Zone area energy storage operation dynamic early warning control method and system

Abstract

The invention discloses a zone area energy storage operation dynamic early warning control method and system, and relates to the technical field of anomaly detection, and the method comprises the following steps: obtaining operation data of an energy storage system, constructing a data trend offset vector, and extracting trend offset features; dynamically matching the trend offset features with a perturbation anomaly feature database pre-constructed based on historical data, and outputting a preliminary anomaly early warning signal; performing disturbance analysis by acquiring synchronous phasor measurement data, identifying disturbance response characteristics, and correcting the preliminary abnormal early warning signal in combination with trend similarity; extracting a target abnormal feature corresponding to the corrected abnormal early warning signal, and outputting an early warning credibility score in combination with a preliminary abnormal early warning signal correction result; according to the invention, the dynamic sensing of the operation state of the energy storage system and the reliability evaluation of the abnormal early warning are realized, the early warning accuracy and the system response capability are effectively improved, and the problems of recognition lag and high false alarm rate are solved.

Description

Technical Field

[0001] The present invention relates to the field of anomaly detection technology, and more specifically, to a method and system for dynamic early warning control of energy storage operation in a substation. Background Art

[0002] With the rapid development of new energy and distributed energy, energy storage systems are becoming a key component of modern power systems as an important technical means to enhance grid regulation capabilities, ensure power supply reliability, and promote energy transformation. Especially at the substation level, energy storage systems can effectively mitigate load fluctuations, optimize energy management, improve power quality, and contribute to the construction of smart grids. However, energy storage equipment is subject to many complex dynamic changes during operation, such as fluctuations in operating parameters, sudden disturbances, and equipment aging. These factors can lead to performance degradation and even safety hazards in energy storage systems, seriously affecting grid stability and operational safety. Therefore, how to monitor and accurately identify abnormal conditions in energy storage systems in real time and issue early warning signals in a timely manner has become a core technical challenge to ensure the safe and efficient operation of energy storage systems.

[0003] For example, the invention patent with announcement number CN114995208A discloses an energy storage safety system and a control method for the energy storage safety system. The control method of the energy storage safety system obtains the real-time temperature inside the energy storage box of the battery energy storage system and the gas concentration of the combustible gas released by the battery thermal runaway in the battery energy storage system. Then, when the real-time temperature is greater than the first warning temperature and / or the gas concentration is greater than the first warning concentration, a first warning signal is generated, and finally the first warning signal is sent to the alarm. In this method, starting from the gas concentration of the combustible gas and the real-time temperature inside the energy storage box, problems such as untimely warning of battery thermal runaway, inability to accurately judge battery thermal runaway, inability to prevent the occurrence of battery thermal runaway, and excessively long alarm signal chains are solved.

[0004] For example, the invention patent with announcement number: CN115328009A discloses a monitoring and management system for a gravity energy storage system, which relates to the field of gravity energy storage technology. The present invention is to solve the problem that there is no integrated monitoring and control management system for gravity energy storage systems in the prior art. The data acquisition unit of the present invention is used to collect gravity energy storage side data, grid side operation data and environmental data; the monitoring and early warning unit is used to make early warning judgments based on the collected data, and generate early warning information when an early warning situation occurs; the data analysis unit is used to perform gravity energy storage side performance analysis, energy storage and grid operation relationship analysis, and protection risk analysis based on the indicators obtained from the collected data and early warning information, and generate analysis results and decision instructions; the comprehensive display and reporting unit is used to display the real-time data collected by the data acquisition unit, the early warning information generated by the monitoring and early warning unit, and the analysis results and decision instructions output by the data analysis unit. The present invention monitors, manages and controls the gravity energy storage system in real time, so that the power grid operates smoothly and improves energy conversion rate and utilization efficiency.

[0005] The above disclosed technical solutions have at least the following technical problems: Existing technologies for extracting time-series features from energy storage system operating data are incomplete. Most focus solely on static indicators or short-term fluctuations, lacking the ability to extract dynamic trend shift features such as adaptive segmentation, trend mutation intensity, and significance scoring. This makes it difficult to capture the complex evolution of energy storage operating conditions. To address these issues, the present invention proposes a solution. Summary of the Invention

[0006] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a method and system for dynamic early warning control of substation energy storage operation. By constructing a dynamic matching mechanism between trend offset characteristics and perturbation templates, and combining phasor disturbance analysis with multimodal data fusion, the present invention realizes dynamic perception of the operating status of the energy storage system and credibility assessment of abnormal early warning, effectively improving the accuracy of early warning and system responsiveness, and solving the problems of recognition lag and high false alarm rate of traditional methods.

[0007] To achieve the above object, the present invention provides the following technical solutions: A method for dynamic early warning control of energy storage operation in a substation area includes the following steps: acquiring operating data of an energy storage system, constructing a data trend offset vector, and extracting trend offset features of the data trend offset vector; performing similarity matching on the trend offset features with a dynamically updated historical abnormal feature database, and outputting a preliminary abnormality early warning signal when the matching degree exceeds a preset threshold; performing disturbance analysis on synchronized phasor measurement data, identifying disturbance response features, and correcting the preliminary abnormality early warning signal based on trend similarity; The target abnormality features corresponding to the corrected abnormal warning signal are extracted, and combined with the preliminary abnormal warning signal correction results, the warning credibility score is output.

[0008] In a preferred embodiment, the method for extracting trend offset features is as follows: based on the preprocessed time series data, the change rate and slope change of the data points are extracted, and the time series data is divided into several trend candidate sub-segments through an adaptive segmentation algorithm; each trend candidate sub-segment is processed by a time series clustering algorithm, and the trend feature vector of each sub-segment is extracted; the trend feature vectors of any two adjacent sub-segments are measured for difference, a trend mutation intensity value is generated, and a significance score is output in combination with the trend feature; high-significance trend offset points are screened according to the significance score, adjacent offset points are connected to generate a trend offset main axis, and the trend offset feature corresponding to each offset point is extracted.

[0009] In a preferred embodiment, the trend offset feature is matched with the dynamically updated historical anomaly feature database for similarity. When the matching degree exceeds a preset threshold, a preliminary anomaly warning signal is output, specifically as follows: along the trend offset main axis, the high-significance trend offset points in the current sliding time window are sorted in time series, the trend offset features of each point are extracted and a trend offset sequence is formed; the trend offset sequence is compared with the anomaly sequence in the anomaly feature database for similarity, a matching confidence score is generated, and a preliminary anomaly warning signal is output based on the warning matching threshold.

[0010] In a preferred embodiment, a similarity comparison is performed between the trend offset sequence and the anomaly sequence in the anomaly feature database to generate a matching confidence score, specifically as follows: within a sliding time window, points with significant scores are screened to form a trend offset point set; the trend offset point set is arranged in chronological order to generate a current trend offset path, and the relative increments of the main axis coordinates of adjacent points are calculated to form a relative incremental path; the baseline relative path in the pre-stored historical anomaly database is read; the path direction consistency score and the main axis incremental similarity score of the current relative incremental path and the baseline relative path are calculated; the direction consistency score and the incremental similarity score are weighted averaged to output the final matching confidence score.

[0011] In a preferred embodiment, the preliminary abnormal warning signal is corrected by obtaining synchronous phasor measurement data for disturbance analysis, identifying disturbance response characteristics and combining trend similarity, specifically as follows: obtaining phasor measurement data of the energy storage node and its adjacent topological nodes within the preliminary abnormal warning window, and synchronously collecting the boundary entropy change trend of the power grid; performing disturbance synchronization analysis on the phasor measurement data of the adjacent topological nodes, extracting mutation characteristics and identifying disturbance response characteristics; matching the current boundary entropy change trend with the boundary entropy trajectory of the historical abnormal evolution sample set through a dynamic time warping method, and outputting trend similarity; fusing the disturbance response characteristics and trend similarity to correct the preliminary abnormal warning signal, and the correction includes promoting, maintaining, marking or downgrading the preliminary abnormal warning signal.

[0012] In a preferred embodiment, the current boundary entropy change trend is matched with the boundary entropy trajectory of the historical abnormal evolution sample set through a dynamic time warping method, and the trend similarity is output, specifically as follows: within the warning time period, the boundary entropy evolution sequence under the sliding window is extracted based on the key operating parameters of the energy storage system, and recorded as the current boundary entropy sequence; the historical abnormal evolution sample set is read, where each sample contains a historical boundary entropy change sequence; based on the dynamic time warping algorithm, the current boundary entropy sequence and the historical boundary entropy change sequence are elastically aligned in time series, and the corresponding matching distance is output; each matching distance is converted into a trend similarity score through an exponential decay function, and the maximum trend similarity score is screened; the maximum trend similarity score is compared with the set threshold, and the trend similarity is output.

[0013] In a preferred embodiment, the target abnormal features corresponding to the corrected abnormal warning signal are extracted, and combined with the preliminary abnormal warning signal correction result, a warning credibility score is output, including: obtaining multimodal time series data in the time window corresponding to the corrected abnormal warning signal, and constructing a multimodal time series data input set; performing cross-combination processing on the multimodal time series data to generate a cross-feature space; extracting key cross-features from the cross-feature space based on a statistical analysis method; performing embedding and reconstruction analysis on the key cross-features in the abnormal window through kernel principal component analysis, and outputting a residual score; extracting disturbance behavior features in phasor measurement data, and generating a disturbance confidence score through discrete counting index mapping; fusing the residual score and the disturbance confidence score, and outputting a warning credibility score.

[0014] In a preferred embodiment, the residual score and the disturbance confidence score are integrated to output the warning credibility score, specifically: the residual score, trend similarity score and disturbance confidence score are mapped to fuzzy linguistic variables through fuzzy membership functions; a fuzzy logic rule set is designed based on domain knowledge and used as a priori knowledge constraint; under the prior knowledge constraint, a decision tree algorithm is used to construct a conditional probability model; the conditional probability weight of each rule is calculated based on historical data, and the Bayesian weighted fusion trigger rule is used to output the final warning credibility score.

[0015] A dynamic early warning and control system for energy storage operation in a substation area includes a feature extraction module, a matching module, a correction module and an output module, and there are connections between the modules; the feature extraction module is used to obtain the operating data of the energy storage system, construct a data trend offset vector, and extract the trend offset feature; the matching module is used to dynamically match the trend offset feature with a perturbation anomaly feature database pre-built based on historical data, and output a preliminary abnormal warning signal; the correction module is used to perform disturbance analysis by acquiring synchronized phasor measurement data, identify disturbance response features, and correct the preliminary abnormal warning signal in combination with trend similarity; the output module is used to extract the target abnormal feature corresponding to the corrected abnormal warning signal, and output a warning credibility score in combination with the correction result of the preliminary abnormal warning signal.

[0016] The technical effects and advantages of the present invention's method and system for dynamic early warning control of substation energy storage operation are as follows: 1. The present invention realizes dynamic, accurate and reliable abnormal early warning control of the operating status of the energy storage system by integrating multi-source heterogeneous data analysis and time series intelligent algorithms. It has many technical advantages such as foresight, sensitivity and robustness, and has important engineering application value and broad promotion prospects. Secondly, through adaptive segmentation and time series clustering algorithms, the time series data is structured and processed to accurately capture the significant trend deviation points in the data changes, and then the trend deviation main axis is constructed. This method effectively explores the deep dynamic change characteristics in the operation of the energy storage system, breaks through the limitations of traditional static threshold monitoring methods in the weak recognition ability of complex abnormal behaviors, and provides a solid dynamic perception foundation for the early warning mechanism.

[0017] 2. The present invention realizes real-time recognition of abnormal patterns by dynamically matching with a pre-built perturbation anomaly template library. The dual indicators of path direction consistency and principal axis increment similarity are used to measure the high-dimensional similarity of the trend offset sequence and the historical abnormal trajectory, and the matching confidence is output in combination with the weighted average algorithm, thereby improving the accuracy and real-time performance of anomaly detection. This mechanism has strong scalability and can adapt to abnormal behaviors of different forms in a variety of operating scenarios. On the basis of the preliminary warning, the system further introduces the phasor measurement data of the energy storage node and its topological neighborhood for disturbance synchronization analysis, combines the boundary entropy evolution trend with the dynamic time regularization matching of the historical evolution samples, and double-checks the credibility of the preliminary warning signal. At this stage, by integrating the dual evidence mechanism of disturbance response and entropy evolution, the robustness and reliability of the warning results are effectively improved, and false alarms and missed alarms are avoided. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 The figure is a flow chart of a method for dynamic early warning control of energy storage operation in a substation according to the present invention.

[0019] Figure 2This is a structural diagram of a dynamic early warning control system for energy storage operation in a substation according to the present invention. DETAILED DESCRIPTION

[0020] The following will provide a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0021] Example 1, Figure 1 The present invention provides a method for dynamic early warning control of energy storage operation in a substation, comprising the following steps: S1, obtain the operating data of the energy storage system, construct a data trend offset vector, and extract the trend offset feature of the data trend offset vector.

[0022] In this embodiment, the operating data of the energy storage system is obtained, a data trend offset vector is constructed, and the trend offset feature of the data trend offset vector is extracted, as follows: S11: Acquire original multi-dimensional parameter time series data of the energy storage system operation, wherein the parameters include voltage, current, SOC, temperature, etc.; Preprocessing of multi-dimensional parameter time series data, including time alignment, missing value filling and noise filtering operations; Based on the preprocessed data, the data change rate and slope changes are extracted. Based on the adaptive segmentation algorithm (such as the sliding window-based change point detection method, which calculates the statistical data change rate and slope changes within the sliding window and divides the boundary when the change amplitude exceeds the set threshold), the entire time series is divided into several trend candidate sub-segments. Each sub-segment corresponds to a relatively stable or continuously changing trend segment. The sub-segment length is automatically adjusted according to the change intensity to form unequal length divisions. Applying a time series clustering algorithm to each sub-segment to extract its representative trend features and form a set of trend feature vectors. The trend features include average slope, average slope sign change, slope standard deviation jump, etc. The difference between the cluster center vectors of any two adjacent sub-segments is measured to calculate the trend mutation intensity, which is used to quantify the trend jump amplitude and preliminarily characterize the degree of trend deviation; A trend mutation significance evaluation function is constructed based on the trend mutation intensity, the average slope sign change (for judging trend direction reversal), and the slope standard deviation jump (for judging volatility). As a cognitive salience function, the significance score of each potential trend deviation point is calculated to quantify its importance to the system's attention. Apply the cognitive salience function to the full time series trend feature sequence to extract highly significant trend offset points and form the trend offset main axis. For example, set a significance threshold, filter out offset points with significance scores higher than the set threshold, and organize them in chronological order to form the trend offset main axis. The main axis represents an ordered set of key evolutionary jump points in the time series, which is used for subsequent key feature extraction and event focusing. For each offset point in the trend offset main axis, its associated extracted trend offset features are extracted. The extracted trend offset features include the offset time (obtained directly from the main axis), the offset direction (determined by the change in the sign of the slope of the previous and next sub-segments), the offset amplitude (quantified by the difference in sub-segment eigenvalues), the local volatility (obtained by the standard deviation of the slope of the previous and next segments), the offset duration (estimated by the segment width window) and the significance level (inherited salience function score), and a trend offset vector is constructed.

[0023] The cognitive salience function is specifically as follows:

[0024] Where: Normalize the slope standard deviation jump (between 0 and 1). is the slope standard deviation jump of trend offset point i, reflecting the degree of trend instability. It is the reversal sign of the trend deviation point i. If a positive or negative trend switch occurs, it is 1, otherwise it is 0. and is the weight coefficient, is the trend jump amplitude of trend offset point i, is the base of natural logarithms, is the Softmax exponential scaling factor, is the comprehensive significance score of the potential trend shift point i, Is an index variable that represents each point of all trend deviation points.

[0025] S2, performs similarity matching between the trend deviation feature and the dynamically updated historical anomaly feature database. When the matching degree exceeds the preset threshold, a preliminary anomaly warning signal is output.

[0026] In this embodiment, the trend deviation feature is matched with the dynamically updated historical abnormal feature database for similarity. When the matching degree exceeds a preset threshold, a preliminary abnormal warning signal is output, as follows: From a large amount of historical energy storage system operation data, various typical abnormal evolution processes (such as abnormal power oscillation, abnormal SOC rise, frequent charge-discharge switching, etc.) are screened out. Key parameter evolution fragments are extracted in the form of time series slices to form a "perturbation anomaly feature database." Each template contains a sequence of key feature points in the direction of the trend deviation main axis, amplitude gradient characteristics, change rate curves, and feature jump point labels. For real-time running data, along the extracted trend shift axis, the highly significant trend shift points within the current sliding time window are sorted in time series, and their trend shift features are extracted to form a trend shift sequence; Compare the current trend offset sequence with each abnormal template sequence in the template library to generate a matching confidence score; Set a warning matching threshold. When the matching confidence score of any abnormal template exceeds the warning matching threshold, it is considered that the current window trend has typical abnormal evolution characteristics, and a preliminary abnormal warning signal is output.

[0027] The similarity comparison between the current trend offset sequence and each abnormal template sequence in the template library is performed as follows: Within the sliding time window, based on the calculated significance score, points with scores higher than the set threshold are selected to form a trend shift point set. Each point contains: timestamp, significance score, jump direction, and main axis position coordinates (indicating the projection position of the point on the trend main axis); Arrange the offset point sequence in chronological order to form the current trend offset path, which can be regarded as an "event flow trajectory" in two-dimensional space; Output the relative increment of the principal axis coordinates (the principal axis position difference of each adjacent projection coordinate in the trend offset path) according to the current trend offset path, forming a relative increment path to eliminate the absolute value difference; Read the relative path saved in the pre-built historical anomaly template. Each template contains a fixed-length relative offset increment and direction sequence. Calculate the path similarity based on the current relative incremental path and each template path, and output the path direction consistency score and the main axis incremental similarity score (obtained using the weighted cosine similarity algorithm); According to the path direction consistency score and the main axis incremental similarity score, the final matching confidence score is output based on the weighted average algorithm.

[0028] The path direction consistency scores are as follows:

[0029] The main axis incremental similarity score is as follows:

[0030] The matching confidence scores are as follows:

[0031] Where: is the path direction consistency score, is the total number of trend deviation points, It is the trend jump direction label, which indicates the trend jump direction of the offset point. It usually takes the value of +1 (up) or −1 (down). is the direction of the i-th trend jump point in the j-th template path, Is a consistency indicator function. If the current direction is consistent with the template direction, the value is 1, otherwise it is 0. It is used to calculate the average consistency. is the principal axis incremental similarity score, is the increment of the i-th position in the j-th template path, which is the same as the current path Correspondingly, used to calculate the inner product and norm, Is the spindle offset increment, which represents the position difference of the i-th offset point relative to the previous offset point in the spindle direction, and is used to form the path vector. and is the Euclidean norm of the sequence, are the lengths of the current path and template path increment vectors, is the matching confidence score between the current path and the j-th template path, and is the weight coefficient.

[0032] S3, performs disturbance analysis by acquiring synchronized phasor measurement data, identifies disturbance response characteristics and corrects the preliminary abnormal warning signal in combination with trend similarity.

[0033] In this embodiment, by acquiring synchronized phasor measurement data for disturbance analysis, identifying disturbance response characteristics and combining trend similarity, the preliminary abnormal warning signal is corrected as follows: Get the corresponding preliminary abnormal warning window 、 The following data: Synchronous phasor measurement data of the energy storage node and its adjacent nodes, including voltage phase angle, frequency and voltage amplitude (sampling frequency ≥ 10Hz); the current operating parameter sequence in 、 Real-time boundary entropy evolution trend within the segment; boundary entropy evolution sample set formed by historically confirmed abnormal evolution paths; The phasor measurement data of the topological nodes adjacent to the energy storage node is 、 Perform disturbance synchronization analysis within a time period to extract mutation characteristic quantities, including voltage phase angle change rate, system frequency jump rate, and voltage amplitude change rate; During the disturbance synchronization analysis process, if any of the mutation feature quantities mutates and highly coincides with the warning time period (for example, the mutation time difference is less than 200ms and the jump direction is consistent), it is considered a "system-level consistency disturbance" and the preliminary warning signal is marked as a high-confidence warning, entering the subsequent processing process. If there is no significant synchronous response behavior among the neighboring nodes, it is judged that the disturbance may be a local isolated event, reducing the system-level confidence. Obtain the boundary entropy change trend of the current operating parameters, and use the dynamic time warping method to perform similarity matching between the boundary entropy change trend of the current operating parameters and the boundary entropy trajectories of multiple historical abnormal evolution samples, output the trend similarity, and determine whether the boundary entropy trend is consistent with the historical abnormal evolution sample set. If consistent, mark it as "having an abnormal evolution trend"; If the similarity between the current boundary entropy trend and the historical typical anomaly evolution process trend is higher than the set threshold, the current disturbance is considered to have a continuous evolution trend and marked as a trend development warning. Otherwise, it is regarded as an isolated anomaly or noise disturbance. The results of phasor mutation judgment and boundary entropy comparison are combined to modify the preliminary abnormal warning signal, as follows: If both the PMU phasor mutation and boundary entropy evolution are satisfied, it is upgraded to a high-confidence warning; If only the PMU mutation is met, the preliminary warning level will be maintained; If only the boundary entropy consistency is met, it is marked as a "trend warning" state and awaits subsequent observation confirmation; If none of the conditions are met, the current warning signal will be downgraded or suppressed.

[0034] The dynamic time warping method is used to perform similarity matching between the boundary entropy change trend of the current operating parameters and the boundary entropy trajectories of multiple historical abnormal evolution samples, as follows: Extract the boundary entropy evolution sequence of any window length from the key operating parameters of the energy storage system (such as SOC, current, and voltage) within the warning period, and record it as the current boundary entropy sequence. This sequence reflects the entropy structure evolution of the current parameter change trend. Construct a historical abnormal evolution sample set. The sample data comes from the operation trajectory of the energy storage system that has been confirmed as a fault or abnormal event in the past operation. Each sample contains a historical boundary entropy change sequence. The current boundary entropy evolution sequence and the historical boundary entropy change sequence are normalized, and the minimum-maximum normalization method is used to map each entropy value to the [0,1] interval to eliminate the amplitude difference between different samples and maintain the consistency of the evolution trend structure; Based on the dynamic time warping (DTW) algorithm, the current boundary entropy sequence is elastically aligned with each historical boundary entropy change sequence in the sample set, and the matching distance is output; Each matching distance is converted into a trend similarity score through an exponential decay function; Count all trend similarity scores, obtain the maximum similarity value, and compare it with the set threshold; If the maximum similarity value exceeds the set threshold, the current trend is considered to be consistent with the historical abnormal evolution path and is determined to have an abnormal evolution trend; If the maximum similarity value is lower than the set threshold, the current trend does not have development consistency and is regarded as an isolated disturbance or short-term fluctuation.

[0035] The exponential decay function is as follows:

[0036] Where: It is represented as the trend similarity score converted from the k-th matching distance, To adjust the trend sensitivity parameter, the value range is usually 0.5~1.5. It represents the registration distance between the current boundary entropy change sequence and the kth historical boundary entropy change sequence.

[0037] S4, extract the target abnormality features corresponding to the corrected abnormal warning signal, combine them with the preliminary abnormal warning signal correction result, and output the warning credibility score.

[0038] In this embodiment, the target abnormality feature corresponding to the corrected abnormal warning signal is extracted, and combined with the preliminary abnormal warning signal correction result, the warning credibility score is output, which is as follows: Obtain operating parameter data within the time window corresponding to the corrected abnormal warning signal, including but not limited to the voltage, current, state of charge (SOC), power, and its rate of change of the energy storage node. Combined with synchronized phasor measurement parameters (frequency, voltage phase angle) and system event records (such as sudden load changes and equipment switching signals), a multi-modal time series data input set is constructed. Performing cross-combination processing on the multimodal data to generate high-order derived features to form a cross-feature space; Using the constructed cross-feature space, statistical analysis or dimensionality reduction methods (such as kernel principal component analysis, isolation forest, or Shapley value methods) are used to identify the feature dimensions most sensitive to changes in system behavior during the abnormal window. Key cross-features (such as SOC change rate × current change rate, power change slope × voltage phase angle difference, and voltage phase angle difference between energy storage nodes and adjacent nodes × phase angle jump synchronization) are extracted as core inputs for subsequent embedding reconstruction and disturbance feature extraction, improving the model's resolution in the context of weak signals or multi-source disturbances. Based on kernel principal component analysis, the key cross-features within the abnormal window are embedded and reconstructed. The reconstruction error can be used to quantify the degree of structural consistency destruction of the system state and output a residual score. Extract disturbance behavior features from the collected phasor measurement data and generate disturbance confidence scores through discrete counting index mapping. The disturbance behavior features include the number of short-term sudden changes in voltage phase angle, the duration of continuous oscillation of system frequency, and the frequency of sharp swings in voltage amplitude. The residual score, trend similarity score and disturbance confidence score are mapped into fuzzy linguistic variables through fuzzy membership functions, which are defined as "low", "medium" and "high". The fuzzy membership function can be constructed using trigonometric functions, trapezoidal functions or Gaussian functions to ensure smooth transition of the scores near the critical values. Several fuzzy logic rules are designed based on fuzzy language variables to express the warning credibility judgment conditions based on expert experience, including: If the trend consistency score is "high" and the perturbation confidence score is "high", the confidence level is "high"; If the perturbation confidence score is "low" and the residual score is "low", the confidence level is "low"; If the trend consistency score is "medium" and the perturbation confidence score is "low", the confidence level is "medium"; Fuzzy logic rules are used as a priori judgment basis, and a conditional probability model is constructed through the decision tree algorithm to evaluate the credible probability of abnormal events developing under the triggering of each rule; The conditional probability weight corresponding to each rule is obtained based on the maximum likelihood estimation method through historical data or statistical learning methods, and the judgment results of all triggering rules are fused using the Bayesian weighting method to output the final warning credibility score.

[0039] The specific calculation formula of the fuzzy membership function is as follows:

[0040] Where, is the membership function value, is the current input value (such as perturbation confidence score, residual score, trend similarity score), is the peak position of the membership function, is the standard deviation.

[0041] It should be noted that discrete counting indicators refer to discretized characteristic quantities such as the frequency of occurrence or duration of system disturbance behavior after event-level identification within a specific time window.

[0042] Example 2, Figure 2The present invention provides a dynamic early warning control system for substation energy storage operation, comprising a feature extraction module, a matching module, a correction module and an output module, with connections between the modules; the feature extraction module is used to obtain operating data of the energy storage system, construct a data trend offset vector, and extract trend offset features; the matching module is used to dynamically match the trend offset features with a perturbation anomaly feature database pre-constructed based on historical data, and output a preliminary abnormal warning signal; the correction module is used to perform disturbance analysis by acquiring synchronized phasor measurement data, identify disturbance response features, and correct the preliminary abnormal warning signal in combination with trend similarity; the output module is used to extract the target abnormal feature corresponding to the corrected abnormal warning signal, and output a warning credibility score in combination with the correction result of the preliminary abnormal warning signal.

[0043] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0044] The above embodiments may be implemented in whole or in part through software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments may be implemented in whole or in part in the form of a computer program product.

[0045] Those skilled in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel 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.

[0046] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0047] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0048] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for dynamic early warning control of energy storage operation in a substation, characterized in that: The steps include: Acquire operating data of the energy storage system, construct a data trend offset vector, and extract trend offset features of the data trend offset vector; The trend deviation feature is matched with the dynamically updated historical abnormal feature database for similarity. When the matching degree exceeds the preset threshold, a preliminary abnormal warning signal is output; By acquiring synchronized phasor measurement data for disturbance analysis, the disturbance response characteristics are identified and combined with trend similarity to correct the preliminary abnormal warning signal; The target abnormality features corresponding to the corrected abnormal warning signal are extracted, and combined with the preliminary abnormal warning signal correction results, the warning credibility score is output.

2. The method for dynamic early warning control of energy storage operation in a substation according to claim 1, characterized in that: The method for extracting the trend shift feature is as follows: Based on the preprocessed time series data, the change rate and slope change of the data points are extracted, and the time series data is divided into several trend candidate sub-segments through an adaptive segmentation algorithm; The time series clustering algorithm is used to process each trend candidate sub-segment and extract the trend feature vector of each sub-segment; The difference between the trend feature vectors of any two adjacent sub-segments is measured to generate a trend mutation intensity value, and the significance score is output in combination with the trend features; Highly significant trend shift points are screened according to the significance scores, adjacent shift points are connected to generate the trend shift axis, and the trend shift features corresponding to each shift point are extracted.

3. The method for dynamic early warning control of energy storage operation in a substation according to claim 2, characterized in that: The trend deviation feature is matched with the dynamically updated historical abnormal feature database for similarity, and when the matching degree exceeds a preset threshold, a preliminary abnormal warning signal is output, as follows: Along the trend shift axis, the highly significant trend shift points within the current sliding time window are sorted in time series, and the trend shift features of each point are extracted to form a trend shift sequence. The trend deviation sequence is compared with the abnormal sequence in the abnormal feature database for similarity, a matching confidence score is generated, and a preliminary abnormal warning signal is output based on the warning matching threshold.

4. The method for dynamic early warning control of energy storage operation in a substation according to claim 3, characterized in that: The trend shift sequence is compared with the abnormal sequence in the abnormal feature database for similarity, and a matching confidence score is generated, as follows: In the sliding time window, points with significance scores higher than the set threshold are screened to form a set of trend deviation points; Arrange the trend offset point set in time sequence to generate the current trend offset path, calculate the relative increments of the principal axis coordinates of adjacent points, and form a relative increment path; Read the reference relative path in the pre-stored historical anomaly database; Calculating a path direction consistency score and a principal axis increment similarity score between the current relative incremental path and the reference relative path; The direction consistency score and the incremental similarity score are weighted averaged to output the final matching confidence score.

5. The method for dynamic early warning control of energy storage operation in a substation according to claim 4, characterized in that: The method involves obtaining synchronized phasor measurement data for disturbance analysis, identifying disturbance response characteristics, and combining trend similarity to correct the preliminary abnormal warning signal, as follows: Acquire phasor measurement data of the energy storage node and its adjacent topological nodes within the preliminary abnormal warning window, and simultaneously collect the entropy change trend of the grid boundary; Perform disturbance synchronization analysis on the phasor measurement data of adjacent topological nodes to extract mutation characteristics and identify disturbance response characteristics; The dynamic time warping method is used to match the current boundary entropy change trend with the boundary entropy trajectory of the historical abnormal evolution sample set, and the trend similarity is output; The disturbance response characteristics and trend similarity are integrated to modify the preliminary abnormal warning signal, and the modification includes promoting, maintaining, marking or degrading the preliminary abnormal warning signal.

6. The method for dynamic early warning control of substation energy storage operation according to claim 5, characterized in that: The dynamic time warping method is used to match the current boundary entropy change trend with the boundary entropy trajectory of the historical abnormal evolution sample set, and the trend similarity is output as follows: During the warning period, the boundary entropy evolution sequence under the sliding window is extracted based on the key operating parameters of the energy storage system and recorded as the current boundary entropy sequence; Read the historical anomaly evolution sample set, where each sample contains a historical boundary entropy change sequence; Based on the dynamic time warping algorithm, the current boundary entropy sequence and the historical boundary entropy change sequence are elastically aligned in time series, and the corresponding matching distance is output; Each matching distance is converted into a trend similarity score through an exponential decay function, and the maximum trend similarity score is filtered; The maximum trend similarity score is compared with the set threshold and the trend similarity is output.

7. The method for dynamic early warning control of energy storage operation in a substation according to claim 6, characterized in that: The extracted target abnormality features corresponding to the corrected abnormal warning signal are combined with the preliminary abnormal warning signal correction result to output the warning credibility score, including: Obtain the multimodal time series data within the time window corresponding to the corrected abnormal warning signal and construct a multimodal time series data input set; Perform cross-combination processing on multimodal time series data to generate a cross feature space; Extract key cross-features from the cross-feature space based on statistical analysis methods; The kernel principal component analysis is used to embed and reconstruct the key cross features in the abnormal window and output the residual score; Extract disturbance behavior features from phasor measurement data and generate disturbance confidence scores through discrete counting index mapping; The residual score and disturbance confidence score are integrated to output the warning credibility score.

8. The method for dynamic early warning control of energy storage operation in a substation according to claim 7, characterized in that: The residual score and the disturbance confidence score are integrated to output the warning credibility score, which is specifically: The residual score, trend similarity score and disturbance confidence score are mapped into fuzzy linguistic variables through fuzzy membership function; Design fuzzy logic rule sets based on domain knowledge and use them as prior knowledge constraints; Under the constraints of prior knowledge, the decision tree algorithm is used to construct a conditional probability model; The conditional probability weight of each rule is calculated based on historical data, and the trigger rules are fused using Bayesian weighting to output the final warning credibility score.

9. The method for dynamic early warning control of substation energy storage operation according to claim 4, characterized in that: The path direction consistency scores are as follows: The main axis incremental similarity score is as follows: The matching confidence scores are as follows: Where: is the path direction consistency score, is the total number of trend deviation points, It is the trend jump direction label, indicating the trend jump direction of the offset point. is the direction of the i-th trend jump point in the j-th template path, Is a consistency indicator function. If the current direction is consistent with the template direction, the value is 1, otherwise it is 0. is the principal axis incremental similarity score, is the increment of the i-th position in the j-th template path, is the spindle offset increment, which indicates the position difference of the i-th offset point relative to the previous offset point in the spindle direction. and is the Euclidean norm of the sequence, are the lengths of the current path and template path increment vectors, is the matching confidence score between the current path and the j-th template path, and is the weight coefficient.

10. A system using a method for dynamic early warning control of energy storage operation in a substation according to any one of claims 1 to 9, characterized in that: It includes feature extraction module, matching module, correction module and output module, and there are connections between modules; A feature extraction module is used to obtain the operating data of the energy storage system, construct a data trend offset vector, and extract trend offset features; A matching module is used to dynamically match trend deviation features with a database of perturbation anomaly features pre-built based on historical data, and output preliminary anomaly warning signals; The correction module is used to perform disturbance analysis by acquiring synchronized phasor measurement data, identify disturbance response characteristics, and correct the preliminary abnormal warning signal in combination with trend similarity; The output module is used to extract the target abnormality features corresponding to the corrected abnormal warning signal, and output the warning credibility score based on the preliminary abnormal warning signal correction results.

Images