A high-efficiency energy router protection and control method based on an ANPC topology structure

Abstract

The application relates to a high-efficiency energy router protection and control method based on an ANPC topology structure. The method comprises the following steps: acquiring operation condition data, constraint parameter data and midpoint voltage ripple data corresponding to a rail transit system; implementing zero net load and zero momentum coding on spatial vector modulation output under the ANPC to generate a three-level self-conservation sequence and a time scale buffer index; performing congruence staggered reconstruction on power echoes according to the index to obtain equivalent oversampling echo matrix data; performing row conformal scattering tomography on the matrix data to output distance branch reflection distribution data and positioning coordinate initial value data; performing persistent homology judgment on the reflection distribution data to obtain abnormal morphology data; and fusing the positioning coordinate initial value data and the abnormal morphology data to carry out a zero-regret game planning linkage strategy to generate localized action protection data to control the energy router. The method can effectively ensure the resilience of the energy router protection and control.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of automatic control, in particular to a high-efficiency energy router protection and control method based on an ANPC topology structure. BACKGROUND

[0002] In rail transit, the energy router protection and control method of the traditional technology adopts "relay protection + hierarchical control", that is, overcurrent, over / under voltage, direction / differential and other elements are set for traction network / third rail and converter side according to sections, and speed-break-delay and reclosing are used, and IGBT / SiC rapid shutdown and DC circuit breaker are used for millisecond-level isolation; the secondary side uses PI / droop voltage and current regulation, and the upper SCADA / ATS issues power and switching sequence. However, the traditional technology has poor adaptability to bidirectional power flow caused by regenerative braking, frequent topology changes caused by train movement and fault current caused by converter current limiting, and fixed value failure causes misoperation / refusal to operate, thereby affecting the resilience of energy router protection and control. SUMMARY

[0003] Therefore, it is necessary to provide a high-efficiency energy router protection and control method based on an ANPC topology structure, which can effectively ensure the resilience of energy router protection and control, in view of the above technical problems.

[0004] In a first aspect, the application provides a high-efficiency energy router protection and control method based on an ANPC topology structure, comprising:

[0005] Obtaining operation condition data, constraint parameter data and midpoint voltage ripple data corresponding to a rail transit system; the energy router of the rail transit system adopts an ANPC topology structure;

[0006] According to the operation condition data, the constraint parameter data and the midpoint voltage ripple data, zero net load zero momentum coding is performed on space vector modulation output data under the ANPC topology structure, to obtain three-level self-conserved emission sequence data and time scale cache index data;

[0007] The power echo original data corresponding to the time scale cache index data is reconstructed by congruence interleaving to obtain equivalent oversampling echo matrix data;

[0008] The equivalent oversampling echo matrix data is subjected to conformal scattering tomography to obtain distance branch reflection distribution data and positioning coordinate initial value data;

[0009] The distance branch reflection distribution data is subjected to persistent homology judgment analysis to obtain abnormal morphology data;

[0010] The energy router of the rail transit system is controlled according to the positioning action protection data.

[0011] In a second aspect, the application further provides a high-efficiency energy router protection and control device based on an ANPC topology structure, comprising:

[0012] A system data acquisition module is configured to acquire operation condition data, constraint parameter data and midpoint voltage ripple data corresponding to a rail transit system; the energy router of the rail transit system adopts an ANPC topology structure;

[0013] A system data encoding module is configured to perform zero-payload zero-momentum encoding on spatial vector modulation output data under the ANPC topology structure according to the operation condition data, the constraint parameter data and the midpoint voltage ripple data, to obtain three-level self-conserved emission sequence data and time scale cache index data;

[0014] A backwave data reconstruction module is configured to perform congruence interleaving reconstruction on power backwave original data corresponding to the time scale cache index data, to obtain equivalent oversampling backwave matrix data;

[0015] A backwave data tomography module is configured to perform shape-preserving scattering tomography on the equivalent oversampling backwave matrix data, to obtain distance branch reflection distribution data and positioning coordinate initial value data;

[0016] A distribution data judgment module is configured to perform persistent homology judgment analysis on the distance branch reflection distribution data, to obtain abnormal morphology data;

[0017] A protection data planning module is configured to perform zero-regret game planning on an energy routing linkage strategy of the rail transit system according to the positioning coordinate initial value data and the abnormal morphology data, to obtain positioning action protection data; the positioning action protection data is used to control the energy router.

[0018] In a third aspect, the application further provides a computer device comprising a memory and a processor; the memory stores a computer program; and the processor implements any step of the high-efficiency energy router protection and control method based on an ANPC topology structure when executing the computer program.

[0019] The high-efficiency energy router protection and control method, device and computer equipment based on the ANPC topology structure, by implementing zero payload zero momentum coding on the space vector modulation output under the ANPC topology, generates a three-level self-conservative emission sequence, significantly suppresses the midpoint voltage drift and ripple, reduces the neutral point circulating current and switching stress, thereby reducing the EMI and DC bus capacitor stress; the power echo is reconstructed by congruence interleaving to obtain an equivalent oversampling matrix, and combined with the conformal scattering tomography and persistent homology judgment analysis, high-resolution and strong anti-noise online identification and accurate positioning of impedance discontinuity, poor contact and insulation degradation and other abnormalities in the rail transit energy routing network are realized; further, the zero regret game is used for linkage decision of the positioning coordinates and abnormal form, and a localized action protection is quickly given, and the energy router is closed-loop controlled, the strategy stable convergence and safety redundancy are ensured under the multiple constraints and uncertain conditions, the resilience of the energy router protection and control can be effectively ensured, and the safety, reliability and scalability of the system are significantly enhanced. BRIEF DESCRIPTION OF DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments or the related art, the drawings needed to be used in the embodiments or the related art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0021] Figure 1 An application environment diagram of the high-efficiency energy router protection and control method based on the ANPC topology structure in one embodiment;

[0022] Figure 2 A flowchart of the high-efficiency energy router protection and control method based on the ANPC topology structure in one embodiment;

[0023] Figure 3 A flowchart of the first equivalent oversampling echo matrix data obtaining method in one embodiment;

[0024] Figure 4 A flowchart of the second equivalent oversampling echo matrix data obtaining method in one embodiment;

[0025] Figure 5 A flowchart of the folded frequency band sparse spectrum obtaining method in one embodiment;

[0026] Figure 6 A flowchart of the first distance branch reflection distribution data and positioning coordinate initial value data obtaining method in one embodiment;

[0027] Figure 7A flowchart of a method for obtaining second distance branch reflection distribution data and positioning coordinate initial value data in an embodiment is shown in the figure;

[0028] Figure 8 A flowchart of a method for obtaining abnormal morphology data in an embodiment is shown in the figure;

[0029] Figure 9 A flowchart of a method for obtaining stable topology signature data in an embodiment is shown in the figure;

[0030] Figure 10 A structural block diagram of an efficient energy router protection and control device based on an ANPC topology structure in an embodiment is shown in the figure;

[0031] Figure 11 An internal structural diagram of a computer device in an embodiment is shown in the figure. DETAILED DESCRIPTION

[0032] In order to make the purposes, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0033] An efficient energy router protection and control method based on an ANPC topology structure provided by an embodiment of the present application can be applied in an application environment as shown in the figure. Figure 1 A terminal 102 communicates with a server 104 through a network. A data storage system can store data required to be processed by the server 104. The data storage system can be integrated on the server 104, or placed on a cloud or other network server. The server 104 can be implemented by an independent server or a server cluster composed of multiple servers.

[0034] In an exemplary embodiment, as shown in the figure, Figure 2 An efficient energy router protection and control method based on an ANPC topology structure is provided. The method is taken as an example to be described by a server in Figure 1 The following steps 202 to 212 are included. Among them:

[0035] Step 202, obtaining running condition data, constraint parameter data and midpoint voltage ripple data corresponding to a rail transit system.

[0036] Step 204, according to the running condition data, the constraint parameter data and the midpoint voltage ripple data, zero net load and zero momentum coding is performed on the space vector modulation output data under the ANPC topology structure, to obtain three-level self-conserved emission sequence data and time scale buffer index data.

[0037] Step 206, congruence interleaving reconstruction is performed on the power echo original data corresponding to the time scale cache index data to obtain equivalent oversampling echo matrix data.

[0038] Step 208, conformal scattering tomography is performed on the equivalent oversampling echo matrix data to obtain distance branch reflection distribution data and positioning coordinate initial value data.

[0039] Step 210, persistent homology judgment analysis is performed on the distance branch reflection distribution data to obtain abnormal morphology data.

[0040] Step 212, according to the positioning coordinate initial value data and the abnormal morphology data, zero-regret game planning is performed on the energy routing linkage strategy of the rail transit system to obtain localization action protection data.

[0041] Among them, the operating condition data is the real-time measurement and control reference information such as voltage, current, DC bus, electric temperature and load instruction collected under a unified time base and used to describe the current device and line state.

[0042] Among them, the constraint parameter data is a set of boundary conditions from design and field configuration, including line geometry and branch topology, inter-stage selectivity curve, EMC and CBTC suppression band, power and thermal limit, etc.

[0043] Among them, the midpoint voltage ripple data is the time fluctuation record and statistical characteristics of the midpoint potential of the ANPC split bus relative to the two end buses, which is used to evaluate the midpoint balance and injection constraints.

[0044] Among them, the energy router is a power electronic device that allocates, limits and isolates energy between the traction network, energy storage and section, and links with the protection system.

[0045] Among them, the ANPC topology structure is a three-level power topology using active neutral point clamping, which has a split DC bus and redundant switching state to realize midpoint voltage control and fault tolerance.

[0046] Among them, the space vector modulation output data is the reference duty ratio, zero vector allocation and switching sequence information of each switching arm calculated by SVPWM / DPWM under the current operating condition.

[0047] Among them, the zero net load zero momentum coding is a constraint coding method that shapes the reference switching sequence online to make the DC net charge injection and equivalent momentum in any short window zero at the same time.

[0048] Among them, the three-level self-conservation emission sequence data is a detection sequence composed of only positive half, zero and negative half three-level symbols after coding and does not change the average power transmission.

[0049] Wherein, the time index data is metadata indexing each detection window start-stop time, phase group and code word identification, channel and length to be cached, and quality and check information.

[0050] Wherein, the power echo raw data is a set of unprocessed waveform samples of port voltage / current collected synchronously in the detection window according to the index.

[0051] Wherein, the congruence interleaving reconstruction is a reconstruction method for restoring the folded multi-phase samples to an efficient sampling sequence of unified time base by using coprime phase interleaving sampling and index registration.

[0052] Wherein, the equivalent oversampling echo matrix data is a high sampling rate echo two-dimensional data set and its quality metric organized by "channel x time" after reconstruction.

[0053] Wherein, the conformal scattering tomography is a sparse inversion of echoes under conformal mapping and physical feasibility constraints to obtain the imaging process of reflection and impedance change along the distance and branch.

[0054] Wherein, the distance-branch reflection distribution data is a discretized atlas data expressing the reflection intensity, phase and equivalent impedance increment of each branch along the distance position.

[0055] Wherein, the positioning coordinate initial value data is a preliminary estimate of the kilometer marker, branch identifier and its uncertainty of the suspected anomaly derived from the imaging result and line geometry mapping.

[0056] Wherein, the persistent homology pattern analysis is a judgment process for classifying and evaluating the persistence and morphology of the reflection structure based on the construction and stabilization of time series topological features.

[0057] Wherein, the anomaly morphology data is a comprehensive output of the probability, evolution speed, intensity level and evidence confidence of arc light, ground, moisture, open circuit and other types.

[0058] Wherein, the energy routing linkage strategy is a collaborative action rule and execution sequence of circuit breakers, bypass switches and energy channels under the constraints of device selectivity and operation.

[0059] Wherein, the zero-regret game planning is a robust decision-making method for optimizing the linkage strategy under uncertainty and worst-case assumptions to ensure that the long-term loss is not worse than the optimal benchmark.

[0060] Wherein, the localized action protection data is an executable control instruction set containing target equipment, action type and timing, limit setting, linkage object, rollback condition and audit summary.

[0061] Specifically, the three-phase port voltage / current, DC bus voltage and midpoint voltage ripple, and operating condition data are collected at the energy router with a unified time base (PTP time synchronization); at the same time, constraint parameter data (including CBTC suppression band, EMC limit, line geometry / branch topology priori, circuit breaker selectivity constraint, thermal / junction temperature limit, etc.) are loaded from the engineering database; the physical structure of each energy router in the rail transit system uses ANPC three-level topology as the hardware premise of the processing link to write metadata header, and all data are packaged as Context{condition, constraint, midpoint ripple, HW: ANPC} according to timestamp, channel, effective bit width and confidence.

[0062] With operating condition data, constraint parameter data, and midpoint voltage ripple data as boundary conditions, the reference switching sequence of the space vector modulator under the ANPC topology is read, and after inserting zero net load zero momentum coding (ZNLM) in the modulation back end, the ANPC zero vector redundancy and midpoint clamping path are used to preferentially realize charge self-balancing and equivalent momentum cancellation in any short time window, while imposing constraints on spectrum leakage, trackside communication protection band, gate slope, device thermal balance, and EMC limit, and converging within a fixed time limit by small-scale integer solution and pre-coding this hybrid search. The three-level self-conservative transmission sequence data composed of only three-level symbols and the timestamp cache index data are obtained.

[0063] With the timestamp cache index data as the only instruction, the corresponding power echo original fragments are triggered and read by synchronous collection according to the probe window, phase group, and channel list indicated by the timestamp cache index data, and the samples are congruently registered and labeled with the phase-time slot identifier in the timestamp cache index data. After nearest neighbor grid correction and cross-channel time alignment, abnormal samples are removed and gaps are filled using the check value in the timestamp cache index data. After the quality of the above-processed samples is verified, the folding components are unfolded and the time axis is rearranged according to the interleaving relationship of the phase groups, obtaining high-efficiency sampling sequences under a unified clock, and each channel sequence is regularized into an equivalent super-sampling echo matrix data of "channel x time". If the quality index is below the threshold, the window is marked as unusable and the original fragment is retained for auditing and subsequent playback.

[0064] With the equivalent oversampled echo matrix data as the only observation, and combining the line geometry topology prior in the constraint parameter data, the multi-branch traction network is conformally parameterized, the actual line is mapped to the regular calculation domain according to the branch and node, and the corresponding forward operator snapshot is generated. On this basis, without introducing any external data, the sparse tomography solution with barrier certificate is executed, including using passivity, energy dissipation and midpoint stability as hard constraints, using distance-branch group sparsity and topology continuity as soft constraints, combining multi-resolution solution from coarse to fine and windowed time consistency check, to suppress false images caused by multi-path false peaks and geometric mismatch. After the solution is completed, the distance-branch reflection distribution data is given by the inverse mapping of the conformal mapping, and according to the forward operator sensitivity and residual, certificate residual, and re-projection consistency, the positioning coordinate initial value data with confidence interval is generated.

[0065] With the distance-branch reflection distribution data as the only judgment object, and only referring to the time tag cache index data for time sequencing and window alignment, without introducing any new data, the distance-branch reflection distribution data of each window is constructed according to the branch and distance grid to form a time-advancing topology sequence, and according to the amplitude threshold and spatial adjacency relationship defined in the constraint parameter data to form a hierarchical filtering sequence, to extract persistent and scale-consistent reflection structures. The obtained barcode is stabilized and ordered, and short-lived false peaks and multi-path ghosting are removed, and evidence checking is performed through back-projection consistency and residual threshold. On the stable topology signature that passes the check, combined with the spatiotemporal evolution rules of the fault prototype library, the judgment is completed, and abnormal morphology data is output, including the probability distribution of abnormal type, evolution speed and intensity level, positioning confidence interval, and auditable evidence summary; if the stability or evidence strength is insufficient, it is automatically downgraded to a conservative category and marked for review.

[0066] According to the positioning coordinate initial value data and abnormal morphology data, and with the constraint parameter data as the constraint basis, a candidate action set is generated according to the branch and kilometer marker, and a fast replacement model is used to evaluate the impact of each candidate on voltage drop, short-circuit current, thermal margin, EMC and communication suppression band within a very short prediction window, and candidates that do not meet the inter-stage selectivity, cross-device or line safety boundary, or conflict with the current traffic organization are removed one by one. Under the uncertainty of abnormal type probability and growth rate, a zero-regret robust optimization is performed on the remaining candidates, and a scheme that can still guarantee local isolation and service continuity in the worst case is preferentially selected. Positioning action protection data is generated on the optimization result, which clearly specifies the target device, action type and sequence, start and end time points and duration, power and current limits, bypass or rerouting path, linkage object and communication identifier, and is accompanied by rollback strategy and unlocking conditions. Finally, the criterion summary, key input snapshot, prediction-observation consistency and threshold check are written into the evidence metadata as the positioning action protection data.

[0067] The position action protection data is sent to each energy router and related circuit breaker / bypass unit through the station TSN / IEC 61850 GOOSE / SV, and the local controller executes selective tripping, current limiting, bypassing or energy rerouting according to the instruction, and returns the execution confirmation and monitoring fragments in real time, and updates the running condition protection closed loop archiving.

[0068] In the above-mentioned high-efficiency energy router protection and control method based on the ANPC topology structure, zero net load and zero momentum coding is performed on the space vector modulation output under the ANPC topology to generate a three-level self-conserved emission sequence, which significantly suppresses the midpoint voltage drift and ripple, reduces the neutral point circulating current and switching stress, thereby reducing the EMI and DC bus capacitor stress; the power echo is reconstructed by congruence interleaving to obtain an equivalent oversampling matrix, and combined with conformal scattering tomography and persistent homology judgment analysis, high-resolution and strong anti-noise online identification and accurate positioning of impedance discontinuity / contact failure / insulation degradation and other abnormalities in the rail transit energy routing network are realized; further, the positioning coordinates and abnormal morphology are linked to make decisions by using zero regret game, and the position action protection is quickly given to close-loop control the energy router, which ensures stable convergence and safety redundancy of the strategy under multiple constraints and uncertain conditions, and can effectively ensure the resilience of energy router protection and control, significantly enhancing the safety, reliability and scalability of the system.

[0069] In one exemplary embodiment, as shown in Figure 3 The power echo original data corresponding to the time scale cache index data is reconstructed by congruence interleaving to obtain equivalent oversampling echo matrix data, including steps 302 to 304. Wherein:

[0070] Step 302, according to the time scale cache index data, the power echo original data is decoded by grid nearest point congruence registration to obtain congruence registration sample sequence data.

[0071] Step 304, according to the three-level self-conserved emission sequence data, the congruence registration sample sequence data is reconstructed by sparse congruence de-stacking to obtain equivalent oversampling echo matrix data.

[0072] Wherein, the grid nearest point congruence registration decoding is to map the echo original sample to a unified time grid according to the time scale cache index, and correct the jitter and offset by the nearest neighbor grid rule to give the sample a unique high-order time index.

[0073] Wherein, the congruence registration sample sequence data is a sample sequence that is rearranged by "channel-unified time index-phase label" after registration decoding, and is attached with quality marks such as coverage and erasure bits and check digest.

[0074] The sparse congruence solution unfolding reconstruction is a method for recovering an efficient sampling echo matrix with unified time base by unfolding the folded time slots and performing sparse refinement under physical and time constraints, taking the congruence registration samples as observations and the three-level self-conservative transmission sequence as priori.

[0075] Specifically, the start and end time of the detection window, the phase group identifier, the time slot mode, the channel list and the check information recorded in the time tag cache index data are used to extract the corresponding power echo raw data from the acquisition cache, and then the integrity of the power echo raw data is verified and abnormal samples are removed according to the continuity and check information in the time tag cache index data. Further, the de-skew and micro-jitter correction is performed according to the channel list and time slot structure given by the time tag cache index data, so that each channel is aligned to a unified reference time base. Using the phase group and time slot identifier in the time tag cache index data, each sample in the corrected power echo raw data is mapped to its congruence category and decoded by the nearest lattice point, conflicts and overlaps are solved, and the samples are "aligned" to the standard time grid. Erasure markers are generated for missing or unreliable time slots to preserve traceable vacancies, and finally the congruence registration sample sequence data is reordered and output in the order of "channel-unified time sequence-phase label".

[0076] Using the congruence registration sample sequence data as observations, the self-correlation baseline and phase transition position are extracted from the three-level self-conservative transmission sequence data, the folding-unfolding dictionary and consistency checking rules consistent with the congruence sampling relationship are constructed, and the fast self-calibration of amplitude and phase scale is completed. According to the iterative process of "coarse first and then fine", the initial high sampling time sequence consistent with the congruence relationship is generated, and then the sparsification and threshold refinement are performed under the constraints of physical feasibility and time causality. The consistency rules are used for back projection verification round by round until the residual error and quality indicators meet the threshold or the iteration limit is reached. Finally, the high sampling reconstruction sequence of each channel is regularized to the equivalent oversampling echo matrix data of "channel x time".

[0077] In this embodiment, by unifying the dispersed and folded multi-phase echo samples to the same time grid and reconstructing them into a high-efficiency sampling echo matrix without increasing the ADC sampling rate and without modifying the hardware, the visibility of short-time details and high-frequency information is significantly improved. The method is robust to jitter, a small amount of sample loss and cross-channel offset, and can reduce false peaks and misjudgments through registration verification and back projection consistency, thereby improving imaging and positioning stability. At the same time, quality metadata such as coverage, residual error and uncertainty are output, which facilitates auditing and rollback, and can achieve higher resolution, lower false operation rate and better real-time performance without increasing EMC risk and system cost.

[0078] In one exemplary embodiment, as Figure 4As shown, according to the three-level self-conserved emission sequence data, the sparse congruence deconvolution reconstruction is performed on the congruence registration sample sequence data to obtain equivalent super-sampling echo matrix data, including steps 402 to 406. Among them:

[0079] Step 402, joint data type cyclic lifting autocorrelation modeling is performed on the three-level self-conserved emission sequence data and the congruence registration sample sequence data to obtain autocorrelation reference data and congruence sparse dictionary data.

[0080] Step 404, number theory mask sparse deconvolution is performed on the autocorrelation reference data and the congruence sparse dictionary data to obtain folded frequency band sparse spectrum.

[0081] Step 406, inverse congruence phase continuity reconstruction is performed on the folded frequency band sparse spectrum to obtain equivalent super-sampling echo matrix data.

[0082] Among them, the joint data type is to regard two or more aligned data sources as a whole input to model and solve synchronously, so that timing, constraints and checks are consistently executed in the same framework.

[0083] Among them, the cyclic lifting autocorrelation modeling is a modeling method for incrementally updating and robustly refining autocorrelation structure in a rolling time window, which is used to obtain a reference template that is resistant to noise and can track drift.

[0084] Among them, the autocorrelation reference data is a stable reference information set obtained by autocorrelation modeling, including main peak position, peak width, sidelobe shape and delay characteristics, etc., which is used as a criterion for subsequent matching and deconvolution.

[0085] Among them, the congruence sparse dictionary data is a sparse dictionary entry set organized by congruence time slots and phase labels, each entry corresponding to a reconstructable response atom and its index relationship.

[0086] Among them, the number theory mask sparse deconvolution is a deconvolution method that uses number theory masks to construct higher-differentiation observation channels and impose sparse constraints to separate the folded and aliased components into real components.

[0087] Among them, the folded frequency band sparse spectrum is a sparse weight set obtained in the folded frequency band coordinate system after deconvolution, which describes the amplitude-phase contribution and confidence of each atomic component.

[0088] Among them, the inverse congruence phase continuity reconstruction is a reconstruction process that performs phase unfolding and continuity correction according to the sparse spectrum coefficients along the congruence branch, and maps back to the unified time grid to generate high equivalent sampling echo matrix.

[0089] Specifically, the three-level self-conserved emission sequence data and the congruence registration sample sequence data are strictly aligned and synchronized with the time scale and the window, and the reference pulse and the noise baseline are extracted from the known symbol segment in the processed three-level self-conserved emission sequence data. Further, the registration samples are grouped and aggregated according to the phase group and the congruence label, and the rolling estimation of the autocorrelation spectrum is performed for each window in a cyclic lifting manner, so that the main peak position, the peak width, the sidelobe shape and the time delay drift characteristics are stably extracted and the reusable autocorrelation reference data are formed. Under this reference, the congruence label and the time grid are mapped to a plurality of “response atoms”, and a congruence sparse dictionary is generated with the channel, the branch, the window and the phase group as the key, and the effective time period, the confidence and the quality mark are attached to each column.

[0090] According to the autocorrelation reference data and the congruence sparse dictionary data, the reference peak position and the dictionary column label are derived to derive a number theory mask sequence, the congruence sparse dictionary data is mapped to a group of high-differentiation mask domain observation channels according to the congruence grouping, and the most stable feature in the autocorrelation reference data is taken as the starting point to obtain initial sparse data. An iterative process of “rough positioning first and then fine sparsification” is adopted, and under the constraints of group continuity and time consistency, the initial sparse data is alternately sparsified and the weight is adaptively updated, and the initial sparse data with erasure marks or high abnormal probability is implemented with confidence weight reduction and local rollback. Each round performs a back projection consistency and cross-channel coherence check, and if the residual and stability thresholds are met, it stops early, otherwise it automatically adjusts the threshold and step size to continue to refine, obtains a folded band sparse spectrum, and simultaneously gives quality and audit metadata such as residual, sparsity, mask coherence, confidence interval and processing trajectory.

[0091] According to the mapping index in the folded band sparse spectrum (including metadata such as dictionary mapping index, confidence, residual and processing trajectory), each sparse component is restored to the corresponding congruence branch and time slice, and the order relationship within the branch and between the branches is established. Under the constraint of quality mark, the branches after the relationship is established are executed with phase unfolding and continuity correction, the phase alignment and amplitude calibration are preferentially completed for high-confidence components, and the calibration step size is adaptively adjusted according to the residual and coherence score. After completing the intra-branch calibration, cross-branch splicing and boundary connection are performed according to the time axis, conservative interpolation is used for the segments with erasure bits or low confidence, and uncertainty is explicitly marked. Finally, the back projection consistency and coherence of the channel results are reviewed, the segments that do not meet the threshold are automatically downgraded or rolled back to the previous stable version, and the reconstructed sequence that passes the review is regularized to an equivalent oversampling echo matrix data of “channel x time”.

[0092] In this embodiment, by jointly modeling the emission sequence and the registration sample, a robust reference and sparse dictionary are formed in the same time base, significantly suppressing jitter and noise; then by number theory mask sparse unfolding, the folded and overlapped components are effectively split, without improving the ADC sampling rate and without changing the hardware, to restore high-frequency details and reduce false peaks and misjudgments; finally, the inverse congruence phase continuous reconstruction ensures the coherence of amplitude and phase over time, and outputs a high-efficiency sampling echo matrix with uncertainty annotation, which can realize higher resolution, stronger robustness and better real-time performance.

[0093] In an exemplary embodiment, as shown in Figure 5 The autocorrelation reference data and the congruence sparse dictionary data are subjected to number theory mask sparse unfolding to obtain a folded frequency band sparse spectrum, including steps 502 to 508. Among them:

[0094] Step 502, the primitive root seed of the autocorrelation reference data and the congruence sparse dictionary data is derived and mapped to obtain number theory mask sequence data.

[0095] Step 504, the number theory mask sequence data and the congruence registration sample sequence data are subjected to primitive root sequence joint shaping to obtain mask domain measurement matrix data and mask domain observation vector data.

[0096] Step 506, the mask domain measurement matrix data and the mask domain observation vector data are subjected to non-convex sparse unfolding to obtain an initial folded frequency band sparse spectrum.

[0097] Step 508, the initial folded frequency band sparse spectrum and the mask domain measurement matrix data are subjected to split consistency projection to obtain a folded frequency band sparse spectrum.

[0098] Among them, the primitive root seed derivation is to extract the most discriminative reference features according to the dictionary grouping of the autocorrelation reference data and the congruence sparse dictionary data, and to generate reproducible mask seed indexes by using the number theory primitive root rule.

[0099] Among them, the congruence mapping is to classify the masks or samples into corresponding congruence time slots and groups according to the time scale and phase label, so as to establish a one-to-one index relationship.

[0100] Among them, the number theory mask sequence data is a set of ordered mask segments and their effective intervals and quality marks obtained by the primitive root seed through congruence mapping, which are used to distinguish folded components.

[0101] Among them, the primitive root sequence joint shaping is to jointly optimize and sort multiple mask segments under a unified time scale to improve the discriminability between channels and suppress mutual coherence.

[0102] Among them, the mask domain measurement matrix data is matrix data generated by the corresponding relationship between the shaped mask and the sparse dictionary, which is used to represent the mapping of the observation channel to the sparse coefficient.

[0103] Wherein, the mask domain observation vector data is a set of ordered and traceable observation vectors formed by projecting and aggregating the congruence registered samples by mask channel.

[0104] Wherein, the non-convex sparse deconvolution is an inversion process of splitting and restoring the aliasing components by non-convex sparse criterion under observation consistency and structural constraint.

[0105] Wherein, the initial folded frequency band sparse spectrum is the sparse coefficient distribution and its confidence and residual markers in the folded coordinate system obtained by the first deconvolution.

[0106] Wherein, the split consistency projection is a refinement method for obtaining stable and auditable solutions by step-by-step projection and alternating iteration of constraint satisfaction and observation consistency through operator splitting.

[0107] Specifically, after the time window and label consistency registration of the autocorrelation reference data and the congruence sparse dictionary data as input, according to the stable and high-discriminative features in the autocorrelation reference data and the congruence grouping, phase label and effective interval of the congruence sparse dictionary data, the "primitive root seed" which can be uniquely reproduced is derived and the candidate mask segment with fixed order and length is generated accordingly. On this basis, the candidate segments are executed congruence mapping and conflict-free sorting according to the dictionary grouping, overlapping and low-confidence segments are removed and necessary coverage interval is supplemented to form the number theory mask sequence data which is one-to-one corresponding in each time window and grouping; at the same time, the effective period, source pointer, conflict processing record and quality score are attached to each mask.

[0108] After the one-to-one registration of the number theory mask sequence data and the congruence registered sample sequence data according to time window and label, the mask segments are mapped to the corresponding sample subsets and the alignment, amplitude and phase normalization and erasure bit processing are completed. In the organization mode of "mask segment x dictionary grouping", weighted projection and aggregation are performed on the selected samples, and the mask domain measurement matrix data is stacked according to the unified index and order rule, and the projection results corresponding to it are concatenated by channel and window to generate mask domain observation vector data; in the generation process, quality metadata such as coverage, coherence, gap filling and abnormal weight reduction are recorded synchronously.

[0109] According to the quality mark and the erasing bit, the mask domain measurement matrix data and the mask domain observation vector data are set confidence weight and available sample set, and coarse initialization data satisfying observation consistency is obtained; then, the non-convex sparse solution folding process of "first sparse and then refined" is used to iteratively update the coarse initialization data, wherein in the process of iterative update, in each round, the high confidence channel of the coarse initialization data is preferentially contracted, the low confidence and abnormal fragments of the coarse initialization data are down-weighted and the update amplitude is limited, and group continuity and time consistency are applied to suppress false peaks and breaks. At the end of each round, back projection verification and coherence check are performed, if not up to standard, the threshold and step size are adaptively adjusted for continuous iteration, when the residual and stability threshold or the iteration upper limit is reached, the initial folding band sparse spectrum is obtained, and audit metadata such as residual statistics, sparsity, coherence score, confidence interval and convergence flag are synchronously given.

[0110] According to the quality mark, the initial folding band sparse spectrum and the reliable section of the mask domain measurement matrix data are locked and the high confidence principal component is frozen, and after establishing the feasible region constraint for the boundary and low confidence coefficient, the consistency projection is performed, the sparse spectrum is projected back to the solution set consistent with the observation and the components not matching the measurement matrix are corrected. Then, the sequence of "principal component preservation, boundary refinement and false peak suppression" is cyclically iterated, in each round of the sequence of cyclic iteration, the local over-dense or isolated sparse spectrum is first purified, then the adjacent sections are smoothly connected and the energy is redistributed, and finally the back projection residual and stability index are used to determine whether to converge or to continue to refine by adaptively tightening the threshold. After the iteration is completed, the final version of the folding band sparse spectrum is output, and audit metadata such as residual statistics, reserved coefficient proportion, boundary correction record and effectiveness mark are generated at the same time.

[0111] In the embodiment, by using the primitive root mask to make the folding component "divisible", and then integrating the mask and the sample into a one-to-one measurement-observation pair, the non-convex sparse solution folding is used to split the aliasing and stabilize the boundary and observation consistency through split consistency projection, so that a clean folding band sparse spectrum is obtained. Without increasing the ADC sampling rate and without modifying the hardware, the equivalent resolution and the weak reflection detection rate can be significantly improved, the false peak and the misjudgment can be reduced, the auditable results with quality marks can be output, and the robustness and real-time performance of imaging positioning can be enhanced.

[0112] In one exemplary embodiment, as shown in Figure 6 As shown in FIG. 6, the equivalent oversampling echo matrix data is subjected to conformal scattering tomography to obtain distance branch reflection distribution data and positioning coordinate initial value data, including steps 602 to 604. Wherein:

[0113] Step 602: joint conformal parameterization mapping is performed on the equivalent oversampling echo matrix data and the line geometry topology template data in the constraint parameter data to obtain conformal mapping domain observation vector and conformal forward operator data.

[0114] Step 604, under the constraint of the conformal forward operator data, barrier certificate constraint sparse tomography is performed on the conformal mapping domain observation vector to obtain distance branch reflection distribution data and positioning coordinate initial value data.

[0115] Wherein, the line geometry topology template data is the priori given line node, branch connection relationship, relative length proportion and boundary condition, etc. for constraining the structure framework of mapping and imaging.

[0116] Wherein, the joint conformal parameterized mapping is to simultaneously perform coordinate transformation on actual line topology and corresponding observation data under the same mapping, keeping the local angle and adjacency relationship unchanged to enter the regular calculation domain.

[0117] Wherein, the conformal mapping domain observation vector is the space-time consistent observation data column formed on each edge and node after re-indexing the equivalent oversampling echoes to the regular calculation domain according to the conformal mapping.

[0118] Wherein, the conformal forward operator data is the response relationship matrix obtained by discretization in the mapping domain, describing the contribution and sensitivity of the reflection change of each branch at each position to the corresponding observation channel.

[0119] Wherein, the barrier certificate constraint sparse tomography is an imaging process of performing observation consistent and sparse priority inversion to reconstruct the line reflection distribution under the physical feasibility of passivity and energy dissipation as barrier constraints.

[0120] Specifically, taking the equivalent oversampling echo matrix data as the observation, the line geometry topology template data in the constraint parameter data is checked for consistency and simplified to generate a calculable topology graph that maintains the node-branch connection relationship and relative length proportion. Under the premise of not changing the topology adjacency of the calculable topology graph, conformal parameterization is completed to map the actual line to a unified regular calculation domain, and each channel echo is re-indexed to the corresponding edge and node in sections to form the conformal mapping domain observation vector data. At the same time, the conformal forward operator data is automatically generated according to the same mapping and the given boundary conditions, the response corresponding relationship of each branch discrete position and observation channel is determined, and quality metadata such as coverage, distortion and mapping residual error are simultaneously output.

[0121] According to the sensitivity of the conformal forward operator data, the suspected reflection segments are quickly screened out on the coarse grid corresponding to the observation vector data in the conformal mapping domain as the initial solution. Then, under the constraints of the barrier certificate composed of passivity, energy dissipation, and midpoint stability, the structure sparsity of "branch grouping and distance continuity" is taken as the main line to gradually refine, and the consistency of back projection and multipath suppression are checked while solving, and the conservative contraction and amplitude limiting strategy is adopted for the boundary and low confidence segment, until the residual and the certificate margin meet the threshold value at the same time, and the distance branch reflection distribution data is obtained according to the distance dispersion of each branch, and the positioning coordinate initial value data (including kilometer mark, branch identification, and uncertainty) is generated through the inverse mapping consistent with the conformal mapping, while the convergence state, evidence strength, and quality mark are attached.

[0122] In this embodiment, by first using the line geometry topology template for joint conformal mapping, the line and observation are synchronized to regulate to reduce geometric mismatch and multipath distortion, and then barrier certificate sparse chromatography is performed under the constraint of the forward operator to limit the solution to the physically feasible domain and suppress false peaks and noise, finally the distance-branch reflection distribution and positioning coordinate initial value (including uncertainty) are output without changing the hardware, which can significantly improve the imaging resolution, positioning accuracy, and result stability.

[0123] In one exemplary embodiment, as shown in Figure 7 As shown in FIG. 7, under the constraint of the conformal forward operator data, barrier certificate constrained sparse chromatography is performed on the observation vector in the conformal mapping domain to obtain distance branch reflection distribution data and positioning coordinate initial value data, including steps 702 to 706. Among them:

[0124] Step 702, the conformal mapping domain observation vector data and the conformal forward operator data are subjected to positive-real-dissipation barrier function, to obtain barrier domain objective function data and feasible set operator data.

[0125] Step 704, the barrier domain objective function data and the feasible set operator data are subjected to sparse chromatography to obtain initial distance branch sparse reflection data.

[0126] Step 706, the initial distance branch sparse reflection data is subjected to semi-positive definite spectral domain projection reconstruction to obtain distance branch reflection distribution data and positioning coordinate initial value data.

[0127] The positive-real-dissipation barrier function is to convert physical constraints such as passivity, energy non-increasing, and stability into computable barrier clauses, which are used to limit the solution to the physically feasible domain during the solving process.

[0128] The barrier domain objective function data is the unified optimization objective and its parameterized description after combining the observation consistency term and the positive-real-dissipation barrier clause.

[0129] Wherein, the feasible set operator data is a set of rules and operation interfaces for checking the projection solution to the feasible domain, including entry / exit criteria, boundary correction and audit metadata.

[0130] Wherein, the sparse tomography is a process of inverting the reflection field to image under the observation consistency and structural sparse constraints, to reconstruct the spatial distribution with a small number of non-zero components.

[0131] Wherein, the initial distance branch sparse reflection data is the sparse reflection coefficient estimation and its confidence and residual label along each branch and distance discrete position obtained by the first iteration of sparse tomography.

[0132] Wherein, the semi-positive spectral domain projection reconstruction is a refinement step of performing semi-positive projection and physical consistency correction on the initial sparse solution in the spectral domain, and reflecting it to the spatial domain to obtain a consistent reflection distribution and positioning initial value.

[0133] Specifically, the conformal mapping domain observation vector data and conformal forward operator data are checked for consistency and dimensionless, confirming that the observation channels, time windows and operator discrete grid points correspond one by one; Then, according to the physical feasibility principle (such as passivity, energy non-increasing, causality and midpoint stability, etc.), the data after consistency checking and dimensionless are converted into computable barrier clauses and violation penalty rules item by item, and the trigger conditions, priorities and exemption lists are established according to the channels, branches and distance grids. On this basis, the barrier domain objective function data containing observation consistency terms and barrier clauses are combined, and the executable feasible set operator data is generated, which includes the feasible domain entry-exit criteria, boundary correction process, residual and certificate margin check interface, and the metadata fields required for audit.

[0134] Without introducing any external information, first, generate an initial support set as a starting point on a coarse grid according to the feasible set rules, and then enter the iterative refinement. Iterative refinement is promoted in the fixed order of "support update - amplitude and phase update - feasible domain correction". In the support update stage, high-confidence entries are preferentially retained and discrete false peaks are removed. In the amplitude and phase update stage, the barrier domain objective function is used as the only criterion for shrinkage and reweighting, taking into account group consistency and distance continuity. In the feasible domain correction stage, the feasible set operator is called to complete the projection and violation repair of constraints such as passivity and energy dissipation, and the non-compliant entries are de-weighted or frozen to prevent rebound. At the end of each round, perform back-projection consistency and barrier margin check. If the residual converges and the margin meets the standard, terminate early, otherwise continue iteration with adaptive tightening of the threshold; Finally, the initial distance branch sparse reflection data is obtained, and quality metadata such as support set, residual statistics, barrier margin, anomaly repair record and convergence flag are generated at the same time.

[0135] The frozen initial distance branch sparse reflection data (including index key, confidence mark and processing track) is a high-confidence main component, and the semi-positive projection and energy redistribution in the spectral domain are applied to the boundary and low-confidence component to quickly remove abnormal components that do not meet the physical feasibility and suppress isolated false peaks. Subsequently, the continuity modification and local smoothing are performed in the order of branch and distance to make the data topologically coherent and temporally stable. Finally, the mapping index embedded in the data is used to map the result back to the spatial domain to generate distance branch reflection distribution data, and the positioning coordinate initial value data is obtained through peak cluster aggregation and gravity estimation. At the same time, the uncertainty and effectiveness mark are output according to the projection gap and processing track, and if the quality threshold is not met, the solution is automatically degraded or rolled back to the previous step.

[0136] In the embodiment, the passive / dissipation physical constraints are embedded and solved by observing and forward operator barrier function, which suppresses unrealistic solutions; then a small amount of key reflections are extracted by sparse tomography, which significantly reduces noise and multipath false peaks, and improves weak fault detection rate; the semi-positive spectral domain projection is used to reconstruct the physical feasibility and boundary stability, and the distance-branch reflection distribution and high-confidence positioning coordinate initial value are output, which can achieve higher resolution, lower misjudgment and better real-time performance without changing the hardware.

[0137] In one exemplary embodiment, as shown in Figure 8 a persistent homotopy pattern analysis is performed on the distance branch reflection distribution data to obtain abnormal morphology data, including steps 802 to 806. Among them:

[0138] Step 802, construct the time sequence zigzag persistent homotopy of distance branch reflection distribution data and time tag buffer index data to obtain persistent barcode data.

[0139] Step 804, centroid stabilization analysis is performed on the persistent barcode data to obtain stable topological signature data.

[0140] Step 806, minimum description length causal pattern analysis is performed on the stable topological signature data to obtain abnormal morphology data.

[0141] Among them, the time sequence zigzag persistent homotopy is a zigzag rule of "add-remove-add" of the topological structure of each window in time sequence, which tracks its appearance and disappearance to extract the topological elements that exist continuously.

[0142] Among them, the persistent barcode data is a barcode-like representation obtained by persistent homotopy, which records the start and end time, duration and intensity of each topological element and other time evolution information.

[0143] Among them, the centroid stabilization analysis is to calculate the weighted centroid after time alignment and consistency of the barcode, and to remove jitter and short-lived false peaks by continuity and back projection verification to obtain a robust representative.

[0144] Wherein, the stable topological signature data is a long-term, cross-scale consistent topological feature set retained after stabilization screening and centroid calculation, used for downstream judgment.

[0145] Wherein, the minimum description length causal judgment analysis is to select the shortest description length and the smallest residual in the candidate causal model to output the abnormal type and its probability according to the principle of "explaining the most observations with the least information".

[0146] Specifically, after window alignment and time normalization by referring to the time tag cache index data, the reflection map in each time window is constructed into a topological structure according to branch and distance grid (adjacent grid and same branch adjacency are connected to form a candidate reflection cluster, and a unified threshold is used). Then, the structures of adjacent windows are connected in the zigzag rule of "add-remove-add" in time sequence, recording the whole process of each topological element from appearance, persistence to disappearance. In the concatenation process, three constraints of minimum survival time, cross-window consistency and adjacency continuity are applied to remove temporary false peaks and isolated noise, and a conservative union strategy is used for boundary change to avoid false deletion, finally outputting the persistent barcode data, in which each persistent element is labeled with start and end window, branch and distance range, duration, stability and audit mark.

[0147] The persistent barcode data is unified to the same time grid and threshold system in different windows and scales, and the barcodes are associated as continuous "tracks" according to branch and homology dimension as keys. For each track, order consistency and time continuity constraints are executed, combined with birth-death threshold, hysteresis and gap compensation to eliminate jitter and temporary breakage, then the track centroid is calculated with duration, spatial support range and intensity stability as weights, and the mutation segment is verified for back projection consistency and abnormal weight reduction. According to the admission rules of stability and evidence strength, unstable elements are screened out, and the stable topological signature data organized by branch and dimension is output, with centroid position, time span, stability index and audit mark of generated track.

[0148] Without introducing any new information, a set of candidate explanation models (such as persistent gradual increase, intermittent recurrence, burst-decay, migration diffusion, etc. spatiotemporal trajectory paradigm) is generated from the signed spatial support, duration, occurrence order and concurrent relationship in the stable topological signature data, and a corresponding causal narrative and residual error description is established for each candidate explanation model. The overall description length and evidence consistency of the candidate explanation models are evaluated one by one according to the principle of "explaining the most phenomena with the least information", and the simplest and residual smallest model is selected as the main judgment, and the relative inferiority of the remaining candidates forms a type probability distribution. Before the final result is determined, key segments are checked for re-throwing and stability, and uncertain or conflicting sections are down-weighted and conservative conclusions are output, generating abnormal morphology data including abnormal type probability, evolution speed and intensity level, fixed position confidence interval and uncertainty.

[0149] In this embodiment, by not introducing new data and not interrupting the existing process, long-term stable topological elements are first extracted from the "distance-branch reflection distribution" using time series zigzag persistent homology, naturally filtering out transient noise and multipath false peaks; then the representative signature with consistent time and order is realized through centroid stabilization, significantly reducing jitter and false tracking; finally, the abnormal type and evolution intensity with explainability and uncertainty are output with the minimum description length causal judgment type, which can achieve higher weak fault detection rate, lower false alarm rate, more stable position confidence and auditable results, while maintaining real-time performance and zero modification to existing hardware.

[0150] In one exemplary embodiment, as shown in Figure 9 centroid stabilization analysis is performed on the persistent barcode data to obtain stable topological signature data, including steps 902 to 906. Among them:

[0151] Step 902, time registration normalization is performed on the persistent barcode data and time index data to obtain registered barcode array data.

[0152] Step 904, multi-scale aggregation integration is performed according to the registered barcode array data to obtain multi-scale barcode cluster data.

[0153] Step 906, weighted centroid synthesis is performed according to the multi-scale barcode cluster data to obtain initial topological centroid data.

[0154] Step 908, consistency backfilling stabilization is performed according to the initial topological centroid data to obtain stable topological signature data.

[0155] Among them, time registration normalization is to unify the barcodes of different windows to the same absolute time grid by using time index, to standardize the start and end time and step length and to make traceable interpolation or erasure mark for the gaps.

[0156] Wherein, the registration barcode array data is a two-dimensional barcode set rearranged by "barcode number x uniform time grid", with quality metadata such as source, branch-distance range and integrity.

[0157] Wherein, the multi-scale aggregation integration is based on single-scale pre-aggregation, combined with top-down and bottom-up consistency rules, merging cross-scale and co-evolution barcodes into stable sets and removing outliers.

[0158] Wherein, the multi-scale barcode cluster data is the aggregated barcode cluster and its member list, time span, spatial support and stability score, etc. descriptive data.

[0159] Wherein, the weighted centroid synthesis is to aggregate the positions and intensities of the members in the cluster according to the weights of duration, coverage, stability and cross-scale consistency, to generate continuous representative centroid trajectories.

[0160] Wherein, the initial topological centroid data is the centroid trajectory obtained after weighted synthesis, as well as its start and end, position and intensity, smoothness, conflict processing record and uncertainty label.

[0161] Wherein, the consistency backfilling stabilization is to project the centroid trajectory back to the registration array to verify the order and continuity, perform gap compensation and boundary convergence, and eliminate non-compliant fragments to output stable topological signatures.

[0162] Specifically, taking the persistent barcode data as the main body and referencing the time tag cache index data to calibrate the time reference, first use the time tag cache index data to map each window barcode to the same absolute time axis, unify the start time, sampling step and threshold semantics; then correct the alignment for cross-window offset and drift, perform traceable interpolation or erasure marking for gaps, and clean up duplicate or overlapping records to avoid double counting. After verification, the registration barcode array data is generated by rearranging according to the rule of "barcode number x uniform time grid", and quality metadata such as source window, branch-distance range, coverage and integrity label are attached to each barcode.

[0163] The registration barcode array data is pre-aggregated within a single scale according to branch-distance adjacency and time continuity, merging overlapping and co-evolution members and removing obvious outliers. Then project the coarse-scale cluster into a candidate skeleton from top to bottom, and construct a multi-scale hierarchy by supplementing, refining and removing false points from the skeleton with fine-scale evidence from bottom to top. During this process, unify the member identity and time boundary to eliminate duplicates and breaks. For areas with conflicts, prioritize coverage, persistence and consistency to retain and redraw boundaries, and reduce the weight or remove unstable sets. For each aggregation result, generate multi-scale barcode cluster data, including cluster identification, member list, time span, spatial support range and stability score, etc. quality metadata.

[0164] Without introducing any new information, first, based on the existing metadata of the duration, coverage, stability score, spatial support range, and cross-scale consistency of the multi-scale barcode cluster data, weights are generated for each cluster and its members, and the members with low weights are removed. On this basis, the weighted members are aggregated in time sequence and branch-distance coordinate pairs for position and intensity, obtaining continuous centroid trajectories. During the aggregation process, boundary convergence and gap compensation are performed for short jumps, conflict resolution and merging are performed for overlapping centroids according to priority, conservative interpolation is performed for low-confidence segments, and uncertainty is explicitly marked, obtaining initial topological centroid data, which contains the start and end time, representative position and intensity, trajectory smoothing index, conflict processing record, and quality label of each centroid trajectory.

[0165] According to the initial topological centroid data (third step output) and the registered barcode array data, each centroid trajectory is back-projected to a unified time grid and corresponding barcode members are verified one by one to check the order consistency, time continuity, and branch-distance adjacency relationship. For the found start and end misplacement, short break and boundary leakage, order repair, gap compensation and boundary convergence are performed in sequence according to the order of "first correction, then rejection". Those that cannot meet the consistency are down-weighted or removed. Then, according to the stability, coverage and back-projection residual, the threshold is set to admit, and the trajectories that pass the verification are fixed and written into the source pointer, repair record and validity label, and finally the stable topological signature data organized by branch and cohomology dimension is output.

[0166] In this embodiment, by normalizing different windows of barcodes to the same time grid through time registration, the timing deviation and gap are eliminated; then through multi-scale aggregation integration, the elements that evolve across scales and in the same direction are aggregated into stable clusters, significantly suppressing short false peaks and outlier segments; then through weighted centroid synthesis, continuous and representative initial trajectories are obtained, improving the distinguishability of weak signals; finally, the stable topological signature data is output by consistency back-filling to correct the order and continuity and remove the non-compliant segments, which can significantly reduce false positives, improve the reliability and auditability of the judgment, and maintain good real-time performance without changing the hardware.

[0167] It should be understood that although each step in the flowchart involved in the above-described embodiments is shown in sequence according to the direction of the arrow, these steps are not necessarily executed in the order indicated by the arrow. Unless otherwise specified herein, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other orders. Moreover, at least some of the steps in the flowchart involved in the above-described embodiments can include multiple steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily sequential, but can be executed alternately or alternately with at least part of other steps or steps or stages in other steps.

[0168] Based on the same inventive concept, the embodiments of the present application also provide an ANPC topology-based high-efficiency energy router protection and control device for implementing the above-mentioned ANPC topology-based high-efficiency energy router protection and control method. As shown in Figure 10 The device provides a solution implementation similar to the implementation described in the above method, so the specific limitations of one or more ANPC topology-based high-efficiency energy router protection and control device embodiments provided below can refer to the above limitations of an ANPC topology-based high-efficiency energy router protection and control method, and will not be repeated here.

[0169] Each module in the above ANPC topology-based high-efficiency energy router protection and control device can be realized by software, hardware, and combinations thereof, in whole or in part. The above modules can be embedded in or independent of the processor in the computer device in hardware form, or can be stored in the memory in the computer device in software form, so as to be called and executed by the processor to perform the operations corresponding to each of the above modules.

[0170] In an exemplary embodiment, a computer device, which can be a server, is provided, and its internal structure diagram can be as shown in Figure 11 The computer device includes a processor, a memory, an input / output interface (I / O), and a communication interface.

[0171] Those skilled in the art can understand that Figure 11It should be noted that the structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0172] In one embodiment, a computer device is also provided, including a memory and a processor, the memory storing a computer program, and the processor implementing the steps in the above method embodiments when executing the computer program.

[0173] In one embodiment, a computer readable storage medium is provided, storing a computer program, and the computer program is executed by a processor to implement the steps in the above method embodiments.

[0174] In one embodiment, a computer program product or computer program is provided, including computer instructions stored in a computer readable storage medium. The processor of the computer device reads the computer instructions from the computer readable storage medium, and the processor executes the computer instructions to make the computer device execute the steps in the above method embodiments.

[0175] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant regulations.

[0176] Those of ordinary skill in the art can understand that all or part of the processes in the above method embodiments can be completed by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer readable storage medium, and when executed, can include the processes of the above method embodiments.

[0177] The technical features of the above embodiments can be combined in any way. To make the description concise, not all possible combinations of the technical features in the above embodiments are described, but as long as the combinations of the technical features do not exist contradictions, they should be considered as the scope of the present disclosure.

[0178] The above-described embodiments are merely illustrative of several embodiments of the present application, which are described in more detail and in a specific manner, but should not be construed as limiting the scope of the patent of the present application. It should be noted that, for those of ordinary skill in the art, several modifications and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A high-efficiency energy router protection and control method based on ANPC topology, characterized in that, The method includes: The system acquires operating condition data, constraint parameter data, and midpoint voltage ripple data corresponding to the rail transit system; the energy router of the rail transit system adopts an ANPC topology. Based on the operating condition data, the constraint parameter data, and the midpoint voltage ripple data, zero-charge zero-momentum encoding is performed on the space vector modulation output data under the ANPC topology to obtain three-level self-conserving transmission sequence data and time-stamped buffer index data. The zero-charge zero-momentum encoding is a constraint encoding method that performs online shaping on the reference switching sequence to make the DC net charge injection and equivalent momentum simultaneously zero within any short window. The three-level self-conserving transmission sequence data is a detection sequence composed of positive half, zero, and negative half level symbols after encoding without changing the average power transmission. The time-stamped buffer index data is a metadata index that marks the start and end times of each detection window, phase group and codeword identifier, buffered channels and lengths, and quality and verification information. The original power echo data corresponding to the time-stamped cache index data is reconstructed using congruent interleaving to obtain equivalent oversampled echo matrix data. The congruent interleaving reconstruction is a reconstruction method that uses coprime phase interleaving sampling and index registration to restore folded multi-phase samples into a unified time-based high equivalent sampling sequence. The equivalent oversampled echo matrix data is a high sampling rate echo two-dimensional dataset and its quality metric organized by channel multiplication by time after reconstruction. Conformal scattering tomography is performed on the equivalent oversampled echo matrix data to obtain range branch reflection distribution data and initial location coordinate data. The conformal scattering tomography is a sparse inversion of the echo under conformal mapping and physical feasibility constraints to obtain the imaging process of reflection and impedance changes along the range and branch. The range branch reflection distribution data is a discretized spectrum data expressing the reflection intensity, phase, and equivalent impedance increment of each branch along the range position. The initial location coordinate data is a preliminary estimate of the kilometer markers, branch markers, and uncertainties of suspected anomalies derived from the imaging results and the geometric mapping of the line. Persistent homology pattern analysis is performed on the distance branch reflection distribution data to obtain anomalous morphology data; the persistent homology pattern analysis is a pattern analysis process based on the construction and stabilization of time series topological features to classify and evaluate the persistence and morphology of reflection structures; the anomalous morphology data is a comprehensive output of the probability, evolution rate, intensity level and evidence confidence of types such as arcing, grounding, dampness, and open circuit. Based on the initial positioning coordinate data and the abnormal morphology data, zero-regret game planning is performed on the energy routing linkage strategy of the rail transit system to obtain location-based action protection data; the location-based action protection data is used to control the energy router; the zero-regret game planning is a robust decision-making method that optimizes the linkage strategy under uncertainty and worst-case assumptions and ensures that long-term losses are not inferior to the optimal benchmark.

2. The method according to claim 1, characterized in that, The process of performing congruential interleaving reconstruction on the original power echo data corresponding to the time-stamped cache index data to obtain equivalent oversampled echo matrix data includes: Based on the time-stamped cache index data, the raw power echo data is subjected to grid nearest point congruent registration decoding to obtain congruent registration sample sequence data; the congruent registration sample sequence data is a sample sequence that has been reordered by channel-unified time index-phase label after registration decoding, and is accompanied by quality markers such as coverage and erase bits and a check summary; Based on the three-level self-conserving transmission sequence data, sparse congruent de-overlap reconstruction is performed on the congruent registration sample sequence data to obtain the equivalent oversampled echo matrix data; the sparse congruent de-overlap reconstruction is a method that uses the congruent registration sample as the observation and the three-level self-conserving transmission sequence as the prior, and recovers the unified time-base high equivalent sampled echo matrix by unfolding the folded time slots and performing sparse refinement under physical and time constraints.

3. The method according to claim 2, characterized in that, The step of performing sparse congruent de-overlap reconstruction on the congruent registration sample sequence data based on the three-level self-conserving emission sequence data to obtain equivalent oversampled echo matrix data includes: A joint data-based cyclic boosting autocorrelation model is performed on the three-level self-conserving emission sequence data and the congruent registration sample sequence data to obtain autocorrelation baseline data and congruent sparse dictionary data. Number-theoretic mask sparse de-overlap is performed on the autocorrelation baseline data and the congruent sparse dictionary data to obtain the folded frequency band sparse spectrum. The equivalent oversampled echo matrix data is obtained by performing inverse coherent phase continuous reconstruction on the folded frequency band sparse spectrum.

4. The method according to claim 3, characterized in that, The step of performing number-theoretic mask sparse de-overlap on the autocorrelation baseline data and the congruent sparse dictionary data to obtain the folded frequency band sparse spectrum includes: The number theory mask sequence data is obtained by deriving primitive root seeds and performing congruential mapping on the autocorrelation baseline data and the congruential sparse dictionary data. The number-theoretic mask sequence data and the congruent registration sample sequence data are subjected to primitive root sequence joint shaping to obtain mask domain measurement matrix data and mask domain observation vector data; Non-convex sparse de-stacking is performed on the mask domain measurement matrix data and the mask domain observation vector data to obtain the initial folded frequency band sparse spectrum; The initial folded band sparse spectrum and the mask domain measurement matrix data are subjected to split-consistent projection to obtain the folded band sparse spectrum.

5. The method according to claim 1, characterized in that, The equivalent oversampled echo matrix data is subjected to conformal scattering tomography to obtain range branch reflection distribution data and initial positioning coordinate data, including: A joint conformal parameterization mapping is performed on the equivalent oversampled echo matrix data and the line geometric topology template data in the constraint parameter data to obtain the conformal mapping domain observation vector and the conformal forward operator data. Under the constraints of the conformal forward operator data, barrier certificate-constrained sparse tomography is performed on the conformal mapping domain observation vector to obtain the distance branch reflection distribution data and the initial value data of the positioning coordinates.

6. The method according to claim 5, characterized in that, Under the constraints of the conformal forward operator data, barrier certificate-constrained sparse tomography is performed on the conformal mapping domain observation vector to obtain the range branch reflection distribution data and the initial positioning coordinate data, including: The conformal mapping domain observation vector data and the conformal forward operator data are subjected to positive real-dissipative barrier function transformation to obtain barrier domain objective function data and feasible set operator data. Sparse tomography is performed on the objective function data of the barrier domain and the feasible set operator data to obtain the initial range branch sparse reflection data. The initial distance branch sparse reflection data are reconstructed by semi-positive spectral domain projection to obtain the distance branch reflection distribution data and the initial positioning coordinate data.

7. The method according to claim 1, characterized in that, The persistent homology analysis of the distance branch reflection distribution data yields anomalous morphological data, including: Construct the temporal zigzag persistent coherence of the distance branch reflection distribution data and the time-stamped cache index data to obtain persistent barcode data; Centroid stabilization analysis is performed on the persistent barcode data to obtain stable topological signature data; The stable topological signature data is subjected to minimum description length causal typing analysis to obtain the abnormal morphological data.

8. The method according to claim 7, characterized in that, The centroid stabilization analysis of the persistent barcode data to obtain stable topological signature data includes: The persistent barcode data and the time-stamped cache index data are time-registered and normalized to obtain registered barcode array data; Multi-scale aggregation and integration are performed on the registered barcode array data to obtain multi-scale barcode cluster data; The initial topological centroid data is obtained by weighted centroid synthesis based on the multi-scale barcode cluster data; Based on the initial topology centroid data, consistent backfeeding and stabilization are performed to obtain stable topology signature data.

9. A high-efficiency energy router protection and control device based on ANPC topology, characterized in that, The device includes: The system data acquisition module is used to acquire the operating condition data, constraint parameter data, and midpoint voltage ripple data of the rail transit system; the energy router of the rail transit system adopts an ANPC topology. The system data encoding module is used to perform zero-charge and zero-momentum encoding on the space vector modulation output data under the ANPC topology based on the operating condition data, the constraint parameter data, and the midpoint voltage ripple data, to obtain three-level self-conserving transmission sequence data and time-stamped buffer index data. The zero-charge and zero-momentum encoding is a constraint encoding method that performs online shaping on the reference switching sequence so that the DC net charge injection and equivalent momentum are simultaneously zero within any short window. The three-level self-conserving transmission sequence data is a detection sequence composed of positive half, zero, and negative half level symbols after encoding, without changing the average power transmission. The time-stamped buffer index data is a metadata index that marks the start and end times of each detection window, phase group and codeword identifier, buffered channels and lengths, and quality and verification information. The echo data reconstruction module is used to perform congruent interleaving reconstruction on the original power echo data corresponding to the time-stamped cache index data to obtain equivalent oversampled echo matrix data. The congruent interleaving reconstruction is a reconstruction method that uses coprime phase interleaving sampling and index registration to restore folded multi-phase samples into a unified time-based high equivalent sampling sequence. The equivalent oversampled echo matrix data is a high sampling rate echo two-dimensional dataset and its quality metric organized by channel multiplication by time after reconstruction. The echo data tomography module is used to perform conformal scattering tomography on the equivalent oversampled echo matrix data to obtain distance branch reflection distribution data and initial location coordinate data. The conformal scattering tomography is a sparse inversion of the echo under conformal mapping and physical feasibility constraints to obtain the imaging process of reflection and impedance changes along the distance and branch. The distance branch reflection distribution data is a discretized spectrum data expressing the reflection intensity, phase, and equivalent impedance increment of each branch along the distance position. The initial location coordinate data is a preliminary estimate of the kilometer markers, branch markers, and their uncertainties of suspected anomalies derived from the imaging results and the geometric mapping of the line. The distribution data classification module is used to perform persistent coherence classification analysis on the distance branch reflection distribution data to obtain anomalous morphology data. The persistent coherence classification analysis is a classification process based on the construction and stabilization of time series topological features to classify and evaluate the persistence and morphology of reflection structures. The anomalous morphology data is a comprehensive output of the probability, evolution rate, intensity level, and evidence confidence of types such as arcing, grounding, dampness, and open circuit. The protection data planning module is used to perform zero-regret game planning on the energy routing linkage strategy of the rail transit system based on the initial value data of the positioning coordinates and the abnormal morphology data, so as to obtain location-based action protection data; the location-based action protection data is used to control the energy router; the zero-regret game planning is a robust decision-making method that optimizes the linkage strategy under uncertainty and worst-case assumptions and ensures that long-term losses are not inferior to the optimal benchmark.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 8.

Images