Method for monitoring health state of storage battery pack of transformer substation in real time based on big data analysis
By adaptively initializing the particle filter algorithm, the problem that the number of particles and the range of initial values cannot adapt to the complexity of the battery pack is solved, and high-precision real-time health status monitoring and fault early warning of substation battery packs are realized.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- STATE GRID SHAANXI ELECTRIC POWER CO LTD YULIN POWER SUPPLY CO
- Filing Date
- 2025-12-12
- Publication Date
- 2026-04-21
AI Technical Summary
Existing particle filtering algorithms, by fixing the number of particles and the range of initial values during the initialization phase, cannot adapt to the complexity of battery packs under different operating conditions, resulting in decreased prediction accuracy and uneven particle distribution.
By collecting battery monitoring data from the battery pack, the particle filter algorithm is used to cluster and initialize battery state particles. The number of particles and the range of initial values are adjusted according to the data confidence level to achieve adaptive particle filter algorithm initialization.
The adaptability and accuracy of the particle filter algorithm under different operating conditions have been improved, ensuring the uniformity of particle distribution and enabling real-time health status monitoring and timely detection of potential faults in substation battery banks.
Smart Images
Figure CN121899684A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the technical field of substation battery data analysis, specifically to a method for real-time monitoring of the health status of substation battery banks based on big data analysis. Background Technology
[0002] As a critical backup power source for the power system, the health of substation battery banks is crucial to the safe and stable operation of the power grid. Traditionally, monitoring of substation battery banks mainly relies on manual inspections and periodic maintenance, which suffers from low efficiency and poor real-time performance. With the development of big data technology, real-time monitoring methods based on big data analytics have emerged. This method collects and analyzes large amounts of data during battery bank operation to assess the health status of the battery banks in real time, promptly identify potential faults, and provide strong support for the safe operation of substations.
[0003] When using particle filtering algorithms to estimate battery voltage and temperature, a fixed number of particles during the initialization phase may not be suitable for the complexity of the battery pack under different operating conditions. For example, under aging battery conditions or complex operating conditions, a fixed number of particles may not accurately capture the dynamic changes of the system, leading to a decrease in prediction accuracy. Furthermore, if the initial particle value range is not adjusted according to the current operating state, it may result in uneven or inaccurate particle distribution. For instance, an excessively wide initial value range may introduce too much noise, affecting prediction accuracy; an excessively narrow initial value range may cause particle degradation, failing to cover the true state of the system. Summary of the Invention
[0004] To address the technical issues encountered when using particle filtering algorithms to estimate battery voltage and temperature, where a fixed number of particles during the initialization phase may not be suitable for the complexity of battery packs under different operating conditions, and where uneven or inaccurate particle distribution may result if the initial particle value range is not adjusted according to the current operating state, this invention aims to provide a real-time monitoring method for the health status of substation battery packs based on big data analysis. The specific technical solution adopted is as follows: A method for real-time monitoring of the health status of substation battery banks based on big data analysis, the system comprising: Collect battery monitoring data for each battery in the battery pack at each sampling time; the battery monitoring data includes the voltage and temperature data of the battery at each sampling time; The particle filter algorithm is used to convert battery monitoring data at each sampling time into battery state particles. One battery in the battery pack is randomly selected as a reference battery. All battery state particles of the reference battery are clustered to obtain all particle clusters. The density of each particle cluster is obtained based on the number and spatial distribution of battery state particles in each cluster of the reference battery. All sampling times are evenly divided to obtain all sampling time periods. The data confidence level of each sampling time period is obtained based on the distribution of battery state particles of the reference battery in all particle clusters and the density of each particle cluster in each sampling time period. The initial number of particles in each sampling time period in the initialization phase of the particle filter algorithm is corrected based on the data confidence level of the reference battery in each sampling time period, thus obtaining the number of particles in each sampling time period in the initialization phase of the particle filter algorithm. Based on the modulus of each particle cluster, all particle clusters are divided into a first category and a second category. Based on the number of particles in each sampling time period during the initialization phase of the particle filter algorithm, and the distribution of the first and second category particle clusters in each sampling time period, the range of initial particle values for each sampling time period during the initialization phase of the particle filter algorithm is obtained. The particle filter algorithm is initialized based on the number of particles in each sampling time period and the range of initial particle values during the initialization phase of the particle filter algorithm. The initialized particle filter algorithm is then used to monitor the health status of the substation battery bank in real time.
[0005] Furthermore, the method for obtaining the density includes: The density is obtained according to the density calculation formula, which is as follows:
[0006] In the formula, This indicates the density of each particle cluster; This represents the number of battery state particles in each particle cluster. Indicates the first The cell state particle and the first The Euclidean distance of the battery-state particles.
[0007] Furthermore, the methods for obtaining the data confidence level for each sampling time period include: The data confidence level is obtained according to the data confidence level calculation formula, which is as follows:
[0008] In the formula; Indicates the first Confidence of data for each sampling period; Indicates the first The number of particle clusters to which the battery state particles belong in each sampling time period; Indicates the first Battery state particles in the sampling time period are in the first... The number of particle clusters; Indicates the first Number of battery state particles in each sampling time period; Indicates the first The number of battery-state particles in each particle cluster. Indicates the first The density of individual particle clusters; Represents the maximum value function; Represents a quantity function.
[0009] Furthermore, the method for obtaining the number of particles in each sampling time interval during the initialization phase of the particle filter algorithm includes: The particle count is obtained according to the particle count calculation formula, which is shown below:
[0010]
[0011] In the formula, This indicates the first step in the initialization phase of the particle filter algorithm. Number of particles in each sampling time period; This indicates the first step in the initialization phase of the particle filter algorithm. The initial number of particles in each sampling time period; Indicates the first Confidence of data for each sampling period; Maximum confidence level of data across all sampling time periods; This represents the floor function; This represents an exponential function with the natural constant as its base.
[0012] Furthermore, the method for obtaining the first and second classification clusters includes: Calculate the mean modulus of the battery state particles in each particle cluster to obtain the cluster modulus of each particle cluster; classify the cluster modulus of all particle clusters by the Otsu's method to obtain the first cluster and the second cluster.
[0013] Furthermore, the method for obtaining the range of initial particle values during the initialization phase of the particle filter algorithm for each sampling time period includes: The range of particle initial values is obtained according to the formula for calculating the range of particle initial values, which is as follows:
[0014] In the formula, Indicates the first The initial particle values during the initialization phase of the particle filter algorithm in each sampling time period follow the upper limit of the range. This indicates the first step in the initialization phase of the particle filter algorithm. Number of particles in each sampling time period; This indicates the first step in the initialization phase of the particle filter algorithm. The initial number of particles in each sampling time period; This indicates the number of particle clusters in the first category; Indicates the first category of clusters The cluster modulus of each particle cluster; This indicates the number of particle clusters in the second category cluster; Indicates the second category cluster. The cluster modulus of each particle cluster; Indicates the first The initial particle values during the initialization phase of the particle filter algorithm in each sampling time period follow the lower bound of the range. Represents the maximum value function; This represents the minimum value function.
[0015] A real-time monitoring system for the health status of substation battery banks based on big data analysis is provided. The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor. The system is characterized in that the processor executes the computer program to implement the steps of the real-time monitoring method for the health status of substation battery banks based on big data analysis described above.
[0016] A computer-readable storage medium storing a computer program, characterized in that, when executed by a processor, the computer program implements the steps of the above-described method for real-time monitoring of the health status of substation battery banks based on big data analysis.
[0017] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the steps of the above-described method for real-time monitoring of the health status of a substation battery bank based on big data analysis.
[0018] The present invention has the following beneficial effects: This invention acquires battery monitoring data for each battery in a battery pack at each sampling time. Since battery temperature and voltage data reflect the battery's health status, the battery monitoring data includes the battery's voltage and temperature at each sampling time. This invention employs a particle filtering algorithm, thus converting the battery monitoring data at each sampling time into battery state particles. In the particle filtering algorithm, the battery state particles need to be initialized first, including the number of battery state particles and the interval to which the initial values of the battery state particles conform. Therefore, all battery state particles of the reference battery are first clustered. Since the higher the confidence level of the data during the monitored time period, the more reliable the data generated within that time period, and the relatively stable the operating state of the battery pack, the number of battery state particles can be reduced. After establishing the data confidence level for the sampling period, the initialization process of the particle filter algorithm can be adaptively adjusted based on the data confidence level. This initialization process includes initializing the number of battery state particles and defining the range of initial values for each particle. The relationship between these two is that the initial number of battery state particles randomly selects values from the range of initial values. Therefore, the number of battery state particles required for initialization is obtained from the data confidence level, and then the range of initial particle values is defined based on this number. The particle filter algorithm is initialized based on the number of particles and the range of initial particle values for each sampling period during the initialization phase. The initialized particle filter algorithm is then used to monitor the health status of the substation battery bank in real time. When using the particle filter algorithm to estimate battery voltage and temperature, this invention can adapt to the complexity of the battery bank under different operating conditions during the initialization phase, and allows the range of initial particle values to be adjusted according to the current operating state, thereby ensuring uniform particle distribution. Attached Figure Description
[0019] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0020] Figure 1 The flowchart illustrates a method for real-time monitoring of the health status of substation battery banks based on big data analysis, as provided in one embodiment of the present invention. Detailed Implementation
[0021] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a real-time monitoring method for the health status of substation battery banks based on big data analysis proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.
[0022] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.
[0023] The following description, in conjunction with the accompanying drawings, details a specific scheme for a real-time monitoring method for the health status of substation battery banks based on big data analysis, provided by this invention.
[0024] Please see Figure 1 This illustrates an embodiment of the present invention providing a method for real-time monitoring of the health status of substation battery banks based on big data analysis, specifically including: Step S1: Collect battery monitoring data for each battery in the battery pack at each sampling time; the battery monitoring data includes the voltage and temperature data of the battery at each sampling time.
[0025] The embodiments of the present invention are mainly applied to the health status monitoring scenario of each battery in the battery pack of a substation. Since the temperature data and voltage data of the battery can reflect the health status of the battery itself, the voltage data and temperature data of each battery at each sampling time are first obtained, and the voltage data and temperature data at each sampling time are combined to form the battery monitoring data at the current sampling time. That is, each battery in the battery pack can obtain a set of battery monitoring data.
[0026] In one embodiment of the present invention, the sampling time is set to 1 second. It should be noted that in other embodiments of the present invention, the sampling time can be set by itself and is not limited here.
[0027] Step S2: Use the particle filter algorithm to convert the battery monitoring data at each sampling time into battery state particles; randomly select one battery in the battery pack as a reference battery; cluster all battery state particles of the reference battery to obtain all particle clusters; obtain the density of each particle cluster based on the number and spatial distribution of battery state particles in each particle cluster of the reference battery; divide all sampling times equally to obtain all sampling time periods; obtain the data confidence of each sampling time period based on the distribution of battery state particles of the reference battery in all particle clusters and the density of each particle cluster in each sampling time period; correct the initial number of particles in each sampling time period in the initialization phase of the particle filter algorithm based on the data confidence of the reference battery in each sampling time period to obtain the number of particles in each sampling time period in the initialization phase of the particle filter algorithm.
[0028] Particle filtering is a state estimation technique based on the Monte Carlo method. It approximates the probability distribution of a system's state using a set of randomly sampled particles. In particle filtering, each particle represents a possible state of the system. In the current scenario, this system represents the healthy state of a battery, which is determined by a state vector. This includes battery voltage data. and temperature data As a vector parameter, the battery state vector The representation is as follows:
[0029] Therefore, the battery state vector can be... Transformed into battery state particles, the relationship between battery monitoring data and battery state particles is as follows: one battery in a battery pack corresponds to a set of several battery state vectors obtained from battery monitoring data. Each of them This corresponds to a particle in a battery state.
[0030] In particle filtering algorithms, it's necessary to first initialize the battery state particles, including the number of battery state particles and the range within which their initial values conform. Of these two, the number of battery state particles is more important; a larger number leads to a more accurate distribution, thus improving filtering accuracy. However, increasing the number of battery state particles also significantly increases computational cost, as each particle requires state prediction and weight updates. Therefore, we first cluster all battery state particles of the reference battery. The clustering results are then analyzed to determine the density of the particle clusters within the reference battery, thereby analyzing the total number of battery state particles.
[0031] In one embodiment of this invention, the k-means clustering method is used to cluster all battery state particles of the reference battery. The clustering process employs the elbow method, analyzing the intra-cluster squared error by calculating and plotting the error at different cluster numbers. As the number of clusters increases, the intra-cluster squared error gradually decreases. However, when the number of clusters reaches a certain suitable value, the rate of decrease in the intra-cluster squared error slows significantly. At this point, the points with the largest change in the rate of decrease are considered to have the optimal cluster number. Based on this optimal cluster number, the clustering algorithm is executed to complete the clustering process. It should be noted that in other embodiments of this invention, other clustering methods can also be used for clustering, such as the LOF clustering algorithm, which are not limited or elaborated upon here.
[0032] In one embodiment of the present invention, the method for obtaining the density includes: The density is obtained according to the density calculation formula, which is shown below:
[0033] In the formula, This indicates the density of each particle cluster; This represents the number of battery state particles in each particle cluster. Indicates the first The cell state particle and the first The Euclidean distance of the battery-state particles.
[0034] In the density calculation formula, the Euclidean distance between battery-state particles in a particle cluster is... The shorter the cluster, the more battery-state particles it contains. The more particles there are, the denser the particle clusters become.
[0035] The higher the confidence level of the data within the monitored time period, the more reliable the data generated within that period, and the relatively stable the changes in the battery pack's operating state. In this case, the number of battery state particles can be reduced, as a smaller number of particles can already approximate the system state distribution well, thereby reducing computational costs. Conversely, when the data confidence level is low, it indicates that the battery monitoring data may have significant uncertainty or anomalies, and the changes in the system state may be more complex. In this case, the number of battery state particles needs to be increased to improve filtering accuracy and ensure accurate capture of changes in the system state. Therefore, after obtaining the density of each particle cluster, all sampling times are first segmented to facilitate the analysis of the battery health status in different time periods. Thus, in this embodiment of the invention, each sampling time period is set to 1 hour. It should be noted that in other embodiments of the invention, the sampling time period can be set arbitrarily and is not limited here.
[0036] Since the data acquisition frequency is the same (voltage and temperature data are acquired once per second), the number of particle points that can be obtained in each time period is the same. The higher the data confidence level of the time period, the more concentrated the distribution in the cluster, while the more dispersed the distribution. That is, the battery state particles obtained in each sampling time period tend to be dispersed in multiple particle clusters. At this time, the data confidence level of the sampling time period is lower. Therefore, based on the above description, the data confidence level of each sampling time period is defined.
[0037] In one embodiment of the present invention, the method for obtaining the data confidence level for each sampling time period includes: The data confidence score is obtained according to the data confidence score calculation formula, which is shown below:
[0038] In the formula; Indicates the first Confidence of data for each sampling period; Indicates the first The number of particle clusters to which the battery state particles belong in each sampling time period; Indicates the first Battery state particles in the sampling time period are in the first... The number of particle clusters; Indicates the first Number of battery state particles in each sampling time period; Indicates the first The number of battery-state particles in each particle cluster. Indicates the first The density of individual particle clusters; Represents the maximum value function; Represents a quantity function.
[0039] In the formula for calculating data confidence, at the first... In each sampling time period, the smaller the number of particles belonging to the cell cluster of the battery state particles, the better the result. The smaller the particle dispersion trend of the battery state obtained in each sampling time period, the better. The higher the confidence level of the data within a given sampling period, the better; Indicates the first The maximum number of battery state particles in different particle clusters during each sampling time period, where the maximum number of battery state particles in different particle clusters accounts for the percentage of the maximum number of battery state particles in different particle clusters. The higher the proportion of all battery state particles in the sampling time period, the more it indicates that the first sampling period is larger. All battery state particles in the sampling time period tend to be distributed in the same particle cluster, at which point the... The higher the confidence level of the data in each sampling time period, the better the likelihood of the presence of the first data point in the particle cluster. Analyze the particle clusters of battery state particles obtained from each sampling time period, and identify the particles belonging to the first sampling period in each cluster. The ratios of battery state particles in each sampling time period to the battery state particles in that particle cluster are summed. The larger the sum, the stronger the ratio. The higher the confidence level of the battery state particles in the sampling time period, the better if the first sampling time period is... The higher the density of the particle cluster to which the battery state particles belong during the sampling time period, the more likely the battery state particles in this cluster are to change within a very small range. The higher the confidence level of the data within a sampling time period, the greater the confidence level.
[0040] After obtaining the data confidence level for each sampling time period, the initialization process of particles in the particle filter algorithm can be adaptively changed based on the data confidence level. The initialization process includes the initialization of the number of particles in the battery state and the definition of the interval for the initial value of the particles in the battery state. The relationship between the two is that the number of particles initialized is randomly selected from the interval for the initial value as the particle value. Therefore, the number of particles required for initialization is obtained from the data confidence level, and then the interval for the initial value of the particles is defined based on the number of particles. Thus, in this embodiment of the invention, the initial number of particles in each sampling time period in the initialization stage of the particle filter algorithm is corrected based on the data confidence level of the reference battery in each sampling time period, thereby obtaining the number of particles in each sampling time period in the initialization stage of the particle filter algorithm.
[0041] In one embodiment of the present invention, the method for obtaining the number of particles in each sampling time period during the initialization phase of the particle filter algorithm includes: The particle count is obtained using the particle count calculation formula, which is shown below:
[0042] In the formula, This indicates the first step in the initialization phase of the particle filter algorithm. Number of particles in each sampling time period; This indicates the first step in the initialization phase of the particle filter algorithm. The initial number of particles in each sampling time period; Indicates the first Confidence of data for each sampling period; Maximum confidence level of data across all sampling time periods; This represents the floor function; This represents an exponential function with the natural constant as its base.
[0043] In the formula for calculating the number of particles, the first... The ratio between the data confidence level for a given sampling period and the maximum data confidence level across all sampling periods. The larger it is, the more likely it is to be the first The fewer the number of particles in a sampling time period during the initialization phase of the particle filter algorithm, the better. The smaller; and in existing technologies, the first... The range of particle count for each sampling time period is within Between, so utilize Limit the range of values to Between; utilizing During the initialization phase of the particle filter algorithm, the first... The initial number of particles in the sampling time period is adjusted to obtain the initial number of particles in the initialization phase of the particle filter algorithm. Number of particles in each sampling time period.
[0044] In one embodiment of the present invention, the first step in the initialization phase of the particle filter algorithm is... The initial number of particles for each sampling time period is set to 500. It should be noted that the initial number of particles can be set by the user and is not limited here. The initial number of particles is existing technology and will not be elaborated here.
[0045] Step S3: Based on the modulus of each particle cluster, divide all particle clusters into a first category cluster and a second category cluster; based on the number of particles in each sampling time period during the initialization phase of the particle filter algorithm, and the distribution of the particle clusters of the first and second categories in each sampling time period, obtain the range of initial particle values for each sampling time period during the initialization phase of the particle filter algorithm; initialize the particle filter algorithm based on the number of particles in each sampling time period and the range of initial particle values during the initialization phase of the particle filter algorithm; use the initialized particle filter algorithm to monitor the health status of the substation battery bank in real time.
[0046] Based on the number of particles in the initialization phase of the particle filtering algorithm obtained from the above steps, the initial value of the particles is adaptively changed within a range. First, the upper and lower limits of the range of the initial value of the particles are analyzed. In one embodiment of the present invention, based on the modulus of each particle cluster, all particle clusters are divided into a first category cluster and a second category cluster.
[0047] Preferably, in one embodiment of the present invention, the method for obtaining the first classification cluster and the second classification cluster includes: The mean modulus of the battery state particles in each particle cluster is calculated to obtain the cluster modulus of each particle cluster. The cluster modulus of all particle clusters is classified by the Otsu's method to obtain a first cluster and a second cluster. There is a large difference in the modulus of the particle clusters in the first cluster and the second cluster.
[0048] The upper and lower limits of the initial particle value range are determined based on the first and second classification clusters, thereby obtaining the initial particle value range. Preferably, in one embodiment of the present invention, the method for obtaining the initial particle value range during the particle filter algorithm initialization phase for each sampling time period includes: The range of particle initial values is obtained based on the formula for calculating the range of particle initial values, as shown below:
[0049] In the formula, Indicates the first The initial particle values during the initialization phase of the particle filter algorithm in each sampling time period follow the upper limit of the range. This indicates the first step in the initialization phase of the particle filter algorithm. Number of particles in each sampling time period; This indicates the first step in the initialization phase of the particle filter algorithm. The initial number of particles in each sampling time period; This indicates the number of particle clusters in the first category; Indicates the first category of clusters The cluster modulus of each particle cluster; This indicates the number of particle clusters in the second category cluster; Indicates the second category cluster. The cluster modulus of each particle cluster; Indicates the first The initial particle values during the initialization phase of the particle filter algorithm in each sampling time period follow the lower bound of the range. Represents the maximum value function; This represents the minimum value function.
[0050] In the formula for calculating the initial particle value within a range, the proportion of particles in the initialization phase to the maximum particle count is... The larger the value, the more particles there are in the initialization phase, and therefore the larger the range of the interval. This is combined with the mean modulus of the particle points in the first and second classification clusters. and ,Will or As the upper or lower limit of the interval, the upper and lower limits constitute the interval to which the initial value of the particle follows.
[0051] The initialization phase of the particle filter algorithm utilizes a range-bound parameter between the number of particles and their initial values to complete the initialization process. Specifically, the number of particles for each battery state is set to... The initial values of the particles follow a range set to Therefore, the set of particles in the battery state and the range of initial particle values are obtained, as shown below:
[0052] in, Indicates the number of each battery The initial values of the particles in the battery state, the initial values are in Random selection from the list.
[0053] In one embodiment of the present invention, after the particle filter algorithm is initialized, the next step is to predict the future voltage and temperature based on the initialized particle set. The general process is as follows: Real-time acquisition of the current battery pack's voltage and temperature data is used as observations and input into the particle filter. Each time data is sampled, this data is used to update the particle filter state. Specifically, the particle filter predicts the future state by recursively updating the weights and positions of the particles, thus predicting the voltage and temperature at future time points. That is, after acquiring new voltage and temperature data, for each particle, the system model predicts the next voltage and temperature value of the battery. The predicted value is compared with the actual measured voltage and temperature data, the error is calculated, and the particle weights are adjusted according to the error, so that particles with smaller errors have higher weights, thereby gradually approximating the true system state. The particle filter algorithm weights and averages the weights of each particle to obtain the optimal predicted values of voltage and temperature. These predicted values reflect the possible state of the battery pack at the next time point.
[0054] Once the particle filter algorithm can accurately predict the trends of voltage and temperature changes, these predictions can be combined to monitor the health status of the battery pack in real time.
[0055] Abnormal voltage changes (such as excessively low or high voltage) may indicate battery malfunction or degradation. Excessively high or low temperatures can affect battery performance and even cause damage. By combining predicted data with real-time voltage and temperature data collected via a particle filter, significant discrepancies between the predicted and real-time values may indicate unstable battery health or potential problems. If the predicted voltage or temperature deviates from the normal range, the system will automatically trigger an alarm to alert maintenance personnel to potential battery failure risks. Furthermore, the system will consider generating regular health assessment reports, providing a detailed analysis of battery health based on particle filter predictions, to help maintenance personnel make informed maintenance decisions.
[0056] In summary, battery monitoring data for each battery in the battery pack is collected at each sampling time. This data includes voltage and temperature data. A particle filtering algorithm is used to convert the battery monitoring data at each sampling time into battery state particles. One battery in the battery pack is randomly selected as a reference battery. All battery state particles of the reference battery are clustered to obtain all particle clusters. The density of each particle cluster is obtained based on the number and spatial distribution of battery state particles within each cluster. All sampling times are evenly divided to obtain all sampling time periods. The data confidence level for each sampling time period is obtained based on the distribution of battery state particles in all particle clusters and the density of each cluster. The data confidence level of the sampling time period is used to correct the initial number of particles in each sampling time period during the initialization phase of the particle filter algorithm, thus obtaining the particle count for each sampling time period during the initialization phase of the particle filter algorithm. Based on the modulus of each particle cluster, all particle clusters are divided into a first category cluster and a second category cluster. Based on the particle count in each sampling time period during the initialization phase of the particle filter algorithm, and the distribution of the first and second category clusters in each sampling time period, the range of initial particle values for each sampling time period during the initialization phase of the particle filter algorithm is obtained. The particle filter algorithm is initialized based on the particle count and the range of initial particle values for each sampling time period during the initialization phase of the particle filter algorithm. The initialized particle filter algorithm is then used to monitor the health status of the substation battery bank in real time.
[0057] One embodiment of the present invention provides a real-time monitoring system for the health status of substation battery packs based on big data analysis. The system includes a memory, a processor, and a computer program. The memory is used to store the corresponding computer program, and the processor is used to run the corresponding computer program. When the computer program runs in the processor, it can implement the methods described in steps S1-S3.
[0058] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0059] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.
Claims
1. A method for real-time monitoring of the health status of substation battery banks based on big data analysis, characterized in that, The system includes: Collect battery monitoring data for each battery in the battery pack at each sampling time; the battery monitoring data includes the voltage and temperature data of the battery at each sampling time; The particle filter algorithm is used to convert battery monitoring data at each sampling time into battery state particles. One battery in the battery pack is randomly selected as a reference battery. All battery state particles of the reference battery are clustered to obtain all particle clusters. The density of each particle cluster is obtained based on the number and spatial distribution of battery state particles in each cluster of the reference battery. All sampling times are evenly divided to obtain all sampling time periods. The data confidence level of each sampling time period is obtained based on the distribution of battery state particles of the reference battery in all particle clusters and the density of each particle cluster in each sampling time period. The initial number of particles in each sampling time period in the initialization phase of the particle filter algorithm is corrected based on the data confidence level of the reference battery in each sampling time period, thus obtaining the number of particles in each sampling time period in the initialization phase of the particle filter algorithm. Based on the modulus of each particle cluster, all particle clusters are divided into a first category and a second category. Based on the number of particles in each sampling time period during the initialization phase of the particle filter algorithm, and the distribution of the first and second category particle clusters in each sampling time period, the range of initial particle values for each sampling time period during the initialization phase of the particle filter algorithm is obtained. The particle filter algorithm is initialized based on the number of particles in each sampling time period and the range of initial particle values during the initialization phase of the particle filter algorithm. The initialized particle filter algorithm is then used to monitor the health status of the substation battery bank in real time.
2. The method for real-time monitoring of the health status of substation battery banks based on big data analysis according to claim 1, characterized in that, The method for obtaining the density includes: The density is obtained according to the density calculation formula, which is as follows: ； In the formula, This indicates the density of each particle cluster; This represents the number of battery state particles in each particle cluster. Indicates the first The cell state particle and the first The Euclidean distance of the battery-state particles.
3. The method for real-time monitoring of the health status of substation battery banks based on big data analysis according to claim 1, characterized in that, The methods for obtaining the data confidence level for each sampling time period include: The data confidence level is obtained according to the data confidence level calculation formula, which is as follows: ； In the formula; Indicates the first Confidence of data for each sampling period; Indicates the first The number of particle clusters to which the battery state particles belong in each sampling time period; Indicates the first Battery state particles in the sampling time period are in the first... The number of particle clusters; Indicates the first Number of battery state particles in each sampling time period; Indicates the first The number of battery-state particles in each particle cluster. Indicates the first The density of individual particle clusters; Represents the maximum value function; Represents a quantity function.
4. The method for real-time monitoring of the health status of substation battery banks based on big data analysis according to claim 1, characterized in that, The methods for obtaining the number of particles in each sampling time interval during the initialization phase of the particle filter algorithm include: The particle count is obtained according to the particle count calculation formula, which is shown below: ； In the formula, This indicates the first step in the initialization phase of the particle filter algorithm. Number of particles in each sampling time period; This indicates the first step in the initialization phase of the particle filter algorithm. The initial number of particles in each sampling time period; Indicates the first Confidence of data for each sampling period; Maximum confidence level of data across all sampling time periods; This represents the floor function; This represents an exponential function with the natural constant as its base.
5. The method for real-time monitoring of the health status of substation battery banks based on big data analysis according to claim 1, characterized in that, The methods for obtaining the first and second classification clusters include: Calculate the mean value of the battery state particles in each particle cluster to obtain the cluster modulus of each particle cluster; classify the cluster modulus of all particle clusters by the Otsu's method to obtain the first cluster and the second cluster.
6. The method for real-time monitoring of the health status of substation battery banks based on big data analysis according to claim 1, characterized in that, Methods for obtaining the range of initial particle values during the initialization phase of the particle filter algorithm for each sampling time period include: The range of particle initial values is obtained according to the formula for calculating the range of particle initial values, which is as follows: ； ； In the formula, Indicates the first The initial particle values during the initialization phase of the particle filter algorithm in each sampling time period follow the upper limit of the range. This indicates the first step in the initialization phase of the particle filter algorithm. Number of particles in each sampling time period; This indicates the first step in the initialization phase of the particle filter algorithm. The initial number of particles in each sampling time period; This indicates the number of particle clusters in the first category; Indicates the first category of the first class cluster. The cluster modulus of each particle cluster; This indicates the number of particle clusters in the second category cluster; Indicates the second category cluster. The cluster modulus of each particle cluster; Indicates the first The initial particle values during the initialization phase of the particle filter algorithm in each sampling time period follow the lower bound of the range. Represents the maximum value function; This represents the minimum value function.
7. A real-time health status monitoring system for substation battery banks based on big data analysis, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the real-time monitoring method for the health status of substation battery banks based on big data analysis as described in any one of claims 1 to 6.
8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the real-time monitoring method for the health status of substation battery banks based on big data analysis as described in any one of claims 1 to 6.
9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the real-time monitoring method for the health status of substation battery banks based on big data analysis as described in any one of claims 1 to 6.