Abnormity processing method and device of inverter, and electronic equipment

Abstract

The invention belongs to the technical field of inverters, and provides an inverter exception handling method and device and electronic equipment, and the inverter exception handling method comprises the steps: obtaining a current sequence output by an inverter; performing anomaly identification on the current sequence to obtain at least one group of abnormal time boundaries; determining the abnormal duration of the inverter according to the at least one group of abnormal time boundaries; and abnormal detection is executed based on the abnormal duration and an abnormal threshold value, corresponding early warning information is triggered and output, the abnormal threshold value is obtained by dynamically adjusting a rated current reference value based on working condition factors of the inverter, and the rated current reference value is determined based on the type of the inverter. The reliability of anomaly detection and the pertinence of operation and maintenance response are improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of inverters, and more particularly relates to an abnormality processing method and device of an inverter and an electronic device. BACKGROUND

[0002] As a main device for power conversion, an inverter is widely used in photovoltaic, energy storage, industrial frequency conversion and other fields, and its operation stability directly affects the reliable operation of the entire power system. In actual working conditions, the inverter is easily affected by factors such as operating environment temperature, load fluctuation and equipment aging, and abnormal working conditions such as overcurrent and harmonic distortion occur. If these abnormal working conditions cannot be accurately identified and disposed of in time, it may lead to equipment damage, system shutdown and even safety accidents.

[0003] However, the existing abnormality processing scheme of the inverter has poor adaptability and is prone to false positives or false negatives. Moreover, it is difficult to comprehensively capture different types of abnormal characteristics (such as fast transient abnormality and slow gradual abnormality), resulting in insufficient accuracy of abnormal boundary identification, and further affecting the accuracy of abnormal duration calculation. SUMMARY

[0004] The purpose of the embodiments of the present application is to provide an abnormality processing method and device of an inverter and an electronic device, aiming to solve the technical problem of the reliability of abnormality detection of the conventional abnormality processing scheme of the inverter.

[0005] To achieve the above-mentioned purpose, according to a first aspect of the present application, an abnormality processing method of an inverter is provided, which comprises:

[0006] obtaining a current sequence output by the inverter; performing abnormality identification on the current sequence to obtain at least one group of abnormal time boundaries; determining an abnormal duration of the inverter according to the at least one group of abnormal time boundaries; performing abnormality detection based on the abnormal duration and an abnormal threshold value to obtain a detection result, wherein the abnormal threshold value is obtained by dynamically adjusting a rated current reference value based on working condition factors of the inverter, and the rated current reference value is determined based on a model of the inverter.

[0007] The embodiments of the present application have the beneficial effect that compared with the prior art, the method obtains a current sequence output by the inverter, performs abnormality identification on the current sequence to obtain at least one group of abnormal time boundaries, and then performs abnormality detection on the inverter based on the abnormal duration and an abnormal threshold value obtained by dynamically adjusting a rated current reference value based on working condition factors of the inverter, to obtain a detection result, thereby improving the reliability of abnormality detection and the pertinence of operation and maintenance response.

[0008] According to a second aspect of the present application, an abnormality processing apparatus of an inverter is provided, comprising: an acquisition unit configured to acquire a current sequence output by the inverter; an identification unit configured to perform abnormality identification on the current sequence to obtain at least one set of abnormality time boundaries; a determination unit configured to determine an abnormality duration of the inverter according to the at least one set of abnormality time boundaries; a processing unit configured to perform abnormality detection based on the abnormality duration and an abnormality threshold to obtain a detection result, wherein the abnormality threshold is obtained by dynamically adjusting a rated current reference value based on a working condition factor of the inverter, and the rated current reference value is determined based on a model of the inverter.

[0009] The second aspect and any one of the implementation manners of the second aspect correspond to the first aspect and any one of the implementation manners of the first aspect respectively. The technical effects corresponding to the second aspect and any one of the implementation manners of the second aspect can be referred to the technical effects corresponding to the first aspect and any one of the implementation manners of the first aspect, which will not be described herein.

[0010] According to a third aspect of the present application, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to enable the electronic device to implement the method according to any one of the aspects.

[0011] According to a fourth aspect of the present application, a computer readable storage medium is provided, which stores a computer program, wherein the computer program is executable by a processor to implement the method according to any one of the aspects.

[0012] According to a fifth aspect of the present application, a computer program product is provided, which, when executed on an electronic device, enables the electronic device to perform the method according to any one of the first aspect.

[0013] It can be understood that the beneficial effects of the second aspect to the fifth aspect can be referred to the related description of the first aspect, which will not be described herein. BRIEF DESCRIPTION OF DRAWINGS

[0014] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced below. 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.

[0015] Figure 1 is a flowchart of an optional abnormality processing method of an inverter provided by an embodiment of the present application; Figure 2 is a flowchart of an optional abnormality processing method of an inverter provided by an embodiment of the present application; Figure 3 is a flowchart of an optional abnormality processing method of an inverter provided by an embodiment of the present application; Figure 4 is a flowchart of an optional abnormality processing method of an inverter provided by an embodiment of the present application; Figure 5 is a flowchart of an optional abnormality processing method of an inverter provided by an embodiment of the present application; Figure 6 is a structural diagram of an abnormality processing device of an inverter provided by an embodiment of the present application; Figure 7 is a structural diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0016] In the following description, for the purposes of explanation and not limitation, specific details are set forth, such as particular system configurations, techniques, etc., in order to provide a thorough understanding of the embodiments of the application. However, it will be apparent to those skilled in the art that the present application can be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present application with unnecessary detail.

[0017] It should be understood that when described herein, the term "comprises / comprising" indicates the presence of the described features, integers, steps, operations, elements, and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0018] It should also be understood that, in the description of this application, unless otherwise stated, the " / " used in the specification and appended claims indicates that the related objects are in an "or" relationship. For example, A / B can mean A or B. The "and / or" in this application is merely a description of the relationship between the related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. Furthermore, in the description of this application, unless otherwise stated, "multiple" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0019] Furthermore, to facilitate a clear description of the technical solutions in the embodiments of this application, the terms "first" and "second" are used in the embodiments of this application to distinguish identical or similar items with essentially the same function and effect. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, but are only used for distinguishing descriptions, and the terms "first" and "second" do not necessarily imply that they are different, nor should they be construed as indicating or implying relative importance.

[0020] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0021] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0022] Currently, inverters, as the main equipment for power conversion, are widely used in photovoltaics, energy storage, industrial frequency conversion, and other fields. Their operational stability directly affects the reliable operation of the entire power system. In actual operating conditions, inverters are susceptible to factors such as ambient temperature, load fluctuations, and equipment aging, which can lead to abnormal operating conditions such as overcurrent and harmonic distortion. If these abnormal conditions are not identified and addressed in a timely and accurate manner, they may cause equipment damage, system shutdown, or even safety accidents.

[0023] Existing inverter anomaly handling solutions are mostly based on fixed thresholds or single anomaly identification algorithms, which have obvious drawbacks: on the one hand, fixed thresholds do not take into account changes in equipment operating conditions and aging, resulting in poor adaptability and a tendency to produce false alarms or missed alarms; on the other hand, single algorithms are difficult to fully capture different types of anomaly features (such as rapid transient anomalies and slow gradual anomalies), leading to insufficient accuracy in anomaly boundary identification, which in turn affects the accuracy of anomaly duration calculation and makes it impossible to achieve graded early warning and accurate handling.

[0024] Therefore, there is an urgent need for an inverter anomaly handling solution that can dynamically adapt to changes in operating conditions, accurately identify abnormal boundaries, and implement graded early warning, so as to improve the reliability of anomaly detection and the pertinence of operation and maintenance response.

[0025] This application provides an example of an inverter anomaly handling method. This method achieves accurate judgment of inverter operation anomalies by acquiring the inverter output current sequence, identifying anomalies, calculating the duration, and detecting anomalies, thereby ensuring the stable and reliable operation of the inverter. The specific implementation process of this method is described in detail below with reference to the accompanying drawings.

[0026] Please refer to Figure 1 As shown, Figure 1 A schematic flowchart illustrating an inverter fault handling method provided in this application is shown. This is an example and not a limitation; the method can be applied to or operated in electronic devices. The method includes: S101, obtain the current sequence output by the inverter.

[0027] S102, perform anomaly identification on the current sequence to obtain at least one set of anomalous time boundaries.

[0028] S103, determine the abnormal duration of the inverter based on at least one set of abnormal time boundaries.

[0029] S104, perform anomaly detection based on the anomaly duration and anomaly threshold to obtain the detection results.

[0030] In some embodiments, the abnormal threshold is obtained by dynamically adjusting the rated current reference value based on the inverter's operating conditions, and the rated current reference value is determined based on the inverter model.

[0031] In some embodiments, firstly, but not limited to, the real-time output current value of the inverter can be acquired at a preset sampling frequency using a current sensor electrically connected to the output terminal of the inverter.

[0032] It should be understood that the above sampling frequency can be set according to the operating characteristics of the inverter and the accuracy requirements for anomaly detection. For example, a sampling frequency of 500Hz to 2kHz can be selected to ensure that the collected current data can fully reflect the changing characteristics of the current signal. Then, the continuously collected current values ​​are arranged in chronological order of sampling time to form a current sequence. This current sequence is indexed by time and contains several consecutive current sampling data points.

[0033] To improve the accuracy of anomaly identification, the acquired current sequence can be preprocessed. Preprocessing operations include filtering and denoising, and standardization. Filtering and denoising are used to eliminate irrelevant interference such as sensor noise and power frequency interference in the current signal, while standardization is used to unify the scale of the current data and avoid the impact of differences in data magnitude on subsequent analysis results.

[0034] In some embodiments, anomaly detection algorithms are used to analyze the changing characteristics of the current sequence, locate the start and end times of the anomaly, and form at least one set of anomalous time boundaries. For example, anomaly detection algorithms in the art (such as traditional machine learning algorithms, deep learning algorithms, and signal processing algorithms) can be used to analyze the preprocessed current sequence to identify anomalous intervals in the current sequence that deviate from the normal operating state, and then determine the start and end times corresponding to the anomalous intervals. The two together constitute a set of anomalous time boundaries.

[0035] Based on the actual detection accuracy and reliability requirements, one or more anomaly identification algorithms can be selected to identify anomaly intervals. Each anomaly identification algorithm outputs a set of anomaly time boundaries, ultimately resulting in at least one set of anomaly time boundaries. If multiple anomaly identification algorithms are used to obtain multiple sets of anomaly time boundaries, these sets can be fused to improve the accuracy of anomaly boundary identification.

[0036] In some embodiments, the duration of the inverter's abnormal operating state is calculated based on the abnormal time boundaries obtained in the aforementioned steps, i.e., the abnormal duration of the inverter. For example, if only one set of abnormal time boundaries is obtained, the time difference between the start and end times of the abnormality in that set of boundaries is directly calculated, and this time difference is the abnormal duration of the inverter. If multiple sets of abnormal time boundaries are obtained, different calculation methods can be selected according to actual application requirements: if multiple sets of abnormal time boundaries are fused, the time difference between the final start and end times of the abnormality after fusion is calculated as the abnormal duration of the inverter; if multiple sets of abnormal time boundaries are not fused, the time difference corresponding to each set of abnormal time boundaries can be calculated separately, and then the final abnormal duration can be determined according to preset rules, such as taking the maximum, minimum, or average value of the time differences of each set.

[0037] Finally, based on the calculated abnormal duration and dynamically adjusted abnormal threshold, the degree of inverter abnormality is graded and the detection results are obtained.

[0038] In some embodiments, anomaly detection can be performed based on set multi-level anomaly thresholds and calculated anomaly durations, with each level of detection corresponding to a warning strategy. For example, if the anomaly characteristics of the current sequence meet the judgment conditions of the first level of detection, a corresponding first-level warning message is triggered; if the judgment conditions of the second level of detection are met, a corresponding second-level warning message is triggered, and so on.

[0039] It should be understood that the judgment conditions for each level of detection need to be set in conjunction with the abnormal threshold and the duration of the abnormality. Different judgment conditions correspond to different degrees of abnormality severity. The output warning information should include key information such as the time of the abnormality and the degree of abnormality so that maintenance personnel can take timely countermeasures.

[0040] It should be noted that the specific abnormal threshold is obtained by dynamically adjusting the rated current reference value based on the inverter's operating conditions, and the rated current reference value is determined based on the inverter model.

[0041] In some embodiments, the rated output current parameter value of the inverter model can be queried and used as the rated current reference value. Then, the rated current reference value is dynamically adjusted based on the actual operating conditions of the inverter. These operating conditions include, but are not limited to, the inverter's operating ambient temperature, equipment operating time, and load rate. For example, when the ambient temperature is higher than a preset temperature threshold, the rated current reference value is appropriately lowered to adapt to the equipment's operating characteristics under high-temperature conditions; when the equipment operating time exceeds a preset number of years, the rated current reference value is lowered according to the degree of equipment aging to avoid misjudgments caused by equipment aging. Finally, multiple abnormal thresholds are set based on the dynamically adjusted rated current reference value.

[0042] In some embodiments, such as Figure 2 As shown, anomaly identification of the current sequence yields at least one set of anomalous time boundaries, including the following specific steps: S201, perform multi-dimensional feature extraction on the preprocessed current sequence to obtain the feature vector.

[0043] For example, the feature vector is used to characterize at least one of the current value, gradient, local mean, local standard deviation, dominant frequency, and high-frequency energy; the feature vector can cover time-domain features, frequency-domain features, and time-frequency joint features.

[0044] The time-domain features include instantaneous value features, gradient features, local statistical features, and fluctuation features. The gradient feature is the difference between the current value at the current sampling point and the current value at the previous sampling point. The local statistical features are the mean and standard deviation within the sliding window. The fluctuation features are the local range and coefficient of variation. The frequency-domain features include the main frequency components of the current signal, the proportion of spectral energy distribution, and harmonic features. The time-frequency joint features include the time-frequency matrix features of the current signal, the instantaneous frequency, and the time-frequency energy concentration. The extracted multi-dimensional features are integrated to construct the feature vector of the corresponding current sequence.

[0045] S201, the isolated forest algorithm is used to calculate anomaly scores on feature vectors to obtain anomaly score sequences.

[0046] For example, feature vectors are input into the isolated forest algorithm model, and the isolated forest algorithm is used to calculate anomaly scores for the feature vectors. The isolated forest algorithm constructs multiple isolated trees, recursively partitions the feature vectors, and determines the anomaly score based on the partition depth of the feature vector in the isolated tree. The shallower the partition depth, the higher the anomaly score, and the deeper the partition depth, the lower the anomaly score. The anomaly scores corresponding to all feature vectors are arranged in the time order of the current sequence to obtain the anomaly score sequence.

[0047] S203, adaptively determine dual thresholds based on the statistical characteristics of abnormal scoring sequences.

[0048] For example, the mean and standard deviation of the abnormal scoring sequence can be calculated, and the sum of the mean and twice the standard deviation can be set as the high threshold, while the sum of the mean and one standard deviation can be set as the low threshold. This dual threshold can adaptively match the abnormal distribution characteristics of the current sequence without manual annotation.

[0049] S204, based on the dual thresholds, identify the first set of abnormal start points and the first set of abnormal end points of the current sequence to form the first set of abnormal time boundaries.

[0050] For example, the intervals in the abnormal scoring sequence where the values ​​are below the low threshold can be defined as normal intervals, and the intervals where the values ​​are above the high threshold can be defined as abnormal intervals. Using the abnormal intervals as a benchmark, a search is performed from the normal intervals toward the abnormal intervals to find the first point that exceeds the low threshold and has the largest gradient, which is taken as the starting point of the first set of anomalies. The search is then performed from the abnormal intervals toward the normal intervals to find the last point that exceeds the low threshold and has the smallest gradient, which is taken as the ending point of the first set of anomalies. The times corresponding to the starting point and the ending point of the first set of anomalies together constitute the time boundary of the first set of anomalies.

[0051] In some embodiments, during the process of identifying anomalies in a current sequence to obtain at least one set of abnormal time boundaries, the extracted multi-dimensional features may specifically include three major categories: time-domain features, frequency-domain features, and time-frequency joint features. The sub-types and extraction logic of each type of feature are clearly defined to characterize the operating state features of the current sequence.

[0052] In some embodiments, time-domain features are extracted directly from the time-domain data of the current sequence, which can intuitively reflect the instantaneous changes and local statistical laws of the current signal, specifically including instantaneous value features, gradient features, local statistical features, and fluctuation features.

[0053] Among them, the instantaneous value feature is the real-time current value of each sampling point in the current sequence; the gradient feature is defined as the difference between the current value of the current sampling point and the current value of the previous sampling point, which is used to characterize the instantaneous rate of change of the current signal; the local statistical feature is extracted by the sliding window method, which can be the mean and standard deviation of the current data within a sliding window of a preset length. The length of the sliding window can be set according to the operating frequency and abnormal response requirements of the inverter (e.g., 5 to 20 sampling periods). The mean reflects the average level of the current within the sliding window, and the standard deviation reflects the dispersion of the current within the sliding window; the fluctuation feature includes the local range and the coefficient of variation. The local range is the difference between the maximum and minimum current values ​​within the sliding window, and the coefficient of variation is the ratio of the standard deviation to the mean within the window. Both together characterize the severity of the fluctuation of the current signal.

[0054] In some embodiments, frequency domain features are obtained by performing a Fourier transform on the current sequence and are used to characterize the frequency distribution characteristics of the current signal. These features may include dominant frequency components, spectral energy distribution, and harmonic features. The dominant frequency components are the frequency components with the largest amplitude in the spectrum after the Fourier transform, corresponding to the dominant frequencies of the current signal. The spectral energy distribution is the proportion of energy corresponding to each frequency component to the total spectral energy, with a focus on the energy proportion of the fundamental frequency and each harmonic frequency. Harmonic features include total harmonic distortion (THD) and the amplitude and phase of each harmonic. The total harmonic distortion is the ratio of the square root of the sum of the squares of the effective values ​​of all harmonic components to the effective value of the fundamental component, used to quantify the degree of harmonic contamination of the current signal.

[0055] In some embodiments, joint time-frequency features are extracted using time-frequency analysis methods such as wavelet transform or short-time Fourier transform. These features can simultaneously reflect the time and frequency variation characteristics of the current signal, avoiding the limitations of single time-domain or frequency-domain analysis. Specifically, they include time-frequency matrix features, instantaneous frequency, and energy concentration. The time-frequency matrix features are the statistical characteristics of the time-frequency matrix obtained by short-time Fourier transform (such as matrix mean and entropy). The instantaneous frequency is the instantaneous frequency value of the signal at each sampling time, used to capture the dynamic change of frequency over time. The energy concentration is the proportion of energy in a specific time-frequency region to the total time-frequency energy, used to identify local energy abrupt change regions in the current signal.

[0056] After completing the multi-dimensional feature extraction described above, various features are filtered and integrated to construct a feature vector for input into the anomaly detection algorithm. The feature vector characterizes the operating state of the current sequence and includes at least one or more of the following: current value (corresponding to instantaneous value features), gradient (corresponding to gradient features), local mean (corresponding to local statistical features), local standard deviation (corresponding to local statistical features), dominant frequency (corresponding to major frequency components), and high-frequency energy (corresponding to spectral energy distribution). Based on the actual anomaly detection accuracy requirements, redundant features can be eliminated by ranking feature importance (e.g., feature importance scoring based on random forests), retaining key features to construct a concise feature vector, thus ensuring both detection accuracy and algorithm efficiency.

[0057] In some embodiments, after completing the multi-dimensional feature extraction of the current sequence and constructing the feature vector, the following specific implementation method can be adopted: based on the isolated forest algorithm, anomaly scoring is calculated on the feature vector to obtain an anomaly scoring sequence, which can accurately quantify the degree of anomaly in the current operating state corresponding to each feature vector.

[0058] First, multiple isolated trees are constructed by randomly selecting subsamples. Each isolated tree isolates data points by randomly selecting features and split values.

[0059] For example, subsamples are randomly selected from all extracted feature vectors. The number of subsamples can be set according to the total size of the feature vectors and the efficiency requirements of the algorithm (e.g., selecting 256 to 1024 feature vectors as subsamples for a single isolated tree). Multiple isolated trees are constructed one by one based on the selected subsamples. The construction process of each isolated tree is as follows: randomly select a feature dimension from the multi-dimensional features, and then randomly select a split value within the value range of the feature dimension. The current subsample set is divided into two subsets by the split value. The above process of randomly selecting features, randomly selecting split values, and splitting the sample set is repeated until each subset contains only one feature vector (i.e., reaching the leaf node of the isolated tree) or the depth of the isolated tree reaches a preset maximum value. The preset maximum value can be adaptively set according to the number of subsamples to avoid overfitting due to an excessively deep tree structure.

[0060] Secondly, the average path length of each feature vector across all isolated trees is calculated. Each feature vector to be scored is input into the constructed isolated trees one by one. Starting from the root node of each isolated tree, based on the feature value of the feature vector at each split node, the process traverses downwards along the corresponding branch until a leaf node is reached. The traversal path length of each feature vector in a single isolated tree (i.e., the number of edges traversed from the root node to the leaf node) is recorded. The arithmetic mean of the path lengths of this feature vector across all isolated trees is taken to obtain the average path length corresponding to this feature vector, thus determining the isolation difficulty of the feature vector in the forest. Abnormal feature vectors have shorter average path lengths.

[0061] Next, an anomaly score is calculated based on the average path length of each feature vector. This anomaly score is the power of 2 raised to the ratio of the negative average path length to the standardization factor. A pre-defined scoring formula is used to calculate the anomaly score for each feature vector: the anomaly score is 2 raised to the power of the ratio of the negative average path length to the standardization factor. The standardization factor is the expected value of the path length of leaf nodes in an isolated tree, used to eliminate the influence of different sizes of isolated forests on path length. Its value can be determined theoretically or empirically based on the number of subsamples, ensuring the comparability of scores for different feature vectors.

[0062] Finally, anomaly scores are standardized by mapping the anomaly score of each feature vector to a range of 0 to 1, resulting in an anomaly score sequence. Since the original anomaly scores may vary, to improve the accuracy of threshold determination, the original anomaly score of each feature vector is mapped to a range of 0 to 1. This mapping can be done using linear normalization or sigmoid function normalization. The standardized anomaly scores of all feature vectors are then arranged according to the chronological order of their corresponding current sequence sampling times, yielding the final anomaly score sequence. This sequence can be directly used for determining adaptive dual thresholds and identifying anomaly time boundaries.

[0063] In some embodiments, the dual thresholds determined based on the above-mentioned abnormal scoring sequence include a high threshold and a low threshold, wherein the high threshold is the sum of the mean of the abnormal scoring sequence and twice the standard deviation, and the low threshold is the sum of the mean of the abnormal scoring sequence and one standard deviation. The specific implementation method for identifying the first set of abnormal start points and the first set of abnormal end points using this dual threshold is as follows: First, a dual threshold is used to divide the abnormal and normal intervals in the abnormal scoring sequence: the interval where the abnormal score exceeds the high threshold is defined as the abnormal interval, the interval where the abnormal score is below the low threshold is defined as the normal interval, and the interval where the abnormal score is between the low and high thresholds is the transition interval. The setting of the transition interval can avoid boundary misjudgment due to signal fluctuations.

[0064] Secondly, locate the first set of abnormal starting points: search the abnormal score sequence point by point from the normal interval to the abnormal interval, find the first abnormal score that exceeds the low threshold and has the largest gradient, and take this point as the first set of abnormal starting points; the criterion for the largest gradient here is that the difference between the abnormal score of the current point and the previous sampling point is the largest, and this point can accurately characterize the starting time of the current sequence from the normal state to the abnormal state.

[0065] Finally, the first set of anomaly termination points is located: The anomaly score sequence is searched point-by-point from the anomaly interval towards the normal interval. The point where the last anomaly score exceeds the low threshold and the gradient is minimized is identified as the first set of anomaly termination points. The criterion for minimum gradient is the minimum difference between the anomaly score of the current point and the next sampling point. This point accurately represents the end time when the current sequence recovers from an abnormal state to a normal state. The aforementioned first set of anomaly termination points together constitute the first set of anomaly time boundaries.

[0066] In some embodiments, such as Figure 3 As shown, anomaly identification of the current sequence yields at least one set of anomalous time boundaries, including: S301 calls the preset waveform template library.

[0067] The waveform template library contains standardized waveform templates for the anomaly initiation, anomaly duration, and anomaly termination phases at different time scales.

[0068] S302 uses a dynamic time warping algorithm to calculate the similarity between the current sequence and each waveform template in the waveform template library.

[0069] S303, select the optimal template from the waveform template library based on similarity.

[0070] S304, Analyze the matching path corresponding to the optimal template to identify the second set of abnormal start points and the second set of abnormal end points, so as to form the second set of abnormal time boundaries.

[0071] In some embodiments, when identifying anomalies in a current sequence to obtain at least one set of abnormal time boundaries, a dynamic time warping algorithm combined with a preset waveform template library can be used. The specific implementation process is as follows. This method can accurately match abnormal waveforms at different time scales and improve the reliability of anomaly boundary identification under complex operating conditions: The first step is to call the pre-built waveform template library. This library is a pre-built set of standardized abnormal waveforms used to cover various common abnormal scenarios and abnormal characteristics at different time scales during inverter operation. Specifically, the library contains standardized waveform templates corresponding to the abnormal initiation, duration, and termination stages at different time scales. The time scales can be divided into short-term (e.g., 1-5 sampling periods), medium-term (e.g., 6-30 sampling periods), and long-term (e.g., 31 or more sampling periods) scales based on the abnormal duration, corresponding to fast transient anomalies, regular anomalies, and slow, gradual anomalies, respectively. Each standardized waveform template undergoes preprocessing, including normalization (unifying amplitude scale) and time alignment (unifying template length benchmark). This preprocessing reduces the impact of amplitude differences between the current sequence and the template, as well as time scaling, on matching accuracy.

[0072] The second step involves using a dynamic time warping algorithm to calculate the dynamic time warping distance between the current sequence and each waveform template in the waveform template library, and then using this distance as the similarity score. Since the current sequence and waveform templates may have inconsistent time lengths, the dynamic time warping algorithm can achieve time alignment through flexible matching, thereby accurately calculating the similarity score. The specific steps include: 1) Data preprocessing and adaptation: The preprocessed current sequence (after filtering, denoising and standardization) is matched with each standardized waveform template in the template library to ensure that the two have a basis for matching. If the current sequence is too long, a sliding window can be used to cut out a subsequence with a length close to that of the template for segmented matching. The sliding window step size is set according to the recognition accuracy requirements. 2) Local distance calculation: Construct a local distance matrix between the current sequence and the template sequence corresponding to the target waveform template. Each element in the matrix represents a similarity metric between a data point in the current sequence and a data point in the template sequence. Here, the absolute difference is used as the calculation method for local distance, that is, local distance = |current sequence data point value - template sequence data point value|. The smaller the local distance, the higher the similarity between the two data points. 3) Construction of Cumulative Distance Matrix: The cumulative distance matrix is ​​constructed by recursively calculating based on the local distance matrix. The recursive calculation follows the temporal constraint rule, that is, the cumulative distance of the current position is obtained by selecting the minimum value among the cumulative distances of the three previous positions directly above, directly to the left, and to the upper left of the current position, and summing it with the local distance of the current position. The formula can be expressed as: Cumulative distance (i,j) = Local distance (i,j) + min[Cumulative distance (i-1,j), Cumulative distance (i,j-1), Cumulative distance (i-1,j-1)], where (i,j) is the row and column index of the cumulative distance matrix; 4) Similarity determination: After the cumulative distance matrix is ​​constructed, the element value of the last row and last column of the matrix is ​​taken as the dynamic time warping distance between the current sequence and the target waveform template. The smaller the distance value, the higher the similarity between the current sequence and the target template. That is, the dynamic time warping distance can be directly used as a quantitative evaluation index of similarity.

[0073] The third step involves selecting the optimal template from the waveform template library based on similarity. To improve the accuracy of anomaly identification and avoid potential misjudgments from single template matching, a multi-template selection and fusion method is used to determine the optimal template. The specific process may include, but is not limited to, the following: 1) Preliminary screening of candidate templates: After calculating the dynamic time warping distance between the current sequence and all waveform templates in the template library, sort the distance values ​​in ascending order and select the N waveform templates with the smallest dynamic time warping distance as candidate templates. The value of N can be set according to the size of the template library and the computational efficiency requirements (e.g., N=3~5). 2) Matching quality assessment: The matching quality of each candidate template is assessed. The assessment indicators include the smoothness of the matching path (the fewer the number of path turns, the higher the smoothness), the mean of the local distance (the smaller the mean, the higher the local matching degree), and the matching coverage (the proportion of the length of the matched current sequence to the total sequence length; the higher the proportion, the better the coverage). 3) Weighted fusion to select the optimal template: Based on the above matching quality evaluation index, each candidate template is assigned a weight. The better the matching quality, the higher the weight ratio. The dynamic time warping distance of the candidate templates is weighted and calculated to obtain the weighted distance mean. The candidate template with the smallest weighted distance mean is selected as the optimal template. This optimal template can accurately characterize the anomaly type and time characteristics corresponding to the current sequence.

[0074] The fourth step is to analyze the matching path corresponding to the optimal template, identify the second set of abnormal start points and the second set of abnormal end points, so as to form the second set of abnormal time boundaries.

[0075] In some embodiments, the matching path corresponding to the optimal template records a one-to-one correspondence between current sequence data points and template sequence data points. Based on this matching path, the mapping point of the template starting point corresponding to the abnormal start stage in the optimal template in the current sequence is located, and this mapping point is used as the second set of abnormal start points. At the same time, the mapping point of the template ending point corresponding to the abnormal end stage in the optimal template in the current sequence is located, and this mapping point is used as the second set of abnormal end points. The sampling times corresponding to the second set of abnormal start points and the second set of abnormal end points constitute the second set of abnormal time boundaries. This boundary can be used to complement and verify the abnormal time boundaries obtained by other algorithms, further improving the accuracy of abnormal identification.

[0076] In some embodiments, firstly, based on the calculated dynamic time warping distance, a preset number of waveform templates with the smallest corresponding dynamic time warping distance are selected as candidate templates. The preset number can be set according to the size of the template library and computational efficiency requirements (e.g., 3-5). Then, a matching quality evaluation is performed on each candidate template. The evaluation indicators include the smoothness of the matching path, the mean of the local distance, and the matching coverage. Based on the matching quality evaluation results, weights are assigned to each candidate template. The better the matching quality, the higher the weight ratio. The candidate templates are then weighted and fused to finally obtain the optimal template. This optimal template can accurately characterize the anomaly type and time characteristics corresponding to the current sequence.

[0077] In some embodiments, such as Figure 4 As shown, the dynamic time warping algorithm is used to calculate the dynamic time warping distance between the current sequence and each waveform template in the waveform template library. The specific process is as follows: S401, construct the initial distance matrix between the current sequence and the template sequence corresponding to each waveform template.

[0078] For example, for each waveform template, an initial distance matrix is ​​constructed between the current sequence and the template sequence corresponding to that waveform template. The row and column dimensions of the initial distance matrix match the number of data points in the current sequence and the template sequence, respectively.

[0079] S402, calculate the local distance of the corresponding data point in the initial distance matrix.

[0080] It should be understood that the local distance is the absolute difference between the data points in the current sequence and the data points in the template sequence, i.e., local distance = |current sequence data point value - template sequence data point value|.

[0081] S403, based on the local distance of each corresponding data point, recursively calculate the cumulative distance of each corresponding data point according to the time-series constraints. For example, based on the local distance of each corresponding data point, recursively calculate the cumulative distance of each corresponding data point according to the time-series constraints, and construct the cumulative distance matrix.

[0082] The temporal constraint rule is as follows: the cumulative distance of the current position is obtained by summing the minimum cumulative distance of the three preceding positions (directly above, directly to the left, and to the upper left) with the local distance of the current position. The formula can be expressed as: cumulative distance (i,j) = local distance (i,j) + min[cumulative distance (i-1,j), cumulative distance (i,j-1), cumulative distance (i-1,j-1)], where (i,j) is the row and column index of the cumulative distance matrix.

[0083] S404 uses the dynamic time warping distance as a quantification of similarity, where a smaller dynamic time warping distance indicates higher similarity. For example, based on the cumulative distance matrix obtained through the aforementioned recursive calculation, the element value of the last row and last column of the matrix is ​​taken as the dynamic time warping distance between the current sequence and the corresponding waveform template.

[0084] In some embodiments, when identifying at least one set of anomalous time boundaries in the current sequence, a multi-scale morphological algorithm can also be used. The specific implementation process is as follows: Figure 5 As shown, this method can effectively preserve the boundary features of the current signal and improve the robustness of abnormal boundary identification under noise interference conditions: S501 employs multi-scale structuring elements to perform morphological filtering on the current sequence, resulting in a filtered current sequence. The multi-scale structuring elements include short-time structuring elements for identifying fast transient boundaries and long-time structuring elements for identifying slowly changing boundaries, adapting to the needs of identifying abnormal boundaries with different rates of change. The morphological filtering process specifically utilizes composite morphological operations, sequentially performing erosion, dilation, dilation, and erosion operations. This composite operation effectively eliminates irrelevant interference such as sensor noise and power frequency interference in the current sequence, while preserving the abnormal boundary features of the current signal to the greatest extent possible, avoiding the loss of boundary information.

[0085] S502 uses the extreme points of the morphological gradient of the filtered current sequence as candidate boundary points.

[0086] First, the morphological gradient of the filtered current sequence is calculated. The morphological gradient is defined as the difference between the result of dilation processing and erosion processing of the filtered current sequence. This difference can significantly amplify the boundary region of the current signal, making abnormal boundaries easier to identify. Then, the extreme points (including maxima and minima) in the morphological gradient sequence are detected. These extreme points correspond to the abrupt changes in the current signal and are selected as candidate boundary points.

[0087] S503, hierarchical clustering is performed on the candidate boundary points to obtain the third set of anomaly start points and the third set of anomaly end points, which constitute the third set of anomaly time boundaries.

[0088] For example, a similarity matrix can be constructed based on the gradient magnitude and neighborhood distribution consistency of candidate boundary points. A hierarchical clustering algorithm can be used to cluster the candidate boundary points, and isolated noise points with small gradient magnitudes can be clustered into one class and then eliminated. The effective clusters are retained, and the temporal distribution characteristics of the candidate boundary points in the effective clusters are analyzed. The earliest candidate boundary point in the cluster is taken as the starting point of the third group of anomalies, and the latest candidate boundary point is taken as the ending point of the third group of anomalies. The sampling times corresponding to the starting point and ending point of the third group of anomalies constitute the third group of anomaly time boundaries, which can be fused with the anomaly time boundaries obtained by the aforementioned isolated forest algorithm and dynamic time warping algorithm to further improve the reliability of anomaly identification.

[0089] In some embodiments, when identifying anomalies in current sequences to obtain at least one set of anomalous time boundaries, a multi-scale morphological algorithm can be used. This algorithm can adapt to anomalous boundaries with different rates of change through multi-scale structuring elements, while taking into account both noise suppression and boundary preservation, thus improving the robustness of anomaly identification under complex operating conditions. The specific implementation process is as follows: The first step is to perform morphological filtering on the current sequence using multi-scale structuring elements to obtain the filtered current sequence.

[0090] The multi-scale structural elements include two types of structural elements, each adapted to different types of abnormal boundary identification needs: one type is a short-time structural element used to identify fast transient boundaries. This type of structural element is short in length and can accurately capture the boundary features of rapid changes in current signals, such as the current change boundary corresponding to fast transient anomalies such as inverter short-circuit faults; the other type is a long-time structural element used to identify slow gradual boundaries. This type of structural element is longer in length and can effectively identify the boundary features of slowly changing current signals, such as the current gradual boundary corresponding to slow gradual anomalies such as inverter overload.

[0091] The specific execution process of morphological filtering is as follows: Simultaneously using the aforementioned short-time and long-time structure elements, a composite morphological operation of erosion, dilation, dilation, and erosion is sequentially performed on the preprocessed current sequence (which has undergone preliminary filtering and denoising). This composite operation combination can exert a synergistic effect. The preceding erosion and dilation operations are mainly used to eliminate high-frequency irrelevant interferences such as sensor noise and power frequency interference in the current sequence, while the dilation and erosion operations are used to repair signal details, preserving the abnormal boundary features of the current signal to the greatest extent possible, avoiding the loss of key boundary information during noise suppression, and finally outputting the filtered current sequence.

[0092] The second step is to calculate the morphological gradient of the filtered current sequence to highlight the boundary positions of the current signal.

[0093] In some embodiments, the morphological gradient is calculated as follows: First, the filtered current sequence obtained in the first step is subjected to dilation and erosion processing, respectively. Dilation is used to expand the bright areas (areas with large amplitudes) of the current signal, resulting in a first result; erosion is used to shrink the bright areas of the current signal, resulting in a second result. Then, the difference between the first and second results is calculated. This difference effectively amplifies the areas with drastic amplitude changes in the filtered current sequence, significantly highlighting the boundary positions of the current signal and improving the clarity of boundary identification. Finally, this difference is directly used as the morphological gradient of the filtered current sequence.

[0094] The third step is to screen candidate boundary points and perform hierarchical clustering to obtain the third set of anomalous time boundaries.

[0095] Specifically, the extreme points (including maxima and minima) in the morphological gradient sequence obtained in the second step are detected. These extreme points correspond to abrupt changes in the amplitude of the current signal and are potential anomaly boundary points, thus they are selected as candidate boundary points. Next, hierarchical clustering is performed on the candidate boundary points: a similarity matrix is ​​constructed based on the gradient magnitude and neighborhood distribution consistency of the candidate boundary points, and the candidate boundary points are divided into different cluster groups using a hierarchical clustering algorithm. Clusters with isolated distributions and small gradient magnitudes are considered noise points and are removed. Effective cluster groups with large gradient magnitudes and concentrated neighborhood distributions are retained. The temporal distribution characteristics of the candidate boundary points in the effective cluster groups are analyzed, and the earliest candidate boundary point in each cluster is designated as the starting point of the third anomaly group, and the latest candidate boundary point is designated as the ending point of the third anomaly group.

[0096] The sampling times corresponding to the third set of anomaly start points and the third set of anomaly end points together constitute the third set of anomaly time boundaries. This boundary can be fused and verified with the anomaly time boundaries obtained based on the isolated forest algorithm and the dynamic time warping algorithm, further improving the accuracy and reliability of inverter anomaly boundary identification and providing accurate boundary basis for calculating the anomaly duration.

[0097] In some embodiments, hierarchical clustering of candidate boundary points is performed to obtain a third set of anomaly start points and a third set of anomaly end points. This can be achieved by the following steps: evaluating the confidence level of candidate boundary points based on the gradient magnitude of extreme points and the distribution consistency of extreme points in the neighborhood; retaining extreme points with confidence levels that meet preset requirements as valid candidate boundary points based on the confidence evaluation results; performing hierarchical clustering on the valid candidate boundary points to remove isolated noise points, thereby obtaining a third set of anomaly start points and a third set of anomaly end points to constitute a third set of anomaly time boundaries.

[0098] Specifically, the extreme points (including maxima and minima) in the morphological gradient sequence obtained above are detected. These extreme points correspond to abrupt changes in the amplitude of the current signal and are potential anomaly boundary points, thus they are selected as candidate boundary points. Next, hierarchical clustering is performed on the candidate boundary points to obtain a third set of anomaly start points and a third set of anomaly end points. The specific process is as follows: First, based on the gradient magnitude of the candidate boundary points (i.e., the extreme points mentioned above) and the consistency of the distribution of extreme points in their neighborhoods, a confidence assessment is performed on each candidate boundary point. The larger the gradient magnitude and the more concentrated the distribution of extreme points in the neighborhood, the more significant the boundary feature corresponding to the candidate boundary point, and the higher the confidence level. Subsequently, based on the confidence assessment results, a confidence threshold is set as a preset requirement. Extreme points with a confidence level not lower than the threshold are retained as valid candidate boundary points, while low-quality candidate points with insufficient confidence are eliminated. Finally, hierarchical clustering is performed on the valid candidate boundary points to further eliminate isolated noise points, resulting in a clustered valid boundary group. The temporal distribution characteristics of the candidate boundary points in the valid boundary group are analyzed. The earliest valid candidate boundary point in the cluster is taken as the starting point of the third group of anomalies, and the latest valid candidate boundary point is taken as the ending point of the third group of anomalies.

[0099] The sampling times corresponding to the third set of anomaly start points and the third set of anomaly end points together constitute the third set of anomaly time boundaries. This boundary can be fused and verified with the anomaly time boundaries obtained based on the isolated forest algorithm and the dynamic time warping algorithm, further improving the accuracy and reliability of inverter anomaly boundary identification and providing accurate boundary basis for calculating the anomaly duration.

[0100] In some embodiments, there are multiple sets of abnormal time boundaries. The duration of the inverter's abnormality is determined based on at least one set of abnormal time boundaries, including: Calculate the confidence level for each group of anomalous time boundaries. The confidence level includes at least one of the following: path quality confidence level, gradient magnitude confidence level, and time consistency confidence level. Based on the confidence level, determine the weight of each group of abnormal time boundaries; By weighting the weights of each set of abnormal time boundaries, the final abnormal start point and the final abnormal end point are obtained, thus forming the final abnormal time boundary. The duration of the inverter's anomaly is determined based on the final anomaly time boundary.

[0101] In some embodiments, if multiple sets of abnormal time boundaries are obtained through the above algorithms, the abnormal duration of the inverter can be calculated based on the fusion of multiple sets of boundaries. The specific implementation process is as follows: The first step is to calculate the confidence score for each set of anomaly time boundaries. The confidence score quantifies the reliability of each set of anomaly time boundaries and includes at least one of the following: path quality confidence score, gradient magnitude confidence score, and temporal consistency confidence score. Path quality confidence score is determined based on metrics such as the smoothness and coverage of the matching path during anomaly identification; the smoother the path and the higher the coverage, the higher the confidence score. Gradient magnitude confidence score is determined based on the magnitude of the gradient corresponding to the anomaly boundary point; the larger the gradient magnitude, the more significant the boundary features, and the higher the confidence score. Temporal consistency confidence score is determined based on the degree of overlap between anomaly boundaries obtained by different algorithms in the time dimension; the higher the overlap, the higher the confidence score.

[0102] The second step is to determine the weight of each group of anomalous time boundaries based on their confidence levels. A normalization process is used to convert the confidence levels of each group of anomalous time boundaries into corresponding weights. Specifically, the sum of the confidence levels of all groups of anomalous time boundaries is calculated, and the confidence level of each group of anomalous time boundaries is divided by this sum to obtain the weight for each group. Anomalous time boundaries with higher confidence levels have a larger weight percentage and a greater influence in the fusion calculation.

[0103] The third step is to calculate the final anomaly time boundaries using weighted averages. For each anomaly start point, the time value corresponding to the anomaly start point of each set of anomaly time boundaries is multiplied by the weight of that set, and the sum of all products is obtained to get the time corresponding to the final anomaly start point. For each anomaly end point, the same weighted average calculation method is used, multiplying the time value corresponding to the anomaly end point of each set of anomaly time boundaries by the weight of that set and summing the products to obtain the time corresponding to the final anomaly end point. The final anomaly start point and the final anomaly end point together constitute the final anomaly time boundaries.

[0104] The fourth step is to determine the duration of the inverter's anomaly. The time difference between the final anomaly start point and the final anomaly end point within the final anomaly time boundary is calculated; this time difference is the duration of the inverter's anomaly. By weighted fusion of multiple sets of boundaries, the error in identifying boundaries using a single algorithm can be effectively reduced, improving the accuracy of the anomaly duration calculation.

[0105] In some embodiments, after performing anomaly detection based on the duration of the anomaly and a dynamically adjusted anomaly threshold, and obtaining the detection results, corresponding early warning information can be triggered based on the detection results, thereby achieving differentiated management and control of anomalies of different severity levels and improving the pertinence and efficiency of operation and maintenance response.

[0106] First, in this embodiment, the abnormal threshold is obtained by dynamically adjusting the rated current reference value based on the inverter's operating conditions, and the rated current reference value is determined based on the inverter's model.

[0107] Specifically, you can first look up the rated output current parameter of the inverter based on its specific model and use this parameter as the initial rated current reference value. Then, combine the actual operating conditions of the inverter (such as operating ambient temperature, equipment operating time, load rate, etc.) to dynamically calibrate the rated current reference value. For example, under high temperature conditions, the reference value can be appropriately lowered, and after the equipment ages, the reference value can be adjusted according to the degree of aging to ensure that the threshold is adapted to the actual operating characteristics of the equipment. Based on the calibrated rated current reference value, set a first threshold and a second threshold. The first threshold is greater than the second threshold to distinguish between anomalies of different severity.

[0108] Subsequently, based on the detection results obtained from the anomaly detection, a corresponding early warning process is triggered.

[0109] For example, if any data point in the current sequence exceeds the first threshold, it indicates a serious abnormality in the inverter (such as severe overcurrent), and a Level 1 warning message must be immediately triggered. The output of the Level 1 warning message includes, but is not limited to, local audible and visual alarms on the device, pushing emergency alarm notifications to the operation and maintenance monitoring platform, and can also be linked to the inverter's protection devices to trigger emergency measures such as load reduction and shutdown to prevent the abnormality from escalating and causing equipment damage or safety accidents.

[0110] First, the system checks if any data points in the current sequence exceed the second threshold. If so, it further compares the corresponding abnormal duration with a preset duration. If the abnormal duration exceeds the preset duration, it indicates that the inverter has a moderate abnormality that has persisted for a certain period, triggering a secondary warning message. This secondary warning message can be pushed to the maintenance personnel's terminal, clearly informing them of the time of the abnormality, its duration, and the current value, reminding them to promptly investigate the fault and prevent the abnormality from persisting for an extended period and affecting equipment lifespan. The preset duration can be set according to the inverter's operating characteristics and maintenance response capabilities, such as 5 seconds or 10 seconds.

[0111] For example, if a data point in the current sequence exceeds the second threshold, but the corresponding abnormal duration does not reach the preset duration, it indicates that the inverter has a minor transient anomaly. No emergency action is required at this time, so a real-time alarm is not triggered, but a third warning message must be recorded. This third warning message must include detailed data such as the time of the anomaly, the peak instantaneous current, and the duration of the anomaly. This data is used for equipment health status analysis, provides data support for preventative maintenance, and helps maintenance personnel anticipate potential equipment problems.

[0112] It should be noted that in practical applications, at least one of the above-mentioned anomaly detection methods can be selected to be enabled based on the inverter's usage scenario and operation and maintenance needs. At the same time, the content and output method of early warning information at all levels can be flexibly configured to ensure that the early warning information can be accurately communicated to relevant personnel and that the handling measures are adapted to the degree of anomaly, so as to achieve the purpose of accurate early warning and graded handling.

[0113] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0114] Corresponding to the inverter's fault handling method in the above embodiment, Figure 6 This is a schematic diagram of the structure of an inverter fault handling device provided in an embodiment of this application. This device can be implemented by software, hardware, or a combination of both, forming part or all of a computer device. This computer device can be... Figure 7 The electronic device shown.

[0115] Reference Figure 6 The inverter's fault handling device includes: Acquisition unit 601 is used to acquire the current sequence output by the inverter.

[0116] The identification unit 602 is used to identify anomalies in the current sequence and obtain at least one set of abnormal time boundaries.

[0117] The determination unit 603 is used to determine the abnormal duration of the inverter based on at least one set of abnormal time boundaries.

[0118] The processing unit 604 is used to perform anomaly detection based on the anomaly duration and anomaly threshold to obtain the detection result.

[0119] The abnormal threshold is obtained by dynamically adjusting the rated current reference value based on the inverter's operating conditions, and the rated current reference value is determined based on the inverter's model.

[0120] It is understood that the embodiments of the inverter's fault handling device and any implementation thereof correspond to the embodiments of the inverter's fault handling method and any implementation thereof. The technical effects corresponding to the embodiments of the inverter's fault handling device and any implementation thereof can be found in the technical effects corresponding to the aforementioned embodiments of the inverter's fault handling method and any implementation thereof, and will not be repeated here.

[0121] It should be noted that the inverter fault handling device provided in the above embodiments is only an example of the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0122] The functional units and modules in the above embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of the embodiments of this application.

[0123] It should be noted that the information interaction and execution process between the above-mentioned devices / units are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, and they will not be repeated here.

[0124] This application also provides an electronic device, which includes one or more processors and a memory; The memory is coupled to one or more processors. The memory is used to store computer program code, which includes computer instructions. One or more processors call the computer instructions to cause the electronic device to execute the aforementioned inverter exception handling method.

[0125] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. The electronic device 700 can be a mobile phone, smart screen, tablet computer, wearable electronic device, in-vehicle electronic device, augmented reality (AR) device, virtual reality (VR) device, laptop computer, ultra-mobile personal computer (UMPC), netbook, personal digital assistant (PDA), projector, or a communication device such as a server, storage device, or base station, or a smart car, etc. This application embodiment does not impose any limitations on the specific type of electronic device.

[0126] The memory 701 can be used to store computer software programs 702 and modules. The processor 703 executes various functional applications and data processing of the electronic device by running the software programs and modules stored in the memory 701. The memory 701 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, application programs required for at least one function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the electronic device (such as audio data, telephone book, etc.). In addition, the memory 701 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0127] The processor 703 may include one or more processors such as a central processing unit (CPU), an application processor (AP), and a baseband processor. The processor can serve as the nerve center and command center of the wireless router. The processor 703 can generate operation control signals based on instruction opcodes and timing signals to control instruction fetching and execution. The memory 701 can be used to store executable program code, including instructions. The processor 703 executes various functional applications and data processing of the network device by running the instructions stored in the memory. The memory 701 may include a program storage area and a data storage area, such as storing data for audio signals to be played. For example, the memory may be Double Data Rate Synchronous Dynamic Random Access Memory (DDR) or Flash memory.

[0128] This application also provides a computer-readable storage medium storing computer instructions; when the computer-readable storage medium is used on an electronic device, it causes the electronic device to execute the aforementioned inverter fault handling method.

[0129] The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or can include one or more data storage devices such as servers or data centers that can be integrated with media. The available medium can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media, or semiconductor media (e.g., solid-state disks (SSDs)).

[0130] This application also provides a computer program product containing computer instructions, which, when run on an electronic device, enables the electronic device to execute the aforementioned inverter exception handling method.

[0131] The computer storage medium and computer program product provided in the embodiments of this application are used to execute the methods provided above. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects corresponding to the methods provided above, and will not be repeated here.

[0132] In the above embodiments, implementation can also be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When the computer instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, optical fiber, Digital Subscriber Line, DSL) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access, or a data storage device such as a server or data center that integrates one or more available media. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk drive (HDD), or solid-state drive (SSD), etc., and the storage medium can also include combinations of the above types of memory.

[0133] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0134] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments claimed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0135] In the embodiments provided in this application, it should be understood that the disclosed apparatus / network devices and methods can be implemented in other ways. For example, the apparatus / network device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0136] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0137] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for handling an inverter's malfunctions, characterized in that, include: Obtain the current sequence output by the inverter; The current sequence is anomaly identified to obtain at least one set of anomalous time boundaries; The duration of the abnormality of the inverter is determined based on the at least one set of abnormal time boundaries; Anomaly detection is performed based on the anomaly duration and anomaly threshold to obtain a detection result. The anomaly threshold is obtained by dynamically adjusting the rated current reference value based on the inverter's operating conditions. The rated current reference value is determined based on the inverter's model.

2. The method according to claim 1, characterized in that, The process of identifying anomalies in the current sequence to obtain at least one set of anomalous time boundaries includes: Multi-dimensional feature extraction is performed on the current sequence to obtain a feature vector; The isolated forest algorithm is used to calculate anomaly scores on the feature vectors to obtain an anomaly score sequence. The dual thresholds are adaptively determined based on the aforementioned anomaly scoring sequence; The first set of abnormal start points and the first set of abnormal end points of the current sequence are identified based on the dual thresholds to form the first set of abnormal time boundaries.

3. The method according to claim 2, characterized in that, The multi-dimensional features include time-domain features, frequency-domain features, and joint time-frequency features; The time-domain features include instantaneous value features, gradient features, local statistical features, and fluctuation features. The gradient feature is the difference between the current value at the current sampling point and the current value at the previous sampling point. The local statistical features are the mean and standard deviation within the sliding window. The fluctuation features are the local range and coefficient of variation. The frequency domain features include the main frequency components, spectral energy distribution, and harmonic features. The time-frequency joint features include time-frequency matrix features, instantaneous frequency, and energy concentration. The feature vector is used to represent at least one of the following: current value, gradient, local mean, local standard deviation, dominant frequency, and high-frequency energy.

4. The method according to claim 2, characterized in that, The isolated forest algorithm is used to calculate anomaly scores on the feature vectors, resulting in an anomaly score sequence, including: Multiple isolated trees are constructed by randomly selecting subsamples, and each isolated tree isolates data points by randomly selecting features and split values; Calculate the average path length of each feature vector across all isolated trees; An anomaly score is calculated based on the average path length of each feature vector, and the anomaly score is the power of the ratio of the negative average path length to the standardization factor. The anomaly score of each feature vector is mapped to the interval between 0 and 1 for standardization to obtain the anomaly score sequence.

5. The method according to claim 2, characterized in that, The dual thresholds include a high threshold and a low threshold. The high threshold is the sum of the mean of the abnormal rating sequence and twice the standard deviation, and the low threshold is the sum of the mean of the abnormal rating sequence and one standard deviation. The step of identifying the first set of abnormal start points and the first set of abnormal end points based on the dual thresholds includes: The dual thresholds are used to determine the abnormal intervals and normal intervals in the abnormal scoring sequence, wherein the abnormal interval is the interval in the abnormal scoring sequence where the abnormal score exceeds the high threshold, and the normal interval is the interval in the abnormal scoring sequence where the abnormal score is lower than the low threshold. Search from the normal range to the abnormal range, and find the first point that exceeds the low threshold and has the largest gradient as the first group of abnormal starting points; Search from the abnormal interval to the normal interval, and find the last point that exceeds the low threshold and has the smallest gradient as the end point of the first group of abnormalities.

6. The method according to any one of claims 1 to 5, characterized in that, The process of identifying anomalies in the current sequence to obtain at least one set of anomalous time boundaries includes: Call a preset waveform template library, which contains standardized waveform templates for the abnormal initiation stage, abnormal duration stage and abnormal termination stage at different time scales; The similarity between the current sequence and each waveform template in the waveform template library is calculated using a dynamic time warping algorithm. The optimal template in the waveform template library is selected based on the similarity. The matching path corresponding to the optimal template is analyzed to identify the second set of abnormal start points and the second set of abnormal end points, which constitute the second set of abnormal time boundaries.

7. The method according to any one of claims 1 to 5, characterized in that, The process of identifying anomalies in the current sequence to obtain at least one set of anomalous time boundaries includes: The current sequence is subjected to morphological filtering using multi-scale structuring elements to obtain a filtered current sequence. The extreme points of the morphological gradient of the filtered current sequence are used as candidate boundary points. Hierarchical clustering is performed on the candidate boundary points to obtain a third set of anomaly start points and a third set of anomaly end points, which constitute the third set of anomaly time boundaries.

8. The method according to any one of claims 1 to 5, characterized in that, The anomaly threshold includes a first threshold and a second threshold. After performing anomaly detection based on the anomaly duration and the anomaly threshold to obtain the detection result, the method further includes: If the detection result indicates that any data point in the current sequence exceeds the first threshold, a first-level warning message is triggered. If the detection result indicates that any data point in the current sequence exceeds the second threshold and the corresponding abnormal duration exceeds the preset duration, then a secondary warning message is triggered. If the detection result indicates that any data point in the current sequence exceeds the second threshold, but the corresponding abnormal duration does not reach the preset duration, no output is triggered, but a third warning message is recorded.

9. An inverter fault handling device, characterized in that, include: Acquisition unit, used to acquire the current sequence output by the inverter; An identification unit is used to identify anomalies in the current sequence and obtain at least one set of abnormal time boundaries. A determining unit is configured to determine the abnormal duration of the inverter based on the at least one set of abnormal time boundaries; The processing unit is used to perform anomaly detection based on the anomaly duration and the anomaly threshold, and obtain the detection result, wherein the anomaly threshold is obtained by dynamically adjusting the rated current reference value based on the inverter's operating conditions, and the rated current reference value is determined based on the inverter's model.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it causes the electronic device to implement the method as described in any one of claims 1 to 8.

Images