this class, recorded as Represents the median of the j-th feature point collection data under the i-th class.

The median of each feature in each category can form a vector The data collected on each day can be expressed as a vector as/> Among them/> Represents the q-th feature on the t-th day under the i-th category. According to Legendre's best approximation criterion and taking formula (1-1) as the goal, find the collected data vector closest to the data median vector, and record the corresponding Date d _i,t in this class.

Take m features from the first N categories in the classified data set, and predict the m feature points of the current class (N+1th class) based on the ARMA model. The specific formula is as follows:

Among them, p is the order of the autoregressive term AR, which can be determined from the PACF chart, α _i (i=0,1,K,p) autoregressive coefficient; q is the order of the moving average term MA, which can be determined from the ACF chart, β _j (j=1,2,K,q) is the moving average coefficient, and ξ _t is white noise.

By bringing the measured values of each feature variable into the above steps, you can get the predicted median of each feature of the next class, and the corresponding date is

Based on the measured characteristic values of the last kt days (divided into k segments) in the Nth class and the predicted future median vector, the change curve of each characteristic (expressed by the function _f Forecast for the cycle, and update and iterate the water pump characteristic curve every 15 days for the next quarter. The specific implementation process is described by formula (1-4), and the least square method is used to solve formula (1-4) to obtain the function f _i

The function _fi represents the prediction function of the i-th feature fitting, and the predicted value at any time in the next class is obtained based on the prediction function.

3. Based on the predicted values of each feature in step 2 and the measured values of each data at the current moment, the operating formulas of the water pump at the rated frequency include the flow-head curve equation, the flow-power curve equation and the efficiency curve equation, as follows Show:

H＝aQ ² +bQ+c (1-5)

P＝jQ ² +kQ+l (1-6)

Among them, H is the water pump head, Q is the water pump flow rate, eta is the water pump efficiency, P is the water pump power, a, b, c, j, k, l are the parameters of fitting the quadratic curve. For different industrial scenarios and collected data, data collected at the same frequency and at different frequencies are processed respectively.

The data collected at the same frequency can be used to obtain the flow-head curve equation and flow-power curve equation at a specific frequency through formulas (1-5) (1-6).

The data collected at different frequencies are fitted by combining the least squares processing method with the mechanism model, as shown below:

H ₀ ＝aQ ₀ ² +bQ ₀ i+ci ² (1-9)

P ₀ ＝jQ ₀ ² i+kQ ₀ i ² +li ³ (1-10)

Among them, H ₀ , Q ₀ , P ₀ are the head, flow rate and power when the pump operating frequency is f ₀ , f is the rated frequency, and i is the proportional coefficient between the operating frequency and the rated frequency.

Assume that n data at different frequencies are collected and calculated by the least squares method from (1-5), (1-9) and (1-6), (1-10):

Ax=T (1-11)

By＝W (1-12)

in:

This step can not only correct the characteristic curve of the water pump at the current moment, but also obtain the predicted flow-head and flow-efficiency curves when the water pump is running in the next time unit.

4. Based on the collected pipeline parameters, the least squares method is used to fit the formula (1-14).

in is the head of the pipeline, Q is the flow rate of the pipeline, D and S are the parameters that need to be fitted.

Draw the pipeline characteristic curve and the water pump characteristic curve. Their intersection is the actual working condition point where the water pump is predicted to work.

Embodiments of the present invention can achieve the following functions: each component on the platform should be able to independently interact with systems outside the platform, collect and transmit data in real time, and upload it to the data processing platform;

Based on the data collected in real time, complete the script for automatic fitting and correction of the water pump and pipeline characteristic curves, calculate the real-time pipeline impedance coefficient, and update and correct the original mechanism model of the water pump in real time;

Based on historical data, predict the changes in the water pump characteristic curve;

It is applicable to different industrial environments. The collected data is processed in blocks to obtain a more accurate water pump characteristic curve.

In order to meet the above functions, an external system is required to collect and input data in real time, and update the characteristic curve of the water pump in real time, without the need for additional manual processing by the platform developers, thereby avoiding secondary development of the platform and improving efficiency.

The above embodiments are used to illustrate the present invention, rather than to limit the present invention. Within the spirit of the present invention and the protection scope of the claims, any modifications and changes made to the present invention fall within the protection scope of the present invention.

Claims (5)

1. An adaptive calibration and prediction method for water pump characteristic curves, which is characterized by including the following steps: S1. Collect data on various characteristics of the water pump and pipeline in real time through external equipment, upload the collected data to the designated platform, and store current and historical data; S2. Establish an ARMA model based on classification improvement and predict various characteristics of the water pump; S3. Substitute the predicted values of each characteristic of the water pump in step S2 and the collected values of each characteristic of the water pump at the current moment into the operating law formula of the water pump at the rated frequency, including the flow-head curve equation, the flow-power curve equation and the efficiency curve equation. The characteristic curve of the water pump at the current moment is corrected, and the predicted flow-head and flow-efficiency curves when the water pump is running in the next time unit are obtained; S4. Based on the collected pipeline parameters, the pipeline characteristic curve is obtained by fitting the least squares method. The fitting formula is: Among them/> is the head of the pipeline, Q is the flow rate of the pipeline, D and S are the parameters that need to be fitted; draw the pipeline characteristic curve and the water pump characteristic curve, and the intersection of the two curves is the actual working condition point for predicting the work of the water pump. 2. A water pump characteristic curve adaptive calibration and prediction method according to claim 1, characterized in that, in step S1, data on various characteristics of the water pump and pipeline are collected once a day, and each piece of data corresponds to the water pump and pipeline. The flow rate and head; For each collected data, the average value is taken through multiple collections as the measured value of this piece of data. 3. A water pump characteristic curve adaptive calibration and prediction method according to claim 1, characterized in that, in step S2, the ARMA model based on classification improvement is implemented through the following steps: 1) Classify the data of each characteristic of the water pump collected in step S1 into data at the same frequency, that is, fixed frequency data, and data at different frequencies, that is, variable frequency data; 2) Classify fixed-frequency data and variable-frequency data by quarter, and extract each feature data information in the same category, including the maximum value, minimum value, mean, censored mean and median of each feature, and select one of the features Data information is used to describe the characteristics of this class, recorded as Represents the median of the j-th feature point collection data under the i-th class; 3) Construct the data information of each feature in each category into a vector, and according to the Legendre best approximation criterion, find the collection data vector closest to the feature data information vector, and record the corresponding date d _i,t in this category; 4) Take m features from the first N categories in the classified data set, and predict the m feature points of the current class, that is, the N+1th class, based on the ARMA model. The specific formula is as follows: Among them, p is the order of the autoregressive term AR, α _i (i=0,1,K,p) autoregressive coefficient; q is the order of the moving average term MA, β _j (j=1,2,K, q) is the moving average coefficient, ξ _t is white noise, Defines the boundary of noise; E represents mathematical expectation, Var represents variance,/> Represents any collected water pump data of the same type; 5) Substitute the collected values of each feature variable into the above steps to obtain the predicted data information of each feature of the next class. The corresponding date is 6) Based on the measured characteristic values of the last kt day in the Nth class and the predicted future characteristic data information vector, fit the change curve of each characteristic, express it with the function f _i , select the prediction period, and realize the next prediction The water pump characteristic curve is updated and iterated for each forecast period in the first quarter; the specific formula is as follows: Use the least squares method to solve the formula to obtain the function f _i , where the function f _i represents the prediction function of the i-th feature fitting, and the predicted value at any time in the next class is obtained based on the prediction function. 4. A water pump characteristic curve adaptive calibration and prediction method according to claim 1, characterized in that, in step S3, the flow-head curve equation, the flow-power curve equation and the efficiency curve equation are as follows: H＝aQ ² +bQ+c P＝jQ ² +kQ+l Among them, H is the water pump head, Q is the water pump flow rate, eta is the water pump efficiency, P is the water pump power, a, b, c, j, k, l are the parameters of fitting the quadratic curve. 5. A water pump characteristic curve adaptive calibration and prediction method according to claim 4, characterized in that the data collected at the same frequency can be obtained at a specific frequency through the flow-head curve equation and the flow-power curve equation. Flow-head curve equation and flow-power curve equation; The data collected at different frequencies are fitted by combining the least squares processing method with the mechanism model, as shown below: H ₀ =aQ ₀ ² +bQ ₀ i+ci ² P ₀ ＝jQ ₀ ² i+kQ ₀ i ² +li ³ Among them, H ₀ , Q ₀ , P ₀ are the head, flow rate and power when the pump operating frequency is f ₀ , f is the rated frequency, and i is the proportional coefficient between the operating frequency and the rated frequency.