Train traction system filter reactor anomaly detection method and system and storage medium

By improving the Hilbert transformation and signal envelope screening methods, the complexity of reactor online status monitoring in the train traction system is solved, reliable abnormality detection of reactors is realized, and the accuracy and stability of parameter identification are improved, so as to avoid Use of additional hardware.

CN119087306APending Publication Date: 2024-12-06CRRC QINGDAO SIFANG ROLLING STOCK RESEARCH INSTITUTE CO LTD

Patent Information

Application Number
CN202411217810.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-02
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

In the prior art, in the train traction system, the complex working conditions of the reactor online status monitoring lead to unstable signal harmonic analysis, the additional hardware increases cost and complexity, and is susceptible to the environment, and the complex voltage and current components of the reactor lead to difficulty in online parameter identification. It is large and difficult to achieve reliable reactor abnormality detection.

Method used

Using the improved Hilbert transform algorithm and signal envelope screening method, the reactor voltage signal and current signal are collected, the target resonance frequency is determined, bandpass filtering is performed, and signals that meet the parameter identification conditions are screened, and the least squares method is used to react. Device parameter identification to determine whether there is an abnormality in the reactor state.

Benefits of technology

The reactor abnormality detection based on online parameter identification is realized, which improves the accuracy and stability of parameter identification. It uses existing sensor signals to perform stable and reliable online reactor parameter identification and abnormality monitoring.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119087306A_ABST
    Figure CN119087306A_ABST
Patent Text Reader

Abstract

The invention provides a train traction system filter reactor anomaly detection method and system and a storage medium. Acquiring a voltage signal and a current signal of the reactor; determining a target resonant frequency of the reactor, and determining a target frequency band based on the target resonant frequency; carrying out filtering processing on the collected current signals and voltage signals of the reactor through improved Hilbert transform, and removing voltage and current harmonic components outside a target frequency band to obtain analysis signals corresponding to the voltage and current signals; the instantaneous envelope of the voltage and current signals can be calculated by using the analysis signal, and the voltage signal and the current signal which meet the condition and only reserve the target frequency band component can be screened out based on the instantaneous envelope for electric reactor parameter identification; based on the obtained electric reactor current signal and the electric reactor voltage signal meeting the parameter identification condition, electric reactor parameter identification is carried out by adopting a recursive least square method, and parameters comprise electric reactor resistance and electric reactor inductance; and according to the identification result, whether the state of the reactor is abnormal can be judged.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of electrical detection, and in particular to a method, system and storage medium for detecting abnormality of a filter reactor of a train traction system. Background Art

[0002] In special application scenarios such as train traction systems, reactors are often added in series on the DC side of the main circuit of the converter system. Their main purpose is to limit the rate of change of the DC supply current, reduce the harm caused by system short circuits, and provide sufficient time for overcurrent detection and protection. At the same time, the reactor and the capacitor together form a second-order filter, which has a significant inhibitory effect on the high-frequency oscillation components in the voltage and current on the DC side, ensuring the stability and reliability of the traction system.

[0003] A large number of practices have shown that the main reason for the failure of reactors is short circuit induced by latent faults such as partial discharge inside the reactor, local overheating, and insulation moisture and aging of the reactor. When the inter-turn insulation of the reactor is damaged and an inter-turn short circuit occurs, the temperature at the short-circuit position will rise rapidly, further accelerating the damage to the inter-turn insulation, causing the short-circuit fault to spread rapidly, and eventually causing the reactor to catch fire and burn.

[0004] At present, the online condition monitoring and abnormality detection methods of reactors can be divided into the following categories:

[0005] 1) Methods based on signal harmonic analysis. For example, patent CN111736093A analyzes the third harmonic current of the phase-controlled reactor to detect whether the reactor has a turn-to-turn short-circuit fault. However, harmonic analysis relies on stable operating conditions, while the operating conditions of trains are complex and changeable. The amplitude and harmonic components of the DC side current of the traction system are always in dynamic change, making it difficult for the harmonic analysis method to be stable and effective.

[0006] 2) Methods based on external hardware devices. For example, patents CN 108195461A and CN 110188642A use external vibration sensors to collect vibration signals for analysis, and then determine whether the reactor is faulty. Patent CN111562450A adds temperature sensors to the temperature measurement points of the core column and / or winding hot spots of the reactor, and monitors temperature data for life warning. However, external hardware will not only increase the cost and complexity of the system, but also signals such as vibration and temperature are greatly affected by environmental factors and require high sensor accuracy. These inevitable problems will restrict the application and promotion of such methods.

[0007] 3) Methods based on the measurement of reactor body parameters. For example, patent CN111830426A detects inter-turn short-circuit faults by detecting the change in reactor resistance, but the proposed method is not conducive to online real-time monitoring, and the change in resistance caused by inter-turn short-circuit is generally not as obvious as that of inductance. Currently, the more mature reactor parameter measurement methods are mainly offline methods. Due to the complex harmonic components of the reactor, online parameter identification is more difficult. At present, there is no mature online parameter identification technology that can be applied to online status monitoring of reactors.

[0008] In fact, when the reactor winding has a fault such as a turn-to-turn short circuit, its inductance and resistance parameters will undoubtedly change accordingly. Therefore, if reliable online parameter identification can be achieved, the reactor can be monitored and abnormalities detected based on the parameter identification results. However, the reactor structure is complex, and due to the existence of parameters such as parasitic capacitance, the equivalent inductance and equivalent resistance of the reactor will change with the current frequency. The input power of the reactor comes from the power supply network, and the voltage and current changes caused by external disturbances will also be reflected in the current harmonics flowing through the reactor. In addition, for the train traction system, when the PWM control technology is used to modulate the voltage required by the motor, the AC side voltage output by the inverter contains a large number of harmonic components in addition to the fundamental voltage required for the operation of the traction motor. Correspondingly, the output current of the inverter will also contain a large number of harmonic components, which will be reflected in the current on the input side of the inverter. Therefore, the voltage and current components flowing through the traction system reactor are very complex, and the reactor parameters obtained by different voltage and current component identification will also change. If the voltage and current components for parameter identification cannot be effectively controlled and screened out, stable reactor parameter identification results cannot be obtained, and reliable reactor status monitoring cannot be performed.

[0009] In summary, the current engineering implementation of online status monitoring of traction reactors faces the following challenges: 1) The operating conditions of urban rail vehicles are complex and changeable, making it difficult to apply signal harmonic analysis algorithms; 2) The use of external equipment to collect vibration, temperature and other methods will not only increase the cost and complexity of the system, but will also be easily affected by the environment and have low reliability; 3) The method based on the reactor body parameter test is ideal, but currently there is only a mature offline test method. Since the voltage and current components of the traction system reactor are very complex, online parameter identification is difficult. Summary of the invention

[0010] The object of the present invention is to solve one of the above technical problems and to provide a train traction system filter reactor anomaly detection method, system and storage medium to achieve reliable reactor anomaly detection.

[0011] To achieve the above purpose, in view of the shortcomings of the prior art, the solution adopted by the present invention is:

[0012] A method for detecting abnormality of a filter reactor of a train traction system, comprising:

[0013] S1: Collect the reactor voltage signal V L and the current signal i L ;

[0014] S2: Determine the target resonant frequency f of the reactor O , and determining a target frequency band based on the target resonant frequency;

[0015] S3: Collect the reactor voltage signal V L and the current signal i L Perform improved Hilbert transform respectively to obtain the voltage analysis signal that only retains the target frequency band component after filtering and current analysis signal Among them, the real part V L,fil is the voltage signal after bandpass filtering, the imaginary part is the voltage Hilbert transform signal after bandpass filtering, the real part i L,fil is the current signal after bandpass filtering, the imaginary part is the current Hilbert transform signal after bandpass filtering.

[0016] S4: Calculate the voltage signal envelope using the voltage analysis signal obtained in step S3, and calculate the current signal envelope using the current analysis signal, and select the reactor voltage signal V that meets the parameter identification conditions according to the voltage signal envelope and the current signal envelope. L,fil , Reactor current signal i L,fil .

[0017] S5: Based on the reactor current signal and the reactor voltage signal obtained by screening in S4, the reactor parameters are identified, wherein the parameters include the reactor resistance and the reactor inductance, and according to the identification result, whether the reactor state is abnormal is determined.

[0018] In some embodiments of the present invention, the target resonant frequency f of the reactor is determined O The method is:

[0019] The reactor voltage and current of the LC resonance process are selected for parameter identification, and the target harmonic frequency f can be obtained. O for

[0020]

[0021] Where, L ref and C ref are the inductance reference value and capacitance reference value of the reactor and capacitor, which can generally be obtained from the technical manual, and then the target frequency band [f O -Δf,f O+Δf], where Δf is the set bandwidth adjustment parameter.

[0022] In some embodiments of the present invention, in step S3, the step of improving the Hilbert transform includes:

[0023] S31: voltage signal V L and the current signal i L Perform fast Fourier transform (FFT) to obtain the frequency domain discretized voltage signal V f and frequency domain discretized current signal i f .

[0024] S32: Construct a Hilbert filter H to obtain a frequency-domain discretized voltage signal V f and frequency domain discretized current signal i f Perform Hilbert filtering in the frequency domain to eliminate negative frequency components and obtain:

[0025] V H =V f H;

[0026] i H =i f H;

[0027] Among them, V G is the frequency domain form of the voltage signal after filtering out the negative frequency components, i H is the frequency domain form of the current signal after filtering out the negative frequency components. The constructed Hilbert filter H is:

[0028]

[0029] Where n = 1, 2, ..., N-1, N, where n is the index of the signal sequence and N is the discretized signal V f and i f length, and their time domain signal V L and i L The same length.

[0030] S33: Construct a bandpass filter B, and perform bandpass filtering on the frequency domain signal after Hilbert filtering to obtain:

[0031] i B =V H B;

[0032] i B =i H B;

[0033] Among them, V B is the voltage frequency domain signal after bandpass filtering, i Bis the current frequency domain signal after bandpass filtering. The constructed bandpass filter B is:

[0034]

[0035] Where N L is the lower limit index of the bandpass filter, N H is the upper limit index of the bandpass filter. Specifically, N L and N H The target resonant frequency f obtained in step S2 is O Calculated, that is, the target filter frequency band [f O -Δf,f O +Δf] corresponding to N L and N H for:

[0036]

[0037] Among them, Δf is the set frequency adjustment parameter, f s is the sampling frequency of the time domain signal.

[0038] S34: V B 、i B They can be converted into time domain signals by performing an inverse Fourier transform (iFFT), and the result is an analytical signal in complex form.

[0039] The steps S31 to S34 are integrated to form the improved Hilbert transform, which is expressed as operation H{·}. L The operation is expressed as: H{V L}, for the current signal i L The operation is expressed as: H{V L}. Then for the voltage signal V L and the current signal i L After performing the improved Hilbert transform operation respectively, the corresponding analytical signal can be obtained:

[0040]

[0041] Among them, the real part V L,fil is the voltage signal after bandpass filtering, the imaginary part is the voltage Hilbert transform signal after bandpass filtering, the real part i L,fil is the current signal after bandpass filtering, the imaginary part is the current Hilbert transform signal after bandpass filtering.

[0042] In some embodiments of the present invention, a method for screening reactor voltage and current signals that can be used for parameter identification using signal envelopes includes:

[0043] Using the analytical signal to calculate the corresponding signal envelope, we get:

[0044]

[0045] Among them, v L,env is the reactor voltage signal envelope, i L,env is the reactor current signal envelope.

[0046] Filter valid signals according to the amplitude of the signal envelope, and the valid signals meet the following conditions:

[0047] Condition one:

[0048] v L,env (n)>thr_v,n=1,2,...,N;

[0049] Wherein, thr_v is the effective threshold of the target resonant frequency component amplitude of the reactor voltage;

[0050] Condition two:

[0051] 2πf o (L ref -ΔL) <v L,env (n) / i L,env (n)<2πf o (L ref +ΔL);

[0052] In the formula, f O is the target harmonic frequency, L ref is the rated value of the inductance of the reactor, ΔL is the reasonable actual inductance offset range limit, and n is the index of the signal sequence, n=1,2,…,N.

[0053] When the analytical signals of voltage and current meet the above conditions at the same time, the real part of the analytical signal V L,fil and i L,fil It can be used to identify the reactor parameters.

[0054] In some embodiments of the present invention, the step of identifying the parameters of the reactor includes:

[0055] Get the equivalent transfer function of the reactor:

[0056]

[0057] Express s as the complex frequency domain independent variable z of the discrete time system, that is:

[0058]

[0059] Then the equivalent transfer function of the reactor is expressed in the complex frequency domain:

[0060]

[0061] Where s is the complex frequency domain independent variable of the continuous time system, U(s) is the complex frequency domain representation of the reactor voltage, I(s) is the complex frequency domain form of the reactor current, R L is the reactor resistance, L is the reactor inductance, and t is the sampling period of the discrete time series;

[0062] By transforming the complex frequency domain expression of the reactor's equivalent transfer function from the complex frequency domain to the time domain, we can get the differential equation of the equivalent circuit:

[0063]

[0064] Among them, u k is the reactor voltage sampling value at time k, u k-1 is the reactor voltage sampling value at time k-1, i k is the reactor current sampling value at time k, i k-1 is the sampling value of the reactor current at time k-1;

[0065] Set the intermediate variables b0 and b1:

[0066]

[0067] The difference equation can be converted into the following form:

[0068] y=θ T Φ;

[0069] in, θ T =[b0b1] represents the transpose of θ, y=u k +u k-1 ;

[0070] V L,fil and i L,fil As u k and i k The input signal sequence is iteratively calculated using the least squares method:

[0071] Iterate step 1 and calculate the predicted value:

[0072]

[0073] in, represents the predicted value of y at time k; represents θ at time k-1 T ; φ k represents Φ at time k;

[0074] Iterate step 2 and calculate the prediction error:

[0075]

[0076] Where ε is the prediction error; y k is the actual value of y at time k: y k =u k +u k-1 ;

[0077] Iterate step 3 and calculate the gain vector K:

[0078]

[0079] Among them, K k is the gain vector at time k, P k-1 is the intermediate quantity matrix at time l-1. Before the first iteration, K and P need to be initialized. During the iteration, K and P are updated and gradually converged to the optimal values ​​of K and P. is φ k The transpose of

[0080] Iteration step 4, update the square matrix P:

[0081]

[0082] Where P k is the intermediate quantity matrix at time k; I is the unit matrix;

[0083] Iteration step 5, update the parameter vector θ:

[0084] θ k =θ k-1 +K k ε;

[0085] Repeat steps 1 to 5 until the optimal parameter vector is obtained, that is, the values ​​of b0 and b1 are obtained. Then calculate the reactor resistance identification value R Lb And the reactor inductance identification value L b :

[0086]

[0087] In some embodiments of the present invention, in step S5, the method for determining whether the reactor state is abnormal is:

[0088] The upper and lower thresholds of the reactor resistance, as well as the upper and lower thresholds of the reactor inductance are set. When the identified value of the reactor resistance exceeds the upper or lower threshold of the reactor resistance, and / or the identified value of the reactor inductance exceeds the upper or lower threshold of the reactor inductance, the reactor is judged to be abnormal.

[0089] A train traction system filter reactor abnormality detection system, characterized in that it is used to execute the train traction system filter reactor abnormality detection method described above.

[0090] A computer-readable storage medium stores a computer program for implementing the above-mentioned train traction system filter reactor abnormality detection method when executed.

[0091] The train traction system filter reactor abnormality detection method, system and storage medium provided by the present invention have the following beneficial effects:

[0092] 1. The present invention proposes a new method for detecting abnormality of a reactor based on online parameter identification. Through the innovative improved Hilbert transform algorithm and the signal screening method based on signal envelope, the accuracy and stability of parameter identification are improved from the data source. Based on the online parameter identification results of the reactor, reliable abnormality detection of the reactor is achieved.

[0093] 2. The online abnormality detection method for the train traction system filter reactor proposed in the present invention does not require additional hardware equipment and can achieve stable and reliable online reactor parameter identification by relying on the existing sensor signals of the traction system, thereby achieving reliable online abnormality monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0094] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0095] Figure 1 This is the circuit schematic diagram of the traction system.

[0096] Figure 2 Flowchart for improving the Hilbert algorithm.

[0097] Figure 3 This is a flow chart of the detection method provided by the present invention.

[0098] Figure 4a This is the timing diagram of the voltage acquisition signal;

[0099] Figure 4b This is the timing diagram of the current acquisition signal;

[0100] Figure 5a Voltage analysis signal and envelope diagram based on improved Hibberdt algorithm;

[0101] Figure 5b Current analysis signal and envelope diagram based on improved Hibbert algorithm;

[0102] Figure 6a is the equivalent resistance value obtained by iterative calculation based on the least square method;

[0103] Figure 6b is the equivalent inductance value obtained by iterative calculation based on the least squares method.

[0104] 1-Grid pressure sensor;

[0105] 2-DC side capacitor voltage sensor;

[0106] 3- Bus current sensor;

[0107] 4- traction motor;

[0108] 5- Reactor;

[0109] 6- Inverter unit. DETAILED DESCRIPTION

[0110] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0111] A first embodiment of the present invention provides a train traction system filter reactor abnormality detection method and system, which are used for detecting abnormal states of the reactors of the traction system of railway vehicles.

[0112] The circuit diagram of the traction system is as follows: Figure 1 As shown, it includes: a reactor 5, a capacitor, an inverter unit 6 and a traction motor 4. The reactor 5 is connected to the input end of the inverter unit 6, and the capacitor is connected in parallel to both ends of the input end of the inverter unit 6. The output end of the inverter unit 6 is connected to the traction motor 4.

[0113] S1: Data collection step.

[0114] Collect the reactor voltage signal V L and the current signal i L .

[0115] During online calculation, the voltage across the reactor and the current flowing through the reactor can be obtained through the existing sensors of the traction system. Specifically, the grid voltage data is collected through the grid voltage sensor 1, and the DC side capacitor voltage sensor 2 collects the capacitor voltage data. The voltage flowing through the reactor can be obtained by subtracting the capacitor voltage from the grid voltage, that is:

[0116] V L =V Bus -V FC (1)

[0117] Where V L is the reactor voltage, V Bus and V FC They are the grid voltage and capacitor voltage collected by the grid voltage sensor 1 and the capacitor voltage sensor 2 respectively.

[0118] The current i flowing through the reactor L It is equal to the bus current collected by the bus current sensor 3.

[0119] refer to Figure 4a and Figure 4b , where v L is the original acquisition value of the reactor voltage signal, i L It is the original acquisition value of the reactor current signal.

[0120] S2: Determine the target resonant frequency f of the reactor O , and determine the target frequency band based on the target resonant frequency.

[0121] The purpose of this step is to find the more stable harmonic components in the current flowing through the reactor so as to identify the reactor parameters.

[0122] Without considering the initial state response of the reactor and capacitor, the reactor current i can be obtained L The frequency domain expression of

[0123]

[0124] Where, L and C are the inductance of the reactor 5 and the capacitance of the capacitor respectively; R L and R C are the equivalent resistances of the reactor 5 and the capacitor respectively; i d is the DC side current of the inverter.

[0125] i d The frequency domain expression of the DC component is:

[0126]

[0127] In the formula, I dc is the inverter DC side current i d The magnitude of the DC component.

[0128] Substituting equation (3) into equation (2) and performing inverse Laplace transformation, we can obtain L The corresponding time domain expression is

[0129]

[0130] In the formula, is the LC resonant frequency; is the RC oscillation frequency; is the oscillation damping coefficient. The units of the above variables ω0 and ω1 are radians per second (rad / s).

[0131] From equation (4), it can be seen that since the reactor and capacitor form a second-order LC filter, the inverter DC side current i d The DC component in the filter will excite the inherent oscillation component of the LC filter, that is, the resonant frequency component with an angular frequency of ω0. This component appears in the transient process and then gradually decays to 0.

[0132] Therefore, the reactor voltage and current of the LC resonance process can be selected for parameter identification, and the angular frequency ω0 can be converted into a variable in Hz to obtain the target harmonic frequency f O for

[0133]

[0134] In practical application, the reference values ​​of L and C can be obtained from the technical manuals of the reactor and capacitor. ref and C ref , substituting into formula (5) to estimate f O The reference value is then used to obtain the target frequency band target filter frequency band [f O -Δf,f O +Δf], where Δf is the set bandwidth adjustment parameter.

[0135] S3: Signal processing based on improved Hilbert transform.

[0136] The signal processing step of the present invention adopts an improved Hibbert transform signal processing.

[0137] The traditional Hilbert transform method is: perform a fast Fourier transform (FFT) on the original signal to convert the original signal from a time domain signal to a frequency domain signal; then filter the frequency domain signal in the frequency domain through a Hilbert filter to eliminate negative frequency components; finally, perform an inverse Fourier transform (iFFT) on the filtered frequency domain signal to obtain a complex number analytic signal; the real part of the analytic signal is the original signal restored from the frequency domain back to the time domain, and the imaginary part of the analytic signal is the Hilbert transform signal. Compared with the original signal, the amplitude of the Hilbert transform signal remains unchanged, but the phase shifts by 90°.

[0138] The improved Hilbert transform proposed in the present invention adds a frequency domain bandpass filter before performing inverse Fourier transform on the filtered frequency domain signal, which can directly set the non-target frequency components (i.e., frequency components outside the target frequency band) to zero in the frequency domain, thereby achieving accurate filtering and retaining only the target frequency components. That is, the real part of the obtained analytical signal is the filtered input signal, and the imaginary part is the real part phase-shifted by 90 degrees.

[0139] Specifically, the steps of processing the voltage signal and the current signal using the improved Hilbert transform are as follows:

[0140] S31: Current signal V L And the voltage signal i L Perform fast Fourier transform (FFT) to obtain the frequency domain discretized voltage signal V f and frequency domain discretized current signal i f .

[0141] In step S1, the current signal V L And the voltage signal i L The collected signal is a discrete time domain signal. L And the voltage signal i L Perform Fourier transform respectively to obtain the frequency domain discretized current signal V f And the frequency domain discretized voltage signal i f .

[0142] For a time-domain discretized signal with a length of N, a frequency-domain discretized signal with a length of N can be obtained after performing a fast Fourier transform (FFT) on it.

[0143] In summary: the time domain discrete signal X with a length of N t ={x t [1],x t [2],…,x t [N]}, perform fast Fourier transform (FFT) to obtain its frequency domain form X f ={x f [1],x f [2],…,x f [N]}, where x f [n] is a complex variable. For the voltage signal V L and the current signal i L , perform the above processing respectively, and obtain the frequency domain discretization signal as follows:

[0144] V f = {v f [1],v f [2],…,v f [N]};

[0145] i f ={i f [1],i f [2],…,i f [N]}.

[0146] S32: construct a Hilbert filter H, and use the Hilbert filter H to discretize the signal V f and if Filtering is performed in the frequency domain.

[0147]

[0148] Where n = 1, 2, ..., N-1, N, N is the discretized signal V f and i f length, and their time domain signal V L and i L The same length.

[0149] For the obtained frequency domain discretized voltage signal V f and frequency domain discretized current signal i f , use Hilbert filter H to filter to eliminate negative frequency components and obtain:

[0150] V H =V f ·H; (7-1)

[0151] i H =i f ·H. (7-2)

[0152] V H is the frequency domain form of the voltage signal after filtering out the negative frequency components, i H It is the frequency domain form of the current signal after filtering out the negative frequency components.

[0153] S33: construct a bandpass filter B, and perform bandpass filtering on the frequency domain signal after the Hilbert filtering.

[0154] The target filter band of the bandpass filter is [f O -Δf,f O +Δf]. According to the filtering frequency band, its corresponding index interval in the frequency domain can be calculated, that is, X f ={x f [1],x f [2],…,x f [N]} After bandpass filtering, the frequency domain signal only retains {x f [N L ],x f [N L +1],…,x f [N H ]}, and all other components are set to zero.

[0155] N L is the lower limit index of the bandpass filter, N H is the upper limit index of the bandpass filter. L and N H The target resonant frequency f obtained in step S2 isO Calculate, specifically, N L and N H The target resonant frequency f obtained in step S2 is O Calculated, that is, the target filter frequency band [f O -Δf,f O +Δf] corresponding to N L and N H for:

[0156]

[0157] In the formula, f s is the sampling frequency of the time domain signal.

[0158] Then the bandpass filter B in the frequency domain can be expressed as

[0159]

[0160] Apply bandpass filter B to V H and i H , the non-target frequency components can be set to zero in the frequency domain to obtain:

[0161] V B =V H ·B (10-1)

[0162] i B =i H ·B (10-2)

[0163] V B is the voltage frequency domain signal after bandpass filtering, i B is the current frequency domain signal after bandpass filtering.

[0164] S34: V B 、i B By performing an inverse Fourier transform (iFFT), they can be converted into time domain signals, and the result is a complex analytical signal. The real part of the analytical signal is the original signal after filtering from the frequency domain signal to the time domain, and the imaginary part of the analytical signal is the Hilbert transform signal, which is essentially the original signal after filtering with a phase shift of 90 degrees.

[0165] Integrate steps S31 to S34, such as Figure 2 As shown, it is called the improved Hilbert transform and is expressed as the operation The operation for current is expressed as: H{V L}, the voltage operation is expressed as: H{V L}. Then for the current signal V L And the voltage signal i LAfter performing the improved Hilbert transform operation respectively, we can get:

[0166]

[0167] Among them, the real part V L,fil is the voltage signal after bandpass filtering, the imaginary part is the voltage Hilbert transform signal after bandpass filtering, the real part i L,fil is the current signal after bandpass filtering, the imaginary part is the current Hilbert transform signal after bandpass filtering.

[0168] refer to Figure 5a and Figure 5b , Figure 5a V L,fil and for Figure 4a V L The real and imaginary parts of the analytical signal obtained after the improved Hilbert transform, Figure 5b i L,fil and for Figure 4b i L Real and imaginary parts of the analytic signal after modified Hilbert transform.

[0169] S4: Signal screening.

[0170] The voltage analysis signal obtained in step S3 is used to calculate the voltage signal envelope, and the current analysis signal is used to calculate the current signal envelope. The reactor voltage signal V that meets the parameter identification conditions is screened out according to the voltage signal envelope and the current signal envelope. L,fil , Reactor current signal i L,fil .

[0171] As described in step S2, the present invention selects the reactor voltage and current of the LC resonance process for parameter identification, but the resonant frequency component only appears in the transient process and gradually decays to 0. Therefore, it is necessary to set rules to filter the signal.

[0172] The analytical signal of the voltage and current signal after bandpass filtering obtained in step S3 The corresponding instantaneous envelope can be calculated, and the instantaneous envelope can reflect the amplitude of the resonant frequency component. Therefore, the instantaneous envelope can be used to screen out the reactor current signal and the reactor voltage signal that can be used for parameter identification.

[0173] First, use the analytical signal to calculate the corresponding signal envelope and get:

[0174]

[0175] refer to Figure 5a and Figure 5b , where v L,env is the reactor voltage signal envelope, i L,env is the reactor current signal envelope.

[0176] Filter valid signals based on the amplitude of the signal envelope. The filtering conditions are as follows:

[0177] 1) The envelope of the voltage signal is greater than the threshold, that is

[0178] v L,env (n)>thr_v, n=1,2,...,N (14)

[0179] Wherein, thr_v is the effective threshold of the target resonant frequency component amplitude of the reactor voltage, which can be set according to actual conditions.

[0180] 2) The voltage signal envelope divided by the current signal envelope is within a reasonable range, which can be determined by the rated value of the reactor inductance.

[0181] πf o (L ref -ΔL) <v L,env (n) / i L,env (n)<2πf o (L ref +ΔL) (15)

[0182] In the formula, f O is the target harmonic frequency, L ref is the rated value of the inductance of the reactor, which can be obtained from the reactor technical manual, ΔL is the reasonable actual inductance offset, which can be adjusted according to the actual situation. n is the index of the signal sequence, n = 1, 2, ..., N.

[0183] If the conditions are met, the corresponding V L,fil and i L,fil It can be used for subsequent parameter identification, otherwise it is discarded.

[0184] S5: Reactor parameter identification

[0185] Based on the obtained reactor current signal and reactor voltage signal that can be used for identification, the reactor parameter identification is performed, and according to the identification result, it is judged whether there is an abnormality in the reactor state.

[0186] After obtaining the voltage and current of the reactor that meets the conditions, the recursive least squares method is used to identify the reactor parameters.

[0187] like Figure 1 As shown, the reactor can be equivalent to a series circuit of inductance and resistance, and the equivalent transfer function of the reactor can be expressed as:

[0188]

[0189] Formula (16) is a complex frequency domain function, where s is the complex frequency domain independent variable of the continuous time system, U(s) and I(s) are the complex frequency domain forms of the reactor voltage and reactor current, respectively, and R L and L are the resistance and inductance of the reactor respectively.

[0190] The bilinear transformation is used to discretize equation (15), and s is expressed as the complex frequency domain independent variable z of the discrete time system, that is,

[0191]

[0192] Where T is the sampling period of the discrete time series.

[0193] Substituting formula (17) into formula (16), the Z-transform form of formula (16) is as follows:

[0194]

[0195] Transforming equation (18) from the complex frequency domain to the time domain, the differential equation of the equivalent circuit is as follows:

[0196]

[0197] Among them, u k and u k-1 are the reactor voltage values ​​collected at time k and time k-1, i k and i k-1 They are the reactor current values ​​collected at time k and time k-1 respectively.

[0198] According to formula (19), the iterative equation of the recursive least squares method can be constructed.

[0199] First, replace i in formula (19) k and i k-1 The coefficients of are integrated into the parameters to be estimated, namely:

[0200]

[0201] b0 and b1 are intermediate variables.

[0202] Then, formula (19) can be converted into the following form:

[0203] y=θ T Φ (21)

[0204] In the formula, θ T =[b0b1] represents the transpose of θ, y=uk +u k-1 .

[0205] With formula (20), then v L,fil and i L,fil As u k and i k The input sequence can be iterated, and the iterative steps are:

[0206] Iterate step 1 and calculate the predicted value:

[0207]

[0208] In the formula, represents the predicted value at time k; represents θ at time k-1 T ; φ k represents Φ at time k.

[0209] Iterate step 2 and calculate the prediction error:

[0210]

[0211] Where ε is the prediction error; y k is the actual collected value at time k, that is, y k =u k +u k-1 .

[0212] Iterate step 3 and calculate the gain vector K:

[0213]

[0214] In the formula, K k is the gain vector at time k, P k-1 It is the intermediate quantity matrix at time k-1. K and P need to be initialized before the first iteration. They will be automatically updated as the iteration proceeds and gradually converge to the optimal value. is Φ k The transpose of .

[0215] Iteration step 4, update the square matrix P:

[0216]

[0217] Where P k is the intermediate quantity matrix at time k; I is the unit matrix.

[0218] Iteration step 5, update the parameter vector θ:

[0219] θ k =θ k-1 +K k ε (25)

[0220] Repeat the above 5 steps and iterate all the input sequences to get the optimal parameter vector, that is, the values ​​of b0 and b1. Then, calculate the resistance identification value R of the reactor. Lb And the reactor inductance identification value L b According to formula (20), we can get:

[0221]

[0222] Figure 6a and Figure 6b They are resistance identification results and inductance identification results respectively.

[0223] The reactor resistance R obtained by identification Lb , and the reactor inductance L b , determine whether the reactor is faulty. Specifically, set the upper and lower thresholds of the reactor resistance, and the upper and lower thresholds of the reactor inductance. When the identified value of the reactor resistance exceeds the upper or lower threshold of the reactor resistance, and / or the identified value of the reactor inductance exceeds the upper or lower threshold of the reactor inductance, the reactor is judged to be abnormal.

[0224] A second embodiment of the present invention provides a train traction system filter reactor abnormality detection system, which is used to execute the train traction system filter reactor abnormality detection method described above.

[0225] A third embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon for implementing the above-mentioned train traction system filter reactor abnormality detection method when executed.

[0226] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for detecting abnormality of a filter reactor in a train traction system, characterized in that: include: S1: Collect the reactor voltage signal V L and the current signal i L ; S2: Determine the target resonant frequency f of the reactor O , and determining a target frequency band based on the target resonant frequency; S3: Collect the reactor voltage signal V L and the current signal i L Perform improved Hilbert transform respectively to obtain the voltage analysis signal that only retains the target frequency band component and current analysis signal Among them, the real part V L,fil is the voltage signal after bandpass filtering, the imaginary part is the voltage Hilbert transform signal after bandpass filtering, the real part i L,fil is the current signal after bandpass filtering, the imaginary part is the Hilbert transform signal of the current after bandpass filtering; S4: Calculate the voltage signal envelope using the voltage analysis signal obtained in step S3, and calculate the current signal envelope using the current analysis signal, and select the reactor voltage signal V that meets the parameter identification conditions according to the voltage signal envelope and the current signal envelope. L,fil , Reactor current signal i L,fil ; S5: Based on the reactor current signal and the reactor voltage signal obtained by screening in S4, the reactor parameters are identified, wherein the parameters include the reactor resistance and the reactor inductance, and according to the identification result, whether the reactor state is abnormal is determined.

2. The train traction system filter reactor abnormality detection method according to claim 1, characterized in that: In step S2, the target resonant frequency f of the reactor is determined. O The method is: The reactor voltage and current of the LC resonance process are selected for parameter identification, and the target harmonic frequency f can be obtained. O for Where, L ref and C ref is the inductance reference value and capacitance reference value of the reactor and capacitor, and then the target frequency band [f O -Δf,f O +Δf], where Δf is the set bandwidth adjustment parameter.

3. The train traction system filter reactor abnormality detection method according to claim 1, characterized in that: In step S3, the step of improving the Hilbert transform includes: S31: voltage signal V L and the current signal i L Perform fast Fourier transform to obtain the frequency domain discretized voltage signal V f and frequency domain discretized current signal i f ; S32: Construct a Hilbert filter H to obtain a frequency-domain discretized voltage signal V f and frequency domain discretized current signal i f Perform Hilbert filtering in the frequency domain to eliminate negative frequency components and obtain: V H =V f ·H; i H =i f ·H; Among them, V G is the frequency domain form of the voltage signal after filtering out the negative frequency components, i H is the frequency domain form of the current signal after filtering out the negative frequency components; the constructed Hilbert filter H is: Where n = 1, 2, ..., N-1, N, where n is the index of the signal sequence and N is the discretized signal V f and i f Length; S33: Construct a bandpass filter B, and perform bandpass filtering on the frequency domain signal after Hilbert filtering to obtain: i B =V H ·B; i B =i H ·B; Among them, V B is the voltage frequency domain signal after bandpass filtering, i B is the current frequency domain signal after bandpass filtering; the bandpass filter B is: Where N L is the lower limit index of the bandpass filter, N H is the upper limit index of the bandpass filter, the target filter frequency band [f O -Δf,f O +Δf] corresponding to N L and N H for: Among them, Δf is the set frequency adjustment parameter, f s is the sampling frequency of the time domain signal; S34: V B 、i B Performing inverse Fourier transform can obtain the corresponding analytical signal: Among them, H{V L } is the voltage signal V L The improved Hilbert operation, H{i L } is the voltage signal i L Improved Hilbert operation.

4. The train traction system filter reactor abnormality detection method according to claim 1, characterized in that: In step S4, the method of using the signal envelope to screen the reactor voltage and current signals that can be used for parameter identification includes: Using the analytical signal to calculate the corresponding signal envelope, we get: Among them, v L,env is the reactor voltage signal envelope, i L,env is the current signal envelope of the reactor; Filter valid signals according to the amplitude of the signal envelope, and the valid signals meet the following conditions: Condition one: v L,env (n)>thr_v,n=1,2,...,N; Wherein, thr_v is the effective threshold of the target resonant frequency component amplitude of the reactor voltage; Condition two: 2πf o (L ref -ΔL) <v L,env (n) / i L,env (n)<2πf o (L ref +ΔL); In the formula, f O is the target harmonic frequency, L ref is the rated value of the inductance of the reactor, ΔL is the reasonable actual inductance offset range limit, n is the index of the signal sequence, n = 1, 2, ..., N; When the voltage and current analytical signals satisfy both conditions 1 and 2, the real part of the analytical signal V L,fil and i L,fil It can be used to identify the reactor parameters.

5. The train traction system filter reactor abnormality detection method according to claim 1, characterized in that: In step S5, the step of identifying the reactor parameters includes: Construct the difference equation of the reactor equivalent circuit: Among them, u k is the reactor voltage sampling value at time k, u k-1 is the reactor voltage sampling value at time k-1, i k is the reactor current sampling value at time k, i k-1 is the reactor current sampling value at time k-1; R L is the reactor resistance, L is the reactor inductance, and T is the sampling period of the discrete time series; Set the intermediate variables b0 and b1: The difference equation can be converted into the following form: y=θ T F; in, θ T =[b0b1] represents the transpose of θ, y=u k +u k-1 ; V L,fil and i L,fil As u k and i k The input signal sequence is iteratively calculated using the least squares method: Iteration step 1: Calculate the predicted value: in, represents the predicted value of y at time k; represents θ at time k-1 T ; φ k represents Φ at time k; Iteration step 2: Calculate the prediction error: Where ε is the prediction error; y k is the actual value of y at time k: y k =u k +u k-1 ; Iteration step 3: Calculate the gain vector K: Among them, K k is the gain vector at time k, P k-1 is the intermediate quantity matrix at time k-1. Before the first iteration, K and P need to be initialized. During the iteration, K and P are updated and gradually converged to obtain the optimal values ​​of L and P. is φ k The transpose of Iteration step 4: Update the square matrix P: Where P k is the intermediate quantity matrix at time k; I is the unit matrix; Iteration step 5: Update the parameter vector θ: i k =θ k-1 +K k e; Repeat steps 1 to 5 until the optimal parameter vector is obtained, that is, the values ​​of b0 and b1 are obtained; and Based on the obtained b0 and b1, calculate the reactor resistance identification value R Lb And the reactor inductance identification value L b :

6. The train traction system filter reactor abnormality detection method according to claim 1, characterized in that: In step S5, the method for determining whether the reactor state is abnormal is as follows: The upper and lower thresholds of the reactor resistance, as well as the upper and lower thresholds of the reactor inductance are set. When the identified value of the reactor resistance exceeds the upper or lower threshold of the reactor resistance, and / or the identified value of the reactor inductance exceeds the upper or lower threshold of the reactor inductance, the reactor is judged to be abnormal.

7. A train traction system filter reactor abnormality detection system, characterized in that: Used to execute the train traction system filter reactor abnormality detection method as described in any one of claims 1 to 6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the train traction system filter reactor abnormality detection method as described in any one of claims 1 to 7 is implemented.

Citation Information

Patent Citations

  • On-line monitoring method, device and system of reactor vibration

    CN108195461A

  • System and method for monitoring service life of reactor

    CN111562450A

  • Method and system for detecting turn-to-turn short circuit fault and turn number deviation of reactor

    CN111830426A

Cited By

  • Permanent magnet synchronous motor turn-to-turn short circuit fault diagnosis and positioning method

    CN120629926A