A frequency signal real-time monitoring method based on space reconstruction and principal component analysis

Abstract

The application discloses a frequency signal real-time monitoring method based on space reconstruction and principal component analysis. For the collected time domain signal, space reconstruction and power spectrum density analysis are adopted to obtain a multi-dimensional power spectrum density matrix, then principal component analysis and standardization processing are performed on the power spectrum density matrix, and a time-frequency diagram of a signal feature is drawn according to the screened principal components. The application can obviously reduce the interference of noise signals in a flow field, and can monitor the change of a characteristic signal with time during the measurement process, and has a good practical application prospect.

Description

TECHNICAL FIELD

[0001] The application relates to a frequency signal real-time monitoring method based on space reconstruction and principal component analysis, and belongs to the field of dynamic testing and parameter identification. BACKGROUND

[0002] When dynamic parameters of a flow field are measured and frequency analysis is performed, various signal interferences exist in the flow field, so that accurate frequency characteristics in the flow field cannot be obtained. When frequency characteristics of a disturbance source are mixed in background noise of the flow field, with the increase of a flow speed, noise sources in the flow field gradually increase, energy of noise signal characteristics is too high, and the noise signal characteristics often disturb signal discrimination in the whole frequency band, so that main useful signals cannot be quickly extracted.

[0003] Traditional fast Fourier transform (FFT) and power spectral density analysis (PSD) often obtain a large amount of frequency domain signals, and characteristic frequencies are often submerged, so that the characteristic frequencies cannot be accurately extracted. In general noise reduction methods, wavelet transform, empirical mode decomposition and HHT transform often obtain time domain information under a combination of many frequency signals, and the calculation amount is large, especially when a frequency band occupied by noise signals is very wide, main characteristic frequencies cannot be separated. In addition, when frequency domain characteristics of signals are extracted, a large amount of time domain information is extracted and analyzed after a test is completed, signal analysis is not timely, and real-time extraction of frequency information cannot be met. SUMMARY

[0004] In view of the noise interference, the application provides a frequency dynamic monitoring method based on space reconstruction and principal component analysis. Real-time time domain signals are normalized, a multi-dimensional premultiplied power spectral density matrix is obtained by using space reconstruction and power spectral density analysis, main frequency characteristics are obtained by performing principal component analysis and screening on the premultiplied power spectral density matrix, the main frequency distribution with time is drawn, the premultiplied power spectral density matrix is averaged and standardized, and a time-frequency diagram of flow signal characteristics is obtained.

[0005] To solve the problems in the prior art, the technical scheme adopted by the application is as follows.

[0006] A frequency dynamic monitoring method based on space reconstruction and principal component analysis comprises the following steps.

[0007] Step 1, given the length t and dimension d of read data, real-time time domain signal data p(t) is read;

[0008] Step 2, a multi-dimensional array matrix A(d,t) is constructed according to the real-time time domain signal data read in step 1;

[0009] Step 3, the multi-dimensional array matrix A(d,t) is standardized;

[0010] Step 4, power spectrum density calculation is performed on each row vector of the standardized multi-dimensional array matrix to obtain a power density matrix P d (d,t);

[0011] Step 5, a premultiplied power spectrum density matrix P dr (d,t) is constructed;

[0012] Step 6, principal component analysis is performed on the premultiplied power spectrum density matrix P dr (d,t) and screening is performed according to the contribution rate, the contribution rate threshold can be set according to the actual situation, the components are screened and reconstructed, and the main frequency is obtained;

[0013] Step 7, the premultiplied power spectrum density matrix P dr (d,t) is averaged and standardized;

[0014] Step 8, a main frequency characteristic time-frequency diagram is drawn to analyze the time domain and frequency domain characteristics of the signal;

[0015] Step 9, with real-time acquisition of the signal, the multi-dimensional array is updated, steps 2-8 are repeated, the dynamic change of the frequency domain characteristic signal with time is obtained, and real-time monitoring of the characteristic signal is realized.

[0016] As an improvement, the power spectrum density calculation method in step 4 is a periodogram method, an autocorrelation method or a parameter model spectrum estimation method.

[0017] Further improvement is that the power spectrum density calculation formula in step 4 is as follows:

[0018]

[0019]

[0020] Wherein, T is a period, G T (f) is the Fourier transform of the energy signal g T (t), P d (f) is the power spectrum density, with a unit of W / Hz.

[0021] As an improvement, the contribution rate in step 6 is 0-1, which can be selected according to the actual situation.

[0022] Beneficial effects:

[0023] Compared with the prior art, the frequency dynamic monitoring method based on space reconstruction and principal component analysis can reduce the interference of noise signals in the flow field, can monitor the change of the characteristic signal with time during the measurement process, can realize real-time dynamic extraction of the main frequency characteristics, and can accurately obtain the time-frequency graph in time. BRIEF DESCRIPTION OF DRAWINGS

[0024] Figure 1 The flow chart of the frequency dynamic monitoring method based on space reconstruction and principal component analysis.

[0025] Figure 2 The pre-multiplied power spectrum density time-frequency graph. DETAILED DESCRIPTION

[0026] The specific embodiments of the present application will be further described below in combination with the drawings and test scenarios.

[0027] The test scenario selected in the present application is a flat plate environment air film cooling flow field. The main flow velocity of the low-speed wind tunnel is about 60 m / s. The cold air is mixed with the main flow through the air film hole. The degree of mixing can be represented by the blowing ratio (the ratio of the cold air outlet density flow to the main flow density flow). Different blowing ratios can be obtained by changing the cold air flow. The vortex flow caused by the mixing of the main flow and the cold air will cause periodic disturbance to the pressure of the flow field, and the disturbance frequency will change with the change of the blowing ratio. In the flat plate air film cooling flow field, the wall surface pressure fluctuation signal after the air film hole is monitored. After data processing, the change of the flow field frequency signal at the monitoring position in the flow field can be obtained. The frequency signal can represent the current state of the flow field. Accurate identification of the characteristic signal can provide strong support for monitoring the development of the flow.

[0028] A frequency dynamic monitoring method based on space reconstruction and principal component analysis, comprising the following steps:

[0029] Step 1: According to the requirements of time and frequency resolution, the length t and dimension d of the time domain signal are selected, and a certain segment of time domain signal p(t) of the wall surface pressure fluctuation is read. The longer the data is, the larger the dimension is, and the slower the calculation time will be. It can be adjusted according to the actual situation. In this embodiment, the frequency is 40 kHz, the signal length t is 2500, and d is 16.

[0030] Step 2: The real-time data p(t) of the time domain signal read in step 1 is subjected to space reconstruction to construct a multi-dimensional matrix A(d, t).

[0031] Step 3, standardizing the multi-dimensional array matrix A(d, t).

[0032] Step 4, calculating the power spectral density of each row vector of the standardized multi-dimensional array matrix to obtain the power density matrix P d (d, t) at different frequencies, and the formula of the power spectral density is as follows:

[0033]

[0034]

[0035] Wherein, T is a period, G T (f) is the Fourier transform of the energy signal g T (t), P d (f) is the power spectral density, with the unit of W / Hz. The solving method of the power spectral density of the signal includes the periodogram method, the autocorrelation method and the parameter model spectrum estimation method, etc.

[0036] Step 5, constructing the premultiplied power spectral density matrix P dr (d, t).

[0037] Step 6, carrying out principal component analysis on the premultiplied power spectral density matrix P dr (d, t), and selecting according to the contribution rate (the value is 0-1), the threshold of the contribution rate can be selected according to the actual situation, in the embodiment, the characteristics with the contribution rate greater than 0.8 are selected for data screening and reconstruction, and the main frequency is obtained.

[0038] Step 7, averaging and standardizing the premultiplied power spectral density matrix P dr (d, t);

[0039] Step 8, drawing the time-frequency graph of the main frequency characteristics to analyze the time domain and frequency domain characteristics of the signal.

[0040] Step 9, updating the multi-dimensional array with the real-time acquisition of the signal, repeating steps 2-8 to obtain the dynamic change of the frequency domain characteristic signal with time, and realizing the real-time monitoring of the characteristic signal.

[0041] The above describes the present application and its embodiments in a schematic manner, which is not limited, and the embodiment shown in the drawings is only one of the embodiments of the present application, and the actual structure is not limited thereto. Therefore, if a person skilled in the art is inspired thereby, without departing from the purpose of the present application, similar structural modes and embodiments can be designed without creativity, which should belong to the protection scope of the present application.

Claims

1. A method for frequency dynamic monitoring based on spatial reconstruction and principal component analysis, characterized in that, The specific steps are: Step 1. In the flat plate environment of the air film cooling flow field, according to the resolution requirements of time and frequency, the length t and dimension d of the time domain signal are selected, and a certain time domain signal p(t) of the wall surface pressure fluctuation is read; Step 2. A multi-dimensional array matrix A(d,t) is constructed according to the real-time data of the time domain signal read in step 1; Step 3. The multi-dimensional array matrix A(d,t) is normalized; Step 4, power spectral density calculation is performed on each row vector of the normalized multi-dimensional array matrix to obtain a power density matrix at different frequencies ; Step 5, Constructing the pre-multiplied power spectral density matrix ; Step 6, pre-multiply the power spectral density matrix The principal component analysis is carried out, the components are screened according to the contribution rate, the contribution rate threshold is set according to the actual situation, the components are screened and reconstructed, and then the main frequency is obtained. Step 7, average and normalize the pre-multiplied power spectral density matrix performing averaging and normalization processes; Step 8. The main frequency characteristic time-frequency diagram is drawn, and the time domain and frequency domain characteristics of the signal are analyzed; Step 9. With the real-time acquisition of the signal, the multi-dimensional array is updated, steps 2-8 are repeated, the dynamic change of the frequency domain characteristic signal with time is obtained, and the real-time monitoring of the characteristic signal is realized.

2. The method of claim 1, wherein, The power spectrum density calculation method in step 4 is a periodogram method, an autocorrelation method or a parameter model spectrum estimation method.

3. The method of claim 2, wherein, The power spectrum density calculation formula in step 4 is as follows: ， where T is the period, is the Fourier transform of the energy signal is the power spectral density in W / Hz.​ 4. The method of claim 1, wherein, The contribution rate in step 6 is 0-1.

Images