A method for back-calculating the sound power level of converter transformers based on reliability

Abstract

The present invention relates to a method for inferring the sound power level of a converter transformer based on reliability, which includes the following steps: 1) measuring the sound pressure level data of a receiving point; 2) establishing an acoustic model in noise prediction software, and using it to obtain the sound propagation attenuation; 3) constructing a mathematical model for sound source sound power inversion; 4) using a genetic algorithm in combination with the reliability solution inversion model to obtain the sound power level of the sound source. The above method constructs a mathematical model for inferring the sound power of a sound source according to the ISO 9613‑2 outdoor sound propagation calculation method, and uses a large number of measured value data of on-site measuring points around the converter transformer, combined with reliability theory, to identify and eliminate bad points. Using a genetic algorithm, it is ensured that the comprehensive error between the effective measuring point and the measured value is minimized, and the inversion calculation of the sound power of the converter transformer is realized.

Description

Technical Field

[0001] The invention relates to the field of converter transformer noise prediction, and in particular to a method for improving the noise prediction accuracy of a converter transformer by eliminating bad points in actual measurement points used for inversely estimating the sound power level of a converter transformer through reliability. Background Art

[0002] In recent years, domestic high-voltage direct current transmission has become increasingly perfect, but in the process, the noise of the converter station has begun to attract public attention. As the most important sound source equipment in the converter station, the noise prediction accuracy of the converter transformer directly affects the noise prediction model of the entire converter station in the later stage, so it is particularly important to obtain its sound power level. At present, in the actual construction of the prediction model, the sound power level of the sound source is mostly adjusted and estimated based on the repeated comparison of the sound pressure level of the measuring point near the sound source with the measured value. It cannot ensure that the error between the predicted value and the measured value of all measuring points is within a certain range. Some measuring points are artificially determined to meet the error range, and some points with large errors are partially not considered. This method has no scientific basis, is relatively arbitrary, and the noise prediction accuracy is difficult to guarantee. The theoretical methods for obtaining the sound power level of the sound source are mainly sound pressure method, sound intensity method, and vibration velocity method: among them, the sound intensity method and vibration velocity method have high operation requirements and are not suitable for outdoor actual engineering applications; the sound pressure method is the most commonly used, but when obtaining the sound pressure level of the measuring point required for the sound pressure method, it is inevitable that some measuring point data will have deviations due to environmental factors, resulting in incorrect final results. These measuring points with errors need to be eliminated. Therefore, it is particularly important to provide a method for back-calculating the sound power level of converter transformers based on reliability. Summary of the invention

[0003] The purpose of the present invention is to provide a method for inferring the sound power of a converter transformer based on reliability, in order to improve the noise prediction accuracy of the converter transformer, in order to solve the problem that the sound power level of the converter transformer is difficult to obtain accurately.

[0004] The present invention achieves the purpose of the invention by adopting the following technical solutions:

[0005] A method for improving the accuracy of a noise prediction model for a thermal power plant comprises the following implementation steps:

[0006] Step 1: Measure the sound pressure level data at the receiving point;

[0007] Step 2: Establish an acoustic model in the noise prediction software and use it to obtain the sound propagation attenuation;

[0008] Step 3: Construct a mathematical model for sound source sound power inversion;

[0009] Step 4: Use the genetic algorithm to solve the inversion model to obtain the sound power level of the sound source.

[0010] As a preference, arrange the measuring points reasonably. When the equipment is running stably, use the noise collection device sound level meter to collect the sound pressure level of the measuring points. Pay attention to recording the temperature, wind speed and measuring point height when collecting data.

[0011] Preferably, the geometric dimensions of the acoustic model are established in the software in a 1:1 ratio according to the converter transformer specifications, and the required known parameters include the sound source type, geometric position coordinates, body height parameters, terrain, receiving point position, and reflection order.

[0012] Preferably, the inverse principle of the sound power level is based on the ISO 9613-2 outdoor sound propagation method, which is specifically expressed as:

[0013] L p =L W +D c -A Formula (1)

[0014] Among them, L w is the sound power of the point sound source, unit: dB(A);

[0015] D c It is the sound source directivity correction, unit: dB(A);

[0016] A is various types of attenuation, including atmospheric attenuation, geometric attenuation, ground effect attenuation, shielding attenuation, etc., unit: dB(A).

[0017] Preferably, when calculating the converter transformer sound field, the reflection order is selected as the third order.

[0018] Preferably, the relationship used for reflection calculation is:

[0019] L W,im =L W +10lg(ρ)+D lr Formula (2)

[0020] Among them, L W,im is the sound power of the virtual sound source used for reflection calculation, unit: dB(A);

[0021] ρ is the obstacle reflection coefficient, which can be obtained by consulting the standard;

[0022] D lr is the directivity index of the virtual sound source at the receiving point.

[0023] As a preferred method, multiple outdoor sound sources are calculated, and the sound pressure level at the measurement point is the energy superposition of each sound source at this point. Specifically expressed as:

[0024]

[0025] Among them, L piis the sound pressure contribution of real sound source i to measuring point j, unit: dB(A);

[0026] L pi ' is the sound pressure contribution of the virtual sound source of sound source i to the measuring point j, unit: dB(A);

[0027] L Wi is the sound power of sound source i, unit: dB(A);

[0028] A ij is the attenuation from real sound source i to measuring point j, unit: dB(A);

[0029] A ij ' is the attenuation from the virtual sound source of sound source i to the measuring point j, unit: dB(A).

[0030] As a preference, there are deviations in the results due to the accuracy of the measuring equipment, the sound attenuation calculation, the houses in the noise prediction model, and the terrain settings, but the errors caused by these factors are allowed in engineering calculations, generally 2-3dB.

[0031] As a preference, the reliability is the ratio of the number of measuring points where the error between the measured value and the predicted value is within a certain range to the total number of measuring points, and the expression is:

[0032] R=m' / m Formula (4)

[0033] Among them, m' is the number of measurement points whose error between the measured value and the predicted value is within a given range, and m is the total number of measurement points.

[0034] As a preferred embodiment, the converter transformer is equivalent to a vertical surface sound source, and for the convenience of calculation, the surface sound source is equivalent to nine point sound sources.

[0035] Preferably, a genetic algorithm is selected for solution to transform the nonlinear equations into an optimal solution.

[0036] As a preference, the mean difference of the errors at each measuring point is used as the objective function, and the expression is:

[0037]

[0038] Preferably, the equivalent point sound source is used as a design variable.

[0039] As a preference, the constraint condition is a reliability constraint, and the expression is:

[0040]

[0041] Among them, L Pj is the measured value of measuring point j, unit: dB(A);

[0042] L Pj' is the predicted value of measuring point j, unit: dB(A);

[0043] e is the allowable error between the measured value and the predicted value, unit: dB(A);

[0044] [R] is the given allowable reliability;

[0045] R is the actually calculated reliability;

[0046] con_value is the number of bad pixels counted;

[0047] sum(con_value) is the number of measurement points that currently meet the error range;

[0048] m is the total number of measuring points.

[0049] As a preference, if there is no bad point in the measurement point data, the reliability can be selected as 100%, otherwise there is a bad point, when the reliability is selected as 100%, the equation has no solution. According to the m measurement points and n sound sources mentioned in the previous section, Table 1 lists the number of bad points and the highest reliability that can be set.

[0050] Table 1: Number of bad pixels and maximum reliability that can be set

[0051]

[0052]

[0053] As a preferred solution, the obtained solution is the sound power level of a single point sound source after the converter transformer is equivalent to nine point sound sources. Therefore, the sound power level of the converter transformer is the energy superposition of the sound power levels of the nine point sound sources, and the relationship is:

[0054]

[0055] Among them, LW (i) It is the sound power level of a single point sound source, unit: dB(A).

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

[0057] Based on the ISO 9613-2 outdoor sound propagation calculation method, the present invention proposes a reliability calculation method and a maximum reliability setting, and constructs a mathematical model for inverse sound power. Through different reliability settings, the present invention can identify and eliminate bad points, minimize the comprehensive error between the effective measurement points and the measured values, and thus obtain a more accurate sound power level of the sound source. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 It is a flow chart of a method for inferring the sound power level of a converter transformer based on reliability provided in an embodiment of the present invention;

[0059] Figure 2 is a schematic diagram of measuring point arrangement in an embodiment of the present invention;

[0060] Figure 3 It is a schematic diagram of the converter transformer in the embodiment of the present invention being equivalent to nine point sound sources after being regarded as a vertical surface sound source. DETAILED DESCRIPTION

[0061] The present invention is specifically and in detail explained below in conjunction with specific embodiments and the accompanying drawings.

[0062] The present invention is suitable for improving the work of predicting noise of converter transformers. After establishing an acoustic model, the sound attenuation of each measuring point is obtained through software, and the bad points in the measuring points are identified and eliminated through reliability. Then, the sound power level of the sound source is inferred by using the updated measuring point data, so as to obtain a more accurate sound power level of the sound source, and finally achieve the purpose of improving the noise prediction accuracy of the converter transformer. Figure 1 The flowchart of the method for inferring the sound power level of the converter transformer based on reliability of the present invention is shown. Each step is explained in detail and given an example below:

[0063] Step 1: Measure the sound pressure level data at the receiving point;

[0064] In this embodiment, a total of 72 measuring points are set, and the schematic diagram of the measuring point positions is as follows: Figure 2 As shown, specifically, the measurement points of each group on both sides are symmetrically distributed from R71 to R713 with a spacing of 3m, and are arranged as follows:

[0065] R11-R16: 6m from the center of the transformer body, 10m apart;

[0066] R21-R29, R31-R36: The measuring points are located alternately at the firewall and the center of the transformer, with an interval of 5m. For specific locations, refer to Figure 2 ;

[0067] R41-R43, R51-R53, R61-R63: at the center of converter transformers 1-3, 1-4 and the firewall between them, with each measuring point spaced 5m apart;

[0068] R71-R713: At the center of the two sets of transformers, each measuring point is 5m apart.

[0069] The measured data are shown in Table 2.

[0070] Table 2: Measured values ​​at measuring points (unit: dB(A))

[0071]

[0072] Step 2: Establish an acoustic model in the noise prediction software and use it to obtain the sound propagation attenuation;

[0073] The calculation accuracy of commonly used environmental acoustic prediction software such as SoundPLAN and Cadna / A is related to the accuracy of model input. When modeling, the known parameters required include sound source type, geometric position coordinates, body height parameters, terrain, receiving point location, and reflection order.

[0074] Specifically, the transformer is equivalent to a vertical sound source with a length of 4m and a height of 4.5m; the reflection order is set to the third order; according to the initial equipment sound power level, the sound attenuation of each measuring point is directly obtained in the Cadna / A software. The calculation criteria of the noise prediction software Cadna / A refer to the ISO9613-2 standard. In the noise prediction software Cadna / A, as long as the parameters such as the terrain do not change, the sound power level of the sound source changes no matter how, and the sound attenuation is also a fixed value. After the noise prediction geometric model is built, the location of the measuring point is determined, and the sound attenuation from the sound source i to the measuring point j can be directly obtained in the software.

[0075] Step 3: Construct a mathematical model for sound source sound power inversion;

[0076] The inverse principle of the sound power level is based on the ISO 9613-2 outdoor sound propagation method, which is specifically expressed as:

[0077] L p =L W +D c -A Formula (1)

[0078] Among them, L w is the sound power of the point sound source, unit: dB(A);

[0079] D c It is the sound source directivity correction, unit: dB(A);

[0080] A is various types of attenuation, including atmospheric attenuation, geometric attenuation, ground effect attenuation, shielding attenuation, etc., unit: dB(A).

[0081] Calculate multiple outdoor sound sources, and the sound pressure level at the measurement point is the energy superposition of each sound source at this point. Specifically expressed as:

[0082]

[0083] Among them, L pi is the sound pressure contribution of real sound source i to measuring point j, unit: dB(A);

[0084] L pi ' is the sound pressure contribution of the virtual sound source of sound source i to the measuring point j, unit: dB(A);

[0085] L Wi is the sound power of sound source i, unit: dB(A);

[0086] A ij is the attenuation from real sound source i to measuring point j, unit: dB(A);

[0087] A ij ' is the attenuation from the virtual sound source of sound source i to the measuring point j, unit: dB(A).

[0088] In this embodiment, the sound pressure level value and the sound attenuation measured at the aforementioned measuring point are substituted into the equation (2) in step 3 to obtain a nonlinear equation group related to the sound power of the sound source.

[0089] Step 4: Use the genetic algorithm to solve the inversion model to obtain the sound power level of the sound source.

[0090] In this embodiment, a genetic algorithm is used in combination with reliability to transform the problem of solving a nonlinear system of equations into the problem of finding the extreme value of an objective function. The objective function expression is:

[0091]

[0092] In this embodiment, the sound power level of each set of transformers is the same, and the two sets of transformers have two design variables: x1 and x2. The range of the design variables is 95-110 dB (A).

[0093] In this embodiment, the expression of reliability is:

[0094]

[0095]

[0096] Among them, L Pj is the measured value of measuring point j, unit: dB(A);

[0097] L Pj ' is the predicted value of measuring point j, unit: dB(A);

[0098] e is the allowable error between the measured value and the predicted value, unit: dB(A);

[0099] [R] is the given allowable reliability;

[0100] R is the actually calculated reliability;

[0101] con_value is the number of bad pixels counted;

[0102] sum(con_value) is the number of measurement points that currently meet the error range;

[0103] m is the total number of measuring points.

[0104] Specifically, in this embodiment, e is 2dB; the initial value of each con_value is 0, and when the measurement point meets the error requirement, the con_value value becomes 1. Therefore, in the later program, it is only necessary to call out the con_value data to know whether each measurement point is a bad point.

[0105] In this embodiment, the accuracy of the measuring point is determined based on different choices of reliability and combined with the con_value data in the program.

[0106] Specifically, in this embodiment, the reliability is set to 100%, 90%, 80%, 70%, and 60% in sequence, and the objective function values ​​are shown in Table 3. When the reliability is 100% and 90%, the equation has no solution and there are bad points in the measuring points; by calling the con_value data under these three reliabilities, it is determined that the measuring points: R22, R25, R34, R125, R126, R62, R73, R75, R82, and R112 are bad points.

[0107] Table 3: Objective function values ​​under different reliability

[0108] Reliability (%) Objective function value (dBA) 100 NAN 90 NAN 80 1.5667 70 1.5667 60 1.5667

[0109] In this embodiment, the reliability is selected as 80%, and the solution is:

[0110] x1=103.8,x2=106.7

[0111] After energy superposition, the sound power of a single transformer in the first group is Pw1=113.3dB(A), and the sound power of a single transformer in the second group is Pw2=116.2dB(A).

[0112] In this embodiment, the inverse result of the present invention is compared with the sound intensity method which is more accurate in predicting the sound power of the sound source, and the error is only 0.3 dB (A).

[0113] It is obvious to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential features of the present invention. The scope of the present invention is defined by the appended claims rather than the above description, and therefore it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention.

Claims

1. A method for inferring the sound power level of a converter transformer based on reliability, characterized in that: The implementation steps include: Step 1: Measure the sound pressure level data at the receiving point; Step 2: Establish an acoustic model in the noise prediction software and use it to obtain the sound propagation attenuation; Step 3: Construct a mathematical model for sound source sound power inversion; Step 4: Use genetic algorithm combined with reliability solution inversion model to obtain the sound power level of the sound source; The relationship between the sound power level and the sound pressure level in step 3 is: L p =L W +D c -A Among them, L w is the sound power of the point sound source, unit: dB(A); D c It is the sound source directivity correction, unit: dB(A); A is various types of attenuation, including atmospheric attenuation, geometric attenuation, ground effect attenuation, shielding attenuation, unit: dB(A); Calculate multiple outdoor sound sources. The sound pressure level at the measuring point is the energy superposition of each sound source at this point, expressed as: Among them, L pi is the sound pressure contribution of real sound source i to measuring point j, unit: dB(A); L pi ' is the sound pressure contribution of the virtual sound source of sound source i to the measuring point j, unit: dB(A); L Wi is the sound power of sound source i, unit: dB(A); A ij is the attenuation from real sound source i to measuring point j, unit: dB(A); A ij ' is the attenuation from the virtual sound source of sound source i to the measuring point j, unit: dB(A); The reliability in step 4 is the ratio of the number of measurement points where the error between the measured value and the predicted value is within a certain range to the total number of measurement points, and the expression is: R=m′ / m Among them, m' is the number of measurement points where the error between the measured value and the predicted value is within a given range; m is the total number of measuring points; If there is no bad point in the measurement point data, the reliability can be selected as 100%. Otherwise, if there is a bad point, when the reliability is selected as 100%, the equation has no solution. When the highest reliability setting and the number of bad points in the measurement point is S, the reliability is (ms) / m; The genetic algorithm in step 4 transforms the problem of solving the nonlinear equations into the problem of finding the extreme value of the objective function; The mean difference of the errors of each measuring point is taken as the objective function, and the expression is: The vertical surface sound source is equivalent to nine point sound sources and used as design variables; the constraint condition is the reliability constraint, and the expression is: if|L P1 -L P1 ′|≤e; con_value(1)=con_value(1)+1; if|L P2 -L P2 ′|≤e; con_value(2)=con_value(2)+1; if|L Pj -L Pj ′|≤e; con_value(j)=con_value(j)+1; R=(sum(con_value)) / m [R]≤R Among them, L Pj is the measured value of measuring point j, unit: dB(A); L Pj ' is the predicted value of measuring point j, unit: dB(A); e is the allowable error between the measured value and the predicted value, unit: dB(A); [R] is the given allowable reliability; R is the actually calculated reliability; con_value is the number of bad pixels counted; sum(con_value) is the number of measurement points that currently meet the error range; m is the total number of measuring points; The solution obtained in step 4 is the sound power level of a single point sound source after the converter transformer is equivalent to nine point sound sources. Therefore, the sound power level of the converter transformer is the energy superposition of the sound power levels of the nine point sound sources, and the relationship is: Among them, LW (i) It is the sound power level of a single point sound source, unit: dB(A).

2. The method for inferring the sound power level of a converter transformer based on reliability according to claim 1 is characterized in that: The measuring points in step 1 are selected between the two groups of transformers and the distance between them is as regular as possible to facilitate input in the software; the measuring points are avoided as much as possible to avoid reflected sound from buildings.

3. The method for inferring the sound power level of a converter transformer based on reliability according to claim 1 is characterized in that: The geometric dimensions of the acoustic model in step 2 are established in the software according to the specifications of the thermal power plant at a ratio of 1:

1. The required known parameters include the type of sound source, geometric position coordinates, body height parameters, terrain, receiving point position, and reflection order.

4. The method for inferring the sound power level of a converter transformer based on reliability according to claim 1 is characterized in that: The sound attenuation in step 2 is directly obtained according to the noise prediction software Cadna / A.

Images