A method for measuring the amplitude-controlled liquid level of a water supply and drainage network
By using amplitude-controlled liquid level measurement method in the water supply and drainage network, using estimated weighted filtering and probability distribution models to process radar signals, the misjudgment problem of radar level measurement in harsh environments is solved, and the measurement accuracy and reliability are improved .
Patent Information
- Application Number
- CN202210104803.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-01-28
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2042-01-28
AI Technical Summary
The existing radar level measurement methods are limited to harsh environments in the water supply and drainage pipeline network, such as scum, turbulence and turbulence, which leads to fluctuations in the reflection amplitude of the radar wave, impurities in the water adhere to the sensor, causing misjudgment, and the pipeline is narrow. This leads to misjudgment of the well wall reflection, reducing measurement accuracy and reliability.
The amplitude-controlled liquid level measurement method is used to measure multiple times through a radar level meter to extract the peak of the amplitude and distance signal, and the estimated weighted filtering algorithm is used to eliminate scum, turbulence and rapid interference, and the screening function and the amplitude-controlled signal weighted function are processed. The signal is combined with the probability distribution model for Gaussian fusion, eliminates the abnormal peak, and determines the final liquid level value.
It effectively eliminates the interference of scum, turbulence and rapids on radar waves, improves the accuracy and reliability of liquid level measurement, reduces misjudgment, and ensures the accuracy of liquid level measurement in the water supply and drainage pipeline network.
Smart Images

Figure CN114136402B_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of liquid level measurement, and in particular relates to an amplitude-controlled liquid level measurement method for a water supply and drainage network. Background Art
[0002] With the development of water environment treatment technology and the improvement of urban planning and construction level, the operation and management of urban water supply and drainage networks are gradually becoming large-scale and intelligent. Intelligent monitoring equipment and visual intelligent data management platform are the development direction of water supply and drainage network operation in the future. Among them, the measurement of pipe network liquid level and data acquisition are important prerequisites for intelligent management.
[0003] The existing efficient inductive liquid level measurement methods mainly include ultrasonic method and radar method. The ultrasonic method is a contact measurement method that needs to be placed in the water body for a long time. Due to the complex water quality of the water supply and drainage network, especially the poor water quality and many impurities in the sewer network, it is easy for a large number of attachments to quickly attach to the sensor surface and affect the sensitivity of the sensor. Compared with ultrasonic ranging, radar ranging has the advantages of high accuracy, strong anti-interference and long measurement distance. Most of the existing radar ranging is used in relatively open and relatively stable occasions. In the water supply and drainage network, it is limited by its harsh environment. For example, there are often scum, rapids and turbulence, which are different from the calm water surface, resulting in more fluctuations in the radar wave reflection amplitude. There are many impurities in the water that are easy to attach to the surface of the radar sensor and cause misjudgment. The narrow pipeline causes the well wall to reflect the radar wave and cause misjudgment, making radar ranging less used in the water supply and drainage network. Summary of the invention
[0004] In view of the above situation, the present invention provides an amplitude-controlled liquid level measurement method for a water supply and drainage network, which detects the amplitude of a radar wave return signal to determine the measured distance, and includes the following steps:
[0005] S100: Use the radar level meter to measure several times, and each measurement will obtain a series of fluctuating amplitude signals and corresponding distance signals;
[0006] S200: extracting peak values corresponding to the amplitude signals measured several times to form an amplitude matrix; extracting distance data corresponding to the distance signals measured several times to form a distance matrix;
[0007] S300: using the time interval of each measurement, performing a pre-estimation weighted filtering algorithm on the amplitude matrix and the distance matrix to obtain accurate amplitude and distance, so as to eliminate the interference of abnormal states of scum, turbulence and rapids on radar waves;
[0008] S400: The accurate amplitude and distance obtained in step S300 are processed by a screening function and an amplitude control signal weighting function to eliminate the influence of abnormal peaks caused by attachments on the well wall or the surface of the liquid level meter, obtain the final amplitude matrix and distance matrix, and determine the final liquid level value.
[0009] Optionally, in step S100, each time the radar level meter measures, a series of fluctuating amplitude signals and corresponding distance signals are obtained, the time interval between each measurement is the same, and multiple series of fluctuating amplitude signals and corresponding distance signals obtained corresponding to multiple measurements are recorded.
[0010] Optionally, in step S200, the peak value a corresponding to the amplitude signal measured several times is extracted. 00 、a 01 , …a 0i , the peak value is the amplitude that is greater than both the closest amplitude on the left and the closest amplitude on the right. All peak values are sorted in descending order and recorded as A0. A0={a 00 ,a 01 ,…a 0i};
[0011] (2) In the first measurement, extract the distance data x corresponding to the amplitude peak in A0 00 、x 01 ,…x 0i , denoted as X0, X0={x 00 ,x 01 ,…x 0i};
[0012] (3) In the subsequent measurements, the peak value a corresponding to the amplitude peak is extracted from the amplitude signal of the second measurement. 10 、a 11 , …a 1i , after sorting in descending order, it is recorded as A1, A1={a 10 ,a 11 ,…a 1i}, and so on, in the jth measured amplitude signal, extract the peak value a corresponding to the amplitude peak j-10 、a j-11 , …a j-1i , sorted in descending order and recorded as A j-1 ={a j-10 ,a j-11 ,…a j-1i}, in the amplitude signal measured for the j+1th time, extract the peak value a corresponding to the amplitude peak j0 、a j1 , …a ji , sorted in descending order and recorded as A j , A j ={aj0 ,a j1 ,…a ji}, the set of peak values corresponding to all amplitude peaks is denoted as A,
[0013]
[0014] Extract the distance data x corresponding to the amplitude peak in A1 10 、x 11 ,…x 1i , denoted as X1, X1={x 10 ,x 11 ,…x 1i}, and so on, extract A j-1 The distance data x corresponding to the amplitude peak in j-10 、x j-11 ,…x j-1i , denoted as X j-1 , X j-1 ={x j-10 ,x j-11 ,…x j-1i}, extract A j The distance data x corresponding to the amplitude peak in j0 、x j1 ,…x ji , denoted as X j , X j ={x j0 ,x j1 ,…x ji}, the set of distances corresponding to all amplitude peaks is recorded as X,
[0015]
[0016] Optionally, step S300 specifically includes the following steps:
[0017] S301: Compare the peak values of two measurements before and after, and divide them by the time interval between the two measurements to obtain the amplitude peak change rate; compare the distances of two measurements before and after, and divide them by the time interval between the two measurements to obtain the distance change rate;
[0018] S302: predicting the peak value of the next measurement according to the amplitude peak change rate and the probability distribution model; predicting the distance of the next measurement according to the distance change rate and the probability distribution model;
[0019] S303: Gaussian fusion is performed on the measured value of the amplitude peak and the predicted value of the amplitude peak obtained in step S302 to determine the probability distribution of the amplitude peak. Gaussian fusion is performed on the measured value of the distance and the predicted value of the distance obtained in step S302 to determine the probability distribution of the distance, so as to obtain an accurate amplitude peak set and distance set.
[0020] Further optionally, in step S301, the amplitude peak change rate between the j+1th measurement and the jth measurement is:
[0021] △A j =(A j -A j-1 ) / t=[(a j0 -a j-10 ) / t (a j1 -a j-11 ) / t … (a ji -a j-1i ) / t]=[△a j0 △a j1 …△a ji ]
[0022] Among them, △a j0 =(a j0- a j-10 ) / t,△a j1 =(a j1- a j-11 ) / t,△a ji =(a ji- a j-1i ) / t,
[0023] The distance change rate is:
[0024] V j =(X j -X j-1 ) / t=[(x j0 -x j-10 ) / t (x j1 -x j-11 ) / t … (x ji -x j-1i ) / t]=[v j0 v j1 …v ji ]
[0025] Among them, △v j0 =(x j0- x j-10 ) / t,△v j1 =(x j1- x j-11 ) / t,△v ji =(x ji- x j-1i ) / t; t is the time interval between two measurements.
[0026] Further optionally, in step S302, the peak value of the next measurement is predicted to be:
[0027]
[0028] The predicted distance for the next measurement is:
[0029]
[0030] Among them, F j is the probability distribution model;
[0031] Further optionally, in step S303, the measured value of the amplitude peak and the predicted value of the amplitude peak obtained in step S302 are Gaussian fused to determine the probability distribution of the amplitude peak, specifically:
[0032]
[0033] Among them, A j+1 is the measured value of the j+2th amplitude peak. In the j+2th measured amplitude signal, the peak value a corresponding to the amplitude peak is extracted. j+10 、a j+11 , …a j+1i , sorted in descending order and recorded as A j+1 , A j+1 ={a j+10 ,a j+11 ,…a j+1i}, △A j+1 is the peak change rate of the amplitude between the j+2th and j+1th measurements, a j+1 and a j are all known values,
[0034] △A j+1 =(A j+1 -A j ) / t=[(a j+10 -a j0 ) / t (a j+11 -a j1 ) / t … (a j+1i -a ji ) / t]=[△a j+10 △a j+11 …△a j+1i ];
[0035] Among them, △a j+10 =(a j+10- a j0 ) / t,△a j+11 =(a j+11- a j1 ) / t,△a j+1i =(a j+1i- a ji ) / t,
[0036] The distance measurement value and the distance prediction value obtained in step S302 are Gaussian-fused to determine the probability distribution of the distance, specifically:
[0037]
[0038] Among them, X j+1 is the j+2th distance measurement value, extract A j+1 The distance data x corresponding to the amplitude peak in j+10 、x j+11 ,…x j+1i , denoted as X j+1 , X j+1 ={x j+10 ,x j+11 ,…x j+1i},V j+1 is the distance change rate between the j+2th and j+1th measurements, X j+1 and X j are all known values,
[0039] V j+1 =(X j+1 -X j ) / t=[(x j+10 -x j0 ) / t (x j+11 -x j1 ) / t … (x j+1i -x ji ) / t]=[v j+10 v j+11 …v j+1i ]
[0040] Among them, v j+10 =(x j+10- x j0 ) / t,v j+11 =(x j+11- x j1 ) / t,v j+1i =(x j+1i- x ji ) / t,
[0041] The F ’ j =F j ; K1 and K2 are the total amplitude peak covariance and the total distance covariance, respectively, K1=[K a K a ],K2=[K v K v ];
[0042] Among them, K a =Pa j 'Hj T (H j Pa j 'H j T +R j ) -1 , K v =Px j 'H j T (H j Px j 'H j T +R j ) -1 ;
[0043] The R j is the error covariance model of the radar level meter itself, specifically:
[0044]
[0045] Among them, x c , v c These are the calibration values of the radar level gauge probe after it is manufactured and tested. c0 , v c0 is the average of the calibration values.
[0046] The distance covariance components caused by the noise models of scum, turbulence and rapids are accumulated, and we can get:
[0047]
[0048] Among them, Qx f is the covariance component caused by scum; Qx t is the covariance component caused by turbulence; Qx j The covariance component caused by the rapids;
[0049] Among them, the covariance matrix Px of the distance estimate j-1 for: ,
[0050] Covariance , where x is the distance data X obtained for the jth measurement j-1 ={x j-10 ,x j-11 ,…x j-1i}, is the average value of the distance data obtained by the jth measurement;
[0051] Covariance , where v is the distance change rate V for the j-th and j-1-th measurements j-1 =[vj-10 v j-11 …v j-1i ], V j-1 The average value of the elements in ;
[0052] Covariance , .
[0053] Further optionally, the Qxf is determined according to the composition, thickness and density of the scum. ; ;
[0054] Among them, b, c, and d are all coefficients. In terms of values, b is 0.1-0.5, c is 0.05-0.1, and d is 0.01-0.03.
[0055] The amplitude peak covariance components caused by the noise models of scum, turbulence and rapids are accumulated to obtain:
[0056]
[0057] Among them, Qa f is the covariance component caused by scum; Qa t is the covariance component caused by turbulence; Qa j The covariance component caused by the rapids;
[0058] Further optionally, the ; ;
[0059] Among them, e, h, and k are all coefficients. In terms of values, e is 0.2-0.3, h is 0.02-0.05, and k is 0.05-0.1.
[0060] Among them, the covariance matrix Pa of the amplitude peak estimate j-1 for:
[0061] Covariance , where a is the peak value A of the amplitude peak obtained for the jth measurement j-1 ={a j-10 ,a j-11 ,…a j-1i}, is the average value of the peak values of the amplitude peaks obtained by the jth measurement;
[0062] Covariance , where △a is the distance change rate △A for the j-th and j-1-th measurements j-1=[△a j-10 △a j-11 … △a j-1i ], = △A j-1 The average value of the elements in ;
[0063] Covariance , covariance .
[0064] The accurate distance set obtained through the above process is: X'={x0,x1,…,x i}; The set of peak values of the amplitude peaks corresponding to X' is A'={a0,a1,…,a i}
[0065] The double correction is achieved by double prediction of the distance signal and the amplitude signal, combined with the above filtering for abnormal conditions such as scum, turbulence and rapids in the pipe network.
[0066] Optionally, step S400 specifically includes the following steps:
[0067] S401: sorting the accurate amplitudes measured once in descending order, and sorting the corresponding accurate distances in order of amplitude;
[0068] S402: Filter outliers through a screening function;
[0069] S403: Calculate the probability distribution of each amplitude peak through the amplitude weighted model, and form a probability distribution set for this measurement;
[0070] S404: The elements in the probability distribution set of the measurement are re-sorted in descending order, and the corresponding amplitude matrix and distance matrix are synchronously sorted along with the probability distribution matrix to obtain the final amplitude matrix and distance matrix, and the final liquid level value is calculated.
[0071] Optionally, in step S401, the accurate amplitudes A′ measured once are sorted in descending order to obtain an intermediate amplitude set A″, and the accurate distances X′ corresponding to A′ are sorted in the order corresponding to A″ to obtain an intermediate distance set X″, and A″ and X″ are respectively written in matrix form:
[0072]
[0073]
[0074] Among them, a0, a1, a i is the amplitude data of A″, x0, x1, x i The distance data is X″.
[0075] Although the subscripts of the elements in the above matrices A″, X″ and A′X′ are the same, the actual values they represent are different. The elements in A″ and X″ are sorted values.
[0076] Optionally, in step S402, the screening function is:
[0077]
[0078] Filter out x i <0 and x i >a i elements.
[0079] Optionally, in step S403, by using A″, X″ and the screening function f(x i ), amplitude weighted model g(a i ) calculates the probability distribution of a single amplitude peak:
[0080] P'x i ={f(x i )×g(a i )
[0081] Amplitude weighted model g(a i ) specifically: ,
[0082] Thus, the probability distribution set of this measurement is determined: P'x i ={Px0,Px1,…Px i}, written in matrix form, we have P″x=[Px0,Px1,…Px i ].
[0083] In step S404, the elements in the probability distribution set of the measurement are re-sorted in descending order, and the corresponding amplitude matrix and distance matrix are synchronously sorted following the probability distribution matrix to obtain the final amplitude matrix and distance matrix:
[0084]
[0085] Take the peak value a0 of the amplitude peak corresponding to the maximum probability Px0 in the above matrix, that is, the first column a0 of the matrix as the final liquid level value. BRIEF DESCRIPTION OF THE DRAWINGS
[0086] Figure 1 This is the radar signal diagram received in the first measurement;
[0087] Figure 2 The accurate distance signal and amplitude signal diagram obtained in step S300;
[0088] Figure 3 The final distance signal and amplitude signal diagram obtained in step S400. DETAILED DESCRIPTION
[0089] This embodiment provides an amplitude-controlled liquid level measurement method for a water supply and drainage network, which detects the amplitude of a radar wave return signal to determine the measured distance, and includes the following steps:
[0090] S100: Use the radar level meter to measure several times, and each measurement will obtain a series of fluctuating amplitude signals and corresponding distance signals;
[0091] S200: extracting peak values corresponding to the amplitude signals measured several times to form an amplitude matrix; extracting distance data corresponding to the distance signals measured several times to form a distance matrix;
[0092] S300: using the time interval of each measurement, performing a pre-estimation weighted filtering algorithm on the amplitude matrix and the distance matrix to obtain accurate amplitude and distance, so as to eliminate the interference of abnormal states of scum, turbulence and rapids on radar waves;
[0093] S400: The accurate amplitude and distance obtained in step S300 are processed by a screening function and an amplitude control signal weighting function to eliminate the influence of abnormal peaks caused by attachments on the well wall or the surface of the liquid level meter, obtain the final amplitude matrix and distance matrix, and determine the final liquid level value.
[0094] In step S100, each time the radar level meter measures, a series of fluctuating amplitude signals and corresponding distance signals are obtained. The time interval between each measurement is the same, and multiple series of fluctuating amplitude signals and corresponding distance signals obtained by multiple measurements are recorded.
[0095] In step S200, the peak values corresponding to the amplitude signals measured several times are extracted to form an amplitude matrix, which is specifically:
[0096] (1) In the first measurement, in a series of fluctuating amplitude signals, extract the peak value a corresponding to the amplitude peak 00 、a 01 , …a 0i , the peak value is the amplitude that is greater than both the closest amplitude on the left and the closest amplitude on the right. All peak values are sorted in descending order and recorded as A0. A0={a 00 ,a 01 ,…a 0i};
[0097] (2) In the first measurement, extract the distance data x corresponding to the amplitude peak in A0 00 、x 01 ,…x 0i , denoted as X0, X0={x 00 ,x 01 ,…x0i},like Figure 1 As shown;
[0098] (3) In the subsequent measurements, the peak value a corresponding to the amplitude peak is extracted from the amplitude signal of the second measurement. 10 、a 11 , …a 1i , after sorting in descending order, it is recorded as A1, A1={a 10 ,a 11 ,…a 1i}, and so on, in the jth measured amplitude signal, extract the peak value a corresponding to the amplitude peak j-10 、a j-11 , …a j-1i , sorted in descending order and recorded as A j-1 ={a j-10 ,a j-11 ,…a j-1i}, in the amplitude signal measured for the j+1th time, extract the peak value a corresponding to the amplitude peak j0 、a j1 , …a ji , sorted in descending order and recorded as A j , A j ={a j0 ,a j1 ,…a ji}, the set of peak values corresponding to all amplitude peaks is denoted as A,
[0099]
[0100] Extract the distance data x corresponding to the amplitude peak in A1 10 、x 11 ,…x 1i , denoted as X1, X1={x 10 ,x 11 ,…x 1i}, and so on, extract A j-1 The distance data x corresponding to the amplitude peak in j-10 、x j-11 ,…x j-1i , denoted as X j-1 , X j-1 ={x j-10 ,x j-11 ,…x j-1i}, extract A j The distance data x corresponding to the amplitude peak in j0 、x j1 ,…x ji , denoted as X j , X j ={x j0 ,x j1 ,…xji}, the set of distances corresponding to all amplitude peaks is recorded as X,
[0101]
[0102] Step S300 specifically includes the following steps:
[0103] S301: Compare the peak values of two measurements before and after, and divide them by the time interval between the two measurements to obtain the amplitude peak change rate; compare the distances of two measurements before and after, and divide them by the time interval between the two measurements to obtain the distance change rate;
[0104] S302: predicting the peak value of the next measurement according to the amplitude peak change rate and the probability distribution model; predicting the distance of the next measurement according to the distance change rate and the probability distribution model;
[0105] S303: Gaussian fusion is performed on the measured value of the amplitude peak and the predicted value of the amplitude peak obtained in step S302 to determine the probability distribution of the amplitude peak. Gaussian fusion is performed on the measured value of the distance and the predicted value of the distance obtained in step S302 to determine the probability distribution of the distance, so as to obtain an accurate amplitude peak set and distance set.
[0106] Further optionally, in step S301, the amplitude peak change rate between the j+1th measurement and the jth measurement is:
[0107] △A j =(A j -A j-1 ) / t=[(a j0 -a j-10 ) / t (a j1 -a j-11 ) / t … (a ji -a j-1i ) / t]=[△a j0 △a j1 …△a ji ]
[0108] Among them, △a j0 =(a j0- a j-10 ) / t,△a j1 =(a j1- a j-11 ) / t,△a ji =(a ji- a j-1i ) / t,
[0109] The distance change rate is:
[0110] V j =(X j -X j-1) / t=[(x j0 -x j-10 ) / t (x j1 -x j-11 ) / t … (x ji -x j-1i ) / t]=[v j0 v j1 …v ji ]
[0111] Among them, △v j0 =(x j0- x j-10 ) / t,△v j1 =(x j1- x j-11 ) / t,△v ji =(x ji- x j-1i ) / t; t is the time interval between two consecutive measurements, and in this embodiment, the value of t is 500ms.
[0112] Further optionally, in step S302, the peak value of the next measurement is predicted to be:
[0113]
[0114] The predicted distance for the next measurement is:
[0115]
[0116] Among them, F j is the probability distribution model, ;
[0117] Further optionally, in step S303, the measured value of the amplitude peak and the predicted value of the amplitude peak obtained in step S302 are Gaussian fused to determine the probability distribution of the amplitude peak, specifically:
[0118]
[0119] Among them, A j+1 is the measured value of the j+2th amplitude peak. In the j+2th measured amplitude signal, the peak value a corresponding to the amplitude peak is extracted. j+10 、a j+11 , …a j+1i , sorted in descending order and recorded as A j+1 , A j+1 ={a j+10 ,a j+11 ,…a j+1i}, △A j+1 is the peak change rate of the amplitude between the j+2th and j+1th measurements, a j+1 and a jare all known values,
[0120] △A j+1 =(A j+1 -A j ) / t=[(a j+10 -a j0 ) / t (a j+11 -a j1 ) / t … (a j+1i -a ji ) / t]=[△a j+10 △a j+11 …△a j+1i ];
[0121] Among them, △a j+10 =(a j+10- a j0 ) / t,△a j+11 =(a j+11- a j1 ) / t,△a j+1i =(a j+1i- a ji ) / t,
[0122] The distance measurement value and the distance prediction value obtained in step S302 are Gaussian-fused to determine the probability distribution of the distance, specifically:
[0123]
[0124] Among them, X j+1 is the measured value of the j+2th distance, the measured value of the j+2th distance, extract A j+1 The distance data x corresponding to the amplitude peak in j+10 、x j+11 ,…x j+1i , denoted as X j+1 , X j+1 ={x j+10 ,x j+11 ,…x j+1i},V j+1 is the distance change rate between the j+2th and j+1th measurements, X j+1 and X j are all known values,
[0125] V j+1 =(X j+1 -X j ) / t=[(x j+10 -x j0 ) / t (x j+11 -x j1 ) / t … (x j+1i -x ji) / t]=[v j+10 v j+11 …v j+1i ]
[0126] Among them, v j+10 =(x j+10- x j0 ) / t,v j+11 =(x j+11- x j1 ) / t,v j+1i =(x j+1i- x ji ) / t,
[0127] The F ’ j =F j ; K1 and K2 are the total amplitude peak covariance and the total distance covariance, respectively, K1=[K a K a ],K2=[K v K v ];
[0128] Among them, K v =Px j 'H j T (H j Px j 'H j T +R j ) -1 , K a =Pa j 'H j T (H j Pa j 'H j T +R j ) -1 ;
[0129] The R j is the error covariance model of the radar level meter itself, specifically:
[0130]
[0131] Among them, x c , v c These are the calibration values of the radar level gauge probe after it is manufactured and tested. c0 , v c0 is the average value of the calibration value. In this embodiment, x c is [1003 1005 1002 999 998], v cis [0.11 0.105 0.108 0.1120.111], x0 is 1001, v c0 It is 0.109.
[0132] The distance covariance components caused by the noise models of scum, turbulence and rapids are accumulated, and we can get:
[0133]
[0134] Among them, Qx f is the distance covariance component caused by scum; Qx t is the distance covariance component caused by turbulence; Qx j is the distance covariance component caused by the rapids;
[0135] Among them, the covariance matrix Px of the distance estimate j-1 for: ,
[0136] Covariance , where x is the distance data X obtained for the jth measurement j-1 ={x j-10 ,x j-11 ,…x j-1i}, is the average value of the distance data obtained by the jth measurement;
[0137] Covariance , where v is the distance change rate V for the j-th and j-1-th measurements j-1 =[v j-10 v j-11 …v j-1i ], V j-1 The average value of the elements in ;
[0138] Covariance , .
[0139] Further optionally, the Qxf is determined according to the composition, thickness and density of the scum. ; ;
[0140] Among them, b, c, and d are all coefficients. In terms of values, b is 0.1, c is 0.05, and d is 0.01.
[0141] The amplitude peak covariance components caused by the noise models of scum, turbulence and rapids are accumulated to obtain:
[0142]
[0143] Among them, Qa f is the amplitude covariance component caused by scum; Qa t is the amplitude covariance component caused by turbulence; Qa j is the amplitude covariance component caused by the rapids;
[0144] Further optionally, the ; ;
[0145] Among them, e, h, and k are all coefficients. In terms of values, e is 0.2, h is 0.02, and k is 0.05.
[0146] Among them, the covariance matrix Pa of the amplitude peak estimate j-1 for
[0147] Covariance , where a is the peak value A of the amplitude peak obtained for the jth measurement j-1 ={a j-10 ,a j-11 ,…a j-1i}, is the average value of the peak values of the amplitude peaks obtained by the jth measurement;
[0148] Covariance , where △a is the distance change rate △A for the j-th and j-1-th measurements j-1 =[△a j-10 △a j-11 … △a j-1i ], = △A j-1 The average value of the elements in ;
[0149] Covariance , covariance .
[0150] Through the above process, we can get the accurate distance set: X'={x0,x1,…,x i}; The set of peak values of the amplitude peaks corresponding to X' is A'={a0,a1,…,a i},like Figure 2 shown.
[0151] The double correction is achieved by double prediction of the distance signal and the amplitude signal, combined with the above filtering for abnormal conditions such as scum, turbulence and rapids in the pipe network.
[0152] Optionally, step S400 specifically includes the following steps:
[0153] S401: sorting the accurate amplitudes measured once in descending order, and sorting the corresponding accurate distances in order of amplitude;
[0154] S402: Filter outliers through a screening function;
[0155] S403: Calculate the probability distribution of each amplitude peak through the amplitude weighted model, and form a probability distribution set for this measurement;
[0156] S404: The elements in the probability distribution set of the measurement are re-sorted in descending order, and the corresponding amplitude matrix and distance matrix are synchronously sorted along with the probability distribution matrix to obtain the final amplitude matrix and distance matrix, and the final liquid level value is calculated.
[0157] Optionally, in step S401, the accurate amplitudes A′ measured once are sorted in descending order to obtain an intermediate amplitude set A″, and the accurate distances X′ corresponding to A′ are sorted in the order corresponding to A″ to obtain an intermediate distance set X″, and A″ and X″ are respectively written in matrix form:
[0158]
[0159]
[0160] Among them, a0, a1, a i is the amplitude data of A″, x0, x1, x i The distance data is X″.
[0161] Although the elements in the above matrices A″, X″ and A′X′ have the same subscripts, they actually represent different values. The elements in A″ and X″ are sorted values.
[0162] Optionally, in step S402, the screening function is:
[0163]
[0164] Filter out x i <0 and x i >a i elements.
[0165] Optionally, in step S403, by using A″, X″ and the screening function f(x i ), amplitude weighted model g(a i ) calculates the probability distribution of a single amplitude peak:
[0166] P'x i ={f(x i )×g(a i )
[0167] Amplitude weighted model g(a i ) specifically: ,
[0168] Thus, the probability distribution set of this measurement is determined: P'x i ={Px0,Px1,…Px i}, written in matrix form, we have P″x=[Px0,Px1,…Px i ].
[0169] In step S404, the elements in the probability distribution set of the measurement are re-sorted in descending order, and the corresponding amplitude matrix and distance matrix are synchronously sorted following the probability distribution matrix to obtain the final amplitude matrix and distance matrix, such as Figure 3 As shown:
[0170]
[0171] Take the peak value a0 of the amplitude peak corresponding to the maximum probability Px0 in the above matrix, that is, the first column a0 of the matrix as the final liquid level value, which is 1000mm.
[0172] Figure 1-Figure 3 The horizontal axis is the distance in mm, and the vertical axis is the amplitude.
Claims
1. A method for measuring the liquid level of a water supply and drainage network by amplitude control, which detects the amplitude of the radar wave return signal to determine the measured distance. It is characterized in that The following steps are involved: S100: Use the radar level meter to measure several times, and each measurement will obtain a series of fluctuating amplitude signals and corresponding distance signals; S200: extracting peak values corresponding to the amplitude signals measured several times to form an amplitude matrix; extracting distance data corresponding to the distance signals measured several times to form a distance matrix; S300: using the time interval of each measurement, performing a pre-estimation weighted filtering algorithm on the amplitude matrix and the distance matrix to obtain accurate amplitude and distance, so as to eliminate the interference of abnormal states of scum, turbulence and rapids on radar waves; S400: The accurate amplitude and distance obtained in step S300 are processed by a screening function and an amplitude control signal weighting function to eliminate the influence of abnormal peaks caused by attachments on the well wall or the surface of the liquid level meter, obtain the final amplitude matrix and distance matrix, and determine the final liquid level value; In step S100, each time the radar level meter measures, a series of fluctuating amplitude signals and corresponding distance signals are obtained, and the time interval between each measurement is the same. Multiple series of fluctuating amplitude signals and corresponding distance signals obtained by multiple measurements are recorded. In step S200, the peak values corresponding to the amplitude signals measured several times are extracted to form an amplitude matrix, which is specifically: (1) In the first measurement, in a series of fluctuating amplitude signals, extract the peak value a corresponding to the amplitude peak 00 、a 01 , …a 0i The peak value is the amplitude that is greater than both the amplitude closest to the left and the amplitude closest to the right. All peak values are sorted in descending order and recorded as A 0 , A 0 ={a 00 ,a 01 ,…a 0i }; (2) In the first measurement, extract A 0 The distance data x corresponding to the amplitude peak in 00 、x 01 ,…x 0i , denoted as X 0 , X 0 ={x 00 ,x 01 ,…x 0i }; (3) In the subsequent measurements, the peak value a corresponding to the amplitude peak is extracted from the amplitude signal of the second measurement. 10 、a 11 , …a 1i , sorted in descending order and recorded as A 1 , A 1 ={a 10 ,a 11 ,…a 1i }, and so on, in the jth measured amplitude signal, extract the peak value a corresponding to the amplitude peak j-10 、a j-11 , …a j-1i , sorted in descending order and recorded as A j-1 ={a j-10 ,a j-11 ,…a j-1i }, in the amplitude signal measured for the j+1th time, extract the peak value a corresponding to the amplitude peak j0 、a j1 , …a ji , sorted in descending order and recorded as A j , A j ={a j0 ,a j1 ,…a ji }, the set of peak values corresponding to all amplitude peaks is denoted as A, , Extract A 1 The distance data x corresponding to the amplitude peak in 10 、x 11 ,…x 1i , denoted as X 1 , X 1 ={x 10 ,x 11 ,…x 1i }, and so on, extract A j-1 The distance data x corresponding to the amplitude peak in j-10 、x j-11 ,…x j-1i , denoted as X j-1 , X j-1 ={x j-10 ,x j-11 ,…x j-1i }, extract A j The distance data x corresponding to the amplitude peak in j0 、x j1 ,…x ji , denoted as X j , X j ={x j0 ,x j1 ,…x ji }, the set of distances corresponding to all amplitude peaks is recorded as X, ; Step S300 specifically includes the following steps: S301: Compare the peak values of two measurements before and after, and divide them by the time interval between the two measurements to obtain the amplitude peak change rate; compare the distances of two measurements before and after, and divide them by the time interval between the two measurements to obtain the distance change rate; S302: predicting the peak value of the next measurement according to the amplitude peak change rate and the probability distribution model; predicting the distance of the next measurement according to the distance change rate and the probability distribution model; S303: Gaussian fusion is performed on the measured value of the amplitude peak and the predicted value of the amplitude peak obtained in step S302 to determine the probability distribution of the amplitude peak, and Gaussian fusion is performed on the measured value of the distance and the predicted value of the distance obtained in step S302 to determine the probability distribution of the distance, so as to obtain an accurate amplitude peak set and distance set; In step S301, the peak change rate of the amplitude between the j+1th measurement and the jth measurement is: △A j =(A j -A j-1 ) / t=[(a j0 -a j-10 ) / t (a j1 -a j-11 ) / t … (a ji -a j-1i ) / t]=[△a j0 △a j1 …△a ji ]; Among them, △a j0 =(a j0- a j-10 ) / t,△a j1 =(a j1- a j-11 ) / t,△a ji =(a ji- a j-1i ) / t, t is the time interval between two measurements; The distance change rate is: V j =(X j -X j-1 ) / t=[(x j0 -x j-10 ) / t (x j1 -x j-11 ) / t … (x ji -x j-1i ) / t]=[v j0 v j1 …v ji ]; where, △v j0 =(x j0- x j-10 ) / t, △v j1 =(x j1- x j-11 ) / t, △v ji =(x ji- x j-1i ) / t; In step S302, the peak value of the next measurement is predicted to be: ; The predicted distance for the next measurement is: ; Among them, F j is the probability distribution model, ; In step S303, the measured value of the amplitude peak and the predicted value of the amplitude peak obtained in step S302 are Gaussian fused to determine the probability distribution of the amplitude peak, specifically: ; Among them, in the amplitude signal measured for the j+2th time, the peak value a corresponding to the amplitude peak is extracted j+10 、a j+11 , …a j+1i , sorted in descending order and recorded as A j+1 , A j+1 ={a j+10 ,a j+11 ,…a j+1i }, △A j+1 is the peak change rate of the amplitude between the j+2th and j+1th measurements, △A j+1 =(A j+1 -A j ) / t=[(a j+10 -a j0 ) / t (a j+11 -a j1 ) / t … (a j+1i -a ji ) / t]=[△a j+10 △a j+11 …△a j+1i ]; where, △a j+10 =(a j+10- a j0 ) / t, △a j+11 =(a j+11- a j1 ) / t, △a j+1i =(a j+1i- a ji ) / t, The distance measurement value and the distance prediction value obtained in step S302 are Gaussian-fused to determine the probability distribution of the distance, specifically: ; Among them, F ’ j =F j , X j+1 is the j+2th distance measurement value, extract A j+1 The distance data x corresponding to the amplitude peak in j+10 、x j+11 ,…x j+1i , denoted as X j+1 , X j+1 ={x j+10 ,x j+11 ,…x j+1i },V j+1 is the distance change rate between the j+2th and j+1th measurements, V j+1 =(X j+1 -X j ) / t=[(x j+10 -x j0 ) / t (x j+11 -x j1 ) / t … (x j+1i -x ji ) / t]=[v j+10 v j+11 …v j+1i ] Among them, v j+10 =(x j+10- x j0 ) / t,v j+11 =(x j+11- x j1 ) / t,v j+1i =(x j+1i- x ji ) / t,K 1 and K 2 They are the total amplitude peak covariance and the total distance covariance respectively; The K 1 =[K a K a ], K 2 =[K v K v ]; Among them, K v =Px j 'H j T (H j Px j 'H j T +R j ) -1 , K a =Pa j 'H j T (H j Pa j 'H j T +R j ) -1 ; The R j is the error covariance model of the radar level meter itself, specifically: ; Among them, x c , v c These are the calibration values of the radar level gauge probe after it is manufactured and tested. c0 , v c0 is the average value of the calibration value; The distance covariance components caused by the noise models of scum, turbulence and rapids are accumulated, and we can get: ; Among them, Qx f is the distance covariance component caused by scum; Qx t is the distance covariance component caused by turbulence; Qx j is the distance covariance component caused by the rapids; Among them, the covariance matrix Px of the distance estimate j-1 for: , Covariance , where x is the distance data X obtained for the jth measurement j-1 ={x j-10 ,x j-11 ,…x j-1i }, is the average value of the distance data obtained by the jth measurement; Covariance , where v is the distance change rate V for the j-th and j-1-th measurements j-1 =[v j-10 v j-11 …v j-1i ], V j-1 The average value of the elements in ; Covariance , ; The Qxf is determined according to the composition, thickness and density of the scum. ; ; ; Among them, b, c, and d are all coefficients. In terms of value, b is 0.1, c is 0.05, and d is 0.01; The amplitude peak covariance components caused by the noise models of scum, turbulence and rapids are accumulated to obtain: ; Among them, Qa f is the amplitude covariance component caused by scum; Qa t is the amplitude covariance component caused by turbulence; Qa j is the amplitude covariance component caused by the rapids; Said ; ; ; Among them, e, h, and k are all coefficients. In terms of value, e is 0.2, h is 0.02, and k is 0.05; Among them, the covariance matrix Pa of the amplitude peak estimate j-1 for: ; Covariance , where a is the peak value A of the amplitude peak obtained for the jth measurement j-1 ={a j-10 ,a j-11 ,…a j-1i }, is the average value of the peak values of the amplitude peaks obtained by the jth measurement; Covariance , where △a is the distance change rate △A for the j-th and j-1-th measurements j-1 =[△a j-10 △a j-11 … △a j-1i ], = △A j-1 The average value of the elements in ; Covariance , covariance ; The accurate distance set obtained through the above process is: X'={x 0 ,x 1 ,…,x i }; The set of peak values of the amplitude peaks corresponding to X' is A'={a 0 ,a 1 ,…,a i }, The double correction is achieved by double prediction of the distance signal and the amplitude signal, combined with the above filtering for abnormal conditions such as scum, turbulence and rapids in the pipe network.
2. The liquid level measurement method according to claim 1, It is characterized in that Step S400 specifically includes the following steps: S401: sorting the accurate amplitudes measured once in descending order, and sorting the corresponding accurate distances in order of amplitude; S402: Filter outliers through a screening function; S403: Calculate the probability distribution of each amplitude peak through the amplitude weighted model, and form a probability distribution set for this measurement; S404: The elements in the probability distribution set of the measurement are re-sorted in descending order, and the corresponding amplitude matrix and distance matrix are synchronously sorted along with the probability distribution matrix to obtain the final amplitude matrix and distance matrix, and the final liquid level value is calculated.
3. The liquid level measurement method according to claim 2, It is characterized in that In step S401, the accurate amplitudes A′ measured once are sorted in descending order to obtain an intermediate amplitude set A″, and the accurate distances X′ corresponding to A′ are sorted in the order corresponding to A″ to obtain an intermediate distance set X″; Write A″ and X″ in matrix form respectively: A″=[a 0 a 1 …a i ] X″=[x 0 x 1 …x i ] Among them, a 0 、a 1 、a i is the amplitude data of A″, x 0 、x 1 、x i The distance data is X″.
4. The liquid level measurement method according to claim 3, It is characterized in that In step S403, by using A″, X″ and the screening function f(x i ), amplitude weighted model g(a i ) calculates the probability distribution of a single amplitude peak: P’x i ={f(x i )×g(a i )}。 5. The liquid level measurement method according to claim 4, It is characterized in that In step S404, the maximum probability P in the final amplitude matrix and distance matrix is obtained. x0 The corresponding peak value a 0 is the final liquid level value.
