Online harmonic feature extraction method and system for multi-component periodic signal identification, and computer readable storage medium

Abstract

The invention discloses an online harmonic feature extraction method and system for multi-component periodic signal identification and a computer readable storage medium, and relates to the field of signal analysis and signal processing. The technical problems to be solved are that a periodic signal harmonic feature extraction method in the prior art is low in recognition precision, sensitive to noise, complex in decomposition process, poor in real-time performance and difficult to meet the signal online recognition requirement in a complex industrial control system. According to the technical key points, a wave trap with an adjustable center frequency is designed as an extreme value function, so that the purposes of dynamically identifying the frequency of a periodic signal and identifying and outputting the frequency of the periodic signal in real time are achieved; according to the invention, a band-pass filter with an adjustable center frequency is designed as an extreme value function, and a signal envelope and phase sensitivity detection composite module is combined, so that the amplitude of a periodic signal is dynamically identified, and the amplitude of the periodic signal is identified and output in real time. The method is used for effectively extracting the periodic signal output by the controller of the motor control system so as to identify the periodic interference.

Description

Technical Field

[0001] The invention relates to the field of signal analysis and signal processing, and in particular to an online harmonic feature extraction method for multi-component periodic signal recognition. Background Art

[0002] In modern signal processing and control systems, real-time recognition of periodic signals is a core requirement in many applications. Whether in the fields of communication, automatic control, industrial automation, medical signal processing, etc., accurate recognition of periodic signals is of great significance for ensuring the stability of the system and optimizing performance. For example, in communication systems, accurate signal recognition helps to improve the efficiency and stability of data transmission; in automated production, identifying external interference signals can prevent premature damage to equipment and improve production efficiency; in the medical field, the analysis of periodic physiological signals can provide effective support for disease diagnosis and treatment.

[0003] A major challenge in identifying periodic signals is that the signal frequency and amplitude may change dynamically over time, which makes it difficult for traditional signal processing methods to cope with it. The classic Fourier Transform (FT) is a widely used frequency domain analysis tool that can decompose time domain signals into several sinusoidal components. However, the Fourier Transform has the following limitations:

[0004] (1) Fourier transform requires global signal data for analysis and cannot draw valid conclusions when the signal has not been fully acquired, which makes it only suitable for offline processing;

[0005] (2) The Fourier transform assumes that the signal is stable. Therefore, when faced with non-stationary signals or signals whose frequency changes over time, it can only identify the average frequency, which greatly reduces the recognition effect.

[0006] In order to solve the shortcomings of Fourier transform in processing dynamic and non-stationary signals, researchers proposed short-time Fourier transform (STFT) and Hilbert-Huang transform (HHT). Short-time Fourier transform captures the local frequency characteristics of the signal by dividing the signal into short time windows and then performing Fourier transform on the signal in each window. However, there is an unavoidable trade-off between the selection of window width and time resolution, which limits its application in high-precision real-time signal processing. Hilbert-Huang transform decomposes the signal into a series of intrinsic mode functions (IMF) through empirical mode decomposition (EMD) and describes the time-varying characteristics of the signal by analyzing the instantaneous frequency. This method performs well in nonlinear and non-stationary signal processing, but it is sensitive to noise and the decomposition process is complicated, resulting in poor real-time performance.

[0007] Therefore, it is urgent to propose an online harmonic feature extraction method to meet the needs of online signal identification in complex industrial control systems. Summary of the invention

[0008] The technical problems to be solved by the present invention are:

[0009] The existing periodic signal harmonic feature extraction method has low recognition accuracy, is more sensitive to noise, has a complex decomposition process, and has poor real-time performance, making it difficult to meet the needs of online signal recognition in complex industrial control systems.

[0010] The present invention provides a technical solution to solve the above technical problems:

[0011] To solve the above technical problems, the present invention provides an online harmonic feature extraction method for multi-component periodic signal recognition. The harmonic feature extraction method runs in real time on the hardware platform of the motor control system, and online identifies the harmonic frequency and amplitude of the multi-component periodic signal, including the following steps:

[0012] Step 1, design an extreme value function D with an adjustable center frequency, which is used to preliminarily identify the periodic signal c input to the extreme value function D, and output the real-time frequency identification value (approximate value) of a single component in the periodic signal c;

[0013] Step 2: Establish an extreme value search closed-loop control structure based on the extreme value function D;

[0014] Step 3, selecting parameters of each link of the extreme value search closed-loop control structure to ensure that the extreme value search closed-loop control structure satisfies the time scale separation principle to ensure its stability;

[0015] Step 4, design an extreme value function S that combines the signal envelope and phase sensitive detection composite module, which is used to preliminarily identify the periodic signal c input to the extreme value function S, and output a real-time amplitude identification value (approximate value) for a single component in the periodic signal c;

[0016] Step 5: introduce a signal processor to filter out the extra periodic components introduced by the excitation signal into the extreme value search closed-loop control structure;

[0017] So far, the construction of the control structure for identifying the frequency and amplitude of a single component of the periodic signal c has been completed;

[0018] Step 6: Connect n control structures designed from steps 1 to 5 in parallel to form n identification channels, and select a different center frequency u in each identification channel. i The initial value u i0 , i = 1, 2, ..., k, k is the number of frequency components, and different parameters are set for each identification channel to achieve real-time harmonic feature extraction of multi-component signals.

[0019] Furthermore, the method for real-time frequency identification of a single component in the periodic signal c described in step 1 specifically comprises the following steps:

[0020] Step 1.1, design the extreme value function D as a band-stop filter with a center frequency u. When the center frequency u of the extreme value function is consistent with the real-time frequency of the input periodic signal c, the output of the extreme value function D has a minimum value of 0;

[0021] Step 1.2: Adjust the center frequency u of the extreme value function D in real time to suppress the periodic signal in a specific frequency range so that the output of D maintains a minimum value.

[0022] Furthermore, the specific steps of the process of establishing an extreme value search closed-loop control structure based on the extreme value function D in step 2 include:

[0023] Step 2.1, use the extreme value function D designed in step 1 as the function with extreme value in the extreme value search algorithm;

[0024] Step 2.2, analyzing the dynamic characteristics of the extreme value function D, and calculating the phase lag range generated during the signal processing;

[0025] Step 2.3, according to the phase lag range calculated in step 2.2, design a phase compensation link K, connect n phase compensation links in series to compensate for the phase lag, and the specific number n of the phase compensation links K is determined by the actual input signal;

[0026] Step 2.4: Design the integration phase The turning frequency is ω b High-pass link The two frequencies are ω e The periodic signal Asin(ω e t) and Bsin(ω e t) as the excitation signal, establish an extreme value search closed-loop control structure, where A and B are two frequencies of ω e The amplitude of the periodic signal, s is the complex variable in the Laplace transform.

[0027] Furthermore, the method for selecting parameters of each link of the extreme value search closed-loop control structure that satisfies the time scale separation principle in step 3 specifically includes the following steps:

[0028] Step 3.1: Calculate the maximum response time of the periodic signal c according to the frequency range of the input periodic signal c, denoted as τ c ;

[0029] Step 3.2: By choosing a reasonable high-pass link corner frequency ω as described in step 2.4 b , ensuring the response time τ of the extreme value search closed-loop control structure b <τ c ;

[0030] Step 3.3: By selecting a reasonable periodic excitation signal frequency ω as described in step 2.4 e , ensuring the response time τ of the excitation signal e <τ b ;

[0031] Step 3.4: Ensure the response time τ of the extreme value function D by designing a reasonable order of the extreme value function D d <τ e .

[0032] Furthermore, the method for real-time amplitude identification of a single component in the periodic signal c described in step 4 specifically comprises the following steps:

[0033] Step 4.1, design the extreme value function S as a bandpass filter with a center frequency u. When the center frequency u of the extreme value function S is consistent with the real-time frequency of the input periodic signal c, the output of the extreme value function S has a maximum value, and the maximum value is consistent with the real-time amplitude of the input periodic signal c;

[0034] Step 4.2: Use the center frequency u and the input periodic signal c to construct a signal c that is orthogonal to the input periodic signal c. ⊥ ;

[0035] Step 4.3: According to the trigonometric identity, use the input periodic signal c and its orthogonal signal c ⊥ Get a constant value that does not contain periodic components

[0036] Step 4.4: Design link P based on phase-sensitive detection technology to cope with the noise in the environment and the high-frequency components introduced in step 4.3, and preliminarily obtain the real-time amplitude approximation of the input periodic signal c.

[0037] Furthermore, the method for filtering the extra periodic components introduced by the excitation signal in the extreme value search closed-loop control structure in step 5 specifically includes the following steps:

[0038] Step 5.1: According to the periodic excitation signal frequency ω described in step 2.4 e , design the notch filters H1 and H2 with the first and third harmonics as the center frequencies;

[0039] Step 5.2: Extract the center frequency u of the extreme value function D described in step 1.1, and pass it through the series-connected notch filters H1 and H2 to obtain the real-time estimated frequency of the input periodic signal c.

[0040] Step 5.3: Extract the approximate amplitude value estimated in step 4.4 and pass it through the series-connected notch filters H1 and H2 to obtain the real-time estimated amplitude of the input periodic signal c.

[0041] Furthermore, the number n of identification channels is greater than or equal to the number k of frequency components.

[0042] Furthermore, the extreme value search closed-loop control structure in step 2.4 includes:

[0043] Input periodic signal c, first excitation signal Asin(ω e t), the second excitation signal Bsin(ω e t), extreme value function D, extreme value function S, phase detection link P, phase compensation link K, series-connected notch filters H1 and H2, integrator, and transition frequency ω b High-pass link, adder and multiplier;

[0044] The connection relationship between the modules is as follows:

[0045] The input periodic signal c and the signal u are two input signals of the extreme value function S, and the output signal of the extreme value function S is used as the input signal of the phase detection link P, forming branch 1;

[0046] The input periodic signal c and the signal u are two input signals of the extreme value function D, and the output signal of the extreme value function D is used as the input signal of the phase compensation link K to form branch 2;

[0047] The branch 1 is connected in parallel with the branch 2;

[0048] The output signal of the phase detection link P and the signal u are the input signals of the series-connected notch filters H1 and H2;

[0049] The turning frequency is ω b The input signal of the high-pass link is the output signal of the phase compensation link K, and its output signal is consistent with the second excitation signal Bsin(ω e t) as the integrator input signal through the multiplier;

[0050] The output signal of the integrator is related to the first excitation signal Asin(ω e t) The signal u is calculated by the adder;

[0051] The signal u is an identifiable variable, and the signal u is one of the input signals of the extreme value function D, the extreme value function S and the series-connected notch filters H1 and H2, and is also the center frequency of the band-stop filter in step 1.1 and the band-pass filter in step 4.1;

[0052] The output amplitude of the series-connected notch filters H1 and H2 is estimated in real time With real-time estimated frequency

[0053] The present invention also provides an online harmonic feature extraction system for multi-component periodic signal identification, which has a program module corresponding to the steps of the method described in any one of the above technical solutions, and executes the steps in the above-mentioned online harmonic feature extraction method for multi-component periodic signal identification during operation.

[0054] The present invention also provides a computer-readable storage medium, which stores a computer program, and the computer program is configured to implement the steps of the online harmonic feature extraction method for multi-component periodic signal identification of the method described in any of the above technical solutions when called by a processor.

[0055] Compared with the prior art, the beneficial technical effects of the present invention are:

[0056] (1) Since the frequency of a periodic signal does not contain extreme value information, the present invention realizes dynamic identification of the frequency of a periodic signal by designing a notch filter with an adjustable center frequency as an extreme value function, thereby achieving the purpose of real-time identification and output of the frequency of the periodic signal;

[0057] (2) Since the amplitude of a periodic signal does not contain extreme value information, the present invention realizes dynamic identification of the amplitude of a periodic signal by designing a bandpass filter with an adjustable center frequency as an extreme value function and combining a signal envelope and a phase sensitive detection composite module, thereby achieving the purpose of real-time identification and output of the amplitude of a periodic signal;

[0058] (3) A signal processor is introduced to assist in filtering out the periodic signal introduced by the excitation signal in the extreme value search closed-loop control structure, thus ensuring the purity of the identification results;

[0059] (4) The parallel structure is used to process multiple signal channels, which expands the application field of multi-component signal recognition, and is particularly suitable for scenarios with complex industrial automation systems containing interference signals of various periodic forms.

[0060] The method of the present invention is applied to a precision servo mechanical system with multiple periodic component interferences. Taking identification accuracy as an indicator, it is verified that the method of the present invention can achieve the purpose of real-time identification of the frequencies and amplitudes of multiple periodic component interferences in a noisy environment.

[0061] The method of the invention is used for effectively extracting the periodic signal output by the controller of the motor control system to identify the periodic interference. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a flow chart of an online harmonic feature extraction method for multi-component periodic signal recognition in an embodiment of the present invention;

[0063] Figure 2 A block diagram of an online harmonic feature extraction method for multi-component periodic signal recognition in an embodiment of the present invention;

[0064] Figure 3 It is a simulation result curve of the Fourier transform method (prior art) in the simulation comparison experiment of the embodiment of the present invention;

[0065] Figure 4 It is a simulation result curve of the Hilbert-Huang transform method (prior art) in the simulation comparison experiment of the embodiment of the present invention;

[0066] Figure 5 It is a simulation result curve of the support vector machine method (existing technology) in the simulation comparison experiment of the embodiment of the present invention;

[0067] Figure 6 It is a simulation result curve of the method of the present invention in the simulation comparison experiment of the embodiment of the present invention;

[0068] Figure 7 It is a schematic diagram of the method of the present invention used for online harmonic feature extraction of disturbance signals of a motor control system in an embodiment of the present invention;

[0069] Figure 8 This is a multi-component periodic signal feature extraction result curve for online harmonic feature extraction of disturbance signals in a motor control system using the method of the present invention in an embodiment of the present invention. DETAILED DESCRIPTION

[0070] In order to enable those skilled in the art to better understand the scheme of the present invention, exemplary implementations or embodiments of the present invention will be described below in conjunction with the accompanying drawings. Obviously, the described implementations or embodiments are only implementations or embodiments of a part of the present invention, not all of them. Based on the implementations or embodiments of the present invention, all other implementations or embodiments obtained by ordinary technicians in the field without creative work should fall within the scope of protection of the present invention.

[0071] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0072] Example 1

[0073] like Figure 1 and Figure 2 As shown, the present invention provides an online harmonic feature extraction method for multi-component periodic signal recognition, the harmonic feature extraction method runs in real time on the hardware platform of the motor control system, and identifies the harmonic frequency and amplitude of the multi-component periodic signal online, including the following steps:

[0074] Step 1, design an extreme value function D with an adjustable center frequency, which is used to preliminarily identify the periodic signal c input to the extreme value function D, and output the real-time frequency identification value (approximate value) of a single component in the periodic signal c;

[0075] The method for real-time frequency identification of a single component in a periodic signal c specifically comprises the following steps:

[0076] Step 1.1, design the extreme value function D as a band-stop filter with a center frequency u. When the center frequency u of the extreme value function is consistent with the real-time frequency of the input periodic signal c, the output of the extreme value function D has a minimum value of 0;

[0077] Step 1.2: Adjust the center frequency u of the extreme value function D in real time to suppress the periodic signal in a specific frequency range so that the output of D maintains a minimum value.

[0078] Step 2: Establish an extreme value search closed-loop control structure based on the extreme value function D;

[0079] The specific steps of the process of establishing an extreme value search closed-loop control structure based on the extreme value function D include:

[0080] Step 2.1, use the extreme value function D designed in step 1 as the function with extreme value in the extreme value search algorithm;

[0081] Step 2.2, analyzing the dynamic characteristics of the extreme value function D, and calculating the phase lag range generated during the signal processing;

[0082] Step 2.3, according to the phase lag range calculated in step 2.2, design a phase compensation link K, connect n phase compensation links in series to compensate for the phase lag, and the specific number n of the phase compensation links K is determined by the actual input signal;

[0083] Step 2.4: Design the integration phase The turning frequency is ω b High-pass link The two frequencies are ω e The periodic signal Asin(ω e t) and Bsin(ω e t) as the excitation signal, establish Figure 2 The extreme value search closed-loop control structure shown in the figure, where A and B are two frequencies ω e The amplitude of the periodic signal, s is the complex variable in the Laplace transform.

[0084] The extreme value search closed-loop control structure in step 2.4 includes:

[0085] Input periodic signal c, first excitation signal Asin(ω e t), the second excitation signal Bsin(ω e t), extreme value function D, extreme value function S, phase detection link P, phase compensation link K, series-connected notch filters H1 and H2, integrator, and transition frequency ω b High-pass link, adder and multiplier;

[0086] The connection relationship between the modules is as follows:

[0087] The input periodic signal c and the signal u are two input signals of the extreme value function S, and the output signal of the extreme value function S is used as the input signal of the phase detection link P, forming branch 1;

[0088] The input periodic signal c and the signal u are two input signals of the extreme value function D, and the output signal of the extreme value function D is used as the input signal of the phase compensation link K to form branch 2;

[0089] The branch 1 is connected in parallel with the branch 2;

[0090] The output signal of the phase detection link P and the signal u are the input signals of the series-connected notch filters H1 and H2;

[0091] The turning frequency is ω b The input signal of the high-pass link is the output signal of the phase compensation link K, and its output signal is consistent with the second excitation signal Bsin(ω e t) as the integrator input signal through the multiplier;

[0092] The output signal of the integrator is related to the first excitation signal Asin(ω e t) The signal u is calculated by the adder;

[0093] The signal u is an identifiable variable, and the signal u is one of the input signals of the extreme value function D, the extreme value function S and the series-connected notch filters H1 and H2, and is also the center frequency of the band-stop filter in step 1.1 and the band-pass filter in step 4.1;

[0094] The output amplitude of the series-connected notch filters H1 and H2 is estimated in real time With real-time estimated frequency

[0095] Step 3, selecting parameters of each link of the extreme value search closed-loop control structure to ensure that the extreme value search closed-loop control structure satisfies the time scale separation principle to ensure its stability;

[0096] The method for selecting parameters of each link of the extreme value search closed-loop control structure that meets the time scale separation principle specifically comprises the following steps:

[0097] Step 3.1: Calculate the maximum response time of the periodic signal c according to the frequency range of the input periodic signal c, denoted as τ c ;

[0098] Step 3.2: By choosing a reasonable high-pass link corner frequency ω as described in step 2.4 b , ensuring the response time τ of the extreme value search closed-loop control structure b <τ c ;

[0099] Step 3.3: By selecting a reasonable periodic excitation signal frequency ω as described in step 2.4 e , ensuring the response time τ of the excitation signal e <τ b ;

[0100] Step 3.4: Ensure the response time τ of the extreme value function D by designing a reasonable order of the extreme value function D d <τ e .

[0101] Step 4, design an extreme value function S that combines the signal envelope and phase sensitive detection composite module, which is used to preliminarily identify the periodic signal c input to the extreme value function S, and output a real-time amplitude identification value (approximate value) for a single component in the periodic signal c;

[0102] The method for real-time amplitude identification of a single component in a periodic signal c specifically comprises the following steps:

[0103] Step 4.1, design the extreme value function S as a bandpass filter with a center frequency u. When the center frequency u of the extreme value function S is consistent with the real-time frequency of the input periodic signal c, the output of the extreme value function S has a maximum value, and the maximum value is consistent with the real-time amplitude of the input periodic signal c;

[0104] Step 4.2: Use the center frequency u and the input periodic signal c to construct a signal c that is orthogonal to the input periodic signal c. ⊥ ;

[0105] Step 4.3: According to the trigonometric identity, use the input periodic signal c and its orthogonal signal c ⊥ Get a constant value that does not contain periodic components

[0106] Step 4.4: Design link P based on phase-sensitive detection technology to cope with the noise in the environment and the high-frequency components introduced in step 4.3, and preliminarily obtain the real-time amplitude approximation of the input periodic signal c.

[0107] Step 5: introduce a signal processor to filter out the extra periodic components introduced by the excitation signal into the extreme value search closed-loop control structure;

[0108] The method for filtering the extra periodic components introduced by the excitation signal in the closed-loop control structure of the extreme value search comprises the following specific steps:

[0109] Step 5.1: According to the periodic excitation signal frequency ω described in step 2.4 e , design the notch filters H1 and H2 with the first and third harmonics as the center frequencies;

[0110] Step 5.2: Extract the center frequency u of the extreme value function D described in step 1.1, and pass it through the series-connected notch filters H1 and H2 to obtain the real-time estimated frequency of the input periodic signal c.

[0111] Step 5.3: Extract the approximate amplitude value estimated in step 4.4 and pass it through the series-connected notch filters H1 and H2 to obtain the real-time estimated amplitude of the input periodic signal c.

[0112] At this point, the construction of the control structure for identifying the frequency and amplitude of a single component of the periodic signal c is completed.

[0113] Step 6: Connect n control structures designed from steps 1 to 5 in parallel to form n identification channels, and select a different center frequency u in each identification channel. i The initial value u i0 , i = 1, 2, ..., k, k is the number of frequency components, and different parameters are set for each identification channel to achieve real-time harmonic feature extraction of multi-component signals;

[0114] The number of identification channels n is greater than or equal to the number of frequency components k.

[0115] In order to simulate the experimental environment, numerical simulations were carried out under multiple environments for the Fourier transform method, the Hilbert-Huang transform method, the support vector machine method and the method of the present invention to discuss the noise sensitivity, the recognition ability of time-varying frequency signals and the real-time performance of the above methods.

[0116] The Fourier transform method is an offline method that must be analyzed after obtaining complete signal data. The simulation results of the Fourier transform method are as follows: Figure 3 As shown, Figure 3 (a) is the simulation experiment result curve for identifying constant frequency signal; Figure 3 (b) is the curve of simulation experiment result for identifying time-varying frequency signal; Figure 3 (c) is the simulation experiment result curve for identifying time-varying and then constant frequency signals. Figure 3 The simulation results shown show that this method:

[0117] 1. Insensitive to noise, it has almost the same recognition ability in a noisy environment as in a noise-free environment;

[0118] 2. It cannot reflect the time characteristics of the signal, that is, it cannot process time-varying signals and can only output its average frequency;

[0119] 3. Identification is not real-time and must be performed offline.

[0120] The Hilbert-Huang transform method is also an offline method, which must be analyzed after obtaining complete signal data. The simulation results of the Hilbert-Huang transform method are as follows: Figure 4 As shown, Figure 4 (a) is the simulation experiment result curve for identifying constant frequency signal; Figure 4 (b) is the curve of simulation experiment result for identifying time-varying frequency signal; Figure 4 (c) is the simulation experiment result curve for identifying time-varying and then constant frequency signals. Figure 4 The simulation results shown show that this method:

[0121] 1. Very sensitive to noise, and it is almost impossible to obtain effective results in a noisy environment;

[0122] 2. It can reflect the time characteristics of the signal and process time-varying signals, but there is a "boundary effect", that is, undesirable divergent results are produced at the beginning, end or intersection of the data;

[0123] 3. Identification is not real-time and must be performed offline.

[0124] The support vector machine method is an online method that requires a large amount of data to learn first, and then it can achieve online recognition. The simulation results of the support vector machine method are as follows Figure 5 As shown, Figure 5 (a) is the curve of the simulation experiment result of identifying constant frequency signal in the absence of noise; Figure 5 (b) is the simulation experiment result curve of identifying time-varying and then constant frequency signal in the absence of noise; Figure 5 (c) is the simulation experiment result curve of identifying constant frequency signal in the case of noise; (d) is the simulation experiment result curve of identifying time-varying and then constant frequency signal in the case of noise. Figure 5 The simulation results shown show that this method:

[0125] 1. Insensitive to noise, it has almost the same recognition ability in a noisy environment as in a noise-free environment;

[0126] 2. Able to reflect the time characteristics of signals and process time-varying signals;

[0127] 3. Recognition has a "certain" real-time nature. It can achieve online recognition after learning a large number of data sets, but the learning process is extremely time-consuming.

[0128] The method of the present invention is an online method that can output the identification value in real time as the data is output. The simulation results of the method of the present invention are as follows: Figure 6 As shown, Figure 6 (a) is the constant frequency signal simulation experiment result curve for identifying two components; Figure 6 (b) is the simulation experiment result curve for identifying the frequency signal that first changes in time and then becomes constant. Figure 6 The simulation results shown show that the method of the present invention:

[0129] 1. Noise has a slight impact on recognition time, but does not affect recognition accuracy;

[0130] 2. Able to reflect the time characteristics of signals and process time-varying signals;

[0131] 3. The recognition is real-time and can output the recognition value in real time with the acquired data;

[0132] 4. Capable of processing multi-component signals.

[0133] contrast Figure 3 to Figure 6The experimental results shown in the figure show that the method of the present invention can realize real-time identification and output of the frequency and amplitude of multi-component periodic signals. Its identification accuracy is less affected by noise and can process time-varying signals. It is suitable for application in complex industrial automation systems. Its effect on online harmonic feature extraction of multi-component periodic signals is significantly better than the existing technical methods.

[0134] Example 3

[0135] like Figure 7 As shown, the method of the present invention is applied to the online harmonic feature extraction of the interference signal of the motor control system containing two known components of interference, with the identification accuracy as the index. The identified interference signal is the voltage / current interference signal that affects the motor angular position output.

[0136] The process of applying the method of the present invention to the hardware platform of the motor control system is as follows:

[0137] The method of the present invention is configured in the industrial control computer according to steps 1 to 6, wherein the number of identification channels n is set to 2. The motor control system is made to run at a variable speed first and then at a constant speed. The method of the present invention extracts the control signal of the motor control system controller as the signal to be identified c, outputs the identification value in real time, and obtains the multi-component periodic signal feature extraction result curve as shown in FIG. Figure 8 shown.

[0138] From Figure 8 It can be seen from the experimental result curve shown that the method of the present invention can adapt to the change of interference frequency and realize real-time identification of interference frequency; at the same time, it can also realize real-time identification of interference amplitude in a relatively short time; in the same experiment, facing the interference of two components, the method of the present invention can separately identify the interference of the two components.

[0139] Example 4

[0140] The present invention also provides an online harmonic feature extraction system for multi-component periodic signal identification, which has a program module corresponding to the steps of the method described in any one of the technical solutions of Example 1 and Example 2, and executes the steps of the above-mentioned online harmonic feature extraction method for multi-component periodic signal identification during operation.

[0141] Example 5

[0142] The present invention also provides a computer-readable storage medium, which stores a computer program, and the computer program is configured to implement the steps of the online harmonic feature extraction method for multi-component periodic signal identification of the method described in any one of the technical solutions of Example 1 and Example 2 when called by a processor.

[0143] Although the present invention is disclosed as above, the protection scope of the present invention is not limited thereto. Those skilled in the art may make various changes and modifications without departing from the spirit and scope of the present invention, and these changes and modifications will fall within the protection scope of the present invention.

Claims

1. An online harmonic feature extraction method for multi-component periodic signal recognition, characterized in that: The harmonic feature extraction method runs in real time on the hardware platform of the motor control system, and identifies the harmonic frequency and amplitude of the multi-component periodic signal online, including the following steps: Step 1, design an extreme value function D with an adjustable center frequency, which is used to preliminarily identify the periodic signal c input to the extreme value function D, and output the real-time frequency identification value of a single component in the periodic signal c; Step 2: Establish an extreme value search closed-loop control structure based on the extreme value function D; Step 3, selecting parameters of each link of the extreme value search closed-loop control structure to ensure that the extreme value search closed-loop control structure satisfies the time scale separation principle to ensure its stability; Step 4, design an extreme value function S that combines the signal envelope and phase sensitive detection composite module, which is used to preliminarily identify the periodic signal c input to the extreme value function S, and output the real-time amplitude identification value of a single component in the periodic signal c; Step 5: introduce a signal processor to filter out the extra periodic components introduced by the excitation signal into the extreme value search closed-loop control structure; So far, the construction of the control structure for identifying the frequency and amplitude of a single component of the periodic signal c has been completed; Step 6: Connect n control structures designed from steps 1 to 5 in parallel to form n identification channels, and select a different center frequency u in each identification channel. i The initial value u i0 , i = 1, 2, ..., k, k is the number of frequency components, and different parameters are set for each identification channel to achieve real-time harmonic feature extraction of multi-component signals.

2. An online harmonic feature extraction method for multi-component periodic signal identification according to claim 1, characterized in that: The method for real-time frequency identification of a single component in the periodic signal c described in step 1 specifically comprises the following steps: Step 1.1, design the extreme value function D as a band-stop filter with a center frequency u. When the center frequency u of the extreme value function is consistent with the real-time frequency of the input periodic signal c, the output of the extreme value function D has a minimum value of 0; Step 1.2: Adjust the center frequency u of the extreme value function D in real time to suppress the periodic signal in a specific frequency range so that the output of D maintains a minimum value.

3. An online harmonic feature extraction method for multi-component periodic signal identification according to claim 2, characterized in that: The specific steps of the process of establishing an extreme value search closed-loop control structure based on the extreme value function D in step 2 include: Step 2.1, use the extreme value function D designed in step 1 as the function with extreme value in the extreme value search algorithm; Step 2.2, analyzing the dynamic characteristics of the extreme value function D, and calculating the phase lag range generated during the signal processing; Step 2.3, according to the phase lag range calculated in step 2.2, design a phase compensation link K, connect n phase compensation links in series to compensate for the phase lag, and the specific number n of the phase compensation links K is determined by the actual input signal; Step 2.4: Design the integration phase The turning frequency is ω b High-pass link The two frequencies are ω e The periodic signal Asin(ω e t) and Bsin(ω e t) as the excitation signal, establish an extreme value search closed-loop control structure, where A and B are two frequencies of ω e The amplitude of the periodic signal, s is the complex variable in the Laplace transform.

4. An online harmonic feature extraction method for multi-component periodic signal identification according to claim 3, characterized in that: The method for selecting parameters of each link of the extreme value search closed-loop control structure that meets the time scale separation principle described in step 3 specifically includes the following steps: Step 3.1: Calculate the maximum response time of the periodic signal c according to the frequency range of the input periodic signal c, denoted as τ c ; Step 3.2: By choosing a reasonable high-pass link corner frequency ω as described in step 2.4 b , ensuring the response time τ of the extreme value search closed-loop control structure b τ c ; Step 3.3: By selecting a reasonable periodic excitation signal frequency ω as described in step 2.4 e , ensuring the response time τ of the excitation signal e <τ b ; Step 3.4: Ensure the response time τ of the extreme value function D by designing a reasonable order of the extreme value function D d <τ e .

5. An online harmonic feature extraction method for multi-component periodic signal identification according to claim 4, characterized in that: The method for real-time amplitude identification of a single component in the periodic signal c described in step 4 specifically comprises the following steps: Step 4.1, design the extreme value function S as a bandpass filter with a center frequency u. When the center frequency u of the extreme value function S is consistent with the real-time frequency of the input periodic signal c, the output of the extreme value function S has a maximum value, and the maximum value is consistent with the real-time amplitude of the input periodic signal c; Step 4.2: Use the center frequency u and the input periodic signal c to construct a signal c that is orthogonal to the input periodic signal c. ⊥ ; Step 4.3: According to the trigonometric identity, use the input periodic signal c and its orthogonal signal c ⊥ Get a constant value that does not contain periodic components Step 4.4: Design link P based on phase-sensitive detection technology to cope with the noise in the environment and the high-frequency components introduced in step 4.3, and preliminarily obtain the real-time amplitude approximation of the input periodic signal c.

6. The online harmonic feature extraction method for multi-component periodic signal identification according to claim 5, characterized in that: The method for filtering the extra periodic components introduced by the excitation signal in the extreme value search closed-loop control structure in step 5 specifically comprises the following steps: Step 5.1: According to the periodic excitation signal frequency ω described in step 2.4 e , design the notch filters H1 and H2 with the first harmonic and the third harmonic as the center frequency; Step 5.2: Extract the center frequency u of the extreme value function D described in step 1.1, and pass it through the series-connected notch filters H1 and H2 to obtain the real-time estimated frequency of the input periodic signal c. Step 5.3: Extract the approximate amplitude value estimated in step 4.4 and pass it through the series-connected notch filters H1 and H2 to obtain the real-time estimated amplitude of the input periodic signal c.

7. An online harmonic feature extraction method for multi-component periodic signal identification according to claim 6, characterized in that: The number n of the identification channels is greater than or equal to the number k of the frequency components.

8. An online harmonic feature extraction method for multi-component periodic signal identification according to claim 7, characterized in that: The extreme value search closed-loop control structure in step 2.4 includes: Input periodic signal c, first excitation signal Asin(ω e t), the second excitation signal Bsin(ω e t), extreme value function D, extreme value function S, phase detection link P, phase compensation link K, series-connected notch filters H1 and H2, integrator, and transition frequency ω b High-pass link, adder and multiplier; The connection relationship between the modules is as follows: The input periodic signal c and the signal u are two input signals of the extreme value function S, and the output signal of the extreme value function S is used as the input signal of the phase detection link P, forming branch 1; The input periodic signal c and the signal u are two input signals of the extreme value function D, and the output signal of the extreme value function D is used as the input signal of the phase compensation link K to form branch 2; The branch 1 is connected in parallel with the branch 2; The output signal of the phase detection link P and the signal u are the input signals of the series-connected notch filters H1 and H2; The turning frequency is ω b The input signal of the high-pass link is the output signal of the phase compensation link K, and its output signal is consistent with the second excitation signal Bsin(ω e t) as the integrator input signal through the multiplier; The output signal of the integrator is related to the first excitation signal Asin(ω e t) The signal u is calculated by the adder; The signal u is an identifiable variable, and the signal u is one of the input signals of the extreme value function D, the extreme value function S and the series-connected notch filters H1 and H2, and is also the center frequency of the band-stop filter in step 1.1 and the band-pass filter in step 4.1; The output amplitude of the series-connected notch filters H1 and H2 is estimated in real time With real-time estimated frequency 9. An online harmonic feature extraction system for multi-component periodic signal recognition, characterized in that: The system has a program module corresponding to the steps of the method described in any one of claims 1 to 8, and executes the steps of the online harmonic feature extraction method for multi-component periodic signal identification when running.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program is configured to implement the steps of the online harmonic feature extraction method for multi-component periodic signal identification according to any one of claims 1 to 8 when called by a processor.

Images