A highway power quality detection method and system based on a smart grid

Abstract

The application relates to the technical field of smart grids, and discloses a highway power quality detection method and system based on a smart grid, which comprises the following steps: acquiring power characteristic time sequence data of a to-be-detected power grid, and extracting a convergence state feature of an adaptive filtering module in a filtering process of original power time sequence data; identifying whether the adaptive filtering module is in a transient adjustment stage according to the fluctuation degree of the power characteristic time sequence data and the convergence state feature; generating a power quality evaluation index according to the identification result of the transient adjustment stage; and determining a control enabling state for the power grid load based on the power quality evaluation index, and performing a load adjustment operation according to the control enabling state. The application significantly improves the operation reliability and service continuity of a highway power supply network.

Description

Technical Field

[0001] This application relates to the field of smart grid technology, and more specifically, to a method and system for detecting power quality on highways based on smart grids. Background Technology

[0002] In practical applications such as highway service areas, existing technologies typically employ distributed sensor networks to collect voltage and current data, and then use adaptive filtering algorithms to process the data to remove steady-state background noise, thereby calculating power quality health indicators. When the grid load is relatively stable, the adaptive filter can converge to the optimal solution and output an accurate filtered signal.

[0003] However, when multiple high-power DC fast charging piles in the service area are activated simultaneously within a short period, the total harmonic distortion (THD) of the current in the distribution branch increases sharply, and the statistical characteristics of the input signal change drastically. At this time, the adaptive filtering module detects signal feature mismatch and inevitably initiates a new round of iterative convergence to adapt to the new noise environment. During the transition period of filter coefficient adjustment, the filter itself is in an unsteady state, and its output filtered signal not only fails to accurately represent the real power disturbance, but also adds ringing or overshoot distortion from the filter's own adjustment process. Existing conventional detection systems typically send the real-time output of this filter directly to the health index calculation module for short-term aggregation, which results in the distorted intermediate signal being directly calculated as a sharply rising, artificially high health index. Because the existing logic lacks a mechanism for sensing and isolating the internal convergence state of the algorithm, this artificially high index instantly exceeds the load adjustment threshold, causing the system to misjudge a serious power quality problem, and subsequently issue commands to reduce power or interrupt charging to the charging piles that are being activated, resulting in an erroneous interruption of the charging service.

[0004] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this application discloses a highway power quality detection method and system based on smart grids. It aims to solve the technical problem that when the concentrated activation of fast charging pile clusters in highway service areas causes drastic changes in the statistical characteristics of the power grid signal, the distorted signal output by the adaptive filtering algorithm during the iterative convergence transition period is incorrectly used to generate power quality evaluation indicators, leading to system misjudgment and erroneous triggering of load adjustment commands.

[0006] The technical solution of this application is as follows: In a first aspect, this application discloses a method for detecting power quality on highways based on a smart grid, comprising: Acquire the time-series data of the electrical energy characteristics of the power grid to be tested, and extract the convergence state characteristics of the adaptive filtering module during the filtering process of the original electrical time-series data; Based on the fluctuation level and convergence characteristics of the time series data of electrical energy features, identify whether the adaptive filtering module is in the transient adjustment stage; Based on the identification results of the transient adjustment phase, power quality evaluation indicators are generated; Based on power quality evaluation indicators, the control enable status for the grid load is determined, and load regulation operations are performed according to the control enable status.

[0007] This technical solution can effectively isolate transient distortion signals of the algorithm by sensing the internal convergence state of the adaptive filter, avoiding the overshoot or ringing during the filter iteration process that leads to artificially high power quality evaluation indicators, thereby preventing the false triggering of load regulation commands.

[0008] Optionally, the convergence state characteristics include the weight update magnitude and the error energy ratio; The steps for extracting the convergence state features of the adaptive filtering module include: Extract the weight vector of the adaptive filtering module at the current time and the weight vector at the previous time, and calculate the norm of the vector difference between the weight vector at the current time and the weight vector at the previous time as the weight update magnitude. The instantaneous error signal is obtained by subtracting the filtered output signal of the adaptive filtering module from the electrical energy characteristic time series data during the filtering process. The short-term error energy is calculated based on the instantaneous error signal of the adaptive filtering module within a preset first number of sampling points; The long-term error energy is calculated based on the instantaneous error signal of the adaptive filtering module within a preset second number of sampling points; The error energy ratio is obtained by calculating the ratio of short-term error energy to long-term error energy.

[0009] Optionally, the power characteristic time series data includes the effective value of the current; the steps to identify whether the adaptive filtering module is in the transient adjustment phase based on the fluctuation degree and convergence state characteristics of the power characteristic time series data include: First-order difference calculation is performed on the effective value of current in the time series data of electrical energy characteristics; If the absolute value of the first-order difference calculation result at the current moment exceeds the preset load change threshold, and the weight update magnitude exceeds the preset convergence critical threshold, then the adaptive filtering module is determined to enter the transient adjustment phase. or, If the error energy ratio at the current moment exceeds the preset upper limit threshold, the adaptive filtering module is determined to enter the transient adjustment phase. After the adaptive filtering module enters the transient adjustment phase, if the weight update amplitude falls back to below the convergence critical threshold, and the error energy ratio recovers to below the upper limit threshold, and continues for a preset anti-jitter time window, the transient adjustment phase is determined to end.

[0010] Optionally, the steps for generating power quality evaluation indicators based on the identification results of the transient adjustment phase include: If the identification result indicates that the system is not in the transient adjustment phase, the power quality evaluation index is calculated based on the real-time output signal obtained after filtering the power characteristic time series data by the adaptive filtering module. If the identification result indicates that the system is in a transient adjustment phase, the real-time output signal of the adaptive filtering module is smoothed, or the power characteristic time series data is called to generate a limited power quality evaluation index.

[0011] Optionally, the steps of determining the control enable state for the grid load based on power quality evaluation indicators and performing load regulation operations according to the control enable state include: If the identification result is not in the transient adjustment phase, and any power quality evaluation index exceeds the preset abnormal threshold, the control enable state will be configured to allow action. Otherwise, configure the control enable state to disable the action; Load regulation operations are performed based on the configured control enable status.

[0012] Optionally, power quality evaluation indicators include macro-level evaluation indicators and micro-level evaluation indicators; The steps for generating constrained power quality evaluation indicators by calling time-series data of power characteristics include: For macro-level evaluation indicators, the output of the bypass adaptive filtering module is used to calculate the minimum guarantee evaluation value based on the original sampled value of the power characteristic time series data, which serves as the macro-level evaluation indicator. For micro-evaluation indicators, the historical steady-state average of the total harmonic distortion rate of multiple sets of continuous currents before entering the transient adjustment stage is smoothed by calling the time series data of power characteristics, and the smoothed historical steady-state average is used as the micro-evaluation indicator.

[0013] Optionally, after determining the control enable state for the grid load, the method further includes: The deviation of power quality evaluation indicators from abnormal thresholds within a preset time period is statistically analyzed to form a deviation data sequence; Calculate the cumulative deviation of the overall indicator based on the deviation data sequence; The severity score of exceeding the limit is obtained by mapping the cumulative deviation value; By combining power quality evaluation indicators and whether the adaptive filtering module is in a transient adjustment phase, the abnormal sources of the power grid load are classified. If the adaptive filtering module is in the transient adjustment phase and the power characteristic time series data shows load change characteristics, the abnormal source is classified as algorithm transition fluctuation. If the adaptive filtering module is in the transient adjustment phase and the power characteristic time series data does not show load change characteristics, the abnormal source is classified as short-term data disturbance. If the adaptive filtering module is not in the transient adjustment phase, and the power quality evaluation index exceeds the abnormal threshold and the duration reaches the preset abnormal time window, the abnormal source is classified as a power grid abnormality. If the adaptive filtering module is not in the transient adjustment phase, and the power quality evaluation index exceeds the abnormal threshold but the duration does not reach the preset abnormal time window, the abnormal source is classified as instantaneous index fluctuation. Otherwise, it is determined that the power grid is operating normally and there is no source of abnormality.

[0014] Integrate the severity score of exceeding limits, the classification of anomaly sources, and the control enable status to generate a comprehensive judgment result vector and output it.

[0015] Optionally, the steps for performing load regulation operations based on the configured control enable state include: If the configured control enable state is disabled and the anomaly source is classified as algorithmic over-fluctuation, then load adjustment action will not be triggered. If the configured control enable state is allowed to operate and the abnormal source is classified as a power grid abnormality, then an adjustment command with a power limit target value and a power drop slope is issued to the load side of the power grid under test according to the severity score of the over-limit, so as to adjust the load power; If any power quality evaluation index still exceeds the abnormal threshold after the adjustment command is executed, or if a voltage drop in the grid under test is detected to exceed the preset range, then a power switching action will be executed.

[0016] Optionally, the method further includes: The duration of the transient adjustment phase is monitored. If the duration exceeds the preset time threshold, the restrictions on the power quality evaluation indicators are lifted, so that the power quality evaluation indicators follow the output trend of the adaptive filtering module. The system synchronously calculates the increment of the effective current value during the transient adjustment phase. If the increment of the effective current value exceeds the preset transformer capacity warning threshold, a capacity warning signal is triggered and the power quality evaluation index is provided.

[0017] Secondly, this application also discloses a highway power quality detection system based on a smart grid, used to perform the steps in the above method, including: The data acquisition module is used to acquire the time series data of the electrical energy characteristics of the power grid to be tested, and to extract the convergence state characteristics of the adaptive filtering module during the filtering process of the original electrical time series data. The state recognition module is used to identify whether the adaptive filtering module is in the transient adjustment stage based on the fluctuation degree and convergence state characteristics of the power characteristic time series data. The indicator generation module is used to generate power quality evaluation indicators based on the identification results of the transient adjustment phase. The load regulation module is used to determine the control enable status for the grid load based on power quality evaluation indicators, and to perform load regulation operations according to the control enable status.

[0018] In summary, this application provides a method and system for detecting highway power quality based on a smart grid. The method achieves precise perception of the algorithm's internal convergence state by embedding a state recognition and dynamic enabling mechanism between the adaptive filtering link and the health index calculation module. When the concentrated activation of fast-charging pile clusters in highway service areas causes drastic changes in the statistical characteristics of the power grid signal, it can accurately identify the re-convergence transition period of the adaptive filter and implement limited acceptance and baseline verification for index calculations during this period, effectively preventing the erroneous transmission of algorithm transient distortion to load adjustment commands. This solution not only ensures the system's real-time online evaluation capability under dynamic operating conditions but also achieves a shift from passive, blind response to proactive, precise defense through multi-dimensional anomaly source classification and graded response strategies, significantly improving the operational reliability and service continuity of the highway power supply network. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating a method for detecting highway power quality based on a smart grid, provided in an embodiment of this application.

[0020] Figure 2 This is a schematic diagram of a highway power quality detection system based on a smart grid, provided as an embodiment of this application.

[0021] Figure 3 A schematic diagram illustrating the process of generating power quality evaluation indicators provided in this application embodiment.

[0022] Labeling Explanation: 210, Data Acquisition Module; 220, Status Recognition Module; 230, Indicator Generation Module; 240, Load Adjustment Module. Detailed Implementation

[0023] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0024] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0025] In practical applications such as highway service areas, distributed sensor networks are typically used to collect voltage and current data, and adaptive filtering algorithms are employed to process the data and remove steady-state background noise, thereby calculating power quality health indicators. When the grid load is relatively stable, the adaptive filter can converge to the optimal solution and output an accurate filtered signal. However, when multiple high-power DC fast charging piles in the service area are activated simultaneously within a short period, the total harmonic distortion (THD) of the current in the distribution branch increases sharply, and the statistical characteristics of the input signal change drastically. At this time, the adaptive filtering module detects signal feature mismatch and inevitably initiates a new round of iterative convergence to adapt to the new noise environment. During the transition period of filter coefficient adjustment, the filter itself is in a non-steady state, and its output filtered signal not only fails to accurately represent the real power disturbance but also adds ringing or overshoot distortion from the filter's own adjustment process. If the real-time output of this filter is directly fed into the health index calculation module, the distorted intermediate signal will be directly calculated as a sharply rising, artificially high health index, leading the system to misjudge a serious power quality problem and incorrectly issue commands to reduce power or interrupt charging to the charging piles that are starting up. Therefore, how to prevent the distorted signal output by the adaptive filtering algorithm during the iterative convergence transition period from being used in real time to generate power quality health indices, thereby causing erroneous load adjustment commands, is the specific technical problem currently faced.

[0026] In this regard, firstly, referring to Figure 1 This application discloses a method for detecting power quality on highways based on smart grids, comprising: S1. Obtain the time series data of the electrical energy characteristics of the power grid to be detected, and extract the convergence state characteristics of the adaptive filtering module during the filtering process of the original electrical time series data. S2. Based on the fluctuation level and convergence characteristics of the power characteristic time series data, identify whether the adaptive filtering module is in the transient adjustment stage; S3. Based on the identification results of the transient adjustment phase, generate power quality evaluation indicators; S4. Based on power quality evaluation indicators, determine the control enable status for the power grid load, and perform load regulation operations according to the control enable status.

[0027] An adaptive filtering module is an algorithm module that can automatically adjust its filtering parameters (such as weight coefficients) based on the statistical characteristics of the input signal. It is commonly used to separate characteristic signals reflecting the true power quality from raw electrical signals containing complex background noise. More specifically, this module does not use a fixed set of parameters, but continuously corrects its internal weights based on the current input signal, reference signal, and output error within each sampling update cycle. Taking a low-voltage power distribution scenario in a service area as an example, when the load is stable at night, the sources of background harmonics are relatively fixed, and the adaptive filtering module gradually forms a stable set of weights to suppress long-term steady-state noise. However, when multiple DC fast charging piles are put into operation successively, the number, amplitude, and phase relationship of harmonic sources may change, and the module's internal weights will be re-migrated to re-establish an adaptation relationship to the new noise environment. Therefore, the adaptive filtering module is both a signal purification unit and a dynamic processing unit with an internal state evolution process.

[0028] Convergence state characteristics refer to feature parameters that reflect whether the internal algorithm iteration process of the adaptive filtering module has reached a stable state, and are used to characterize the reliability of the current output signal of the filter. These characteristics do not directly describe whether the power grid itself is abnormal, but rather whether the filter has stabilized. For example, when the weight changes are small over several consecutive update cycles and the error signal energy distribution tends to be stable, the filter can be considered close to convergence; conversely, when the weights change significantly and the short-term error energy is significantly higher than the long-term error energy, it indicates that the filter is still in the process of adapting to the new environment, and its output reliability is low.

[0029] The transient adjustment phase refers to the transitional period during which the adaptive filter module drastically adjusts its internal parameters to adapt to the new environment when faced with drastic changes in the statistical characteristics of the input signal. During this phase, the filter output is typically accompanied by distortion. Taking the centralized startup of a fast-charging pile cluster as an example, the external current waveform will undergo a step change and harmonic structure reconstruction within a short period, rendering the previously established noise model of the filter inapplicable and requiring readjustment. Even if there is no serious fault in the power grid itself during this phase, the filter output may exhibit abnormal spikes, overshoot, or oscillations for a short time.

[0030] Control enable status refers to a logical permission flag indicating whether the system is allowed to issue actual control actions (such as power reduction or disconnection) for grid loads (e.g., charging piles). This status is not equivalent to the final control command itself, but rather a logical gate located at the front end of the control link. Only when the control enable status allows action will the subsequent load regulation module convert the evaluation result into an actual control message; if the control enable status prohibits action, even if the evaluation index fluctuates at a certain moment, it will not directly trigger physical control, thus avoiding malfunctions.

[0031] To address the above steps, when acquiring the time-series data of the electrical characteristics of the power grid under test, raw voltage and current signals can be collected using distributed voltage and current transformers deployed on the low-voltage side of the main line of the distribution transformer in the service area and the branch circuits of each DC fast charging station. After analog-to-digital conversion, these signals can be organized into a fixed-length sliding time window queue.

[0032] Specifically, the following acquisition method can be adopted: A set of voltage and current transformers is installed at the low-voltage side busbar to reflect changes in total load. Branch current transformers are installed at the entrance of each fast-charging pile branch to identify the source of local load abrupt changes. The analog outputs from each sensor are isolated and conditioned before being sent to the edge acquisition terminal, which performs synchronous sampling according to a unified clock and adds a timestamp, measurement point number, and loop identifier to each set of sampled values. The resulting power characteristic time-series data includes not only the original instantaneous voltage and current values, but also further derived effective current values, effective voltage values, harmonic correlation statistics, and fluctuation characteristic values ​​within the window. A fixed-length sliding time window queue refers to continuously retaining the most recent continuous sampling data and removing the earliest sampling point when a new sampling point arrives, ensuring that subsequent analysis is always based on the latest time period data. In actual implementation, the edge acquisition terminal can establish a circular buffer in memory for each measurement point, storing data fragments from several recent basic analysis windows sequentially, so that the status identification module can call them at any time.

[0033] Simultaneously, it is necessary to extract the convergence state characteristics of the adaptive filtering module when processing these data. In one possible example, the convergence state characteristic can be the trend of the change in the absolute value of the instantaneous error, which can be achieved by calculating the moving average of the absolute values ​​of the instantaneous errors at multiple consecutive sampling points. The instantaneous error can be understood as the deviation between the current output of the filter and the target reference. If the system uses the original electrical timing data as input and the estimated background component as reference, the instantaneous error can be obtained from the difference between the filter output and the corresponding reference quantity. To avoid a single sampling spike directly affecting the judgment, the absolute value of the error at a single sampling point is not used directly. Instead, the absolute values ​​of the errors at the most recent consecutive sampling points are averaged to obtain a smoother error change trend. When the input signal undergoes a sudden change, this moving average will increase significantly, thus reflecting the change in the internal state of the filter.

[0034] Furthermore, to prevent misjudgments caused by sampling anomalies, communication jitter, or momentary sensor glitches, basic data cleaning can be performed before sending the raw voltage and current signals to the adaptive filtering module. This includes removing obviously out-of-bounds invalid sampling points, interpolating nearby sampling points for missing points, and rearranging data with misaligned timestamps. After these processing steps, the extraction results of convergence state features are more stable and can more realistically reflect the convergence process within the filter.

[0035] When identifying whether the adaptive filtering module is in a transient adjustment phase, it is necessary to make a comprehensive judgment by combining the fluctuation level of the power characteristic time series data and the aforementioned convergence state characteristics. Specifically, the fluctuation level can be characterized by calculating the variance or standard deviation of the power characteristic time series data within the current time window. Here, the fluctuation level describes whether the external power grid signal has changed significantly, while the convergence state characteristics describe whether the internal filter is readapting. These correspond to external physical changes and internal algorithm changes, respectively. Only by combining these two types of information can we avoid misjudgments caused by relying solely on one type of characteristic. For example, if there is only external fluctuation while the filter remains stable, it should not be easily concluded that it has entered the transient adjustment phase; conversely, if the internal error of the filter suddenly increases, but there is no significant load change in the external power grid, it is more likely to be a short-term interference or local anomaly, rather than a typical transition process caused by the concentrated start-up of fast charging piles. When the fluctuation level exceeds the set normal fluctuation range, and the convergence state characteristics (such as the moving average of the absolute value of the instantaneous error) also exceed the preset error threshold, it can be identified that the adaptive filtering module is currently in a transient adjustment phase, adjusting parameters to adapt to new signal characteristics.

[0036] The normal fluctuation range and error threshold can be pre-calibrated using historical operating data. One feasible calibration method is to continuously collect time-series data of electrical characteristics and corresponding convergence state characteristics during multiple periods of relatively stable load in the service area, without concentrated start-stop events. The upper bound of fluctuation and the upper bound of error under normal operating conditions are then statistically analyzed, and the boundary value higher than normal background fluctuation but lower than significantly abnormal fluctuation is set as the identification threshold. To adapt to different service area sizes, transformer capacities, and fast-charging pile configurations, the threshold does not need to be a uniform fixed value; instead, parameter files can be established for each site. Thus, during actual deployment, only the preset threshold configuration for the corresponding site needs to be called to complete the identification of the transient adjustment phase.

[0037] When generating power quality evaluation indicators, different generation strategies are adopted based on the identification results of the transient adjustment phase. Specifically, if the identification results indicate that the filter is not in the transient adjustment phase, it means that the filter output is reliable, and indicators such as voltage deviation and total harmonic distortion of current can be directly calculated using the filter's output signal. If the identification results indicate that the filter is in the transient adjustment phase, in order to avoid algorithm distortion leading to artificially high indicators, the update of the indicators can be paused, and the indicator values ​​of the previous moment can be directly output as the current power quality evaluation indicators.

[0038] To more clearly illustrate the implementation of this step, the index generation process can be understood as having two paths: a normal calculation branch and a restricted calculation branch. In the normal calculation branch, following the conventional power quality analysis process, the filtered stable output is called to calculate various evaluation indices, and the results are written to the index buffer at the current moment. In the restricted calculation branch, the current instantaneous output of the filter is no longer directly used as the basis for index calculation; instead, a conservative strategy is prioritized to avoid amplifying and propagating transient distortions. To ensure a smooth switching process, when the identification result switches from non-transient to transient, the last reliable index value before the switch is latched, and this latched value is continuously called throughout the transient period until the identification result returns to non-transient. This avoids abrupt changes in indices at the switching boundary.

[0039] When determining the control enable state and performing load regulation operations, decisions are made based on generated power quality evaluation indicators. In some implementations, the generated power quality evaluation indicators are compared with preset safety thresholds. If the indicators exceed the safety thresholds, the control enable state is configured to a triggered state, and an adjustment command to reduce the output power is issued to the corresponding fast charging pile; if the indicators are within the safety thresholds, normal operation is maintained.

[0040] Specifically, the generation of the control enable state can adopt a hierarchical decision-making approach. First, the indicator judgment unit checks each current power quality evaluation indicator to preliminarily determine whether any limits have been exceeded. Then, the control permission unit, combined with the identification results from the transient adjustment phase, performs a secondary screening of this preliminary judgment. Finally, the execution interface unit decides whether to send a control message to the load side based on the final control enable state. This setup allows the evaluation indicators to reflect whether the grid state warrants attention, while the control enable state determines whether to allow the result of attention to be translated into a physical action. The responsibilities of both are clearly defined, facilitating engineering implementation and subsequent maintenance.

[0041] Through the above technical solution, this application introduces a sensing mechanism for the convergence state of the adaptive filtering module into the traditional power quality detection link. Its working principle is that when the simultaneous activation of external fast-charging pile clusters causes a sudden change in the grid signal, it no longer blindly trusts the real-time output of the filter. Instead, it extracts convergence state characteristics and combines them with the degree of data fluctuation to determine whether the filter has entered a transient adjustment phase. Once a transient adjustment phase is identified, the generation method of the power quality evaluation index is adjusted according to the identification result, thereby cutting off the direct transmission of the filter's transient distortion signal to the health index. Determining the control enable state based on this state-constrained evaluation index effectively avoids artificially high indexes caused by iterative oscillations within the algorithm. This prevents the system from misjudging power quality anomalies, thus avoiding erroneous triggering of load adjustment commands to interrupt normal charging services and ensuring the continuous and reliable operation of highway charging facilities.

[0042] Optionally, a highway power quality detection method based on smart grids includes convergence state characteristics including weight update magnitude and error energy ratio; The steps for extracting the convergence state features of the adaptive filtering module include: Extract the weight vector of the adaptive filtering module at the current time and the weight vector at the previous time, and calculate the norm of the vector difference between the weight vector at the current time and the weight vector at the previous time as the weight update magnitude. The instantaneous error signal is obtained by subtracting the filtered output signal of the adaptive filtering module from the electrical energy characteristic time series data during the filtering process. The short-term error energy is calculated based on the instantaneous error signal of the adaptive filtering module within a preset first number of sampling points; The long-term error energy is calculated based on the instantaneous error signal of the adaptive filtering module within a preset second number of sampling points; The error energy ratio is obtained by calculating the ratio of short-term error energy to long-term error energy.

[0043] In this specific implementation, the weight update magnitude directly characterizes the degree of drastic adjustment of the filter's internal parameters. Specifically, the weight vector at the current moment is extracted from the weight vector at the previous moment. The corresponding elements of the two are subtracted to obtain a difference vector. Then, each element of the difference vector is squared, and all squared values ​​are summed. Finally, the square root (Euclidean norm) of the sum is taken as the weight update magnitude. Specifically, after each weight update by the adaptive filtering module, the current weight vector is written to the state buffer, and a copy of the weights from the previous update cycle is retained. The state recognition module reads these two sets of weights to obtain the current update magnitude. If a multi-channel filtering structure is used, the weight update magnitude of each channel can be calculated separately, and the maximum, average, or weighted result can be taken as a unified convergence state feature input.

[0044] Simultaneously, to reflect changes in the energy distribution of the error, short-term error energy and long-term error energy are calculated, and their ratio is obtained. The first sampling point reflects whether the error has suddenly increased within a recent short period, while the second sampling point reflects the normal level of the error over a longer period. To ensure a significant timescale difference, the second sampling point should be larger than the first. In practice, separate short-term and long-term error buffers can be established: the short-term buffer continuously stores the instantaneous error signal over a recent short period, while the long-term buffer continuously stores the instantaneous error signal over a recent longer period. The error signals in the two buffers are squared, accumulated, and averaged respectively to obtain the short-term and long-term error energy. If the short-term error energy is significantly higher than the long-term error energy, it indicates that the filter has recently experienced significant mismatch and is more likely in a transient adjustment phase.

[0045] The preset first and second sampling points can be obtained through site calibration. Specifically, under various operating conditions such as normal operation in the service area, single-pile start-up, dual-pile staggered start-up, and multi-pile centralized start-up, instantaneous error signals are continuously recorded. The impact of different time window lengths on the stability of the error energy ratio is compared. A short window length that can promptly reflect sudden changes and is not easily affected by single-pulse interference is selected as the first sampling point, and a longer window length that can represent the background error level is selected as the second sampling point. This set of window configurations is then written into the site parameter table. In this way, the calculation basis for the error energy ratio is not arbitrarily set, but rather derived from the calibration results of actual operating data.

[0046] In some preferred embodiments, a method for detecting highway power quality based on a smart grid includes power characteristic time-series data comprising RMS current values; the step of identifying whether an adaptive filtering module is in a transient adjustment phase based on the fluctuation degree and convergence characteristics of the power characteristic time-series data includes: First-order difference calculation is performed on the effective value of current in the time series data of electrical energy characteristics; If the absolute value of the first-order difference calculation result at the current moment exceeds the preset load change threshold, and the weight update magnitude exceeds the preset convergence critical threshold, then the adaptive filtering module is determined to enter the transient adjustment phase. or, If the error energy ratio at the current moment exceeds the preset upper limit threshold, the adaptive filtering module is determined to enter the transient adjustment phase. After the adaptive filtering module enters the transient adjustment phase, if the weight update amplitude falls back to below the convergence critical threshold, and the error energy ratio recovers to below the upper limit threshold, and continues for a preset anti-jitter time window, the transient adjustment phase is determined to end.

[0047] In this specific implementation, the root mean square (RMS) value of the original current signal within the current basic analysis window is extracted and subtracted from the RMS value of the original current signal within the previous basic analysis window to obtain the first-order difference value. When the absolute value of this difference value exceeds the preset physical threshold for load mutation and the weight update amplitude exceeds the convergence critical threshold, it indicates that a high-dynamic load mutation has occurred externally and the internal algorithm is undergoing drastic adjustments. At this point, it is determined that the system has entered the transient adjustment phase. Here, the basic analysis window can be understood as the smallest analysis unit used to update the effective current value. In actual deployment, it can be consistent with the power frequency analysis cycle of the sampling system or the window length preset in the site configuration file. To avoid the differential results from window boundary misalignment, the effective current value can be calculated by continuously splicing adjacent windows end to end, ensuring that each window uses a unified time base.

[0048] Furthermore, even if external abrupt changes are not obvious, if the error energy ratio exceeds the upper limit threshold, it indicates a severe mismatch within the algorithm, and the algorithm is also judged to enter the transient adjustment phase. The reason for this setting is that in some cases, load changes may not manifest as a significant step change in the total current. For example, multiple branches may cancel each other out, local harmonic source changes may be greater than fundamental frequency changes, or the sampling point may be located in a local branch rather than the main input line. In these cases, relying solely on the difference in the effective current value may miss detections, while the error energy ratio can still reflect the degree of mismatch within the filter.

[0049] Regarding the exit mechanism, the weight update amplitude must decrease and the error energy ratio must recover. This must continue for a full anti-jitter time window before the transient adjustment phase can be considered complete. This anti-jitter time window is used to absorb the slight back-and-forth fluctuations of the filter at the end of the re-establishment of steady-state convergence, preventing frequent switching between the initial recovery and subsequent short-term fluctuations. This anti-jitter time window can be obtained through calibration using historical operating data. Specifically, it can be calculated by statistically analyzing the duration of sustained stability required for the filter to recover from significant mismatch to stable output during multiple fast-charging station start-up and shutdown events, and selecting a time length that covers most of the recovery tail-end jitter as the site configuration parameter.

[0050] The preset load mutation threshold, preset convergence critical threshold, and preset upper limit threshold can be established as follows: First, collect multiple sets of known operating condition data in the target service area, including normal stable operation data, single pile start-up data, multi-pile staggered start-up data, multi-pile concentrated start-up data, and short-term disturbance data caused by non-load reasons. Then, calculate the first-order difference of the current effective value, the weight update amplitude, and the error energy ratio for these data respectively, and observe the distribution range of the three types of characteristics under various operating conditions. Finally, select the boundary value that can better distinguish between normal fluctuations and transient adjustments, and write it into the site threshold configuration table. In this way, the thresholds are derived from actual operating condition statistics, rather than arbitrary settings detached from the scenario.

[0051] Optionally, a method for detecting power quality on highways based on smart grids includes the following steps for generating power quality evaluation indicators based on the identification results during the transient adjustment phase: If the identification result indicates that the system is not in the transient adjustment phase, the power quality evaluation index is calculated based on the real-time output signal obtained after filtering the power characteristic time series data by the adaptive filtering module. If the identification result indicates that the system is in a transient adjustment phase, the real-time output signal of the adaptive filtering module is smoothed, or the power characteristic time series data is called to generate a limited power quality evaluation index.

[0052] In this specific implementation scheme, when not in the transient adjustment phase, it indicates that the adaptive filtering module has converged, and its output signal has high reliability. At this time, the adaptive filtering module is directly used to calculate power quality evaluation indicators for the real-time output signal after filtering the power characteristic time-series data. Specifically, the filtered clean real-time signal is first used as the calculation data source. This signal has been filtered to remove steady-state background noise and can accurately represent the electrical operating state of the power grid. Then, basic electrical parameters such as voltage RMS value, current RMS value, power, and power factor are calculated sequentially according to a fixed data window. Subsequently, various power quality sub-indicators such as voltage deviation, frequency deviation, three-phase unbalance, current total harmonic distortion rate, voltage fluctuation, and flicker are solved by combining the basic electrical parameters. Finally, the various indicators are integrated and output according to preset specifications to obtain a complete power quality evaluation result for subsequent threshold judgment, operating status identification, and data reporting.

[0053] During the transient adjustment phase, the instantaneous distorted output of the filter is no longer used directly; instead, a restricted processing branch is initiated. Specifically, the real-time output signal of the adaptive filtering module can be smoothed, for example, by using a moving average filtering algorithm to reduce overshoot spikes; or the filter can be bypassed directly, and the original power characteristic time-series data can be used to generate restricted power quality evaluation indicators. This smoothing is not intended to restore the filter's true output, but rather to suppress high-frequency spurious fluctuations introduced by the algorithm itself during transients, keeping indicator changes within a controllable range. To avoid the smoothing itself masking serious faults, the smoothed result can be compared in parallel with the original power characteristic time-series data: if both show abnormal trends, the anomaly is more likely to originate from the power grid itself; if only the filter output is abnormal while the original power characteristic time-series data shows no corresponding change, it is more likely due to excessive algorithm distortion. The method for calculating power quality evaluation indicators using the smoothed real-time output signal is described in the previous section on calculating power quality evaluation indicators using real-time output signals.

[0054] In some preferred embodiments, a method for detecting power quality on highways based on a smart grid includes the following steps: determining the control enable state for the grid load based on power quality evaluation indicators, and performing load regulation operations according to the control enable state: If the identification result is not in the transient adjustment phase, and any power quality evaluation index exceeds the preset abnormal threshold, the control enable state will be configured to allow action. Otherwise, configure the control enable state to disable the action; Load regulation operations are performed based on the configured control enable status.

[0055] In this specific implementation, the configuration of the control enable state is strictly tied to the identification results of the transient adjustment phase. The control enable state is only configured to allow action when the identification result clearly indicates that the system is not in the transient adjustment phase (i.e., the filter output is completely reliable) and the power quality evaluation index does indeed exceed the abnormal threshold. In any other case (including when the system is in the transient adjustment phase or the index does not exceed the threshold), the control enable state is forcibly configured to prohibit action. Subsequently, the response execution unit will only actually issue load adjustment commands upon receiving an enable state that allows action.

[0056] Preset anomaly thresholds can be established using a sub-index calibration method. Specifically, historical operating data and maintenance records are collected for voltage deviation, harmonic, and current fluctuation indicators, respectively. Fluctuation ranges confirmed not to affect equipment safety and business continuity are designated as normal ranges, while ranges requiring intervention are designated as abnormal ranges. A suitable boundary value between these two ranges is then selected as the anomaly threshold. For distribution systems of different service areas and capacity levels, corresponding threshold configuration files can be established and loaded per site during system deployment. This ensures that the triggering criteria for control enable states match the actual operating characteristics of the sites, reducing errors caused by applying fixed thresholds across different scenarios.

[0057] Optionally, a method for detecting highway power quality based on smart grids, wherein power quality evaluation indicators include macro-level evaluation indicators and micro-level evaluation indicators; The steps for generating constrained power quality evaluation indicators by calling time-series data of power characteristics include: For macro-level evaluation indicators, the output of the bypass adaptive filtering module is used to calculate the minimum guarantee evaluation value based on the original sampled value of the power characteristic time series data, which serves as the macro-level evaluation indicator. For micro-evaluation indicators, the historical steady-state average of the total harmonic distortion rate of multiple sets of continuous currents before entering the transient adjustment stage is smoothed by calling the time series data of power characteristics, and the smoothed historical steady-state average is used as the micro-evaluation indicator.

[0058] In this specific implementation scheme, for macroscopic evaluation indicators such as voltage deviation, the adaptive filter can be temporarily bypassed to extract the raw voltage data. A baseline assessment value is then calculated within a preset sliding window, serving as the macroscopic evaluation indicator during the transition period. This baseline assessment value can be the effective value, mean, or other statistical quantity that stably reflects the overall power supply status, formed by the raw sampled values ​​within the window. The baseline assessment value is calculated by collecting all raw voltage sampling points within the window according to a set sampling period, sequentially performing calculations for the effective value, arithmetic mean, extreme values, and other conventional statistical quantities to obtain the baseline assessment value. Since macroscopic indicators are mainly used to identify large-scale changes in power supply status such as voltage dips and sustained high or low voltage levels, they can still maintain good usability even without relying on high-precision filtering separation.

[0059] For micro-level evaluation indicators such as total harmonic distortion (THD) in electrical energy characteristic time-series data, multiple historical steady-state averages of THD over a period before entering the transient adjustment phase can be extracted, smoothed, and used as the current micro-level evaluation indicator. These multiple historical steady-state averages can be understood as several continuously stored micro-level indicator results before entering the transient phase, when the system was confirmed to be in a stable convergence state. The purpose of smoothing is not simply to replicate historical values, but to apply constrained variation rules to the historical steady-state averages, ensuring the micro-level indicator maintains a continuous state during the transient period. For example, the variation amplitude between adjacent output results can be limited to an allowable boundary determined by the historical natural fluctuation range, thereby preventing sudden spikes in the micro-level indicator that do not conform to physical laws during the transient period.

[0060] To ensure a clear source for historical steady-state means, each time the identification result indicates the data is not in a transient adjustment phase, the currently calculated micro-evaluation index is written to the steady-state history cache, and a steady-state label is attached to each data set. When the identification result switches to a transient adjustment phase, the index generation module selects only the most recent consecutive sets from the historical data with steady-state labels as input for smoothing, without using data generated during the transient period. This ensures that the constrained micro-evaluation index is based on a reliable historical foundation.

[0061] In some preferred embodiments, a method for detecting power quality on highways based on a smart grid, after determining the control enable state for the grid load, further includes: The deviation of power quality evaluation indicators from abnormal thresholds within a preset time period is statistically analyzed to form a deviation data sequence; Calculate the cumulative deviation of the overall indicator based on the deviation data sequence; The severity score of exceeding the limit is obtained by mapping the cumulative deviation value; By combining power quality evaluation indicators and whether the adaptive filtering module is in a transient adjustment phase, the abnormal sources of the power grid load are classified. If the adaptive filtering module is in the transient adjustment phase and the power characteristic time series data shows load change characteristics, the abnormal source is classified as algorithm transition fluctuation. If the adaptive filtering module is in the transient adjustment phase and the power characteristic time series data does not show load change characteristics, the abnormal source is classified as short-term data disturbance. If the adaptive filtering module is not in the transient adjustment phase, and the power quality evaluation index exceeds the abnormal threshold and the duration reaches the preset abnormal time window, the abnormal source is classified as a power grid abnormality. If the adaptive filtering module is not in the transient adjustment phase, and the power quality evaluation index exceeds the abnormal threshold but the duration does not reach the preset abnormal time window, the abnormal source is classified as instantaneous index fluctuation. Otherwise, it is determined that the power grid is operating normally and there is no source of abnormality.

[0062] Integrate the severity score of exceeding limits, the classification of anomaly sources, and the control enable status to generate a comprehensive judgment result vector and output it.

[0063] In this specific implementation plan, the deviation of each evaluation indicator from the abnormal threshold within a preset time period is first statistically analyzed to form a deviation data sequence. This deviation can be understood as the degree to which the current indicator exceeds the threshold. If the indicator does not exceed the threshold at a certain moment, the deviation at that moment can be recorded as zero or not included in the accumulation. Subsequently, the deviation data sequence is accumulated chronologically to obtain the overall cumulative deviation value of the indicator. This cumulative value not only reflects whether the limit has been exceeded, but also reflects how long the exceedance lasted and by how much. Therefore, it better reflects the overall severity of the anomaly than the limit-exceeding judgment at a single moment.

[0064] When mapping the severity of exceedances to cumulative deviation values, a scoring mapping table can be pre-established. This table can be created by: collecting historical operational events and labeling them as minor fluctuations, general anomalies, significant anomalies, and severe anomalies; then calculating the cumulative deviation value ranges corresponding to each level of event; finally, establishing a correspondence between these ranges and the scoring ranges to form a site-callable mapping configuration. For example, a lower cumulative deviation value range can be mapped to a lower score, and a higher range to a higher score. This provides a clear data source for the scores, facilitating subsequent control modules to adopt different response strategies based on the severity level.

[0065] Subsequently, fine-grained anomaly source classification is performed: when in a transient state with load abrupt changes, it is classified as algorithmic transient fluctuation; when in a transient state without load abrupt changes, it is classified as short-term data disturbance; when not in a transient state but the indicator continuously exceeds the limit, it is classified as a real power grid anomaly; when not in a transient state but the limit exceedance time is extremely short, it is classified as instantaneous indicator fluctuation. The preset anomaly time window here can also be obtained by historical event calibration, that is, statistically analyzing the shortest verifiable duration of real power grid anomalies, and using it as the time boundary to distinguish between continuous anomalies and instantaneous fluctuations. Finally, the scoring, classification, and enabling status are packaged to generate a comprehensive judgment result vector. This result vector can be output in a structured data format, such as including scoring fields, classification fields, enabling fields, timestamp fields, and measurement point fields, so that downstream modules can directly parse and call it.

[0066] Optionally, a method for detecting power quality on highways based on a smart grid includes the following steps for performing load regulation operations according to a configured control enable state: If the configured control enable state is disabled and the anomaly source is classified as algorithmic over-fluctuation, then load adjustment action will not be triggered. If the configured control enable state is allowed to operate and the abnormal source is classified as a power grid abnormality, then an adjustment command with a power limit target value and a power drop slope is issued to the load side of the power grid under test according to the severity score of the over-limit, so as to adjust the load power; If any power quality evaluation index still exceeds the abnormal threshold after the adjustment command is executed, or if a voltage drop in the grid under test is detected to exceed the preset range, then a power switching action will be executed.

[0067] In this specific implementation plan, when a prohibited action is received and it is due to algorithmic fluctuations, a temporary observation period is initiated to avoid interfering with the charging pile's startup. When a permitted action is received and it indicates a grid anomaly, a single disconnect message is not directly issued. Instead, an adjustment command with a power limit target value and a power decrease slope is issued based on a score indicating the severity of the exceedance. Here, the power limit target value can be understood as the target output level that the load side should gradually converge to, and the power decrease slope can be understood as the allowed rate of change from the current power to the target power. A higher score indicates a more severe anomaly, allowing for a lower target power and a faster decrease process; a lower score allows for a more lenient power limiting strategy. To enable this step, a score-adjustment strategy mapping table can be pre-established. For example, corresponding target power levels and decrease slope levels can be configured for different score ranges, and adjustment commands can be generated directly from the table during runtime.

[0068] The adjustment command can be issued in stages, meaning the control module doesn't reduce the load to the target value all at once, but rather sends new power limiting commands gradually according to a preset adjustment cycle, allowing the charging pile output to decrease smoothly. This reduces the impact of sudden current changes on the power distribution system. If any power quality evaluation indicator still exceeds the abnormal threshold after the adjustment command is executed, or if a voltage drop in the monitored grid exceeds a preset range, a power switching action is performed. This preset range can be calibrated using site power supply safety requirements and historical fault records, serving as a boundary condition for identifying serious power supply risks. The power switching action can be completed by a switching execution unit connected to critical loads, such as switching critical facilities like traffic monitoring, toll communication, or emergency lighting to a backup power supply path to ensure continuous operation of core services.

[0069] In some preferred embodiments, a method for detecting power quality on highways based on a smart grid further includes: The duration of the transient adjustment phase is monitored. If the duration exceeds the preset time threshold, the restrictions on the power quality evaluation indicators are lifted, so that the power quality evaluation indicators follow the output trend of the adaptive filtering module. The system synchronously calculates the increment of the effective current value during the transient adjustment phase. If the increment of the effective current value exceeds the preset transformer capacity warning threshold, a capacity warning signal is triggered and the power quality evaluation index is provided.

[0070] In this specific implementation scheme, when the state identification module first determines that the transient adjustment phase has begun, it starts timing the duration. If the exit condition is not met in subsequent continuous detections, the transient duration is continuously accumulated. When this duration exceeds the set maximum tolerable time threshold, it is determined that the current operating condition is no longer suitable for long-term use of the restricted calculation branch, because it is now more necessary to restore sensitivity to changes in the actual power grid state. Therefore, the completely frozen or strongly restricted indicator handling method is forcibly exited, and the restrictions are gradually lifted, allowing the indicators to follow the output trend of the filter. This gradual lifting of restrictions can adopt a phased recovery strategy, for example, first restoring the real-time update of macroscopic indicators, then restoring the trend tracking of some microscopic indicators, and finally restoring the normal calculation of all indicators after confirming that the filter is basically stable. This can avoid the secondary jump caused by suddenly switching from a completely restricted state to a completely open state.

[0071] When simultaneously calculating the increment of the effective current value during the transient adjustment phase, the effective current value at the moment of entering the transient state can be used as the reference value, and the effective current value of each analysis window during the transient period can be continuously recorded. The current value is then compared with the reference value to obtain the cumulative increment. If this increment exceeds a preset transformer capacity warning threshold, a capacity warning signal is triggered, and the power quality evaluation index is provided. This capacity warning threshold can be preset based on the rated capacity of the site's distribution transformer, its normal operating margin, and maintenance strategies, or it can be calibrated using historical high-load operating data to form a site-specific warning configuration. In this way, even if the filter has not fully converged, the risk of approaching the capacity limit can be identified in a timely manner based on changes in the original physical quantities.

[0072] Secondly, referring to Figure 2 This application also provides a smart grid-based highway power quality detection system for performing the steps in the above method, including: The data acquisition module 210 is used to acquire the time series data of the electrical energy characteristics of the power grid to be detected, and to extract the convergence state characteristics of the adaptive filtering module during the filtering process of the original electrical time series data. The state recognition module 220 is used to identify whether the adaptive filtering module is in the transient adjustment stage based on the fluctuation degree and convergence state characteristics of the power characteristic time series data. The indicator generation module 230 is used to generate power quality evaluation indicators based on the identification results of the transient adjustment phase. The load regulation module 240 is used to determine the control enable state for the grid load based on power quality evaluation indicators, and to perform load regulation operations according to the control enable state.

[0073] In this specific implementation scheme, the system is divided into four core functional modules. The data acquisition module 210 can consist of an edge computing terminal and a high-speed data acquisition card deployed on-site, responsible for interfacing with the distributed sensor network, buffering time-series data in real time, and extracting the internal state of the filter. To ensure that data from different measurement points can be used for joint analysis, the data acquisition module 210 can also include a unified clock synchronization unit, a measurement point identification management unit, and a cache management unit. The unified clock synchronization unit is used to ensure that the sampling data of the main line and each branch is aligned in time, the measurement point identification management unit is used to distinguish the sources of different loops, and the cache management unit is used to maintain the original sampling cache area and the steady-state historical cache area.

[0074] The state recognition module 220, as the core of the logic judgment, incorporates a mutation feature analysis and state machine switching algorithm, and outputs the current time window marker in real time. More specifically, this module can include an external fluctuation analysis subunit, an internal convergence analysis subunit, and a state machine management subunit: the external fluctuation analysis subunit is responsible for calculating external features such as the difference, variance, or standard deviation of the current effective value; the internal convergence analysis subunit is responsible for calculating internal features such as the weight update amplitude and error energy ratio; and the state machine management subunit outputs state markers such as steady state and transient adjustment stages based on the entry conditions, hold conditions, and exit conditions.

[0075] The index generation module 230 contains multiple calculation engines, capable of dynamically switching between normal and restricted calculation branches based on the input state flags. This module may further include a normal index calculation subunit, a restricted index generation subunit, and an index caching subunit. The normal index calculation subunit is responsible for calculating complete evaluation indices when the filter is stable. The restricted index generation subunit is responsible for generating restricted indices during transient periods by calling smoothing results or raw electrical energy characteristic time-series data. The index caching subunit is responsible for storing the current index, the previous time-stressed reliable index, and the historical steady-state average.

[0076] The load regulation module 240 integrates a communication protocol stack, responsible for converting the final decision result into specific control messages and sending them to the charging pile main control board or power switching execution unit. This module may include a control enable determination subunit, an anomaly classification subunit, a regulation strategy generation subunit, and an execution interface subunit. The regulation strategy generation subunit generates power limit target values ​​and power reduction slopes based on the severity of exceeding limits, while the execution interface subunit sends control commands according to the communication methods supported by the field equipment.

[0077] The above are merely embodiments of this application and are not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for detecting highway power quality based on a smart grid, characterized in that, The method includes: Acquire the time-series data of the electrical energy characteristics of the power grid to be detected, and extract the convergence state characteristics of the adaptive filtering module during the filtering process of the original electrical time-series data; Based on the fluctuation level of the electrical energy characteristic time series data and the convergence state characteristics, identify whether the adaptive filtering module is in a transient adjustment phase; Based on the identification results of the transient adjustment phase, power quality evaluation indicators are generated; Based on the power quality evaluation indicators, the control enable state for the power grid load is determined, and load regulation operations are performed according to the control enable state.

2. The method for detecting highway power quality based on a smart grid according to claim 1, characterized in that, The convergence state characteristics include the weight update magnitude and the error energy ratio; The step of extracting the convergence state features of the adaptive filtering module includes: Extract the weight vector of the adaptive filtering module at the current time and the weight vector at the previous time, and calculate the norm of the vector difference between the weight vector at the current time and the weight vector at the previous time as the weight update magnitude; The instantaneous error signal is obtained by subtracting the filtered output signal of the adaptive filtering module during the filtering process from the electrical energy characteristic time series data; The short-term error energy is calculated based on the instantaneous error signal of the adaptive filtering module within a preset first number of sampling points; The long-term error energy is calculated based on the instantaneous error signal of the adaptive filtering module within a preset second number of sampling points; The error energy ratio is obtained by calculating the ratio of the short-term error energy to the long-term error energy.

3. The method for detecting highway power quality based on a smart grid according to claim 2, characterized in that, The power characteristic time-series data includes the effective value of the current; the step of identifying whether the adaptive filtering module is in the transient adjustment phase based on the fluctuation degree of the power characteristic time-series data and the convergence state characteristics includes: Perform first-order difference calculation on the effective current value in the electrical energy characteristic time series data; If the absolute value of the first-order difference calculation result at the current moment exceeds the preset load mutation threshold, and the weight update magnitude exceeds the preset convergence critical threshold, then the adaptive filtering module is determined to enter the transient adjustment stage. or, If the error energy ratio at the current moment exceeds the preset upper limit threshold, then the adaptive filtering module is determined to enter the transient adjustment phase. After determining that the adaptive filtering module has entered the transient adjustment phase, if the weight update magnitude falls back to below the convergence critical threshold and the error energy ratio recovers to below the upper limit threshold, and continues for a preset anti-jitter time window, the transient adjustment phase is determined to end.

4. The method for detecting highway power quality based on a smart grid according to claim 1, characterized in that, The step of generating power quality evaluation indicators based on the identification results of the transient adjustment phase includes: If the identification result indicates that the system is not in the transient adjustment phase, the power quality evaluation index is calculated based on the real-time output signal obtained after filtering the power characteristic time series data by the adaptive filtering module. If the identification result indicates that the system is in the transient adjustment phase, the real-time output signal of the adaptive filtering module is smoothed, or the power characteristic time series data is called to generate a limited power quality evaluation index.

5. The method for detecting highway power quality based on a smart grid according to claim 1, characterized in that, The steps of determining the control enable state for the grid load based on the power quality evaluation index, and performing load regulation operations according to the control enable state, include: If the identification result is not in the transient adjustment phase, and any of the power quality evaluation indicators exceeds the preset abnormal threshold, then the control enable state is configured to allow action. Otherwise, configure the control enable state to disable the action; Load regulation operations are performed based on the configured control enable status.

6. The method for detecting highway power quality based on a smart grid according to claim 4, characterized in that, The power quality evaluation indicators include macro-level evaluation indicators and micro-level evaluation indicators; The step of calling the power characteristic time-series data to generate the limited power quality evaluation index includes: For the aforementioned macro-evaluation index, the output of the adaptive filtering module is bypassed, and a minimum guarantee evaluation value is calculated based on the original sampled value of the power characteristic time series data, which serves as the macro-evaluation index. For the micro-evaluation index, the power characteristic time series data is called, and the historical steady-state average of the total harmonic distortion rate of multiple sets of continuous currents before entering the transient adjustment stage is smoothed, and the smoothed historical steady-state average is used as the micro-evaluation index.

7. The method for detecting highway power quality based on a smart grid according to claim 5, characterized in that, Following the step of determining the control enable state for the grid load, the method further includes: The deviation of the power quality evaluation index from the abnormal threshold within a preset time period is statistically analyzed to form a deviation data sequence; Based on the deviation data sequence, calculate the overall cumulative deviation value of the indicator; The severity score of exceeding the limit is obtained based on the cumulative deviation value mapping. By combining the power quality evaluation indicators and whether the adaptive filtering module is in a transient adjustment phase, the abnormal sources of the power grid load are classified. If the adaptive filtering module is in the transient adjustment phase and the power characteristic time series data shows load change characteristics, the abnormal source is classified as algorithm transition fluctuation. If the adaptive filtering module is in the transient adjustment phase and the power characteristic time series data does not show load change characteristics, the abnormal source is classified as a short-term data disturbance. If the adaptive filtering module is not in the transient adjustment phase, and the power quality evaluation index exceeds the abnormal threshold and the duration reaches the preset abnormal time window, the abnormal source is classified as a power grid anomaly. If the adaptive filtering module is not in the transient adjustment phase, and the power quality evaluation index exceeds the abnormal threshold but the duration does not reach the preset abnormal time window, the abnormal source is classified as instantaneous index fluctuation. Otherwise, it is determined that the power grid is operating normally and there is no source of abnormality; The severity score of the violation, the classification of the anomaly source, and the control enable state are integrated to generate a comprehensive judgment result vector and output it.

8. The method for detecting highway power quality based on a smart grid according to claim 7, characterized in that, The steps for performing load regulation operations based on the configured control enable state include: If the configured control enable state is prohibited and the abnormal source is classified as algorithmic over-fluctuation, then the load adjustment action will not be triggered. If the configured control enable state is to allow action and the abnormal source is classified as a power grid abnormality, then an adjustment command with a power limit target value and a power drop slope is issued to the load side of the power grid under test according to the severity score of the over-limit, so as to adjust the load power; If any of the power quality evaluation indicators still exceed the abnormal threshold after the adjustment command is executed, or if a voltage drop in the grid under test is detected to exceed a preset range, then a power switching action is performed.

9. A method for detecting highway power quality based on a smart grid according to claim 4, characterized in that, The method further includes: The duration of the transient adjustment phase is monitored. If the duration exceeds a preset time threshold, the restriction on the power quality evaluation index is lifted, so that the power quality evaluation index follows the output trend of the adaptive filtering module. The effective value increment of the current during the transient adjustment phase is calculated synchronously. If the effective value increment of the current exceeds the preset transformer capacity warning threshold, a capacity warning signal is triggered and the value of the power quality evaluation index is provided.

10. A highway power quality detection system based on a smart grid, used to perform the steps of the method according to any one of claims 1 to 9, characterized in that, include: The data acquisition module is used to acquire the time series data of the electrical energy characteristics of the power grid to be detected, and to extract the convergence state characteristics of the adaptive filtering module during the filtering process of the original electrical time series data. The state recognition module is used to identify whether the adaptive filtering module is in a transient adjustment phase based on the fluctuation level of the electrical energy characteristic time series data and the convergence state characteristics. The index generation module is used to generate power quality evaluation indicators based on the identification results of the transient adjustment phase. The load regulation module is used to determine the control enable state for the grid load based on the power quality evaluation index, and to perform load regulation operations according to the control enable state.

Images