Traction power supply system fault location method and system based on clock synchronization of traveling wave ranging

Abstract

This invention discloses a method and system for traction network fault location based on traveling wave ranging and clock synchronization, belonging to the field of traction network fault location technology. This invention sets up clock subsystems and measurement subsystems in substations, sectioning stations, and AT stations. A first confidence level is generated based on the crystal oscillator type and historical timekeeping error rate of the clock subsystem; a second confidence level is generated based on the fault signal collected by the measurement subsystem; and a third confidence level is generated based on the signal loss duration of the clock subsystem. The fault location is predicted by combining the reception time, the first confidence level, the second confidence level, and the third confidence level. Furthermore, a minimum reflection path is fitted based on the fault location and reflection type, and the actual wave speed is fitted based on the minimum reflection time difference to predict the clock reliability. Then, the local clock is actively calibrated based on the clock reliability. This invention can effectively improve the accuracy of traction network fault location by reducing measurement errors caused by clock asynchrony.

Description

Technical Field

[0001] This invention relates to the field of traction network fault location technology, and in particular to a traction network fault location method and system based on traveling wave ranging clock synchronization. Background Technology

[0002] Distributed traveling wave ranging technology predicts fault location by detecting the fault reception times of multiple nodes. Referring to the fault location method and system for power distribution networks based on multi-terminal traveling wave ranging proposed in Chinese patent application CN202510778674.2, the fault location accuracy is related to time measurement, and wide-area time synchronization performance directly affects the accuracy of traveling wave fault location. Currently, the clocks of distributed traveling wave monitoring devices mainly come from GPS / BeiDou satellite clock synchronization systems, such as the dual-terminal traveling wave fault location method for power distribution lines based on high-precision clock synchronization proposed in Chinese patent application CN202311499722.1. In this method, the traveling wave signal is collected by traveling wave acquisition devices located at N key nodes, and each traveling wave acquisition device synchronizes its time using a traveling wave time difference positioning model. However, GPS / BeiDou clock synchronization in underground subway spaces suffers from signal shielding issues. The underground environment shields more than 90% of GPS / BeiDou satellite signals, preventing underground equipment from directly receiving satellite time sources. Although ground stations can relay satellite signals and transmit time signals through dedicated shielded lines, the synchronization accuracy is affected by network transmission. Therefore, it is necessary to conduct in-depth research on the wide-area time synchronization problem of distributed traveling wave monitoring networks deployed in subway environments to ensure the accurate implementation of distributed traveling wave ranging in complex underground environments. Summary of the Invention

[0003] To address the aforementioned issues, this invention provides a traction network fault location method and system based on traveling wave ranging clock synchronization. By predicting fault locations using fault signals and confidence levels from different clock subsystems, the fault location accuracy is improved. Simultaneously, the local clock is calibrated based on the secondary fault signal, reducing measurement errors caused by clock asynchrony between substations, sectioning stations, and AT stations.

[0004] The objective of this invention can be achieved through the following technical means: a traction network fault location method based on traveling wave ranging clock synchronization, comprising the following steps: Step 1: Deploy clock subsystems and measurement subsystems in substations, sectioning stations, and each AT station, and assign a first confidence level to each clock subsystem; Step 2: Each clock subsystem receives satellite signals and updates its local clock and crystal oscillator discipline function. If a clock subsystem loses satellite signals, it updates its local clock according to the crystal oscillator discipline function and records the duration of signal loss. Step 3: If at least one measurement subsystem receives a fault signal, the measurement subsystem determines the time of receipt of the fault signal according to the local clock, the measurement subsystem collects the secondary fault signal, and determines the minimum reflection time difference and reflection type; Step 4: Generate a second confidence level for the corresponding clock subsystem based on the fault signals of different measurement subsystems, and generate a third confidence level based on the signal loss duration of the clock subsystem; Step 5: Predict the fault location based on the receiving time, theoretical wave velocity, and first confidence level, second confidence level, and third confidence level of at least three sets of measurement subsystems; Step 6: Fit the minimum reflection path by combining the fault location and reflection type, then fit the actual wave speed by combining the reflection type and minimum reflection time difference, and finally request synchronization of the local clock by combining the minimum reflection path and actual wave speed.

[0005] In this invention, in step 1, multiple AT stations are located between substations and section stations. The first confidence level of the section station clock subsystem is greater than the first confidence level of the AT station clock subsystem and less than the first confidence level of the substation clock subsystem. The first confidence level is updated according to the crystal oscillator type and historical timekeeping error rate of the substation, section station, and AT station clock subsystems.

[0006] In this invention, in step 2, a third-order crystal oscillator discipline function is assigned to each clock subsystem. If the clock subsystem receives a satellite signal, it parses the 1PPS pulse from the satellite signal and updates the local clock, collects the crystal frequency of the crystal clock generator, measures the frequency deviation of the crystal frequency relative to the 1PPS pulse, and updates the linear parameters, temperature parameters, and aging parameters of the crystal oscillator discipline function according to the recursive least squares algorithm.

[0007] In this invention, if the clock subsystem does not receive a satellite signal, the crystal oscillator frequency is input to the crystal oscillator discipline function to generate a clock increment, and then the local clock is updated.

[0008] In this invention, in step 3, the reflection time difference between the received fault signal and the fault secondary signal wavefront is calculated, and the minimum reception time measurement subsystem is extracted. The reflection time difference corresponding to the measurement subsystem is the minimum reflection time difference. The reflection type is identified according to the current polarity of the fault signal and the fault secondary signal of the measurement subsystem. The reflection type includes reflection through the fault point and reflection through the node. The node includes substations, substations, and AT stations.

[0009] In this invention, in step 4, the wavefront segment signal sequence of the fault signal at the receiving time is intercepted, and the amplitude, jitter, and signal-to-noise ratio of the wavefront segment signal sequence are extracted. After normalization and geometric averaging, the second confidence level is obtained.

[0010] In this invention, in step 4, the drift error of the clock subsystem and the model error of the crystal oscillator discipline function are calculated according to the signal loss duration, and then a third confidence level is generated.

[0011] In this invention, in step 5, the expected coordinates of the fault point are calculated based on the receiving time of the substation measurement subsystem and the section measurement subsystem. The absolute residual sum function of the fault location is established based on the receiving time and the first confidence level, the second confidence level, and the third confidence level. The final coordinates that minimize the absolute residual sum function are found along the expected coordinates. The final coordinates are the fault location.

[0012] In this invention, in step 6, a first estimate of the minimum reflection path is fitted according to the fault location, a second estimate of the minimum reflection path is generated according to the reflection type, the actual wave speed is predicted according to the second estimate and the minimum reflection time difference, and the clock reliability is predicted according to the first estimate, the second estimate, the actual wave speed and the theoretical wave speed. If the clock reliability is greater than the reliability threshold, the process returns to step 2; otherwise, the process requests synchronization of the local clock of each clock subsystem and returns to step 1.

[0013] A traction network fault location system for implementing the aforementioned traction network fault location method based on traveling wave ranging clock synchronization includes: The traction network includes substations, sectioning stations, and multiple AT stations. Multiple clock subsystems are configured to update the local clocks of the substations, section stations, and multiple AT stations; Multiple measurement subsystems are configured to collect fault signals and secondary fault signals from substations, sectioning stations, and AT stations, and combine them with the local clock to determine the receiving time, reflection type, and minimum reflection time difference; A positioning subsystem is configured to predict the fault location based on the receiving time and a first confidence level, a second confidence level, and a third confidence level, and to request synchronization of the local clock based on the clock reliability.

[0014] The traction network fault location method and system based on traveling wave ranging and clock synchronization of this invention have the following advantages: This invention predicts the fault location based on the receiving times of different clock subsystems and first, second, and third confidence levels. Compared to two-node detection technology, this invention fully utilizes fault signals collected by substations, sectioning stations, and AT stations, improving fault location accuracy. This invention fully utilizes the propagation characteristics of fault signals within the traction network to predict the reflection type of the target node. Simultaneously, it generates the actual wave velocity and predicts clock reliability based on the minimum reflection time difference of the target node. When the clock reliability is low, it synchronizes the local clock, achieving active local clock calibration. The minimum reflection time difference is obtained from the local clock of the same clock subsystem and can be used to predict measurement errors caused by clock asynchrony, further ensuring fault location accuracy. Attached Figure Description

[0015] Figure 1 This is a flowchart of the traction network fault location method based on traveling wave ranging clock synchronization according to the present invention; Figure 2 This is a plan view of the traction network layout based on traveling wave ranging clock synchronization according to the present invention; Figure 3 This is a schematic diagram of the fault signal propagation of the present invention; Figure 4 This is a schematic diagram of the minimum reflection path of the present invention via the fault point; Figure 5 This is a waveform diagram of the fault current signal reflected from the fault point according to the present invention; Figure 6 This is a schematic diagram illustrating the minimum reflection time difference of the present invention via the fault point; Figure 7 This is a schematic diagram of the minimum reflection path via node reflection according to the present invention; Figure 8 This is a waveform diagram of the fault current signal reflected by the node according to the present invention; Figure 9 This is a schematic diagram illustrating the minimum reflection time difference of the present invention via node reflection; Figure 10 This is a block diagram of a traction network fault location system that implements the traction network fault location method based on traveling wave ranging and clock synchronization according to the present invention. Figure 11 This is a block diagram of the clock subsystem of the present invention; Figure 12 This is a block diagram of the measurement subsystem of the present invention.

[0016] The reference numerals in the attached diagram are: substation 11, sectioning station 12, AT station 13, contact wire 14, positive feeder 15, rail 16, return line 17, three-phase line 18, train 20. Detailed Implementation

[0017] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0018] Example 1

[0019] This invention predicts fault locations based on the reception times of different clock subsystems and first, second, and third confidence levels, fully utilizing fault signals collected by substations, sectioning stations, and automatic transmission (AT) stations to improve fault location accuracy. This invention collectively refers to substations, sectioning stations, and AT stations as nodes. Based on the propagation characteristics of fault signals within the traction network, it predicts the reflection type of the target node. Simultaneously, it generates the actual wave velocity and predicts clock reliability based on the minimum reflection time difference of the target node. When clock reliability is low, it synchronizes the local clock, achieving active local clock calibration. For example... Figures 1 to 10 As shown, the present invention provides a traction network fault location method based on traveling wave ranging clock synchronization, comprising the following steps.

[0020] Step 1: Deploy clock subsystems and measurement subsystems in substations, sectioning stations, and each AT station, and assign a first confidence level to each clock subsystem. For example... Figure 2 As shown, the substation 11 of this invention supplies power to the train 20, and the sectioning station 12 is located at the end of the substation 11. The substation 11, contact line 14, and sectioning station 12 form a power supply circuit. The AT station 13, also known as an autotransformer, is used to achieve long-distance, high-efficiency power transmission. Multiple AT stations 13 are located between the substation 11 and the sectioning station 12. The AT stations 13 connect the substation 11, contact line 14, and rail 16, dividing the power supply circuit into different power supply sections. In this embodiment, to control equipment costs, the substation clock subsystem uses a high-precision OCXO crystal oscillator, the sectioning station uses a TCXO crystal oscillator, and the AT station uses a common crystal oscillator. The first confidence level is used to evaluate the clock accuracy of the equipment itself. Typically, the first confidence level C of the sectioning station clock subsystem is... 1,Sec The first confidence level C of the clock subsystem of AT is greater than 1,AT And less than the first confidence level C of the substation clock subsystem 1,Sub C 1,Sub >C 1,Sec >C 1,AT .

[0021] Before the equipment is put into use, the initial value C of the first confidence level is preset according to the crystal oscillator type of the clock subsystem of the substation, sectioning station, and AT station. 1,Sub(0) C 1,Sec(0) C 1,AT(0) For example, C 1,Sub(0) =0.95, C 1,Sec(0) =0.7, C 1,AT(0) =0.5. After the equipment is put into use, the first confidence level is updated based on the historical timekeeping error rate of the clock subsystems of the substation, sectioning station, and AT station. Taking the substation as an example, the drift error of the clock subsystem in the most recent 20 fault events is extracted, and its mean σ is calculated. avg Then the historical timekeeping error rate ε t =σ avg / σ max , σ max This represents the maximum drift error. The first confidence level C of the substation is then updated. 1,Sub C 1,Sub =C 1,Sub(0) (1-ε tThe calculation of the drift error is as described in Example 3. The first confidence level of the substation and the AT station can be updated using the same method. The maximum drift error is set according to the crystal oscillator type and the maximum allowable satellite loss duration. For example, when the maximum allowable satellite loss duration is 1 hour, the maximum drift errors of the substation, substation, and AT station can be set to 38 μs, 1900 μs, and 76000 μs, respectively.

[0022] Step 2: Each clock subsystem receives satellite signals and updates its local clock and crystal oscillator discipline function. If a clock subsystem loses satellite signals, it updates its local clock according to the crystal oscillator discipline function and records the duration of signal loss. Specifically, this invention assigns a third-order crystal oscillator discipline function to each clock subsystem. The crystal oscillator discipline function represents the mapping between the local clock increment and the crystal oscillator frequency. If the clock subsystem receives satellite signals, it parses 1PPS pulses from the satellite signals and updates its local clock, acquires the crystal oscillator frequency of the crystal clock generator, and measures the frequency deviation of the crystal oscillator frequency relative to the 1PPS pulse. As described in Embodiment 2, the linear parameters, temperature parameters, and aging parameters of the crystal oscillator discipline function are updated according to the recursive least squares algorithm. If the clock subsystem does not receive satellite signals, it inputs the crystal oscillator frequency into the crystal oscillator discipline function to generate clock increments and then updates the local clock.

[0023] Step 3: If at least one measurement subsystem receives a fault signal, the measurement subsystem determines the reception time of the fault signal based on the local clock, acquires the secondary fault signal, and determines the minimum reflection time difference and reflection type. When a fault occurs at a certain location on the contact wire of the traction network, a fault signal is generated at the fault location and transmitted along the contact wire to both sides. For example... Figure 3 As shown, the measurement subsystems of substations, sectioning stations, and each AT station will successively receive fault signals. As described in Example 4, the reflection time difference between the received fault signal and the wavefront of the fault secondary signal is calculated. The measurement subsystem with the minimum reception time is extracted; the reflection time difference corresponding to this measurement subsystem is the minimum reflection time difference. The reflection type of the fault secondary signal is identified based on the current polarity of the fault signal and the fault secondary signal of this measurement subsystem. The reflection types include reflection via the fault point and reflection via nodes, where nodes include substations, sectioning stations, and AT stations.

[0024] Step 4: Generate a second confidence level for the corresponding clock subsystem based on the fault signals of different measurement subsystems, and generate a third confidence level based on the signal loss duration of the clock subsystem. In this invention, the second confidence level is used to describe the quality of the fault signal acquired by the corresponding measurement subsystem, enhance the sampling contribution near the fault point, and suppress far-end noise. The third confidence level is used to describe the clock synchronization error of the corresponding clock subsystem, reduce the weight of clock subsystems that have been out of lock for a long time, and avoid timestamp errors. As described in Embodiment 3, this invention extracts the wavefront segment signal sequence of the fault signal at the receiving time, extracts the amplitude, jitter, and signal-to-noise ratio of the wavefront segment signal sequence, and obtains the second confidence level after normalization and geometric averaging. Then, the drift error of the clock subsystem and the model error of the crystal oscillator discipline function are calculated based on the signal loss duration, and the third confidence level is generated.

[0025] Step 5: Predict the fault location based on the reception time of at least three sets of measurement subsystems, theoretical wave velocity, and first, second, and third confidence levels. The theoretical wave velocity is typically 3 × 10⁻⁶. 8 The speed (m / s) can be measured on-site after equipment installation. This invention fully utilizes multi-node redundancy information to improve positioning reliability. As described in Embodiment 3, this invention calculates the expected coordinates of the fault point based on the receiving times of the substation measurement subsystem and the section measurement subsystem. Based on the receiving time and the first confidence level, the second confidence level, and the third confidence level, an absolute residual sum function of the fault location is established. The final coordinates that minimize the absolute residual sum function are found along the expected coordinates; these final coordinates are the fault location. In another embodiment, grid search, recursive estimation based on Kalman filtering, and other methods can also be used to predict the fault location.

[0026] Step 6: Fit the minimum reflection path by combining the fault location and reflection type, then fit the actual wave speed by combining the reflection type and minimum reflection time difference, and finally request synchronization of the local clock by combining the minimum reflection path and actual wave speed. First, extract the node with the minimum reception time from multiple groups of nodes. The minimum reflection path refers to the path difference between the fault signal and the fault secondary signal of that node. After the fault location is determined, the expected length of the minimum reflection path can be fitted according to the fault location in the traction network topology, which is defined as the first estimate of the minimum reflection path. At the same time, the expected length of the minimum reflection path can be generated according to the reflection type, which is defined as the second estimate of the minimum reflection path. The actual wave speed is predicted based on the second estimate of the minimum reflection path and the minimum reflection time difference. The minimum reflection time difference is obtained from the local clock of the same clock subsystem and can be used to predict system errors caused by clock asynchrony. As described in Example 4, this invention predicts the clock reliability based on the first estimate and the second estimate, as well as the actual wave speed and the theoretical wave speed. If the clock reliability is greater than the threshold, return to step 2; otherwise, request synchronization of the local clock of each clock subsystem and return to step 1.

[0027] Example 2

[0028] This embodiment further discloses a preferred method for updating the local clock and the crystal oscillator discipline function. The crystal oscillator discipline function is used to represent the clock increment Δt of the local clock. local In a simpler embodiment, the mapping relationship with the crystal oscillator frequency f can be expressed as a linear crystal oscillator discipline function.

[0029] Assign a third-order crystal discipline function to each clock subsystem. Clock increment Δt local =N / {f[1+θ g +θ l t w +θ T (TT ref )+θ age t w 2 ]}. Where N is the total number of crystal oscillator pulses accumulated since the last update of the local clock based on satellite signals. θ g For a fixed frequency offset parameter, θ. l θ T θ age These are linear parameters, temperature parameters, and aging parameters, respectively. The linear parameter characterizes the linear rate of change of the crystal oscillator frequency over time, the temperature parameter characterizes the sensitivity of the crystal oscillator frequency to temperature, and the aging parameter characterizes the aging effect of the crystal oscillator frequency over time. For clock subsystems using different crystal oscillator types, the parameter values ​​of the crystal oscillator discipline function differ. For example, for a substation clock subsystem using an OCXO crystal oscillator, during the system initialization phase, θ can be... g θ l θ T θ age Set to 10 respectively -10 10 -16 / s、10 -11 / ℃, 10 -23 / s 2 T, T ref These are the current temperature and the reference temperature, T. ref It is usually set at 25℃. w This represents the cumulative operating time of the local clock.

[0030] Update the local clock. If the clock subsystem receives a satellite signal, it parses the 1PPS pulse from the satellite signal and updates the local clock. Specifically, the clock subsystem parses the current UTC integer second timestamp from the satellite message, then uses the rising edge of the 1PPS pulse as the precise trigger time, records this timestamp as the satellite synchronization time, and writes it into the local clock's seconds register to update the local clock. If the clock subsystem does not receive a satellite signal, it inputs the crystal oscillator frequency f to the crystal oscillator discipline function to generate the clock increment Δt. local Then update the local clock T, T=Tsync +Δt local T sync This is the last satellite synchronization time.

[0031] Update the crystal oscillator discipline function. When the clock subsystem receives a satellite signal, it acquires the crystal oscillator frequency of the crystal clock generator, measures the frequency deviation of the crystal oscillator frequency relative to the 1PPS pulse, and updates the linear parameters, temperature parameters, and aging parameters of the crystal oscillator discipline function according to the recursive least squares algorithm. Specifically, the time interval (1s) between two adjacent 1PPS pulses is defined as one 1PPS cycle. The number N of crystal oscillator pulses output by the local clock in the k-th 1PPS cycle is counted. k Then the actual crystal oscillator frequency f of the local clock k '=N k The crystal oscillator frequency f of the crystal clock generator is collected, and the frequency deviation Δf of the crystal oscillator frequency relative to the 1PPS pulse is then calculated. k =(f k '-f) / f. The frequency deviation Δf k As an observation y k Construct the observation equation y k =φ k T θ k-1 +e k , where φ k Let φ be the eigenvector of the k-th 1PPS period. k =[1,t w,k ,(T k -T ref ),t w,k 2 ] T θ k-1 Let θ be the parameter vector for the (k-1)th 1PPS period. k-1 =[θ l,k-1 , θ T,k-1 , θ age,k-1 ] T e k The measurement noise is for the k-th 1PPS cycle.

[0032] For the k-th 1PPS cycle, k=1,2,…, perform least squares recursion, the specific steps of which are as follows: (1) Based on the eigenvector φ k The covariance matrix P k-1 Calculate the gain vector K k K k =(P k-1 φ k ) / (λ+φ k T P k-1 φk ), where λ is the forgetting factor, typically between 0.95 and 0.995; (2) Based on the eigenvector φ k and parameter vector θ k-1 Calculate the prediction error ε k , ε k =φ k T θ k-1 ; (3) Based on gain vector K k and prediction error ε k Update parameter vector θ k θ k =θ k-1 +K k ε k ; (4) Based on gain vector K k and eigenvector φ k Update covariance matrix P k P k =(1-K k φ k T )P k-1 / λ; After the iteration is complete, the linear parameters, temperature parameters, and aging parameters of the crystal oscillator discipline function are updated based on the parameter vector of the last IPPS cycle. It should be noted that during the initialization phase, P0 = δ -1 I4. δ is the precision parameter, usually taken as a small positive number, such as 1×10. -4 I4 is a 4×4 identity matrix.

[0033] Example 3

[0034] This embodiment further discloses a preferred method for predicting fault location according to the present invention.

[0035] Calculate the second confidence level. The measurement subsystem takes the fault signal reception time t as an example. arr Centered on N, extract the elements that are N in front of it. pre A noise window sequence of N sampling points is truncated to include N. post The sequence of signal windows at each sampling point together constitutes the wavefront signal sequence X=[x1,x2,…,x…]. m ,…x M ]. x m This represents the signal value at the m-th sampling point. M=N pre +N post +1. Extract the magnitude A from the sequence X. Calculate the jitter S. Δt is the sampling interval. Calculate the noise power P. nand signal power P s , , Next, calculate the signal-to-noise ratio (SNR): SNR = 10·lg(P s / P n ).

[0036] Furthermore, since the dimensions and dynamic ranges of amplitude, jitter, and signal-to-noise ratio (SNR) differ significantly, they need to be normalized and mapped to the [0,1] interval. Taking amplitude A as an example, the minimum amplitude A is extracted from the amplitudes of all measurement subsystems. min and the maximum amplitude A max Then the normalized amplitude A' = (AA min ) / (A max -A min Using the same method, the normalized jitter S' and signal-to-noise ratio SNR' can be obtained. The normalized amplitude, jitter, and signal-to-noise ratio are then geometrically averaged to obtain the second confidence level C2 corresponding to the fault signal of this measurement subsystem, where C2 = (A'·S'·SNR'). 1 / 3 C2∈[0,1], the closer C2 is to 1, the higher the quality of the fault signal.

[0037] Calculate the third confidence level. Since the time drift of a crystal oscillator without satellite calibration is typically proportional to the square of the lost time (Allan variance characteristic), in this embodiment, an empirical model is used to calculate the drift error σ1, i.e., σ1 = a1·t. loss +a2·t loss 2 Where a1 and a2 are coefficients related to the crystal oscillator type, such as a1 and a2 of 1×10 for OCXO crystal oscillators. -11 and 1×10 -16 . t loss The duration of signal loss is given. The covariance matrix P of the crystal oscillator discipline function at the termination time of the previous iteration update is extracted. t-1 And construct the feature vector φ at the current time. t The model error σ2 of the crystal oscillator discipline function is calculated by combining the signal loss duration, σ2=(φ t T P t-1 φ t ) 1 / 2 ·t loss The combined time error σ is calculated by combining the drift error σ1 and the model error σ2. t , σ t =(σ1 2 +σ2 2 ) 1 / 2 Then calculate the third confidence level C3, C3=exp(-σ t / σ0), where σ0 is the normalized reference parameter, such as 1 μs. C3∈(0,1], the closer C3 is to 1, the more accurate the local clock's measurement time. When t loss =0, meaning that when the clock subsystem receives a satellite signal, the third confidence level C3 = 1.

[0038] Determine the expected coordinates of the fault location. Let the receiving times of the substation measurement subsystem and the section measurement subsystem be t, respectively. A t B Based on the two-end traveling wave ranging formula, the distance d from the fault point to the substation is calculated, where d = [L]. AB +v0(t A -t B )] / 2. Where L AB Let d be the line length between the substation and the sectioning station, and v0 be the theoretical wave velocity. The expected coordinates x of the fault point can be determined based on the distance d.

[0039] Establish the absolute residual sum function for fault location. Assume there are I groups of measurement subsystems involved in fault location, where I ≥ 3. Based on the first confidence level C of the i-th measurement subsystem... 1,i Second confidence level C 2,i Third confidence level C 3,i Calculate the overall confidence level C a,i C a,i =z1C 1,i +z2C 2,i +z3C 3,i z1, z2, and z3 are preset weighting coefficients, satisfying z1 + z2 + z3 = 1. i = 1, 2, ..., I. An absolute residual sum function is established based on the reception time and the overall confidence level: x i x j Let t be the node coordinates of the corresponding nodes of the i-th and j-th measurement subsystems. i t j The receiving times are the fault signal wavefronts of the i-th and j-th measurement subsystems.

[0040] Determine the fault location. Using the expected coordinate x as the center, define the search interval [x-ΔL, x+ΔL], where ΔL is the search distance, which can be 10% of the total line length or determined based on the theoretical error of the double-ended method. Preset a search step size and traverse the search interval using this step size to obtain multiple candidate coordinates. Calculate the absolute residual and function value J(x) for each candidate coordinate, and select the final coordinate that minimizes the absolute residual and function value as the fault location.

[0041] Example 4

[0042] The reflection types include reflections via fault points and reflections via nodes, where nodes include substations, sectioning stations, and AT stations. This embodiment further discloses a preferred method for determining the reflection type and predicting clock reliability.

[0043] The reflection type of the fault secondary signal is identified based on the current polarity of the fault signal and the fault secondary signal of the measurement subsystem. Figure 5 and Figure 8 The waveforms of the fault current signal reflected from the fault point and the node are shown respectively. The fault signal and the secondary fault signal correspond to the first peak signal and the second peak signal in the waveform diagram. (Refer to...) Figure 5 When the current polarity of the fault signal and the secondary fault signal are the same, it indicates that the secondary fault signal is obtained through reflection from the fault point. (Refer to...) Figure 8 When the current polarity of the fault signal and the secondary fault signal are opposite, it indicates that the secondary fault signal is obtained through node reflection.

[0044] Extract the node n0 with the minimum reception time from multiple sets of nodes. The minimum reflection path refers to the path difference between the fault signal and the secondary fault signal of this node. Let the node coordinates of node n0 be x. n0 The fault location of fault point O is x. O .

[0045] The first estimate of the minimum reflection path is generated based on the topological relationship of the fault location in the traction network. The first distance d1 of the fault signal propagation path at node n0 is calculated, where d1 = |x n0 -x O Assuming the secondary fault signal is obtained through reflection from fault point O, refer to... Figure 4 Calculate the second distance d of the secondary fault signal propagation path of node n0. 2,0 d 2,0 =3|x n0 -x O Then calculate the first distance d1 and the second distance d. 2,0 The path difference Δd0, Δd0=|d1-d 2,0 Assuming the secondary fault signal is obtained via node reflection, extract all neighboring nodes of node n0 from the traction network topology graph. (Refer to...) Figure 7 For adjacent node n i Based on its node coordinates x ni Calculate the second distance d 2,i d 2,i =|x ni -x O |+|x n0 -x ni Then calculate the path difference Δd between reflections from different adjacent nodes. i Δd i=|d1-d 2,i |. i=1,2,…. Find the minimum path difference Δd from the path difference sequence [Δd1, Δd2,…]. min Then the first estimate of the minimum reflection path, d min,1 =min(Δd0,Δd min ).

[0046] A second estimate of the minimum reflection path is generated based on the reflection type. If the secondary fault signal is obtained by reflection from the fault point O, the second estimate d... min,2 =Δd0. If the secondary fault signal is obtained through node reflection, d min,2 =Δd min .

[0047] The actual wave velocity is predicted based on the second estimate and the minimum reflection time difference. (Refer to...) Figure 6 and Figure 9 Calculate the reflection time difference between the fault signal and the fault secondary signal wavefront, extract the measurement subsystem with the minimum reception time, and the reflection time difference corresponding to this measurement subsystem is the minimum reflection time difference Δt. min Based on the second estimate and the minimum reflection time difference Δt min Calculate the actual wave speed v1, v1=d min,2 / Δt min .

[0048] The reliability of the clock is predicted based on the first and second estimates, as well as the actual and theoretical wave velocities. Specifically, based on the first estimate d... min,1 Second estimate d min,2 Calculate the path consistency factor H1, H1 = exp(-|d min,1 -d min,2 | / σ d ), σ d For distance tolerance, for example, 0.5% of the total line length can be taken. Calculate the wave speed rationality factor H2 based on the actual wave speed v1 and the theoretical wave speed v0, H2=exp(-|v1-v0| / σ v ), σ v For wave velocity tolerance, for example, 5×10 6 m / s. The clock reliability H is predicted by combining the path consistency factor H1 and the wave speed rationality factor H2, where H = H1·H2. When the first estimate is highly consistent with the second estimate and the actual wave speed is close to the theoretical relative speed, the clock reliability approaches 1, indicating that the local clock synchronization is good; conversely, if the deviation is too large, the clock reliability approaches 0.

[0049] Example 5

[0050] like Figures 2 to 12As shown, the present invention discloses a traction network fault location system for implementing the traction network fault location method based on traveling wave ranging clock synchronization, comprising: a traction network, multiple clock subsystems, multiple measurement subsystems, and a location subsystem. When a fault occurs in the traction network, fault signals are collected through different measurement subsystems, and the signal reception time is determined through the corresponding clock subsystem, thereby predicting the fault location in the traction network.

[0051] like Figure 2 As shown, the traction network includes a substation 11, a sectioning station 12, multiple AT stations 13, a contact wire 14, a positive feeder 15, rails 16, and a return line 17. Substation 11 connects the three-phase line 18 to the contact wire 14 and the positive feeder 15. The pantograph of the train 20 receives power from the contact wire 14 and returns it to the positive feeder 15 via the return line 17 through the rails 16. In this invention, substation 11, sectioning station 12, and multiple AT stations 13 are collectively referred to as nodes.

[0052] The clock subsystem and measurement subsystem correspond one-to-one with each node. The clock subsystem is configured to update the local clocks of the substations, section substations, and multiple AT substations. (See reference...) Figure 11 The clock subsystem includes a satellite clock generator, a crystal clock generator, a time synchronization unit, and a time synchronization unit. The satellite clock generator receives satellite signals and outputs a 1PPS pulse. The crystal clock generator generates a local high-frequency square wave oscillation signal and outputs the crystal frequency. The time synchronization unit is connected to both the satellite clock generator and the crystal clock generator. When a satellite signal is available, the time synchronization unit generates the satellite synchronization time based on the 1PPS pulse and calculates the frequency deviation using the crystal frequency, updating the linear parameters, temperature parameters, and aging parameters of the crystal discipline function. When the satellite signal is lost, the local clock is compensated for based on the clock increment output by the crystal discipline function. The time synchronization unit broadcasts the updated local clock to the measurement subsystem in IRIG-B code or via the NTP / 1588 network protocol.

[0053] The measurement subsystem is configured to acquire fault signals and secondary fault signals from substation 11, section 12, and AT station 13, and determine the reception time, reflection type, and minimum reflection time difference in conjunction with the local clock. (Refer to...) Figure 12 The measurement subsystem includes a fault signal acquisition unit, a high-speed analog-to-digital converter (ADC), a fault signal processing unit, and a time difference calculation unit. The fault signal acquisition unit acquires the fault signal and secondary fault signals. The high-speed ADC converts the analog signals into digital signals recognizable by subsequent systems. The fault signal processing unit identifies parameters such as the wavefront, amplitude, jitter, and signal-to-noise ratio of the fault signal. The time difference calculation unit records the reflection time difference of the fault signal and secondary fault signals and calculates the minimum reflection time difference; the time difference calculation unit requires access to the local clock signal of the clock subsystem.

[0054] The positioning subsystem receives detection data from the measurement subsystems of different nodes and predicts the fault location accordingly. In this embodiment, the positioning subsystem is configured to predict the fault location based on the reception time and a first confidence level, a second confidence level, and a third confidence level, and to request synchronization of the local clock based on the clock reliability. The positioning subsystem can communicate with each node through a communication channel, which can be an optical fiber to reduce interference from electromagnetic signals from the underground traction network. Furthermore, the subway master clock can be integrated into the communication channel; when the positioning subsystem requests synchronization of the local clock, the subway master clock sends a reference clock signal to the clock subsystems of each node.

[0055] 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 within the protection scope of the present invention.

Claims

1. A traction power network fault location method based on clock synchronization of traveling wave distance measurement, characterized in that, Includes the following steps: Step 1: Deploy clock subsystems and measurement subsystems in substations, sectioning stations, and each AT station, and assign a first confidence level to each clock subsystem; Step 2: Each clock subsystem receives satellite signals and updates its local clock and crystal oscillator discipline function. If a clock subsystem loses satellite signals, it updates its local clock according to the crystal oscillator discipline function and records the duration of signal loss. Step 3: If at least one measurement subsystem receives a fault signal, the measurement subsystem determines the time of receipt of the fault signal according to the local clock, the measurement subsystem collects the secondary fault signal, and determines the minimum reflection time difference and reflection type; Step 4: Generate a second confidence level for the corresponding clock subsystem based on the fault signals of different measurement subsystems, and generate a third confidence level based on the signal loss duration of the clock subsystem; Step 5: Predict the fault location based on the receiving time, theoretical wave velocity, and first confidence level, second confidence level, and third confidence level of at least three sets of measurement subsystems; Step 6: Fit the minimum reflection path by combining the fault location and reflection type, then fit the actual wave speed by combining the reflection type and minimum reflection time difference, and finally request synchronization of the local clock by combining the minimum reflection path and actual wave speed.

2. The method according to claim 1, characterized in that, In step 1, multiple AT stations are located between substations and section stations. The first confidence level of the section station clock subsystem is greater than the first confidence level of the AT station clock subsystem and less than the first confidence level of the substation clock subsystem. The first confidence level is updated based on the crystal oscillator type and historical timekeeping error rate of the substation, section station, and AT station clock subsystems.

3. The method of claim 1, wherein, In step 2, a third-order crystal oscillator discipline function is assigned to each clock subsystem. If the clock subsystem receives a satellite signal, it parses the 1PPS pulse from the satellite signal and updates the local clock. It collects the crystal frequency of the crystal clock generator, measures the frequency deviation of the crystal frequency relative to the 1PPS pulse, and updates the linear parameters, temperature parameters, and aging parameters of the crystal oscillator discipline function according to the recursive least squares algorithm.

4. The traction network fault location method based on traveling wave ranging clock synchronization according to claim 3, characterized in that, If the clock subsystem does not receive a satellite signal, it inputs the crystal oscillator frequency into the crystal oscillator discipline function to generate a clock increment, and then updates the local clock.

5. The traction network fault location method based on traveling wave ranging clock synchronization according to claim 1, characterized in that, In step 3, the reflection time difference between the received fault signal and the fault secondary signal wavefront is calculated, and the minimum reception time measurement subsystem is extracted. The reflection time difference corresponding to this measurement subsystem is the minimum reflection time difference. The reflection type is identified based on the current polarity of the fault signal and the fault secondary signal of this measurement subsystem. The reflection type includes reflection through the fault point and reflection through the node. The node includes substations, substations, and AT stations.

6. The traction network fault location method based on traveling wave ranging clock synchronization according to claim 1, characterized in that, In step 4, the wavefront segment signal sequence of the fault signal at the receiving time is intercepted, and the amplitude, jitter, and signal-to-noise ratio of the wavefront segment signal sequence are extracted. After normalization and geometric averaging, the second confidence level is obtained.

7. The traction network fault location method based on traveling wave ranging clock synchronization according to claim 1, characterized in that, In step 4, the drift error of the clock subsystem and the model error of the crystal oscillator discipline function are calculated based on the signal loss duration, and then a third confidence level is generated.

8. The traction network fault location method based on traveling wave ranging clock synchronization according to claim 1, characterized in that, In step 5, the expected coordinates of the fault point are calculated based on the receiving time of the substation measurement subsystem and the section measurement subsystem. The absolute residual sum function of the fault location is established based on the receiving time and the first confidence level, the second confidence level, and the third confidence level. The final coordinates that minimize the absolute residual sum function are found along the expected coordinates. The final coordinates are the fault location.

9. The traction network fault location method based on traveling wave ranging clock synchronization according to claim 1, characterized in that, In step 6, a first estimate of the minimum reflection path is fitted based on the fault location, a second estimate of the minimum reflection path is generated based on the reflection type, the actual wave speed is predicted based on the second estimate and the minimum reflection time difference, and the clock reliability is predicted based on the first estimate, the second estimate, the actual wave speed, and the theoretical wave speed. If the clock reliability is greater than the reliability threshold, return to step 2; otherwise, request synchronization of the local clock of each clock subsystem and return to step 1.

10. A traction network fault location system for implementing the traction network fault location method based on traveling wave ranging and clock synchronization as described in claim 1, characterized in that, include: The traction network includes substations, sectioning stations, and multiple AT stations. Multiple clock subsystems are configured to update the local clocks of the substations, section stations, and multiple AT stations; Multiple measurement subsystems are configured to collect fault signals and secondary fault signals from substations, sectioning stations, and AT stations, and combine them with the local clock to determine the receiving time, reflection type, and minimum reflection time difference; A positioning subsystem is configured to predict the fault location based on the receiving time and a first confidence level, a second confidence level, and a third confidence level, and to request synchronization of the local clock based on the clock reliability.

Images