Method for monitoring and predicting insulating property of mobile substation

Abstract

The invention discloses a mobile substation insulation performance monitoring and predicting method. The method comprises the steps that S1, a microprocessor controls a digital excitation source to generate an excitation signal synthesized by a plurality of frequency sinusoidal signals; s2, response signals of the excitation signals are synchronously sampled through a sampling circuit, the sampling point number is N, and the sampling process is that the excitation signals generated by the digital excitation source are applied to possible leakage resistance and leakage capacitance through the sampling circuit, so that a detected loop is formed; s3, separating frequency components of response signals from sampling results of the N-point sampling signals by using a spectrum analysis algorithm to obtain amplitudes and phases of injection signals with different frequencies after the injection signals pass through the measured loop; and S4, the leakage resistance and the leakage capacitance of the measured loop are calculated in combination with the sampling circuit structure. According to the invention, the leakage capacitance is measured while the leakage resistance is measured, the change of insulation parameters is predicted by using a data mining algorithm, and the early warning of insulation faults is realized.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of substation insulation performance monitoring, and particularly relates to a mobile substation insulation performance monitoring and prediction method. BACKGROUND

[0002] With the rapid development of the national economy and the year-by-year improvement of people's living standards, the demand for electricity has increased sharply. At the same time, due to power failure, electric leakage, and line insulation failure, electric shock and electrical fire accidents have also increased dramatically, posing a serious threat to people's life and property safety. In order to strengthen the safety and reliability of system power supply, the state has introduced standards, clearly stipulating that IT systems (also known as isolated power supply systems, i.e. neutral point ungrounded power supply systems) should be used for power supply in some important places, and insulation monitoring equipment should be equipped to monitor the insulation performance of the IT system.

[0003] Mobile substations are mainly used in power supply in energy exploration sites, temporary power supply in large-scale activities, emergency power supply in emergencies, and other scenarios, and have high requirements for power supply reliability. Non-grounded IT distribution systems are used to improve power availability and reduce fire risk, and can effectively reduce the probability of insulation fault tripping. In such a system, when one live line has an insulation problem to ground, it will not cause a large current to trip, and only when multiple-phase insulation faults occur will a large current cause tripping. If an insulation fault occurs in the first phase, timely identification and resolution of the insulation fault can prevent tripping problems. Therefore, such a distribution system needs to monitor the insulation performance of the AC system in real time, and once there is a trend of insulation abnormality, an early warning should be given and possible hidden dangers should be investigated. After an insulation abnormality occurs, timely alarm and fault handling are needed.

[0004] Insulation monitoring technology is mainly applied in two scenarios: DC systems and AC systems. DC systems primarily involve insulation monitoring of energy storage and photovoltaic panels; AC systems mainly involve insulation monitoring of IT power distribution systems. Current industry insulation monitoring solutions involve applying a DC voltage between the PE (protective earth) and neutral point, and measuring the circuit current to obtain the insulation impedance value. For example, several AC circuit insulation performance monitoring methods, such as CN202510394130.6 Insulation detection circuit and method for inverter system, CN202411916733.X An online insulation monitoring device and method for motor, and CN 202411868506.4 An insulation monitoring circuit and its detection method, all involve applying a DC voltage to the circuit where insulation needs to be measured, judging the current formed, and calculating the insulation impedance. The hardware and software for implementing this measurement method are relatively simple. However, when the load current noise is large (such as in coal mines where mobile substations are often used, where equipment containing frequency converters and high-power engines is frequently used), if a DC signal is applied to measure the insulation impedance, conventional filtering methods such as RC filter circuits, sampling point arithmetic average method, and extreme value removal averaging method are used to process the DC signal. Although certain noise filtering methods are adopted, it is difficult to filter out the interference in the signal during field implementation. Some noise will still be spread into the measurement result, resulting in large fluctuations in the monitored DC insulation impedance, and in severe cases, false alarms.

[0005] For example, CN113589192B uses a dual-frequency injection method, but it does not consider the problem that if one of the frequencies in the dual-frequency measurement happens to be the same as the interference frequency on the line under test, it will cause serious errors in the calculation results. In use, there is still a certain risk of measurement errors and false alarms. Moreover, since this scheme uses a two-phase phase-locked loop to filter the response signal, if the injection frequency is changed, the waiting time is relatively long. The patent document mentions that each frequency change requires a 5-minute wait, and during this 5-minute interval, new anomalies may have occurred.

[0006] Furthermore, existing solutions primarily focus on impedance measurement without measuring the capacitive reactance of the circuit under test to ground. However, in engineering settings, even when the impedance is within standard requirements, the circuit may still experience significant leakage capacitance to ground, leading to issues such as single-wire electric shock and phase-to-phase short circuits in AC circuits.

[0007] Furthermore, existing solutions cannot predict short-term trends in insulation performance, and alarms are only issued when a fault occurs, which may have already caused some consequences. If the system can predict the trend of later changes in insulation conditions based on existing insulation parameter change curves, the probability of actual faults can be reduced. Summary of the Invention

[0008] The application provides a mobile substation insulation performance monitoring and prediction method.

[0009] The technical solution solving the above technical problems is as follows.

[0010] A mobile substation insulation performance monitoring and prediction method, characterized in that it comprises the following steps.

[0011] S1, a microprocessor controls a digital excitation source to generate an excitation signal synthesized by multiple frequency sinusoidal signals;

[0012] S2, a response signal of the excitation signal is synchronously sampled by a sampling circuit, the number of sampling points is N, N is an integer power of 2, and the sampling process is as follows: the excitation signal generated by the digital excitation source is applied to the possible leakage resistance and leakage capacitance through the sampling circuit, so as to form a measured loop, and the response signal of the measured loop under the excitation signal is input to the sampling circuit for conversion;

[0013] S3, N-point sampling signals are separated from the sampling results by using a frequency spectrum analysis algorithm to obtain the amplitude and phase of the different frequency injection signals after passing through the measured loop;

[0014] S4, the leakage resistance and the leakage capacitance of the measured loop are calculated in combination with the structure of the sampling circuit.

[0015] Compared with the prior art, the application has the following advantages.

[0016] 1. The method solves the false alarm problem caused by interference in the traditional measurement, adopts an alternating current injection excitation signal, and solves the problem that the noise is averaged into the response signal when a direct current excitation signal is used.

[0017] 2. The method injects multiple frequency components at one time, redundantly calculates the consistency of the results, and eliminates the problem that individual frequency points are disturbed by the same frequency signal in the application field.

[0018] 3. The method uses a frequency spectrum analysis algorithm to separate the response results of multiple alternating current signals, overcomes the waiting time for stable phase-locked loop caused by the frequency switching of the hardware phase-locked loop, and prevents the alarm delay caused by the interruption of monitoring.

[0019] 4. After the alternating current injection excitation signal is used, in combination with the alternating current response of the circuit, the leakage capacitance can be measured in addition to the leakage resistance, so as to prevent the leakage capacitance from being too large. In the alternating current circuit, too large leakage capacitance can also cause faults and safety accidents.

[0020] 5. On the basis of real-time measurement result accumulation, insulation parameter future change prediction is realized according to existing data change trend in combination with data mining algorithm, and fault evolution to produce adverse consequences is prevented. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 It is a circuit principle diagram of the mobile substation insulation performance real-time monitoring system in the application.

[0022] Figure 2 It is a flow chart of the mobile substation insulation performance monitoring and prediction method in the application. DETAILED DESCRIPTION

[0023] As Figure 1 , the mobile substation insulation performance real-time monitoring system is composed of a microprocessor CPU, a digital excitation source Es and a sampling circuit, PE is the ground, L1 is a line of a certain phase to be measured to the ground PE insulation impedance, Ro is the leakage resistance to be measured, Co is the leakage capacitance to be measured, the microprocessor CPU controls the digital excitation source Es to generate an alternating current signal output, the digital excitation source Es adopts a function signal generating chip, and the digital excitation source Es can output a superimposed signal of multiple frequency sine signals according to the instruction of the microprocessor CPU.

[0024] The sampling circuit includes a digital-to-analog converter ADC, a first resistor R1 and a second resistor R2, the digital-to-analog converter ADC is electrically connected with the microprocessor CPU, the microprocessor CPU is connected with the digital excitation source Es, one end of the first resistor R1, the second resistor R2 and the digital excitation source Es is electrically connected with the digital-to-analog converter ADC, the other end of the first resistor R1 is connected with the line L1 of a certain phase to be measured to the ground PE insulation impedance, and the other end of the second resistor R2 and the other end of the digital excitation source Es are both connected with the ground.

[0025] As Figure 2 , the monitoring method of the mobile substation insulation performance using the above system includes the following steps:

[0026] S1, the microprocessor CPU controls the digital excitation source Es to generate a plurality of frequency sinusoidal signal synthesized excitation signal Vs. In this embodiment, the microprocessor CPU first controls the digital excitation source Es to output 4 frequency excitation signal Vs, for the frequency selection of excitation signal Vs, try to avoid the integer multiple of 50Hz, to prevent the interference of the grid itself 50Hz frequency of the fundamental and harmonic, also not to select on the frequency point of the interharmonic, in addition to consider the frequency resolution when the selected discrete Fourier transform for spectral analysis, therefore, the selected frequency point is preferably on the integer multiple of the resolution, in addition, the frequency is not too high (such as: more than 100Hz), too low (such as: lower than 2Hz). In this application, the four frequency points of 10Hz, 40Hz, 60Hz and 90Hz are preferentially selected. The amplitude of the excitation signal Vs adopts the signal of Vpp=32V. According to the current selected several frequency points, the frequency resolution is 5Hz.

[0027] S2, the response signal Vi of the excitation signal Vs is synchronously sampled by the sampling circuit, and the number of sampling points is N, N is an integer power of 2, and the sampling process is: the excitation signal generated by the digital excitation source is applied to the possible leakage resistance Ro and leakage capacitance Co through the sampling circuit, so as to form the measured loop, and the response signal Vi of the measured loop under the excitation signal Vs is input to the sampling circuit row conversion.

[0028] In this embodiment, in order to prevent the spectral aliasing effect from occurring in the later spectral analysis, the sampling frequency is at least twice the 16th harmonic of the grid interference, and the fast algorithm (FFT algorithm) of the discrete Fourier transform is used in the algorithm of the later spectral analysis. The number of sampling points is an integer power of 2. Considering the above factors, the sampling frequency fs is selected as 2560Hz. 512 points are collected for once FFT calculation, and the frequency resolution is 5Hz. If the number of points is more, the frequency resolution can be further improved, and the error of spectral aliasing can be reduced, but the amount of operation will also increase, and the number of sampling points for analysis needs to be determined according to the performance of the processor. In order to further reduce the error caused by spectral leakage, the 512-point sampling signal is preferentially selected to be multiplied by the Hanning window window function to form the input sampling point of 512 points for discrete Fourier transform before spectral analysis.

[0029] S3, the N-point sampling signal is separated from the sampling result by using the spectral analysis algorithm to obtain the amplitude and phase of the different frequency injection signals after passing through the measured loop. In this embodiment, the spectral analysis algorithm of the N-point sampling signal is the discrete Fourier transform algorithm, and the discrete Fourier transform algorithm is:

[0030]

[0031] In the above formula, X[k] is the amplitude of the signal at the kth frequency component, j is the imaginary unit, x[n] is the sampling result of the sampling point (512 points), j is the imaginary unit, k is the spectral number of the spectral analysis result, k = 0, 1,..., N-1, and the corresponding frequency value is k times the frequency resolution, and k = 0 represents the direct current component.

[0032] The discrete Fourier transform decomposes the signal into a series of superimposed sine waves, and in this embodiment, N = 512 has been determined to be 8 times of 2, so the fast algorithm of the discrete Fourier transform, i.e., the fast Fourier transform, can be used. The basic idea of the fast Fourier algorithm based on time extraction is to divide the original sampling points of the sampling according to the sampling order (0-511 points) into odd points and even points, and perform discrete Fourier calculation on them respectively, and the expression is:

[0033]

[0034] In the above formula, E[k] and O[k] are the discrete Fourier transforms of the even index and odd index parts of x[n], respectively, and W N -j2π / N is a rotation factor, and since the periodicity, the rotation factor needs to be calculated only once for each pair of E[k] and O[k], thereby reducing the amount of calculation.

[0035] For the implementation of the 512-point discrete Fourier transform, the microprocessor CPU first needs to buffer the synchronous sampling values of the response signal Vi and the excitation signal Vs each for 512 points, and then perform fast Fourier transform on the two signals. Taking the response signal Vi as an example, the input x[n] of the discrete Fourier transform algorithm is divided into 512-point real part and 512-point imaginary part data, and the 512 consecutive sampling results of the excitation signal Vs or the response signal Vi are placed in the real part of the algorithm input, and the initial value of the imaginary part of the algorithm input is 0.

[0036] Through the 512-point discrete Fourier transform, 512-point real part and 512-point imaginary part results are obtained. The real part and the imaginary part with the same order number can calculate a modulus and an angle, which represent the signal parameters of the frequency component multiplied by 5 Hz, the modulus is the amplitude of the frequency component, and the angle is the phase of the frequency component. In the calculation result, the amplitudes U2i and the phases of 20 Hz, 30 Hz, 70 Hz, and 80 Hz are extracted once, the frequency number is the frequency number, and the phase is the phase deviation of the signal of the same frequency of the response signal Vi and the excitation signal Vs relative to the phase of the excitation signal Vs.

[0037] ​S4, after the modulus and phase of each frequency component of each excitation signal Vs and response signal Vi are obtained, the leakage resistance Ro and leakage capacitance Co of the measured circuit are calculated in combination with the sampling circuit structure. In this embodiment, the process of calculating the leakage resistance Ro and leakage capacitance Co of the measured circuit is as follows:

[0038] The frequency component of the excitation signal Vs is fi, the amplitude is U2i, and the phase is i is the number of frequencies, and the voltage U2 of the second resistance R2 is:

[0039]

[0040] In the above formula, ω is the angular frequency, ω = 2π·fi;

[0041] The current Ii flowing through the second resistance R2 is:

[0042]

[0043] The current of the first resistance R1 is the same as that of the second resistance R2, and therefore the total voltage on the first resistance R1 and the second resistance R2 is:

[0044]

[0045] The voltage on the leakage resistance Ro and the leakage capacitance Co is:

[0046]

[0047] In the above formula, Usi is the amplitude of the frequency component signal in Vs calculated by fast Fourier transform.

[0048] Therefore, the reactance Z of the parallel circuit of the leakage resistance (Ro) and the leakage capacitance (Co) is calculated as:

[0049]

[0050] For the parallel circuit of the leakage resistance (Ro) and the leakage capacitance (Co):

[0051]

[0052] According to the above calculation formula, after substituting specific values, the two formulas are converted into the expression of vectors, and by following the above-listed equations at different frequencies, an equation group is formed, and the values of the leakage resistance Ro and the leakage capacitance Co are calculated by using the vector method. In the present embodiment, since four groups of frequency components are used, multiple results can be calculated, and if it is found that the results are inconsistent, it can be determined that a certain group of frequencies is interfered by the same frequency, at which time the alternative frequency (such as 15 Hz, 35 Hz, 65 Hz, and 85 Hz) closest to the self can be selected to avoid the problem that the frequency point of the excitation signal Vs is close to the frequency of the interference signal and is interfered.

[0053] After the values of the leakage resistance Ro and the leakage capacitance Co are calculated, if the change of the insulation performance is to be predicted, the real-time acquisition of many actual parameters on site is involved, such as the electrical parameters of the line insulation material, the working voltage, the environmental parameters, and the like. These parameters are difficult to measure, and many devices need to be added, and it is impossible to predict where the weak point of insulation is. These parameters need to be measured at the weak point of insulation of the line, and therefore, the exponential smoothing method is used in the present embodiment to perform the medium and short-term trend prediction, which is specifically as follows.

[0054] The exponential smoothing method in the present embodiment includes a first exponential smoothing method, and the calculation formula of the first exponential smoothing method is as follows:

[0055]

[0056] In the above formula, S t is the smoothed value at time t, that is, the value at time t predicted; x t is the actual value at time t; S t-1 is the smoothed value at time t-1; and a is a smoothing constant, which is in the range of [0, 1], and (1) at the upper right corner of S represents the result of the first smoothing.

[0057] Further, the exponential smoothing method also includes a second exponential smoothing method, and the calculation formula of the second exponential smoothing method is as follows:

[0058]

[0059] In the above formula, (2) at the upper right corner of S represents the result of the second smoothing.

[0060] Still further, the exponential smoothing method also includes a third exponential smoothing method, and the calculation formula of the third exponential smoothing method is as follows:

[0061]

[0062] In the above formula, (3) at the upper right corner of S represents the result of the third smoothing.

[0063] The cubic exponential smoothing is suitable for data prediction with certain curve trend, so that the medium and short term prediction results can be preferably made by using the cubic exponential smoothing in the embodiment.

[0064]

[0065] Because the method is mainly used for medium and short term prediction, and through comparison and test, the value of alpha is 0.7. In addition, the calculated leakage resistance Ro and leakage capacitance Co are stored, and at least 20 groups of data are stored for the substitution calculation of previous data in the prediction calculation.

[0066] In summary, the method has the following characteristics:

[0067] 1. The method of the application adopts the AC signal injection, and avoids the measurement error caused by the filtering method of the traditional DC measurement when the noise is large.

[0068] 2. The method of the application adopts the multi-frequency composite AC signal to measure the insulation characteristics. The sinusoidal signals of multiple frequency components are superimposed and applied to the measured system, and the discrete Fourier transform is used for frequency spectrum analysis on the response result, so that the response result of each frequency component is obtained at one time, the waiting time is reduced, and the delay alarm of the triggered fault during the waiting period is prevented.

[0069] 3. The method of the application adopts multiple frequency points to inject the measured loop at one time, separates the signal response results of multiple frequency points at one time through the spectrum analysis algorithm, calculates the redundant results through multiple equation groups, checks the consistency of the calculation results of different frequency points, and ensures that no frequency point is disturbed by external interference to cause the result error.

[0070] 4. The method of the application measures the insulation characteristics, and the result not only contains the leakage resistance, but also contains the leakage capacitance. The fault and safety accident caused by the too large leakage capacitance can be prevented.

[0071] 5. The method of the application adopts the data mining algorithm to predict the change trend of the system insulation failure, gives the medium and short term prediction of the fault occurrence, and can give early warning of the insulation fault.

Claims

1. A method for monitoring and predicting the insulation performance of a mobile substation, characterized in that, The method comprises the following steps: S1, a microprocessor (CPU) controls a digital excitation source (Es) to generate an excitation signal (Vs) composed of multiple frequency sinusoidal signals; S2, a response signal (Vi) of the excitation signal (Vs) is synchronously sampled by a sampling circuit, the number of sampling points is N, N is an integer power of 2, and the sampling process is as follows: the excitation signal (Vs) generated by the digital excitation source is applied to a possible leakage resistance (Ro) and a leakage capacitance (Co) through the sampling circuit to form a measured loop, and the response signal (Vi) of the measured loop under the excitation signal (Vs) is input to the sampling circuit for conversion; S3, the N-point sampling signal is separated from the sampling result by using a spectrum analysis algorithm to obtain the amplitude and phase of the different frequency injection signals after passing through the measured loop; S4, the leakage resistance (Ro) and the leakage capacitance (Co) of the measured loop are calculated in combination with the structure of the sampling circuit.

2. The method for monitoring and predicting insulation performance of a mobile substation according to claim 1, characterized in that, In S1, the frequency of the excitation signal (Vs) is less than 100 Hz, and 50 Hz is not used.

3. The method for monitoring and predicting insulation performance of a mobile substation according to claim 1, wherein, In S2, the sampling frequency is at least twice the 16th harmonic of the power grid interference.

4. The method of claim 1, wherein, The spectrum analysis algorithm of the N-point sampling signal uses a discrete Fourier transform algorithm, and the discrete Fourier transform algorithm is as follows: In the formula, X[k] is the amplitude of the signal at the kth frequency component, j is an imaginary unit, x[n] is the sampling result of the sampling point, j is an imaginary unit, k is the spectral number of the spectrum analysis result, k=0, 1…, N-1, and the corresponding frequency value is k times the frequency resolution, and k is 0, representing a direct current component; The fast Fourier algorithm according to time extraction divides the original sampling points into odd points and even points according to the sampling sequence, and the expression is as follows: In the above equation, E[k] and O[k] are the discrete Fourier transforms of the even and odd indexed portions of x[n], respectively, where W N = e -j2π / N is the rotation factor.

5. A method of monitoring and predicting insulation performance of a mobile substation according to claim 4, characterized in that, The input x[n] of the discrete Fourier transform algorithm is divided into N-point real part and N-point imaginary part data, the N continuous sampling results of the excitation signal (Vs) or the response signal (Vi) are placed in the real part of the algorithm input, and the initial value of the imaginary part of the algorithm input is 0; N-point real part and N-point imaginary part results are obtained through N-point discrete Fourier transform, and the real part and the imaginary part with the same sequence number can calculate a modulus and an angle, the modulus is the amplitude of the frequency component, and the angle is the phase of the frequency component.

6. The method of claim 1, wherein the method further comprises: The sampling circuit comprises an analog-to-digital converter (ADC), a first resistor (R1), and a second resistor (R2), one end of the first resistor (R1), the second resistor (R2), and the digital excitation source (Es) is electrically connected with the analog-to-digital converter (ADC), the other end of the first resistor (R1) is connected with a line (L1) to be measured with respect to the earth (PE) insulation impedance, and the other end of the second resistor (R2) and the other end of the digital excitation source (Es) are grounded; In S4, the process of calculating the leakage resistance (Ro) and the leakage capacitance (Co) of the measured loop is as follows: The frequency component of the excitation signal (Vs) is fi, the amplitude is U2i, and the phase is i is the number of the frequency, and the voltage U2 of the second resistor R2 is: In the formula, ω is an angular frequency, and ω=2π·fi; The current Ii flowing through the second resistor (R2) is: The current of the first resistor (R1) is the same as that of the second resistor (R2), and therefore, the total voltage on the first resistor (R1) and the second resistor (R2) is: The voltage across the leakage resistance (Ro) and the leakage capacitance (Co) is: In the above equation, Usi is the amplitude of the signal of this frequency component in Vs calculated by fast Fourier transform; Therefore, the calculation formula of the reactance Z of the parallel circuit of the leakage resistance (Ro) and the leakage capacitance (Co) is: For the parallel circuit of the leakage resistance (Ro) and the leakage capacitance (Co): According to the above calculation formula, the values of the leakage resistance (Ro) and the leakage capacitance (Co) can be calculated.

7. A method of monitoring and predicting the insulation performance of a mobile substation according to any one of claims 1 to 6, characterized in that, S5 is further included, which adopts an exponential smoothing method to perform a medium and short term trend prediction, and the exponential smoothing method includes a first exponential smoothing method, and a calculation formula of the first exponential smoothing method is as follows: S t is the smoothed value of time t, that is, the predicted value at time t; x t is the actual value at time t; S t-1 is the smoothed value at time t-1; a is a smoothing constant, whose value range is [0, 1], and (1) at the upper right corner of S represents the result of once smoothing.

8. A method of monitoring and predicting insulation performance of a mobile substation according to claim 7, characterized in that, The exponential smoothing method further includes a second exponential smoothing method, and a calculation formula of the second exponential smoothing method is as follows: In the above equation, (2) at the top right corner of S represents a result of second smoothing.

9. A method of monitoring and predicting insulation performance of a mobile substation according to claim 8, characterized in that, The exponential smoothing method further includes a third exponential smoothing method, and a calculation formula of the third exponential smoothing method is as follows: In the above equation, (3) at the top right corner of S represents a result of third smoothing.

10. A method of monitoring and predicting insulation performance of a mobile substation according to claim 9, wherein, A recursive relationship expanded by the formula of the first exponential smoothing is as follows: The value of α is 0.7.

Images