Power distribution system busbar insulating medium fault assessment method based on big data analysis

Abstract

A power distribution system busbar insulating medium fault assessment method based on big data analysis belongs to the technical field of power distribution assessment, and comprises the following steps: obtaining real-time data from a high-voltage power distribution system; the obtained real-time data is used for constructing a database; performing data cleaning on the database; establishing a dynamic normalization reference value; calculating a voltage-partial discharge coupling term and an ultrahigh frequency partial discharge summation by means of sliding time window integration; carrying out nonlinear correction on the temperature item by means of a Sigmoid function, wherein the nonlinear correction comprises busbar temperature correction and cabinet temperature correction; and constructing a BDI evaluation formula and setting a BDI judgment threshold value, and outputting busbar insulating medium risk early warning. Through improvement of an overall evaluation method, early detection and early warning of a busbar insulating medium of a high-voltage power distribution system are realized, and potential safety hazards are avoided.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of power distribution evaluation, and particularly relates to a power distribution system busbar insulation medium fault evaluation method based on big data analysis. BACKGROUND

[0002] In the power distribution room scene, the insulation medium of the busbar system in the high-voltage switch cabinet is generally epoxy resin, and the insulation performance of the insulation medium is crucial. In the long-term operation process, the deterioration of the insulation medium often causes major safety accidents, but the deterioration process is nonlinear, and it is often difficult to intuitively observe before the problem occurs. Therefore, it is necessary to introduce a busbar insulation medium evaluation method to facilitate early detection of insulation deterioration. SUMMARY

[0003] The technical problem to be solved by the application is to provide a power distribution system busbar insulation medium fault evaluation method based on big data analysis, which improves the overall evaluation method to enable early detection and early warning of the busbar insulation medium of the high-voltage power distribution system, thereby avoiding safety hazards.

[0004] The technical scheme adopted by the application is: a power distribution system busbar insulation medium fault evaluation method based on big data analysis, the method comprising the following steps: A. Obtain zero sequence voltage U0 and negative sequence voltage U2 from the microcomputer protection device of the high-voltage power distribution system, obtain capacitive reactive power Q and inductive reactive power Q from the smart electric energy meter, obtain ground electric wave value TEV, ultrasonic wave value US, and ultra-high frequency value UHF from the partial discharge sensor, and obtain A-phase temperature T, B-phase temperature T, C-phase temperature T, and environmental temperature T from the wireless temperature measurement sensor. c ind A B C cab The real-time data obtained above is used to construct a database; B. Data cleaning is performed on the database in step A; C. A dynamic normalization reference value is established, including voltage imbalance reference, reactive deviation reference, and partial discharge signal threshold; By means of sliding time window integration, voltage-partial discharge coupling items and ultra-high frequency partial discharge summation are calculated; Then, the temperature items are nonlinearly corrected by means of Sigmoid function, including busbar temperature correction and cabinet temperature correction; D. A BDI evaluation formula is constructed and a BDI determination threshold is set, and a busbar insulation medium risk warning is output.

[0005] Further, the BDI evaluation formula is: ​​​​​​ , In the above formula: t w represents the width of the sliding window, t represents the current time stamp, max(U hist ) represents the maximum value of the voltage history, μ TEV represents the mean value of the ground wave history, σ TEV represents the standard deviation of the ground wave history, max(UHF hist ) represents the maximum value of the ultra-high frequency history, T x represents the temperature of a certain phase of the main bus, x represents a certain phase among the three phases A / B / C, T crit represents the temperature threshold, Max(|Q c - Q ind | hist ) represents the maximum value of the absolute value of the difference between the capacitive reactive power and the inductive reactive power, μ US represents the mean value of the ultrasonic wave history, σ US represents the standard deviation of the ultrasonic wave history.

[0006] Further, the BDI determination threshold includes: when BDI < 1, it is determined that the busbar insulation medium is healthy; when 1 < BDI < 2, it is determined that the busbar insulation medium is early deteriorated; when 2 < BDI < 3, it is determined that the busbar insulation medium is moderately deteriorated; when BDI ≥ 3, it is determined that the busbar insulation medium is severely deteriorated; The determination time of the above threshold lasts at least 10 minutes.

[0007] Further, the data cleaning method of step B includes, in sequence, invalid value filtering, time alignment, missing value processing, and noise suppression.

[0008] Further, the filtering determination standard of the invalid value filtering includes: the effective range of the zero sequence voltage U0 collected is 0 ≤ U0 ≤ 0.15U n , U n is the rated voltage of the system; the effective range of the negative sequence voltage U2 collected is 0 ≤ U2 ≤ 0.08U n ; the effective range of the capacitive reactive power Q c collected is 0 ≤ Q c ≤ 1.2Qrated, Qrated is the rated capacity of the capacitor bank; the effective range of the inductive reactive power Q ind collected is 0 ≤ Q ind≤ 0.8Strans, Qrated is the rated capacity of the capacitor bank; The effective range of the value of the ground wave TEV collected is 10 dBmV ≤ TEV ≤ 80 dBmV, and the frequency band of the ground wave collected is 3-30 MHz; The effective range of the value of the ultrasonic wave US collected is 0.05 mV ≤ US ≤ 50 mV, and the frequency band of the ultrasonic wave collected is 20-40 KHz; The effective range of the value of the ultra-high frequency UHF collected is 0.5 mV ≤ UHF ≤ 300 mV, and the frequency band of the ultra-high frequency collected is 300 MHz-1.5 GHz; The A-phase temperature T of the main busbar A , the B-phase temperature T of the main busbar B and the C-phase temperature T of the main busbar C The effective range of the collected temperature is ambient temperature ≤ T ≤ 90℃; The temperature T in the cabinet cab The effective range of the collected temperature is ambient temperature ≤ Tcab ≤ 70℃.

[0009] Further, the time alignment method is: The interpolation formula of other signals S in the time stamp tᵢ considering the maximum delay of the value of the ground wave TEV, , In the above formula, x represents any one of the zero sequence voltage, the negative sequence voltage, the capacitive reactive power, the inductive reactive power, the ground wave value, the ultrasonic wave value, the A / B / C phase temperature of the main busbar, and the temperature in the cabinet, x(ti) represents the target value after alignment, x(tk) represents the kth original sampling value, n represents the interpolation order, ti represents the alignment target time, tk represents the kth original sampling time, and tj represents the time of traversing all non-k.

[0010] Further, the missing value processing includes temperature data-KNN interpolation processing, electric parameter-ARIMA data processing, and partial discharge sensor data processing; The temperature data-KNN interpolation processing: Three adjacent wireless temperature measurement sensor data, cabinet temperature, and load current are selected to construct a feature matrix, Euclidean distance weighting is adopted, the current weight is set to 0.3, and the temperature weight is set to 0.7, , wherein Tmiss represents the missing temperature to be filled, I i represents the current of the phase where the ith sensor is located, I max represents the rated maximum value of the phase current, k is 3, and T i represents the temperature value of the ith adjacent sensor, and di represents the missing point and the distance of the i-th sensor; Electric parameter-ARIMA data processing: Sensibility reactive power ΔQ ind (t) = 0.5ΔQc(t - 1) + e(t) - 0.2e(t - 1), Capacitive reactive power ΔQ c (t) = 0.5ΔQc(t - 1) + e(t) - 0.2e(t - 1), Zero sequence voltage ΔU0(t) = 0.6ΔU0(t - 1) + e(t) - 0.3e(t - 1), Negative sequence voltage ΔU2(t) = 0.7ΔU2(t - 1) - 0.2ΔU2(t - 2) + e(t); Partial discharge sensor data processing: The pulse matching method is adopted to find similar waveforms in the history library, and the waveforms require that the rising edge is <20 ns, and the pulse width is 50-200 ns, The replacement rule is, .

[0011] Further, the noise suppression method adopts wavelet transform to denoise the data of the database.

[0012] The beneficial effects generated by the present application are: the present application has comprehensive monitoring: simultaneously collecting voltage, current, partial discharge, temperature and other multi-dimensional parameters, realizing stereoscopic evaluation of insulation state, overcoming the limitations of traditional single parameter monitoring; the present application has early warning capability: through multi-parameter intelligent fusion analysis, microcosmic insulation defects invisible to the naked eye can be found in advance, and potential faults can be warned more than 48 hours in advance than conventional methods; the present application has self-adaptive adjustment capability: the system can automatically optimize the alarm threshold according to environmental changes and equipment aging degree, significantly reducing the false alarm rate; the present application has economic advantage: under the premise of ensuring monitoring accuracy, through optimizing the sensor configuration scheme, the hardware cost can be reduced by more than 30%. BRIEF DESCRIPTION OF DRAWINGS

[0013] Figure 1 is a structural diagram of the present application. DETAILED DESCRIPTION

[0014] In order to make the purpose, technical scheme and advantages of the embodiments of the present application more clear, the technical scheme in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application.

[0015] Referring to the drawings Figure 1, High-voltage distribution cabinet is a key power distribution equipment in power system, mainly composed of cabinet, bus system, circuit breaker, transformer and monitoring device. The cabinet is made of steel plate or aluminum alloy, and the inside is divided into incoming line unit, bus chamber and outgoing line unit according to function, and isolation baffle is arranged between units to ensure safe operation. The high-voltage distribution cabinet realizes the functions of power distribution, control and protection through reasonable structure design, and its reliability directly affects the power supply quality. The busbar insulation system is a key component to ensure the safe operation of the distribution cabinet, and the insulation material needs to have good dielectric properties and mechanical strength, and at the same time can withstand long-term electric heating stress, at present, generally uses epoxy resin material to realize, the insulation performance of the insulation medium is very important, in the long-term operation process, the degradation of the insulation medium often causes serious safety accidents, but this degradation process is nonlinear, and it is often difficult to intuitively observe before the problem occurs.

[0016] The application provides a kind of based on big data analysis's distribution system busbar insulation medium fault evaluation method, the method comprises the following steps: A. Obtain zero sequence voltage U0 and negative sequence voltage U2 from the microcomputer protection device of high-voltage distribution system, obtain capacitive reactive power Q and inductive reactive power Q from smart meter c ind , obtain ground electric wave value TEV, ultrasonic value US and ultra-high frequency value UHF from partial discharge sensor, and obtain A-phase temperature T A , B-phase temperature T B , C-phase temperature T C from wireless temperature sensor, and obtain ambient temperature T cab from ambient temperature sensor. The real-time data obtained above is used to construct a database. The detailed method is: using IEC103 protocol, obtaining zero sequence voltage U0 and negative sequence voltage U2 from the microcomputer protection device, using Modbus RTU protocol, obtaining capacitive reactive power Q and inductive reactive power Q from smart meter c ind , obtaining ground electric wave value TEV, ultrasonic value US and ultra-high frequency value UHF from partial discharge sensor, and obtaining A-phase temperature T A , B-phase temperature T B , C-phase temperature T C from wireless temperature sensor, and obtaining ambient temperature T cab from ambient temperature sensor, the above real-time data is used as a data source of algorithm.

[0017] B. Data cleaning is performed on the database in step A. C. A dynamic normalized reference value is established, including voltage imbalance reference, reactive power deviation reference and partial discharge signal threshold. ​​The voltage-partial discharge coupling term and the ultra-high frequency partial discharge summation are calculated by means of the sliding time window integration; The temperature term is further corrected by means of the Sigmoid function, including busbar temperature correction and cabinet temperature correction. D. The BDI evaluation formula is constructed and the BDI judgment threshold is set to output the busbar insulation medium risk warning.

[0018] Further, the BDI evaluation formula is: , In the above formula, t w represents the sliding window width, t represents the current time stamp, max(U hist ) represents the historical maximum value of voltage, μ TEV represents the historical mean value of ground wave, σ TEV represents the historical standard deviation of ground wave, max(UHF hist ) represents the historical maximum value of ultra-high frequency, T x represents the temperature of a certain phase of the main busbar, x represents a certain phase among A / B / C three phases, T crit represents the temperature critical point, Max(|Q c - Q ind | hist ) represents the historical maximum value of the absolute value of the difference between capacitive reactive power and inductive reactive power, μ US represents the historical mean value of ultrasonic wave, σ US represents the historical standard deviation of ultrasonic wave.

[0019] The BDI judgment threshold includes: When BDI<1, it is determined that the busbar insulation medium is healthy; When 1<BDI<2, it is determined that the busbar insulation medium is early deteriorated; When 2<BDI<3, it is determined that the busbar insulation medium is moderately deteriorated; When BDI≥3, it is determined that the busbar insulation medium is seriously deteriorated; The judgment time of the above threshold lasts at least 10 min.

[0020] In specific implementation, U0, U2, TEV, UHF, US, Q c , Q ind , T x , T cab are collected in real time. The dynamic normalized reference value is calculated, the voltage imbalance reference, max( )hist, the sequence of the past 30 days is calculated, rolling. Take the 99% percentile as the benchmark (avoid extreme value interference), Reactive deviation benchmark: max(∣Q c −Q ind ∣ hist ); When Q c or Q ind exceeds 1.2 times the rated value, suspend updating the benchmark, Partial discharge signal threshold: μ TEV +3σ TEV ; Integrate the coupling voltage term and partial discharge term by sliding time window Voltage-partial discharge coupling term: ; UHF partial discharge summation: ; Nonlinearly correct the temperature term using the Sigmoid function Busbar temperature correction: ; Cabinet temperature correction: .

[0021] Build BDI evaluation formula and set BDI judgment threshold; After obtaining the calculation result, the following is the response strategy when BDI is in different intervals: BDI needs to continuously exceed the threshold for more than 10 minutes to trigger the alarm, to avoid transient interference misjudgment.

[0022] The data cleaning method of step B includes in turn: invalid value filtering, time alignment, missing value processing, noise suppression.

[0023] The method of invalid value filtering is to eliminate out-of-range data, and the filtering judgment standard of invalid value filtering is: The effective range of zero sequence voltage U0 collection is 0 ≤ U0 ≤ 0.15U n , U n is the rated voltage of the system; The effective range of negative sequence voltage U2 collection is 0 ≤ U2 ≤ 0.08U n ; The effective range of capacitive reactive power energy Q c collection is 0 ≤ Q c ≤ 1.2Qrated, Qrated is the rated capacity of the capacitor bank; Inductive reactive energy Q ind The effective range of data collection is 0 ≤ Q ind ≤ 0.8Strans, where Qrated is the rated capacity of the capacitor bank; The effective range for TEV (Transient Electric Wave) data acquisition is 10dBmV ≤ TEV ≤ 80dBmV; the acquisition frequency band is 3-30MHz. The effective range for ultrasonic value US acquisition is 0.05mV ≤ US ≤ 50mV, and the ultrasonic acquisition frequency band is 20-40KHz; The effective range for UHF value acquisition is 0.5mV ≤ UHF ≤ 300mV, and the UHF acquisition frequency band is 300MHz-1.5GHz; Main busbar temperature A phase temperature T A Phase B temperature T B and C phase temperature T C The effective range for data collection is ambient temperature ≤ T ≤ 90℃; Cabinet internal temperature T cab The effective range for data collection is when the ambient temperature is ≤ T. cab ≤ 70℃.

[0024] The time alignment method is as follows: Considering the ground wave value TEV with the largest delay, the interpolation formula for other signals S at timestamp tᵢ is... , In the above formula: x represents any one of zero-sequence voltage, negative-sequence voltage, capacitive reactive energy, inductive reactive energy, ground wave value, ultrasonic value, main busbar A / B / C phase temperature, and cabinet temperature; x(ti) represents the aligned target value; x(tk) represents the k-th original sampled value; n represents the interpolation order; ti represents the alignment target time; tk represents the k-th original sampled time; and tj represents the time when all non-k times are traversed.

[0025] Furthermore, the missing value processing includes temperature data-KNN interpolation processing, electrical parameter-ARIMA data processing, and partial discharge sensor data processing; The temperature data is processed using KNN interpolation: A feature matrix is ​​constructed by selecting data from three adjacent wireless temperature sensors, cabinet temperature, and load current. Euclidean distance weighting was used, with the current weight set to 0.3 and the temperature weight set to 0.7. , Where Tmiss represents the missing temperature to be filled, and Ii Ii represents the current of the phase where the ith sensor is located ma x represents the rated maximum value of the phase current, k is 3, T i Ti represents the temperature value of the ith adjacent sensor i Di represents the distance between the missing point and the ith sensor Electric parameter-ARIMA data processing: Inductive reactive power ΔQ ind (t)=ϵ(t)−0.4ϵ(t−1)−0.1ϵ(t−2), Capacitive reactive power ΔQ c (t)=0.5ΔQc(t−1)+ϵ(t)−0.2ϵ(t−1), Zero sequence voltage ΔU0(t)=0.6ΔU0(t−1)+ϵ(t)−0.3ϵ(t−1), Negative sequence voltage ΔU2(t)=0.7ΔU2(t−1)−0.2ΔU2(t−2)+ϵ(t); Pit sensor data processing: Pulse matching method is adopted to find similar waveforms in the history library, and the waveform requirements are that the rising edge is <20 ns, and the pulse width is 50-200 ns, The replacement rule is, .

[0026] The noise suppression method adopts wavelet transform to denoise the data of the database. The specific noise suppression method is: Wavelet transform is used for denoising (DB4 wavelet, 3-layer decomposition); decomposition: 3-layer wavelet decomposition obtains high-frequency coefficients {d1, d2, d3} and low-frequency coefficients a3; threshold processing: di′=sign(di)⋅max(0,∣di∣−0.2σnoise); wherein σnoise=median(∣d1∣) / 0.6745; reconstruction: reconstruct the signal with {d1′, d2′, d3′, a3}; Wavelet layer number Frequency band range (sample rate 10 kHz) Processing mode d1 2.5k-5kHz Hard threshold d2 1.25k-2.5kHz Soft threshold d3 625-1250Hz Preserve low frequency components

[0027] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for evaluating busbar insulation medium fault of power distribution system based on big data analysis, characterized in that, The method comprises the following steps: A. Obtain zero sequence voltage U0, negative sequence voltage U2 from microcomputer protection device of high-voltage power distribution system, obtain capacitive reactive power Q from smart electric energy meter c , inductive reactive power Q ind , obtain ground electric wave value TEV, ultrasonic wave value US and ultra-high frequency value UHF from partial discharge sensor, obtain A-phase temperature T A , B-phase temperature T B , C-phase temperature T C from wireless temperature sensor, obtain ambient temperature T cab from ambient temperature sensor; The real-time data obtained above is used to build a database; B. Data cleaning is performed on the database in step A; C. A dynamic normalized reference value is built, including a voltage imbalance reference, a reactive power deviation reference, and a partial discharge signal threshold value; A voltage-partial discharge coupling term and a UHF partial discharge summation formula are established by means of sliding time window integration; Temperature data is nonlinearly corrected by means of a Sigmoid function, including busbar temperature correction and cabinet temperature correction; D. A BDI evaluation formula is built and a BDI judgment threshold value is set, and a busbar insulating medium risk early warning is output.

2. The method for power distribution system busbar insulation medium fault assessment based on big data analysis according to claim 1, characterized in that: The BDI evaluation formula is: ， In the above formula: t w represents the sliding window width, t represents the current time stamp, max(U hist ) represents the maximum value of the voltage history, μ TEV represents the historical mean of the ground wave, σ TEV represents the historical standard deviation of the ground wave, max(UHF hist ) represents the historical maximum of the ultra-high frequency, Tx represents the temperature of a certain phase of the main bus, x refers to a certain phase among the three phases A / B / C, T crit represents the temperature critical point, Max(|Qc-Q ind | hist ) represents the historical maximum of the absolute value of the difference between the capacitive and inductive reactive power, μ US represents the historical mean of the ultrasonic wave, σ US represents the historical standard deviation of the ultrasonic wave.

3. The method for power distribution system busbar insulation medium fault assessment based on big data analysis according to claim 1, characterized in that, The BDI judgment threshold value comprises: When BDI < 1, it is determined that the busbar insulating medium is healthy; When 1 < BDI < 2, it is determined that the busbar insulating medium is in early degradation; When 2 < BDI < 3, it is determined that the busbar insulating medium is in moderate degradation; When BDI ≥ 3, it is determined that the busbar insulating medium is in serious degradation; The determination time of the above threshold value lasts at least 10 minutes.

4. The method for power distribution system busbar insulation medium fault assessment based on big data analysis according to claim 1, characterized in that, The data cleaning method of step B comprises, in sequence, invalid value filtering, time alignment, missing value processing, and noise suppression.

5. The method for power distribution system busbar insulation medium fault assessment based on big data analysis according to claim 4, characterized in that, The filtering judgment standard of the invalid value filtering comprises: The effective range of the zero-sequence voltage U0 collected is 0 ≤ U0 ≤ 0.15U n , U n is the system rated voltage; The effective range of negative sequence voltage U2 collected is 0 ≤ U2 ≤ 0.08U n ; Qcapacitive reactive power c The effective range of acquisition is 0 ≤ Q c ≤ 1.2Qrated, Qrated being the rated capacity of the capacitor bank; Perceptual reactive power Q ind The effective range of collection is 0 ≤ Q ind ≤ 0.8 Strans, Qrated is the rated capacity of the capacitor bank; The effective range of the ground wave value TEV is 10 dBmV ≤ TEV ≤ 80 dBmV, and the ground wave collection frequency band is 3-30 MHz; The effective range of the ultrasonic wave value US is 0.05 mV ≤ US ≤ 50 mV, and the ultrasonic wave collection frequency band is 20-40 KHz; The effective range of the UHF value UHF is 0.5 mV ≤ UHF ≤ 300 mV, and the UHF collection frequency band is 300 MHz-1.5 GHz; Main bus temperature A phase temperature T A B phase temperature T B and C phase temperature T C The effective range of collection is ambient temperature ≤ T ≤ 90℃; The effective range of the temperature Tcab collected in the cabinet is ambient temperature ≤ T cab ≤ 70℃.

6. The method for power distribution system busbar insulation medium fault assessment based on big data analysis according to claim 4, characterized in that, The method for time alignment is: The interpolation formula of the ground wave value TEV with the longest delay and other signals S at the time stamp tᵢ is considered, ， In the above formula, x represents any one of the zero sequence voltage, the negative sequence voltage, the capacitive reactive power, the inductive reactive power, the ground wave value, the ultrasonic wave value, the main busbar A / B / C phase temperature, and the cabinet temperature, x(ti) represents the aligned target value, x(tk) represents the kth original sampling value, n represents the interpolation order, ti represents the alignment target time, tk represents the kth original sampling time, and tj represents the time of traversing all non-k.

7. The method for power distribution system busbar insulation medium fault assessment based on big data analysis according to claim 4, characterized in that, The missing value processing comprises temperature data-KNN interpolation processing, electric parameter-ARIMA data processing, and partial discharge sensor data processing; The temperature data-KNN interpolation processing comprises: Three adjacent wireless temperature measurement sensor data, cabinet temperature, and load current are selected to build a feature matrix, The Euclidean distance is weighted, the current weight is set to 0.3, and the temperature weight is set to 0.7, ， wherein T miss represents the missing temperature to be compensated, I i represents the current of the phase where the i-th sensor is located, I max represents the rated maximum value of the phase current, k is 3, T i represents the temperature value of the i-th adjacent sensor, d i represents the distance between the missing point and the i-th sensor; The electric parameter-ARIMA data processing comprises: Perceived reactive ΔQ ind (t) = e(t) - 0.4e(t - 1) - 0.1e(t - 2), Capacitive reactive ΔQ c (t) = 0.5 AQc(t - 1) + e(t) - 0.2 e(t - 1), Zero sequence voltage ΔU0(t)=0.6ΔU0(t−1)+ϵ(t)−0.3ϵ(t−1), Negative sequence voltage ΔU2(t)=0.7ΔU2(t−1)−0.2ΔU2(t−2)+ϵ(t) ; The partial discharge sensor data processing comprises: The pulse matching method is adopted to find similar waveforms in the historical library, and the waveform requirements are that the rising edge is <20 ns and the pulse width is 50-200 ns, The replacement rule is 。 8. The method for power distribution system busbar insulation medium fault assessment based on big data analysis according to claim 4, characterized in that, The noise suppression method adopts wavelet transform to denoise data of the database.

Images