Energy storage power station abnormity alarm method, device and equipment based on multi-modal data

Abstract

The invention relates to the technical field of energy storage, and discloses an energy storage power station abnormity alarm method, device and equipment based on multi-modal data, and the method comprises the steps: obtaining the original sampling data of an energy storage power station, and carrying out the sampling frequency alignment processing of the original sampling data based on the resampling frequency, so as to obtain the multi-modal data; performing feature extraction and fusion processing on the multi-modal data to obtain a comprehensive feature vector, and performing dimension reduction processing on the comprehensive feature vector to obtain a key feature vector; and inputting the key feature vector into a pre-trained anomaly monitoring model, and generating an alarm signal under the condition that the energy storage power station is monitored to run abnormally according to an output result of the anomaly monitoring model. The method has the beneficial effects that the multi-modal data of different data sources are kept consistent in the sampling time dimension, the availability of the multi-modal data is improved, and the accuracy of subsequent feature extraction and analysis is also ensured.

Description

Technical Field

[0001] The present application relates to the field of energy storage technology, and in particular to an abnormal alarm method, device and equipment for an energy storage power station based on multimodal data. Background Art

[0002] In order to cope with the current continuous growth in electricity demand and low-carbon requirements, new energy power generation and energy storage technology have become an important part of energy generation. As an important infrastructure of the new power system, energy storage technology is of great significance in promoting green energy transformation, responding to extreme events, ensuring energy supply security and promoting high-quality development of new energy. Therefore, as an important part of the stability of the power grid, the safety and reliability of energy storage power stations are increasingly valued.

[0003] Energy storage power stations store electricity through a large number of battery cells and release electricity when demand peaks to balance the load on the power grid. However, energy storage power stations face many safety risks during operation, such as battery overheating, fire, chemical leakage, etc. In order to ensure the safe operation of energy storage power stations, they must be monitored and warned in a comprehensive and real-time manner. In related technologies, traditional monitoring systems often rely on a single-mode data source, which is prone to false alarms or omissions, and cannot fully reflect the true status of energy storage power stations, making it difficult to detect abnormal situations in a timely manner. Summary of the invention

[0004] The present application provides an abnormal alarm method, device and equipment for an energy storage power station based on multimodal data, which solves the technical problem that the data source modality relied on by the current operation monitoring of the energy storage power station is too single, which easily leads to false alarms or missed alarms, and achieves the technical effect of using the feature fusion of multimodal data to monitor the abnormal operation of the energy storage power station and comprehensively improve the energy safety of the energy storage power station.

[0005] In order to achieve the above objectives, the main technical solutions adopted in this application include:

[0006] In a first aspect, an embodiment of the present application provides an abnormal alarm method for an energy storage power station based on multimodal data, and the abnormal alarm method for an energy storage power station includes:

[0007] Acquire original sampling data of the energy storage power station, and perform sampling frequency alignment processing on the original sampling data based on the resampling frequency to obtain multimodal data;

[0008] Performing feature extraction and fusion processing on the multimodal data to obtain a comprehensive feature vector, and performing dimensionality reduction processing on the comprehensive feature vector to obtain a key feature vector;

[0009] The key feature vector is input into a pre-trained abnormality monitoring model, and when an abnormal operation of the energy storage power station is detected according to an output result of the abnormality monitoring model, an alarm signal is generated.

[0010] The abnormal alarm method for energy storage power station proposed in the embodiment of the present application relies on multiple data acquisition devices to collect and monitor multimodal data related to the operating status of the energy storage power station in real time, and realizes alignment processing of the original sampled data by setting the resampling frequency, so that the multimodal data from different data sources are consistent in the sampling time dimension. This alignment processing not only improves the availability of multimodal data, but also ensures the accuracy of subsequent feature extraction and analysis, and avoids information loss or misjudgment due to data asynchrony.

[0011] In addition, in the feature extraction and fusion processing stage, the various characteristics of multimodal data are comprehensively utilized to obtain a more comprehensive comprehensive feature vector, and on this basis, dimensionality reduction processing is performed to extract key feature vectors, thereby effectively reducing the data dimension, reducing the computational complexity of the subsequent anomaly monitoring model, and improving the anomaly monitoring efficiency. At the same time, the key information of the data is retained as much as possible to ensure the accuracy of anomaly detection.

[0012] Optionally, performing sampling frequency alignment processing on the original sampled data based on the resampling frequency to obtain multimodal data includes:

[0013] Acquire each original sampling frequency of the original sampling data, and use the lowest frequency among the original sampling frequencies as the resampling frequency;

[0014] The original sampled data is adjusted based on the resampling frequency to obtain multimodal data with the same sampling frequency.

[0015] Optionally, the adjusting the original sampled data based on the resampling frequency includes:

[0016] Determine a resampling period corresponding to the resampling frequency;

[0017] When there is one and only one original sampled data in the resampling period, directly using the original sampled data as the multimodal data;

[0018] In the case that there are at least two original sampling data in the resampling period, an average value of all the original sampling data in the resampling period is used as the multimodal data.

[0019] Using the lowest original sampling frequency as the resampling frequency can ensure that at least one data point can be sampled for each modal data, thereby ensuring that data from different sources are aligned and processed on the same time base. This synchronization of sampling frequencies can eliminate the sampling frequency differences of different data acquisition devices and provide more accurate data input for subsequent feature fusion and anomaly monitoring of multimodal data.

[0020] Optionally, the original sampling data is obtained through various data acquisition devices in the energy storage power station, wherein the data acquisition device includes at least one of a battery management system, a smoke sensor, a temperature sensor, and a micro-nano particle sensor.

[0021] Optionally, the extracting and fusing the multimodal data to obtain a comprehensive feature vector includes:

[0022] For the data acquired by the battery management system and after the alignment processing, the change indicators of the battery cell voltage, battery cell current and battery cell temperature are extracted as the first type of features;

[0023] For the data acquired by the smoke sensor and after alignment processing, the average smoke concentration, the maximum smoke concentration and the number of smoke concentration mutations are extracted as the second type of features;

[0024] For the data acquired by the temperature sensor and after alignment processing, the average value of the ambient temperature, the maximum value of the ambient temperature and the ambient temperature fluctuation rate of the energy storage power station are extracted as the third type of features;

[0025] For the data acquired by the micro-nanoparticle sensor and after alignment processing, the average concentration of micro-nanoparticles, the maximum concentration of micro-nanoparticles and the number of mutations in the concentration of micro-nanoparticles are extracted as the fourth type of features;

[0026] The first type of features, the second type of features, the third type of features and the fourth type of features are concatenated to obtain the comprehensive feature vector.

[0027] Optionally, the number of smoke concentration mutations is obtained according to the following steps, including:

[0028] The difference between the smoke concentrations in two adjacent sampling periods is obtained respectively, and the number of times the difference exceeds a preset threshold in a number of consecutive sampling periods is counted to obtain the number of sudden changes in the smoke concentration.

[0029] Through the above feature extraction and fusion processing, more comprehensive comprehensive feature vectors are obtained. These comprehensive feature vectors can better describe the operating status of the energy storage power station, thereby providing more reliable input for subsequent abnormal monitoring and realizing more comprehensive safety monitoring of the energy storage power station.

[0030] Optionally, the performing dimensionality reduction processing on the comprehensive feature vector to obtain a key feature vector includes:

[0031] Performing standardization on the comprehensive feature vector to obtain a standardized feature vector;

[0032] Analyzing the standardized eigenvector based on principal component analysis to obtain a projection matrix;

[0033] The normalized eigenvector is projected onto the projection matrix to obtain the key eigenvector.

[0034] The principal component analysis is used to reduce the dimensionality of the comprehensive feature vector, thereby reducing the dimension of the multimodal data in the process, while retaining the maximum amount of information of the multimodal data. Therefore, the key feature data can not only use the characteristics of multimodal data to fully reflect the operating status of the energy storage power station, but also simplify the data structure on this basis, reduce the computational complexity of the subsequent abnormal monitoring model, and reduce the interference of redundant information.

[0035] Optionally, the anomaly monitoring model is a support vector machine trained based on a supervised learning algorithm.

[0036] The embodiment of the present application utilizes a supervised learning algorithm to train a support vector machine, which has significant scalability, so new positive and negative sample data can be continuously introduced for training, thereby continuously improving the accuracy of abnormal monitoring results.

[0037] In a second aspect, an embodiment of the present application provides an abnormal alarm device for an energy storage power station based on multimodal data, wherein the abnormal alarm device for an energy storage power station comprises:

[0038] A resampling module, used to obtain original sampling data of the energy storage power station, and perform sampling frequency alignment processing on the original sampling data based on the resampling frequency to obtain multimodal data;

[0039] A feature extraction module, used for performing feature extraction and fusion processing on the multimodal data to obtain a comprehensive feature vector, and performing dimensionality reduction processing on the comprehensive feature vector to obtain a key feature vector;

[0040] The alarm module is used to input the key feature vector into a pre-trained abnormality monitoring model, and generate an alarm signal when the energy storage power station is detected to have an abnormal operation according to the output result of the abnormality monitoring model.

[0041] The abnormal alarm device for energy storage power station proposed in the embodiment of the present application relies on multiple data acquisition devices to collect and monitor multimodal data related to the operating status of energy storage power station in real time, and realizes alignment processing of original sampled data by setting the resampling frequency, so that multimodal data from different data sources are consistent in the sampling time dimension. This alignment processing not only improves the availability of multimodal data, but also ensures the accuracy of subsequent feature extraction and analysis, and avoids information loss or misjudgment due to data asynchrony.

[0042] In addition, in the feature extraction and fusion processing stage, the various characteristics of multimodal data are comprehensively utilized to obtain a more comprehensive comprehensive feature vector, and on this basis, dimensionality reduction processing is performed to extract key feature vectors, thereby effectively reducing the data dimension, reducing the computational complexity of the subsequent anomaly monitoring model, and improving the anomaly monitoring efficiency. At the same time, the key information of the data is retained as much as possible to ensure the accuracy of anomaly detection.

[0043] In a third aspect, an embodiment of the present application provides a computer device, including:

[0044] A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the abnormal alarm method for energy storage power station based on multimodal data described in the above embodiment by executing the computer instructions. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the specific implementation methods of the present application or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0046] Figure 1 A schematic diagram of a flow chart of an abnormal alarm method for an energy storage power station based on multimodal data proposed in an embodiment of the present application;

[0047] Figure 2 A schematic diagram of the process of feature extraction, fusion and dimensionality reduction proposed in the embodiment of the present application;

[0048] Figure 3 A schematic diagram of the structure of an abnormal alarm device for an energy storage power station based on multimodal data proposed in an embodiment of the present application;

[0049] Figure 4 The present invention is a schematic diagram of the structure of a computer device used to execute the abnormal alarm method of an energy storage power station proposed in an embodiment of the present application. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0051] As an important infrastructure of the new power system, energy storage technology is of great significance in promoting green energy transformation, responding to extreme events, ensuring energy supply security, and promoting high-quality development of new energy. As an important part of the stability of the power grid, the safety and reliability of energy storage power stations are increasingly valued. Energy storage power stations store electrical energy through a large number of battery units and release electrical energy during peak demand to balance the load of the power grid. However, energy storage power stations face many safety risks during operation, such as battery overheating, fire, chemical leakage, etc. In order to ensure the safe operation of energy storage power stations, they must be monitored and warned in a comprehensive and real-time manner. In the monitoring and early warning scenarios of energy storage power stations, traditional monitoring often relies on a single modality data source, such as relying solely on temperature sensors and smoke alarms, which can easily lead to false alarms or missed alarms and cannot fully reflect the true state of the system.

[0052] With the development of battery management systems (BMS), sensors and other data acquisition equipment, multimodal data acquisition in energy storage power stations has become possible, and different types of sensors can provide richer and more diverse information. However, the fusion and processing of different multimodal data face many technical challenges. For example, different data acquisition devices are affected by factory settings and may have different data acquisition frequencies. Such asynchronous multimodal data in the time dimension is difficult to be fused and applied in the future.

[0053] To this end, the embodiments of the present application can be applied to the supercomputing data middle platform, which is the data processing center of the security monitoring, early warning and control platform. Through the integrated data acquisition capability, it gathers various types of data from the energy storage power station to the storage layer, which is suitable for centralized and efficient processing of massive and scattered data to avoid data islands. The energy storage power station abnormal alarm method based on multimodal data proposed in the embodiment of the present application is deployed on this supercomputing data middle platform. The data of different data acquisition devices are cleaned, processed, converted in real time, and feature fused to form multimodal data, thereby realizing real-time operation monitoring of the energy storage power station and issuing alarms safely and promptly for abnormal operation.

[0054] The embodiment of the present application relies on multimodal data collected in real time by multiple data collection devices, and integrates them to obtain comprehensive feature vectors. By training a machine learning-based detector with positive and negative samples, the real-time feature vectors are received to achieve real-time detection and early warning of energy storage power station scenarios. The complementarity of various types of data is fully utilized to improve the overall monitoring capability and early warning accuracy of the system.

[0055] According to an embodiment of the present application, an embodiment of an abnormal alarm method for an energy storage power station based on multimodal data is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0056] In this embodiment, a method for abnormal alarm of an energy storage power station based on multimodal data is provided, which can be used in the above-mentioned super computing power data center. Figure 1 is a flow chart of an abnormal alarm method for an energy storage power station according to an embodiment of the present application, such as Figure 1 As shown, the process includes the following steps:

[0057] Step S1, obtaining original sampling data of the energy storage power station, and performing sampling frequency alignment processing on the original sampling data based on a resampling frequency to obtain multimodal data.

[0058] In this step, since the real-time sampling frequencies of the modal data are different, in order to align these data in time, a unified resampling frequency is required so that the multimodal data have the same time reference.

[0059] Step S3, performing feature extraction and fusion processing on the multimodal data to obtain a comprehensive feature vector, and performing dimensionality reduction processing on the comprehensive feature vector to obtain a key feature vector.

[0060] In the feature extraction and fusion processing stage, the various characteristics of multimodal data are comprehensively utilized to extract a comprehensive feature vector to comprehensively reflect the operating status of the energy storage power station. At the same time, considering that the comprehensive feature vector may contain a large amount of feature information, but some of the information may be redundant, the most important information in the multimodal data is retained through dimensionality reduction processing to improve the processing efficiency and generalization ability of the subsequent safety monitoring model.

[0061] Step S5: input the key feature vector into a pre-trained abnormality monitoring model, and generate an alarm signal when an abnormal operation of the energy storage power station is detected according to the output result of the abnormality monitoring model.

[0062] The operation of the energy storage power station is monitored in real time through the abnormal monitoring model, where the output of the abnormal monitoring model can be the classification probability of the normal operation and abnormal operation of the energy storage power station. When the abnormal operation of the energy storage power station is detected, an alarm signal will be generated immediately, and the alarm signal can notify the corresponding management personnel in various ways to respond to the abnormal operation.

[0063] The energy storage power station abnormal alarm method provided in this embodiment relies on multiple data acquisition devices to collect and monitor multimodal data related to the operating status of the energy storage power station in real time, and realizes alignment processing of the original sampled data by setting the resampling frequency, so that the multimodal data from different data sources are consistent in the sampling time dimension. This alignment processing not only improves the availability of multimodal data, but also ensures the accuracy of subsequent feature extraction and analysis, and avoids information loss or misjudgment due to data asynchrony.

[0064] In addition, in the feature extraction and fusion processing stage, the various characteristics of multimodal data are comprehensively utilized to obtain a more comprehensive comprehensive feature vector, and on this basis, dimensionality reduction processing is performed to extract key feature vectors, thereby effectively reducing the data dimension, reducing the computational complexity of the subsequent anomaly monitoring model, and improving the anomaly monitoring efficiency. At the same time, the key information of the data is retained as much as possible to ensure the accuracy of anomaly detection.

[0065] In an embodiment of the present application, the sampling frequency alignment processing is performed on the original sampled data based on the resampling frequency to obtain multimodal data, including the following steps:

[0066] Step S110, obtaining the original sampling frequencies of the original sampled data, and taking the lowest frequency among the original sampling frequencies as the resampling frequency;

[0067] Step S120: adjusting the original sampled data based on the resampling frequency to obtain multimodal data with the same sampling frequency.

[0068] Further, the adjusting the original sampled data based on the resampling frequency comprises the following steps:

[0069] Step S121, determining a resampling period corresponding to the resampling frequency;

[0070] Step S122, when there is only one original sampled data in the resampling period, directly use the original sampled data as the multimodal data;

[0071] Step S123: When there are at least two original sampled data in the resampling period, an average value of all the original sampled data in the resampling period is used as the multimodal data.

[0072] The data sources of the original sampling data in the embodiments of the present application are different, so there may be multiple original sampling frequencies, which are respectively recorded as F1, F2, F3, ..., F n etc. Select the lowest frequency F among these original sampling frequencies i As the resampling frequency, to ensure that in the corresponding single resampling period S i Within, the original sampling data of each data source can sample at least one data.

[0073] Specifically, determine the resampling period S corresponding to the resampling frequency i =1 / F i , if there is only one original sample data in S, then the original sample data is directly used as the resampling result. i If there are multiple original sampled data, the average of these data points is taken as the resampling result. In the embodiment of the present application, the high-frequency sampled data is downsampled by taking the average value to achieve synchronization of all multimodal data in the time dimension.

[0074] In an embodiment of the present application, the original sampling data is obtained through various data acquisition devices in the energy storage power station, wherein the data acquisition device includes at least one of a battery management system, a smoke sensor, a temperature sensor, and a micro-nano particle sensor.

[0075] Specifically, the first multimodal data obtained by the battery management system includes parameters such as the battery cell voltage, battery cell current, and battery cell temperature. These data are used to reflect the battery performance in the energy storage power station. Abnormal changes in the battery cell voltage may indicate problems such as battery performance degradation, short circuit or overcharging, and long-term abnormal voltage may also cause battery overheating, thereby damaging the battery or causing a safety accident. Abnormal current may indicate abnormal load, line short circuit, connection problems, etc. In addition, sudden large current changes may be a signal of battery overload or internal battery failure. The increase in battery temperature may be caused by overcharging, over-discharging, increased internal resistance or excessive external ambient temperature. If the temperature exceeds the safe range, it may cause thermal runaway of the battery, causing a serious accident.

[0076] The second multimodal data acquired by the smoke sensor includes the smoke concentration in the air of the energy storage power station. Increased smoke concentration in the energy storage power station is usually a precursor to fire, which may be caused by battery failure, line short circuit, overheating of electrical equipment, etc. Even if the smoke concentration does not reach the alarm standard, a sudden change in concentration may be an early sign of potential fire risk.

[0077] The third multimodal data obtained by the temperature sensor includes the ambient temperature of the energy storage power station. The operating performance of the energy storage power station is closely related to the ambient temperature. Excessive ambient temperature may affect the heat dissipation capacity of the battery and the equipment, resulting in an increase in the operating temperature of the equipment. Conversely, abnormal changes in ambient temperature may also be the result of problems such as equipment overheating or external fire.

[0078] The fourth multimodal data obtained by the micro-nano particle sensor includes the concentration of micro-nano particles in the air of the energy storage power station. The abnormal increase in the concentration of micro-nano particles in the air may indicate problems such as material combustion, material aging, chemical reaction or external environmental pollution. Such changes can be a signal of internal equipment failure in the energy storage power station, such as battery leakage, material combustion, etc.

[0079] Figure 2 It shows the feature extraction and fusion based on the above multimodal data, as well as the subsequent abnormality monitoring process, which is described in detail below.

[0080] The extracting and fusing the multimodal data to obtain a comprehensive feature vector includes:

[0081] For the data acquired by the battery management system and after the alignment processing, the change indicators of the cell voltage, cell current and cell temperature are extracted as the first type of features. The first type of features includes the following indicators:

[0082] Cell voltage change rate ΔV: ΔV = V t -V t-1 ;

[0083] Average voltage

[0084] Voltage fluctuation rate σ V :

[0085] Current change rate ΔI: ΔI=I t -I t-1 ;

[0086] Average current

[0087] Current fluctuation rate σ I :

[0088] Cell temperature fluctuation rateσ T,电芯 :

[0089] Temperature coefficient of variation:

[0090] In the above formula, V t is the cell voltage in the tth resampling period, V t-1 is the cell voltage in the t-1th resampling period, V i is the cell voltage in the i-th resampling cycle, N is the total number of resampling cycles, I t is the cell current in the tth resampling period, I t-1 is the cell current in the t-1th resampling period, I i is the cell current in the i-th resampling period, T i is the cell temperature in the i-th resampling period, is the average cell temperature of all resampling cycles.

[0091] For the data acquired by the smoke sensor and after alignment processing, the average smoke concentration, the maximum smoke concentration and the number of smoke concentration mutations are extracted as the second type of features. The first type of features includes the following indicators:

[0092] Average concentration

[0093] Maximum concentration C max ;

[0094] Number of sudden changes in smoke density N 浓度突变 ;

[0095] In the above formula, C i is the smoke concentration in the i-th resampling period.

[0096] The number of smoke concentration mutations is obtained according to the following steps, including: respectively obtaining the difference in smoke concentration between two adjacent sampling periods, and counting the number of times the difference exceeds a preset threshold in several consecutive sampling periods to obtain the number of smoke concentration mutations.

[0097] In the embodiment of the present application, the first preset threshold ΔC is determined based on actual experience and historical data th , calculate the concentration change ΔC for every two consecutive sampling points i =C i -C i-1 , if ΔC i The absolute value exceeds the threshold ΔC th , it is determined as a sudden change in smoke concentration. In an example of the embodiment of the present application, the number of concentration sudden changes is counted within 10 resampling cycles.

[0098] For the data acquired by the temperature sensor and after alignment processing, the average value, maximum value and fluctuation rate of the ambient temperature of the energy storage power station are extracted as the third type of features. The third type of features includes the following indicators:

[0099] Average temperature

[0100] Maximum temperature T max,场站 ;

[0101] Temperature fluctuation rate σ T,场站 :

[0102] In the above formula, T 场站i is the ambient temperature in the i-th sampling period.

[0103] For the data acquired by the micro-nano particle sensor and after alignment processing, the average concentration of micro-nano particles, the maximum concentration of micro-nano particles and the number of micro-nano particle concentration mutations are extracted as the fourth type of features. The fourth type of features includes the following indicators:

[0104] Average concentration

[0105] Maximum concentration P max

[0106] Concentration fluctuation rate σ P :

[0107] In the above formula, P i is the concentration of micro-nanoparticles in the i-th sampling period.

[0108] The first type of features, the second type of features, the third type of features and the fourth type of features are concatenated to obtain the comprehensive feature vector

[0109]

[0110] Through the above feature extraction and fusion processing, more comprehensive comprehensive feature vectors are obtained. These comprehensive feature vectors can better describe the operating status of the energy storage power station, thereby providing more reliable input for subsequent abnormal monitoring and realizing more comprehensive safety monitoring of the energy storage power station.

[0111] Figure 2 The process of fusion and dimensionality reduction based on the above feature extraction process is shown. On this basis, the dimensionality reduction process of the comprehensive feature vector is performed to obtain the key feature vector, including:

[0112] First, the comprehensive feature vector is normalized to obtain a normalized feature vector. Specifically, the normalization process makes the mean of each feature 0 and the variance 1, thereby eliminating the influence of different dimensions on model training and prediction.

[0113] Subsequently, the standardized feature vectors are analyzed based on the principal component analysis method to obtain a projection matrix. The specific calculation process is as follows:

[0114] Calculate the covariance matrix of the standardized feature vectors. The standardized data matrix is X′, and its covariance matrix calculation formula is:

[0115]

[0116] where n is the number of samples.

[0117] Perform eigenvalue decomposition on the covariance matrix to obtain eigenvalues and eigenvectors:

[0118] C = PΛP -1

[0119] where Λ is the diagonal matrix of eigenvalues, and P is the matrix of eigenvectors.

[0120] Select the eigenvectors corresponding to the top k largest eigenvalues to form the projection matrix:

[0121] W k = [v1, v2, …, v k .

[0122] Finally, project the standardized feature vectors onto the projection matrix to obtain the key feature vectors, that is, obtain the key feature vectors after dimensionality reduction

[0123] Z = X′W k .

[0124] In the embodiment of the present application, the principal component analysis method is used to achieve feature dimensionality reduction. Principal component analysis is a statistical method that transforms a set of variables that may be correlated through an orthogonal transformation into a set of linearly uncorrelated variables. This set of transformed variables is called the principal components. The idea is to map the n-dimensional features onto the m-dimensional space (m < n). These m dimensions are new orthogonal features, called the principal components. These m-dimensional features are reconstructed and not simply obtained by subtracting the n - m-dimensional features from the n-dimensional features. The core idea of PCA is to project the data along the maximum direction, making the data easier to distinguish.

[0125] Through the principal component analysis method, after the comprehensive feature vectors are standardized, covariance matrix calculation, eigenvalue decomposition, and dimensionality reduction processing, the key feature vectors are obtained. Therefore, the feature processing efficiency is improved, the data dimension is reduced, and optimized input data is provided for subsequent classification and anomaly detection algorithms.

[0126] Figure 2The training and application process of the anomaly monitoring model is also shown. The anomaly monitoring model in the embodiment of the present application is a support vector machine (SVM) trained based on a supervised learning algorithm. SVM is a classic supervised learning algorithm used to solve binary and multi-classification problems.

[0127] The support vector machine is trained by positive and negative samples using a supervised learning algorithm. It should be noted that the SVM in the embodiment of the present application adopts a conventional model construction and training method, which will not be repeated here.

[0128] When using support vector machines to monitor abnormal operation of energy storage power stations, if the sample is classified as normal, the multimodal data at the next moment will continue to be received for detection; if the sample is classified as abnormal, a rapid alarm will be issued to promptly notify the energy storage power station operation and maintenance personnel, thereby quickly locating the source of the alarm.

[0129] Accordingly, please refer to Figure 3 The embodiment of the present application provides an abnormal alarm device for an energy storage power station based on multimodal data, and the abnormal alarm device for an energy storage power station includes

[0130] The resampling module 601 is used to obtain the original sampling data of the energy storage power station, and perform sampling frequency alignment processing on the original sampling data based on the resampling frequency to obtain multimodal data;

[0131] A feature extraction module 602 is used to perform feature extraction and fusion processing on the multimodal data to obtain a comprehensive feature vector, and perform dimensionality reduction processing on the comprehensive feature vector to obtain a key feature vector;

[0132] The alarm module 603 is used to input the key feature vector into a pre-trained abnormal monitoring model, and generate an alarm signal when the energy storage power station is detected to have an abnormal operation according to the output result of the abnormal monitoring model.

[0133] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0134] The abnormal alarm device of the energy storage power station in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0135] See also Figure 4 , Figure 4 is a schematic diagram of the structure of a computer device provided in an embodiment of the present application, such as Figure 4 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 4 A processor 10 is taken as an example.

[0136] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0137] The memory 20 stores instructions executable by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiment.

[0138] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0139] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0140] The computer device further comprises a communication interface 30 for the computer device to communicate with other devices or a communication network.

[0141] The embodiment of the present application also provides a computer-readable storage medium. The above method according to the embodiment of the present application can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0142] The embodiment of the present application provides a computer program product, which includes computer instructions, which are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the method of any embodiment of the present application.

[0143] Although the embodiments of the present application are described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present application, and such modifications and variations are all within the scope defined by the appended claims.

[0144] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing the present application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0145] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0146] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0147] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0148] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0149] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0150] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0151] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

[0152] Although the embodiments of the present application have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present application, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. An abnormal alarm method for energy storage power station based on multimodal data, characterized in that: The energy storage power station abnormal alarm method comprises: Acquire original sampling data of the energy storage power station, and perform sampling frequency alignment processing on the original sampling data based on the resampling frequency to obtain multimodal data; Performing feature extraction and fusion processing on the multimodal data to obtain a comprehensive feature vector, and performing dimensionality reduction processing on the comprehensive feature vector to obtain a key feature vector; The key feature vector is input into a pre-trained abnormality monitoring model, and when an abnormal operation of the energy storage power station is detected according to an output result of the abnormality monitoring model, an alarm signal is generated.

2. The abnormal alarm method for energy storage power station based on multimodal data according to claim 1 is characterized in that: The performing sampling frequency alignment processing on the original sampled data based on the resampling frequency to obtain multimodal data includes: Acquire each original sampling frequency of the original sampling data, and use the lowest frequency among the original sampling frequencies as the resampling frequency; The original sampled data is adjusted based on the resampling frequency to obtain multimodal data with the same sampling frequency.

3. The abnormal alarm method for energy storage power station based on multimodal data according to claim 2 is characterized in that: The adjusting the original sampled data based on the resampling frequency includes: Determine a resampling period corresponding to the resampling frequency; When there is one and only one original sampled data in the resampling period, directly using the original sampled data as the multimodal data; In the case that there are at least two original sampling data in the resampling period, an average value of all the original sampling data in the resampling period is used as the multimodal data.

4. The abnormal alarm method for energy storage power station based on multimodal data according to any one of claims 1 to 3, characterized in that: The original sampling data is obtained through various data acquisition devices in the energy storage power station, wherein the data acquisition device includes at least one of a battery management system, a smoke sensor, a temperature sensor, and a micro-nano particle sensor.

5. The abnormal alarm method for energy storage power station based on multimodal data according to claim 4 is characterized in that: The extracting and fusing the multimodal data to obtain a comprehensive feature vector includes: For the data acquired by the battery management system and after the alignment processing, the change indicators of the battery cell voltage, battery cell current and battery cell temperature are extracted as the first type of features; For the data acquired by the smoke sensor and after alignment processing, the average smoke concentration, the maximum smoke concentration and the number of smoke concentration mutations are extracted as the second type of features; For the data acquired by the temperature sensor and after alignment processing, the average value of the ambient temperature, the maximum value of the ambient temperature and the ambient temperature fluctuation rate of the energy storage power station are extracted as the third type of features; For the data acquired by the micro-nanoparticle sensor and after alignment processing, the average concentration of micro-nanoparticles, the maximum concentration of micro-nanoparticles and the number of mutations in the concentration of micro-nanoparticles are extracted as the fourth type of features; The first type of features, the second type of features, the third type of features and the fourth type of features are concatenated to obtain the comprehensive feature vector.

6. The abnormal alarm method for energy storage power station based on multimodal data according to claim 5 is characterized in that: The number of smoke concentration mutations is obtained according to the following steps, including: The difference between the smoke concentrations in two adjacent sampling periods is obtained respectively, and the number of times the difference exceeds a preset threshold in a number of consecutive sampling periods is counted to obtain the number of sudden changes in the smoke concentration.

7. The abnormal alarm method for energy storage power station based on multimodal data according to claim 1 is characterized in that: The step of performing dimensionality reduction processing on the comprehensive feature vector to obtain a key feature vector includes: Performing standardization on the comprehensive feature vector to obtain a standardized feature vector; Analyzing the standardized eigenvector based on principal component analysis to obtain a projection matrix; The normalized eigenvector is projected onto the projection matrix to obtain the key eigenvector.

8. The abnormal alarm method for energy storage power station based on multimodal data according to claim 1 is characterized in that: The anomaly monitoring model is a support vector machine trained based on a supervised learning algorithm.

9. An abnormal alarm device for energy storage power station based on multimodal data, characterized in that: The abnormal alarm device of the energy storage power station comprises: A resampling module, used to obtain original sampling data of the energy storage power station, and perform sampling frequency alignment processing on the original sampling data based on the resampling frequency to obtain multimodal data; A feature extraction module, used for performing feature extraction and fusion processing on the multimodal data to obtain a comprehensive feature vector, and performing dimensionality reduction processing on the comprehensive feature vector to obtain a key feature vector; The alarm module is used to input the key feature vector into a pre-trained abnormality monitoring model, and generate an alarm signal when the energy storage power station is detected to have an abnormal operation according to the output result of the abnormality monitoring model.

10. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the abnormal alarm method for energy storage power station based on multimodal data according to any one of claims 1 to 8 by executing the computer instructions.

Images