CN103926463B - The method for measuring phase difference that a kind of Data extension formula is relevant - Google Patents

Abstract

The present invention relates to signal processing field, the method for measuring phase difference that a kind of Data extension formula is relevant.The applicable object of the present invention is the phase difference measurement of the most same frequency sinusoidal signal, and the present invention comprises the following steps: first, and sampled signal is made cycle pre-estimation；Then, to data length, whether complete cycle judges, the data length of non-integer-period is made Data extension process complete cycle, makes the length of correlation computations just or be approximately complete cycle, thus reaches to improve the purpose of method of correlation Phase Difference Measuring Precision；Finally, method of correlation phase difference calculating formula is utilized to be calculated the phase contrast of signal.The method for measuring phase difference that the present invention relates to realizes simply, noiseproof feature is higher, precision is higher, and eliminating method of correlation Phase Difference Measuring Precision is affected by non-integer-period sampled, can jointly improve the Phase Difference Measuring Precision of method of correlation in conjunction with other method.

Description

The method for measuring phase difference that a kind of Data extension formula is relevant

Technical field

The present invention relates to signal processing field, the method for measuring phase difference that a kind of Data extension formula is relevant.

Background technology

The measurement of phase contrast all has in fields such as fault diagnosis, navigator fix, testing and control, power electronics, communication, radars Important and be widely applied.Such as, the determination of power factor (PF) in electrical network electric energy metrical, relate to phase place between relevant voltage and electric current The accurate measurement of difference；In high-precision radar precision positioning, depend on the high-acruracy survey to same-frequency signal phase difference；At height The Coriolis flowmeter that Accuracy Flow fields of measurement is used widely, the certainty of measurement of phase contrast directly affects fluid mass and measures essence Degree.Therefore, it is achieved the high-acruracy survey of phase contrast is seemed and is even more important.

Method of correlation utilizes two-way to be that cross-correlation function value when zero is directly proportional to the cosine value of its phase contrast with the time delay of frequency sinusoidal signal Principle obtain phase contrast, be widely applied in the signal processing.Due to noise signal generally and useful signal dependency relatively Little, method of correlation has good noise inhibiting ability, it is possible to suppress all and reference signal incoherent various forms noise, but For the interference signal harmony wave interference that dependency is stronger, under Low SNR, measurement error is bigger.Additionally, method of correlation is wanted Ask and signal is carried out the most integer-period sampled, the accurate measurement of phase contrast can be realized under the conditions of integer-period sampled, but at non-complete cycle Under phase sampling condition, certainty of measurement need to improve.For improving the certainty of measurement of phase contrast, in succession occur in that multiple based on relevant former The phase difference measurement new algorithm of reason or innovatory algorithm, mainly have following several:

(1) list of references [1]: improve oneself in a left side, Xu Yang, Cao Xiaolong, etc. changing of the correlation function algorithm of calculating capacitance type equipment dielectric dissipation factor Enter [J]. electric power network technique, the basic thought of 2004,28 (18): 53-57. document [1] methods is to use interpolation to round the method in cycle to integration Interval is modified, and replaces rectangular integration logarithm value integration method to improve with trapezoidal integration, and chooses suitable sample frequency Computational accuracy is improved with sampling number.The method computational accuracy is by integrating range, numerical integration method and numerical value integration step , there is the problem that influence factor is many and calculates complexity in impact.

(2) list of references [2]: Li Jianmin, Zhao Peng, Hou Wen, etc. the error research [J] of phase contrast algorithm based on correlation theory. Northcentral University Journal (natural science edition), 2009,30 (6): 616-619. documents [2] use double method of correlation, need to produce two with frequency, there are 90 ° The reference signal of phase shift, carries out related operation with two reference signals respectively by input and output signal and tries to achieve the phase contrast of signal. The method reduce the phase error that randomness causes, but non-integer-period sampled in the case of phase difference error the biggest.

(3) list of references [3]: Zheng Shengfeng, building Potiria pectinifera (Mukller et Tro Sehel), model essay is brilliant, etc. a kind of method for measuring phase difference based on multiple correlation method [J]. aerospace Measurement technology, 2008,28 (5): 57-60. lists of references [4]: Zheng Shengfeng, Chen Suming, Di Jinhai, etc. a kind of phase based on multiple cross-correlation Potential difference new method of measuring [J]. aerospace measurement technology, 2012,32 (1): 34-40. documents [3] and document [4] utilize the particularity of sinusoidal signal Matter, by signal carries out auto-correlation repeatedly and computing cross-correlation, recycling relative theory solves the phase contrast of signal.The party Method can improve the Phase Difference Measuring Precision in the case of low signal-to-noise ratio, but certainty of measurement is also by sampling, and whether complete cycle is affected.

In sum, the method for measuring phase difference of method of correlation has important research meaning and using value, but existing method exists all Many problems, these methods or there is the problem that amount of calculation is bigger, or cannot be completely eliminated non-integer-period sampled to measurement result Bring the impact of error, need to propose one and calculate simple, be completely eliminated and non-integer-period sampled measurement result brought error shadow The method for measuring phase difference rung.

Summary of the invention

The purpose of the present invention is to propose to a kind of calculating simply, the non-integer-period sampled phase bringing error to affect measurement result can be eliminated Potential difference measuring method, it is adaptable to arbitrarily with the phase difference measurement of frequency sinusoidal signal, solves what existing method of correlation phase difference measurement existed Subject matter, expands its range of application.

The present invention proposes the method for measuring phase difference that a kind of Data extension formula is relevant.

The present invention comprises the following steps:

The first step: frequency or cycle to signal carry out pre-estimation.Assuming that the sampled data obtaining two-way short signal is respectively x=[x₁, x₂..., x_N-1, x_N], y=[y₁, y₂..., y_N-1, y_N], P ∈ Z⁺Being the sampling number in a cycle, phase estimate result is set to P.

Second step: whether complete cycle judges to data.Assuming that N is k divided by the business of P, remainder is m.If m=0, then phase Pass length matched with the complete cycle of signal, became integral multiple relation, and now data are without processing.If m ＞ 0, correlation length is described More than k cycle but not up to k+1 cycle.

3rd step: carry out data process.Can carry out data process in two kinds of situation:

1) carry out the rejecting of data, i.e. according to the periodicity of signal, from sampled data x, y, reject m outside data complete cycle Data [x_kP+1..., x_kP+m], [y_kP+1..., y_kP+m], making correlation length is complete cycle, rejects the signal sequence after data and is:

x_e=[x₁, x₂..., x_kP], y_e=[y₁, y₂..., y_kP]

2) carry out the continuation of data, i.e. according to the periodicity of signal, from sampled data x, y, search P-m data [x_(k-1)P+m+1..., x_kP], [y_(k-1)P+m+1..., y_kP], it being incorporated into former data sequence x, after y, making correlation length is complete cycle, data Signal sequence after continuation is:

x_e=[x₁, x₂..., x_N, x_(k-1)P+m+1..., x_kP], y_e=[y₁, y₂..., y_N, y_(k-1)P+m+1, y_kP]

It is pointed out that for short duration sinusoidal signal, count less owing to obtaining the sampled data of signal, and also carry in signal Noisy interference, rejects, by data, the way that adjustment signal sequence is complete cycle and can improve algorithm measurement precision, but can reduce Data itself with quantity of information；Use the way of Data extension, on the basis of legacy data is counted, add data Quantity of information, while improving arithmetic accuracy, it is possible to improving the noise immunity of algorithm, its practicality is higher.

4th step: utilize method of correlation to calculate phase contrast.The signal x after Data extension processes_e, y_eMake auto-correlation respectively with mutual Closing computing, recycling cross-correlation function calculates phase contrast.Assuming that Δ θ is the phase contrast of two paths of signals, R_xy(0)、R_xx(0)、R_yy(0) It is respectively x_e, y_eCross-correlation and auto-correlation function, then

$\mathrm{\Δ\θ}=\arccos \left(\frac{R_{\mathrm{xy}}\left(0\right)}{\sqrt{R_{\mathrm{xx}}\left(0\right)\sqrt{R_{\mathrm{yy}}\left(0\right)}}}\right)$

Accompanying drawing explanation

Below according to accompanying drawing be embodied as that the present invention is further elaborated.To illustrate containing as a example by white noise sinusoidal signal.

Fig. 1 is method flowchart.

In figure: 1 represents known sinusoidal sequence x (n) and y (n)；2 represent that the frequency to x (n) and y (n) or cycle carry out pre-estimation；3 Representing whether complete cycle judges to data, i.e. with N divided by P, business is k, and remainder is that m, N represent that signal sampling is counted, P ∈ Z⁺ It it is the sampling number in a cycle；4 expressions carry out Data extension process complete cycle to signal x (n) and y (n)；5 represent that utilization is relevant Method phase difference calculating formula tries to achieve the phase contrast of signal；6 represent when m ≠ 0；7 represent as m=0.

Fig. 2 is the x (n) before continuation and y (n).

In figure: signal frequency is 1Hz, sample frequency is 15Hz, and signal to noise ratio is 20dB, sampling number N=20.

Fig. 3 is the x (n) after continuation and y (n).

Detailed description of the invention

The method of the present invention realizes flow process as shown in Figure 1.

The detailed description of the invention of the present invention is as follows:

Illustrate with as a example by frequency sinusoidal signal (shown in Fig. 2) by two-way.

1) to sinusoidal signal x, the frequency of y or cycle carry out pre-estimation.Phase estimate result is set to P ∈ Z⁺, P is a cycle Sampling number.

2) to data, whether complete cycle judges.N is k divided by the business of P, and remainder is m.If m=0, then correlation length and letter Number complete cycle match, become integral multiple relation, now data without process.If m ＞ 0, illustrate that correlation length is more than k cycle But not up to k+1 cycle.

3) Data extension process is carried out.I.e. according to the periodicity of signal, from sampled data x, y, search P-m data [x_(k-1)P+m+1..., x_kP], [y_(k-1)P+m+1..., y_kP], it being incorporated into former data sequence x, after y, making correlation length is complete cycle, data Signal sequence after continuation is: x_e=[x₁, x₂..., x_N, x_(k-1)P+m+1, x_kP], y_e=[y₁, y₂..., y_N, y_(k-1)P+m+1..., y_kP] (as shown in Figure 3).

4) method of correlation phase difference calculating formula is utilized to be calculated the phase contrast of signal.

$\mathrm{\Δ\θ}=\arccos \left(\frac{R_{\mathrm{xy}}\left(0\right)}{\sqrt{R_{\mathrm{xx}}\left(0\right)}\sqrt{R_{\mathrm{yy}}\left(0\right)}}\right)$

Claims (1)

1. the method for measuring phase difference that a Data extension formula is relevant, it is characterised in that: applicable object is arbitrarily with frequency sinusoidal signal Phase difference measurement；

The method comprises the following steps:

The first step: frequency or cycle to signal carry out pre-estimation, it is assumed that the sampled data obtaining two-way short signal is respectively X=[x₁, x₂..., x_N-1, x_N], y=[y₁, y₂..., y_N-1, y_N], P ∈ Z⁺Being the sampling number in a cycle, phase estimate result is set to P；

Second step: whether complete cycle judges to data, it is assumed that N is k divided by the business of P, remainder is m, if m=0, then phase Pass length matched with the complete cycle of signal, became integral multiple relation, and now data are without processing, if m ＞ 0, correlation length are described More than k cycle but not up to k+1 cycle；

3rd step: carry out data process, can carry out data process in two kinds of situation:

1) carry out the rejecting of data, i.e. according to the periodicity of signal, from sampled data x, y, reject m outside data complete cycle Data [x_kP+1..., x_kP+m], [y_kP+1..., y_kP+m], making correlation length is complete cycle, rejects the signal sequence after data and is:

x_e=[x₁, x₂..., x_kP], y_e=[y₁, y₂..., y_kP]；

2) carry out the continuation of data, i.e. according to the periodicity of signal, from sampled data x, y, search P-m data [x_(k-1)P+m+1..., x_kP], [y_(k-1)P+m+1..., y_kP], it being incorporated into former data sequence x, after y, making correlation length is complete cycle, data Signal sequence after continuation is:

x_e=[x₁, x₂..., x_N, x_(k-1)P+m+1..., x_kP], y_e=[y₁, y₂..., y_N, y_(k-1)P+m+1..., y_kP]；

4th step: utilize method of correlation to calculate phase contrast, the signal x after Data extension processes_e, y_eMake auto-correlation respectively with mutual Closing computing, recycling cross-correlation function calculates phase contrast, it is assumed that Δ θ is the phase contrast of two paths of signals, R_xy(0)、R_xx(0)、R_yy(0) It is respectively x_e, y_eCross-correlation and auto-correlation function, then

$\mathrm{\Δ}\mathrm{\θ}=arccos\left(\frac{R_{xy}\left(0\right)}{\sqrt{R_{xx}\left(0\right)}\sqrt{R_{yy}\left(0\right)}}\right).$