An intelligent collection system and method using Beidou communication technology

Abstract

The invention belongs to the technical field of data acquisition and transmission, and discloses an intelligent acquisition system and method using Beidou communication technology; the method comprises: acquiring an operation data set of electric power equipment; preprocessing parameters in the operation data set to obtain preliminary processed data; performing feature extraction and dimensionality reduction processing on the preliminary processed data to obtain key feature vectors; encrypting the key feature vectors to obtain encrypted data; based on the key feature vectors, calculating a data importance score and a real-time requirement score; and obtaining the location information of the electric power equipment by using Beidou satellite positioning; dynamically selecting a Beidou satellite communication mode according to the location information of the electric power equipment, the data importance score and the real-time requirement score; using the Beidou satellite communication mode to send the encrypted data to a ground data center through a Beidou communication network; achieving full coverage, wide connection and easy access of electric power data acquisition, and promoting the application of the electric power Internet of Things.

Description

Technical Field

[0001] The present invention relates to the technical field of data acquisition and transmission, and more specifically, to an intelligent acquisition system and method using Beidou communication technology. Background Art

[0002] The patent application with application publication number CN117544872A discloses an efficient power data collection and transmission system, including a data collection module, a data processing module, a data transmission module and a remote control module. The data collection module is used to collect power data of power equipment. The data processing module processes and analyzes the power data collected by the data collection module. The data transmission module transmits the processed data to the remote control module, and the remote control module remotely controls the power equipment. The data collection module collects power equipment operation data in real time through power lines and wireless communication technology to improve the efficiency and accuracy of data collection. The data processing module cleans, analyzes and processes the received data, and stores the processed data in the cloud for subsequent data analysis and mining. The energy efficiency of the power equipment is evaluated by real-time monitoring of the operation data of the power equipment and its operation status is adjusted to achieve optimized operation of the power equipment and improve the overall energy efficiency of the power system.

[0003] However, there are still many problems to be solved in the existing technology in the collection and transmission of power equipment operation data. In the data collection link, due to the wide variety of power equipment and complex and changeable working conditions, sensors are easily interfered, resulting in uneven quality of collected data, frequent noise and abnormal values, which seriously affect the reliability and availability of data. The data of some key equipment is missing and wrong, making it difficult to accurately grasp the equipment status, which is very likely to cause serious safety accidents. In the data transmission link, the power system is widely distributed and the communication environment is poor. Traditional ground communication methods are difficult to meet the requirements of full coverage and high reliability. Data is faced with multiple security threats such as eavesdropping, tampering, and counterfeiting during transmission. At the same time, due to the limitation of communication resources, some important equipment status data and alarm data cannot be transmitted in time, affecting the real-time and effectiveness of power dispatching. In the data application link, although the massive equipment operation data contains rich information, it lacks effective feature extraction and fusion analysis methods, and the data value is difficult to fully explore, resulting in insufficient basis and poor timeliness for some major decisions. Especially in emergency scenarios such as sudden failures and extreme weather, the data support capability is seriously insufficient, affecting the rapid response and coordinated disposal of the power system.

[0004] In view of this, the present invention proposes an intelligent data collection system and method using Beidou communication technology to solve the above problems. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art and to achieve the above-mentioned purpose, the present invention provides the following technical solutions: an intelligent data collection system using Beidou communication technology, comprising: a data collection module for collecting operation data sets of power equipment;

[0006] A data processing module is used to pre-process the parameters in the running data set using an improved grid algorithm to obtain preliminary processed data;

[0007] The feature extraction module is used to perform feature extraction and dimensionality reduction on the preliminary processed data to obtain key feature vectors;

[0008] A data encryption module is used to encrypt key feature vectors to obtain encrypted data;

[0009] The comprehensive transfer module calculates the data importance score and real-time requirement score based on the key feature vectors, and uses Beidou satellite positioning to obtain the location information of power equipment;

[0010] The dynamic transmission module is used to dynamically select the Beidou satellite communication mode according to the location information, data importance score and real-time requirement score of the power equipment; use the Beidou satellite communication mode to send the encrypted data to the ground data center through the Beidou communication network; and each module is connected by wired and / or wireless means.

[0011] Furthermore, the parameters in the operating data set include basic electrical parameters, mechanical parameters and power quality parameters; the basic electrical parameters include voltage, current, power and frequency; the mechanical parameters include temperature and vibration; the power quality parameters include harmonics and negative sequence components.

[0012] Furthermore, the method of preprocessing the operation data includes:

[0013] Construct a two-dimensional discrete grid space and define the total number of levels of the two-dimensional discrete grid space, corresponding to the grouping of parameters, and assign a unique identifier to each level; construct a grid coordinate system for each level; select parameters in any grouping as the x-axis and y-axis of the grid coordinate system; define the number of grids for each axis in the x-axis and y-axis, and adjust the number of grids to a power of 2; assign data points in the time series data of the parameters to the corresponding grid coordinate system of the corresponding level;

[0014] Calculate the correlation coefficient matrix between parameters; identify directly related parameter pairs based on the correlation coefficient matrix; group the directly related parameter pairs as a group;

[0015] The rules for particle motion are defined in a two-dimensional discrete lattice space, including collision rules and propagation rules;

[0016] The collision rule is that when two particles meet in a grid, they follow the law of conservation of momentum and energy; by calculating the momentum and energy before and after the collision, the new velocity direction and magnitude of the particles after the collision are determined; the propagation rule is that the particles move along a straight line between adjacent grids at a constant speed, and the particles move between adjacent grids in random directions and steps, and a Coulomb force interaction force is introduced;

[0017] A displacement threshold is preset, and particles are iteratively updated. For each time step, pairs of particles on all grids are detected, and the collision rule is applied to calculate the new velocity direction and magnitude after the collision. For each particle, its new position in the two-dimensional discrete grid space is updated according to the propagation rule. For each grid, the weighted average of all particles in it is calculated as the smoothed data value of the grid. If the displacement of the particle is less than the preset displacement threshold, the iteration is terminated, otherwise it enters the next iteration; finally, the smoothed data value of the grid node is mapped back to the original data space to obtain the corresponding smoothing parameter; the corresponding smoothing parameter constitutes the preliminary processed data.

[0018] Furthermore, the calculation method of the correlation coefficient matrix includes:

[0019] Create an n×n matrix, where n is the number of parameters, and for each pair of parameters the time series data and , calculate the correlation coefficient between any data points ;

[0020] ;in, for The data points, for The data points, for The average of the data points in for The average value of the data points in The current time and The time difference of the data points; is the attenuation rate, For the The joint weight of the data points;

[0021] ;in, and is an adjustable parameter; fill the correlation coefficient into the corresponding position of the matrix, and the matrix is ​​symmetrical, and the values ​​on the diagonal are all 1; get the correlation coefficient matrix;

[0022] The method of identifying directly related parameter pairs includes:

[0023] Calculate statistics based on the correlation matrix ;in, is the k-order autocorrelation coefficient, is an adjustable parameter, is the Spearman rank correlation coefficient;

[0024] Determine the critical value based on the preset significance level and degrees of freedom; if If it is greater than the critical value, the corresponding two parameters are directly related parameters.

[0025] Furthermore, the method of performing feature extraction and dimensionality reduction processing includes:

[0026] Calculate the statistical characteristics of the preliminary processed data, including mean, variance, peak value and peak factor; perform Fourier transform on the preliminary processed data and extract frequency domain characteristics, including dominant frequency and frequency distribution; use wavelet transform to perform multi-scale decomposition on the preliminary processed data and extract wavelet coefficients of each scale; record the statistical characteristics, frequency domain characteristics and wavelet coefficients of each scale as data characteristics and form a data feature set; use the data features in the data feature set as samples to construct a random forest model;

[0027] For each decision tree in the random forest model, calculate each data feature In the corresponding decision tree Importance score on ;in, Data features For Node The reduction in the Gini coefficient is Data features For Node The reduction in node impurity, Data features For Node The reduction in the residual sum of squares; , and is the weight coefficient of the corresponding item;

[0028] For the random forest model, the mean of the importance scores of all decision trees is taken as the data feature The final importance score of all data features is Sort in descending order and select the first N1 data features as important features, where N1 is an integer greater than 1; use Laplace eigenmap to reduce the dimension of important features to obtain low-dimensional feature data. The low-dimensional feature data is in matrix form and is standardized, that is, each dimensional feature is subtracted from the mean and divided by the standard deviation; perform singular value decomposition on the low-dimensional feature data, extract the singular vectors corresponding to the maximum p singular values, and concatenate the p singular vectors by column to obtain the key feature vectors.

[0029] Furthermore, the method of constructing the random forest model includes:

[0030] The random forest model consists of several decision trees. For each decision tree, several samples are extracted from the data feature set with replacement to form a sample set. At each node of the decision tree, m data features are randomly selected from the sample set to form a feature subset. The optimal split point of the corresponding node is selected based on the feature subset. According to the optimal split point, the current node is split into two child nodes. For the newly generated child nodes, the splitting process is repeated until the maximum tree depth set in advance is reached.

[0031] Construct N decision trees, each of which uses a different sample set, and each decision tree uses a different feature subset when splitting a node; that is, complete the construction of the random forest model;

[0032] The selection method of the optimal segmentation point includes:

[0033] For each data feature, its value range is divided into several adjacent intervals; the value of each data feature will be mapped to the corresponding interval; for each data feature, a histogram is constructed, and for each interval in the histogram, the number of samples contained therein is counted as the statistical information of the corresponding interval; based on the statistical information of each interval, the Gini coefficient or information gain when the corresponding interval is used as the split point is calculated as a scoring indicator; all intervals of each data feature are traversed to find the interval with the largest scoring indicator as the optimal split point.

[0034] Furthermore, the method of encrypting the key feature vector includes:

[0035] Define an elliptic curve E and a base point G, where the elliptic curve E is an algebraic curve defined on a finite field or a binary extension field, and the base point G is a known point on the elliptic curve E; randomly select an integer d as a private key, where d belongs to the interval [1, n1-1], where n1 is the order of the finite point on the elliptic curve E, and calculate the public key Q=d×G based on the private key; represent the key feature vector as a binary string of length k, and map the binary string to a point Pm on the elliptic curve E; encrypt Pm using the public key Q to obtain an encrypted point C=[k]Pm+Q; where [k] represents a scalar multiplication operation on the elliptic curve, and the point Pm is repeatedly added k times; and the encrypted point C is used as the encrypted data.

[0036] Furthermore, the calculation method of the data importance score and the real-time requirement score includes:

[0037] Based on the key feature vectors, the pre-trained machine learning model is used to detect anomalies in the data and obtain the anomaly probability; the anomaly probability is used as the data importance score;

[0038] Analyze the timestamp of the original data corresponding to the key feature vector; pre-configure the corresponding timeliness window ; Based on the timestamp and timeliness window obtained from the analysis , calculate the timeliness deadline ;in The timestamp obtained by analysis; the timeliness deadline and the timestamp obtained by analysis are used to calculate the real-time requirement score ;in, is the current time, and To adjust the parameters.

[0039] Further, the method of dynamically selecting the Beidou satellite communication mode includes:

[0040] According to the location information of the power equipment, query the currently visible Beidou satellite list, and determine the available Beidou communication mode based on the orbital parameters and signal strength of the visible satellites in the currently visible Beidou satellite list; for each available Beidou communication mode, calculate its applicability score ;in, , , , and The weight of the corresponding factors; Score the data for importance. is the corresponding power consumption in Beidou communication mode, is the communication distance, is the reliability score in Beidou communication mode;

[0041] Among all available Beidou communication modes, the Beidou communication mode with the highest applicability score is selected as the final Beidou satellite communication mode.

[0042] An intelligent collection method using Beidou communication technology, which is implemented based on an intelligent collection system using Beidou communication technology, comprises: S1, collecting operation data sets of power equipment;

[0043] S2, using the improved grid algorithm to pre-process the parameters in the running data set to obtain preliminary processed data;

[0044] S3, performing feature extraction and dimensionality reduction processing on the preliminary processed data to obtain key feature vectors;

[0045] S4, encrypting the key feature vector to obtain encrypted data;

[0046] S5. Based on the key feature vector, calculate the data importance score and real-time requirement score; and use Beidou satellite positioning to obtain the location information of the power equipment;

[0047] S6. Dynamically select the Beidou satellite communication mode based on the location information of the power equipment, the data importance score and the real-time requirement score; use the Beidou satellite communication mode to send the encrypted data to the ground data center through the Beidou communication network.

[0048] The technical effects and advantages of the intelligent collection system and method using Beidou communication technology of the present invention are as follows:

[0049] The present invention effectively improves the reliability and representativeness of data through intelligent data preprocessing and feature extraction methods, laying a solid foundation for subsequent analysis and decision-making; secondly, by using encryption algorithms, the confidentiality and integrity of data are guaranteed from the source of data transmission, greatly reducing the risk of data leakage and tampering; thirdly, by comprehensively considering multiple factors such as data characteristics, equipment status, and communication resources, dynamic optimization of satellite communication modes is achieved, which significantly improves the real-time, reliability and flexibility of data transmission; more importantly, the system has built a complete closed-loop solution, which seamlessly connects data collection, secure transmission, intelligent analysis and other links, forming a coordinated connection of data flow, information flow, and business flow, providing strong data support for real-time monitoring, fault diagnosis, operation and maintenance optimization and other applications of power systems; in addition, it makes full use of the navigation, positioning and communication capabilities of Beidou satellites, breaks through the limitations of traditional ground communications, realizes power data collection coverage, wide connection, and easy access, and promotes the construction and application of the power Internet of Things. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 A schematic diagram of an intelligent data collection system using Beidou communication technology according to the present invention;

[0051] Figure 2 It is a schematic diagram of an intelligent collection method using Beidou communication technology of the present invention. DETAILED DESCRIPTION

[0052] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0053] Example 1

[0054] See also Figure 1 As shown, the intelligent data collection system using Beidou communication technology described in this embodiment includes: a data collection module for collecting operation data sets of power equipment;

[0055] A data processing module is used to pre-process the parameters in the running data set using an improved grid algorithm to obtain preliminary processed data;

[0056] The feature extraction module is used to perform feature extraction and dimensionality reduction on the preliminary processed data to obtain key feature vectors;

[0057] A data encryption module is used to encrypt key feature vectors to obtain encrypted data;

[0058] The comprehensive transfer module calculates the data importance score and real-time requirement score based on the key feature vectors, and uses Beidou satellite positioning to obtain the location information of power equipment;

[0059] The dynamic transmission module is used to dynamically select the Beidou satellite communication mode according to the location information of the power equipment, the data importance score and the real-time requirement score; use the Beidou satellite communication mode to send the encrypted data to the ground data center through the Beidou communication network; each module is connected by wired and / or wireless means to realize data transmission between modules.

[0060] The parameters in the operating data set include basic electrical parameters, mechanical parameters and power quality parameters; the basic electrical parameters include voltage, current, power and frequency; the mechanical parameters include temperature and vibration; the power quality parameters include harmonics and negative sequence components; the parameters in the operating data set are all time series data; the corresponding time series data contain several corresponding parameter data points.

[0061] Voltage is measured by a voltage transformer or a voltage sensor; current is measured by a current transformer or a current sensor; power is measured by an electric power transmitter or a power sensor; frequency is measured by a frequency sensor or by performing spectrum analysis on the voltage / current signal; temperature is measured by a temperature sensor; commonly used temperature sensors include thermocouples, thermistors, and semiconductor temperature sensors; vibration is measured by a vibration sensor; commonly used vibration sensors include accelerometers, velocity sensors, and displacement sensors; harmonics are obtained by measuring the harmonic components of the voltage / current; negative sequence components are obtained by calculating the imbalance of the three-phase voltage / current; the above sensors and measuring equipment convert analog signals into digital signals through signal acquisition circuits and data acquisition cards, and upload the data to the data acquisition module through communication interfaces (such as RS485, Modbus, etc.).

[0062] The methods for preprocessing running data include:

[0063] Construct a two-dimensional discrete grid space and define the total number of levels of the two-dimensional discrete grid space, corresponding to the grouping of parameters, and assign a unique identifier to each level; construct a grid coordinate system for each level, such as using a Cartesian coordinate system or a polar coordinate system; select parameters in any group as the x-axis and y-axis of the grid coordinate system, such as voltage-current or power-frequency; define the number of grids for each axis in the x-axis and y-axis, and adjust the number of grids to a power of 2 to optimize the computational efficiency; assign the data points in the time series data of the parameters to the corresponding grid coordinate system of the corresponding level;

[0064] List all collected parameters, calculate the correlation coefficient matrix between the parameters; identify directly related parameter pairs based on the correlation coefficient matrix; and group the directly related parameter pairs as a group.

[0065] The calculation method of the correlation coefficient matrix includes:

[0066] Create an n×n matrix, where n is the number of parameters, and for each pair of parameters the time series data and , calculate the correlation coefficient between any data points ;

[0067] ;in, for The data points, for The data points, for The average of the data points in for The average value of the data points in The current time and The time difference of the data points; As the decay rate, time decay is introduced to make recent data have a greater impact on the correlation. For the The joint weight of the data points reduces the impact of outliers on the overall correlation.

[0068] ;in, and It is an adjustable parameter used to control the impact of outliers; fill the correlation coefficient into the corresponding position of the matrix, and the matrix is ​​symmetrical, and the values ​​on the diagonal are all 1; obtain the correlation coefficient matrix.

[0069] Ways to identify directly related parameter pairs include:

[0070] Calculate statistics based on the correlation matrix ;in, is the k-order autocorrelation coefficient (indicating the correlation between the time series data and its own data lagged by k time units). Taking into account the autocorrelation of the time series data, is an adjustable parameter, It is the Spearman rank correlation coefficient (a non-parametric correlation measure used to measure the monotonic correlation between two variables), which captures possible nonlinear relationships such as power law relationships and exponential relationships.

[0071] Determine the critical value (find the corresponding statistical distribution table) based on the preset significance level (such as 0.05) and degrees of freedom (n-2); if If it is greater than the critical value, the corresponding two parameters are directly related parameters.

[0072] According to the value of each data point, place a particle at the corresponding grid; assign an initial velocity to each particle, which is 0 or a random value; record the position and value of the data point represented by each particle.

[0073] The particle motion rules defined in the two-dimensional discrete lattice space include collision rules and propagation rules; the collision rule is that when two particles meet in the lattice, they follow the law of conservation of momentum and the law of conservation of energy; by calculating the momentum and energy before and after the collision, the new velocity direction and magnitude of the particles after the collision are determined; in this process, the dissipation parameter is introduced to simulate the partial energy loss of the particles during the collision process, thereby achieving the effect of data smoothing;

[0074] The propagation rule is that particles move between adjacent grids along a straight line at a constant speed, and particles move between adjacent grids in random directions and steps. A Coulomb force interaction is introduced, which makes there an attraction or repulsion between particles, thereby realizing the aggregation or dispersion of local data.

[0075] A displacement threshold is preset and particles are iteratively updated. Specifically, for each time step, pairs of particles on all grids are detected, and the collision rule is applied to calculate the new velocity direction and magnitude after the collision. For each particle, its new position in the two-dimensional discrete grid space is updated according to the propagation rule. For each grid, the weighted average of all particles in it is calculated as the smoothed data value of the grid. If the displacement of the particle is less than the preset displacement threshold, the iteration is terminated, otherwise it enters the next iteration.

[0076] In the iterative process, data smoothing and denoising are achieved through the movement, collision and redistribution of particles; finally, the smoothed data values ​​of the grid nodes are mapped back to the original data space to obtain the corresponding smoothing parameters; the corresponding smoothing parameters constitute the preliminary processed data.

[0077] By adjusting the particle motion rules (such as collision dissipation parameters, gravitational constants, etc.) and initialization conditions, the degree of smoothing and denoising is controlled to better adapt to different data characteristics.

[0078] Methods for feature extraction and dimensionality reduction include:

[0079] Calculate the statistical characteristics of the preliminary processed data, including mean, variance, peak and peak factor; perform Fourier transform on the preliminary processed data and extract frequency domain characteristics, including dominant frequency and frequency distribution; use wavelet transform to perform multi-scale decomposition on the preliminary processed data and extract wavelet coefficients of each scale; record the statistical characteristics, frequency domain characteristics and wavelet coefficients of each scale as data features and form a data feature set; use the data features in the data feature set as samples to construct a random forest model.

[0080] Specifically, the random forest model is composed of several decision trees. For the construction of each decision tree, several samples are extracted from the data feature set with replacement to form a sample set. In this sample set, about 63.2% of the samples are original samples, and the remaining 36.8% are repeated samples. At each node of the decision tree, m data features are randomly selected from the sample set to form a feature subset. The optimal split point of the corresponding node is selected based on the feature subset. According to the optimal split point, the current node is split into two child nodes. For the newly generated child nodes, the splitting process is repeated until the maximum tree depth set in advance is reached.

[0081] Construct N decision trees, each using a different sample set, and each using a different feature subset when splitting a node; N is usually set to several hundred to thousands to obtain sufficient generalization ability; that is, the construction of the random forest model is completed.

[0082] The optimal split point selection methods include:

[0083] For each data feature, its value range is divided into several adjacent intervals (histogram intervals); the division methods include equal width, equal frequency or based on data quantiles; the value of each data feature will be mapped to the corresponding interval; for each data feature, a histogram is constructed, and for each interval in the histogram, the number of samples (data features) contained therein is counted as the statistical information of the corresponding interval; based on the statistical information of each interval, the Gini coefficient or information gain when the corresponding interval is used as the split point is calculated; as a scoring indicator; traverse all intervals of each data feature, and find the interval with the largest scoring indicator as the optimal split point, and the value of the split point corresponding to the optimal split point is the lower boundary or median of the optimal split point.

[0084] For each decision tree in the random forest model, calculate each data feature In the corresponding decision tree Importance score on ;in, Data features For Node The reduction in the Gini coefficient is Data features For Node The reduction in node impurity, Data features For Node The reduction in the residual sum of squares; , and is the weight coefficient of the corresponding item, which is used to balance the weights of different importance measures;

[0085] For the random forest model, the mean of the importance scores of all decision trees is taken as the data feature The final importance score of all data features is Sort in descending order and select the first N1 data features as important features, where N1 is an integer greater than 1;

[0086] The feature importance calculated by the random forest model can well capture the interactions and redundant relationships between features, thereby selecting the feature subset that is most relevant and important to the prediction target.

[0087] The Laplace eigenmap is used to reduce the dimensionality of important features (find the low-dimensional manifold structure where the data is located) to obtain low-dimensional feature data, and key feature vectors are generated based on the low-dimensional feature data.

[0088] Laplacian Eigenmaps is a manifold learning algorithm used to discover low-dimensional manifold structures in high-dimensional data sets. It can project high-dimensional data into low-dimensional space while retaining the local neighborhood structure of the data, thereby achieving dimensionality reduction and visualization of the data.

[0089] The low-dimensional feature data is in matrix form and is standardized, that is, the mean of each dimensional feature is subtracted and divided by the standard deviation. Singular value decomposition (SVD) is performed on the low-dimensional feature data to extract the singular vectors corresponding to the maximum p singular values, and the p singular vectors are concatenated by columns to obtain the key feature vectors. The key feature vectors contain the core feature information of the data and greatly reduce the computational complexity of subsequent analysis.

[0090] According to the nature of the data and the purpose of analysis, the corresponding parameters are adjusted to obtain the best feature representation; at the same time, the feature engineering is optimized; the key feature vector finally obtained will lay the foundation for subsequent encryption, transmission and analysis.

[0091] Methods for encrypting key feature vectors include:

[0092] Define an elliptic curve E and a base point G. The elliptic curve E is an algebraic curve defined on a finite field (an algebraic structure containing only a finite number of elements) or a binary extension field (when the number of elements is a power of 2, it is called a binary extension field). The base point G is a known point on the elliptic curve E. Randomly select an integer d as the private key, d belongs to the interval [1, n1-1], where n1 is the order of the finite point on the elliptic curve E, and calculate the public key Q=d×G based on the private key.

[0093] The key feature vector is represented as a binary string of length k, and the binary string is mapped to a point Pm on the elliptic curve E; the mapping method adopts the Koblitz method or the probabilistic addition chain method, etc.; Pm is encrypted with the public key Q to obtain the encrypted point C=[k]Pm+Q; wherein [k] represents the scalar multiplication operation on the elliptic curve, and the point Pm is repeatedly added k times; the encrypted point C is transmitted as the encrypted data.

[0094] The calculation method of data importance score and real-time requirement score includes:

[0095] Based on the key feature vectors, a pre-trained machine learning model (such as a neural network, support vector machine, etc.) is used to detect anomalies in the data and obtain the anomaly probability. The anomaly probability is used as the data importance score. The higher the score or probability, the more abnormal the data is and the higher its importance.

[0096] Specifically, a large number of historical operating data sets of power equipment are collected, including data under normal and abnormal conditions; these original data are preprocessed, feature extracted and dimension reduced to obtain the corresponding key feature vectors as sample data; at the same time, each sample data is labeled normal or abnormal according to whether the power equipment is abnormal; the labeled sample data are randomly divided into training set and test set according to a certain ratio (such as 7:3); the training set is used for model training, and the test set is used for model evaluation.

[0097] The machine learning model is a support vector machine. Specifically, a kernel function (such as RBF kernel) and kernel parameters are selected, key feature vectors are used as input features, abnormal / normal labels are used as outputs, training data are constructed, and the dual optimization problem of the support vector machine is solved to obtain a classification hyperplane.

[0098] Evaluate the performance indicators of the machine learning model on the test set, such as precision, recall, F1 score, area under the ROC curve, etc., to determine whether the machine learning model meets the requirements; if not, adjust the model parameters and retrain; for models with good performance, save their parameters and structure, and obtain a trained machine learning model.

[0099] It should be noted that the above training process is an iterative process, and the model needs to be continuously adjusted according to the evaluation indicators to obtain the best anomaly detection performance.

[0100] Analyze the timestamp of the original data corresponding to the key feature vector; pre-configure the corresponding timeliness window for different types and working conditions of power equipment ; The timeliness window is obtained by analyzing historical data, indicating that the data has reference value within this time window; based on the timestamp and timeliness window obtained by analysis , calculate the timeliness deadline ;in The timestamp obtained by analysis; the timeliness deadline and the timestamp obtained by analysis are used to calculate the real-time requirement score ;in, is the current time, and To adjust the parameters, the dynamic change and change rate of the timeliness window are controlled respectively; the real-time requirement score is between 0 and 1, and the closer to 1, the higher the real-time requirement.

[0101] The Beidou satellite navigation and positioning system is used to obtain the current geographical coordinates (longitude, latitude, etc.) of the power equipment. Combined with geographic information system (GIS) data, the regional location information of the equipment (such as province, city, county, substation, etc.) is obtained. The location information is used to determine the destination and path selection for data transmission. The encrypted data is used to calculate the data importance score and real-time requirement score, and the precise location information of the power equipment is obtained to provide a basis for the subsequent selection of Beidou satellite communication mode.

[0102] The ways to dynamically select Beidou satellite communication mode include:

[0103] According to the location information of the power equipment, query the currently visible Beidou satellite list, and determine the available Beidou communication mode based on the orbital parameters and signal strength of the visible satellites in the currently visible Beidou satellite list; specifically, query the orbital parameters of each visible satellite from the Beidou ephemeris data (such as broadcast ephemeris or precise ephemeris), including satellite position (coordinates), satellite speed (vector) and time coefficient (used to calculate satellite clock deviation); use the satellite position and the position of the power equipment to calculate the geometric distance of the signal from the satellite to the receiving end, and calculate the signal propagation delay according to the signal propagation speed (speed of light); the signal strength mainly depends on the propagation distance and the satellite transmission power, and query the transmission power of each satellite; estimate the signal strength at the receiving end based on the geometric distance and transmission power; at the same time, it is necessary to consider the attenuation effects of factors such as the atmosphere, ionosphere, and terrain on the signal and correct the signal strength.

[0104] Beidou communication modes have different requirements for communication distance, bandwidth, delay, power and other parameters. The propagation delay and signal strength of the signal from each visible satellite are used to determine whether the conditions for using a certain Beidou communication mode are met. If all conditions for a certain Beidou communication mode are met, the Beidou communication mode is listed as an available Beidou communication mode. Common modes include short message communication mode, data transmission mode, etc.

[0105] For each available BeiDou communication mode, calculate its suitability score ;in, , , , and The weights of the corresponding factors are adjusted according to needs; Score the data for importance. is the corresponding power consumption in Beidou communication mode, is the communication distance, is the reliability score in Beidou communication mode.

[0106] It should be noted that the power consumption evaluation takes into account the power consumption of the communication mode under different working states, including transmitting power, receiving power and idle power. Specifically, query the technical specifications of the communication mode to obtain the nominal power value under each working state; estimate the time proportion of different working states, such as transmitting time proportion, receiving time proportion, idle time proportion, etc.; calculate the weighted average power (transmitting power × transmitting time proportion + receiving power × receiving time proportion + idle power × idle time proportion), and convert the weighted average power into energy consumption (weighted average power × expected communication time); according to the battery capacity of the power equipment, calculate the ratio of energy consumption to battery capacity as the power consumption fraction (1 minus energy consumption divided by battery capacity).

[0107] Query the technical specifications of the communication mode to obtain nominal bit error rate, packet loss rate and other indicators; estimate the interference intensity of the current communication environment, including the influence of other radio signals, weather, terrain and other factors; according to the interference intensity, correct the nominal bit error rate and packet loss rate to obtain the actual bit error rate and packet loss rate; the reliability score is 1 minus the actual bit error rate minus the actual packet loss rate.

[0108] Among all available Beidou communication modes, the Beidou communication mode with the highest applicability score is selected as the final Beidou satellite communication mode; the encrypted data is sent to a preset data center using the selected Beidou satellite communication mode.

[0109] This process comprehensively considers multiple factors such as data importance, real-time requirements, equipment location, available communication resources, etc., and can dynamically select the current optimal Beidou satellite communication mode, improve communication efficiency and reliability, and meet different data transmission needs; at the same time, the process also incorporates the calculation method of suitability scoring to make decisions more objective and quantitative.

[0110] According to the selected Beidou satellite communication mode, the corresponding communication parameters are configured, such as working frequency, transmission power, modulation and demodulation mode, etc., and a link request is sent to the Beidou satellite to establish a satellite communication link; the encrypted data C is encapsulated according to a predetermined format to generate a data packet, and control information such as a synchronization field and an address field are added to the header of the data packet, and error detection coding is performed on the data packet to improve transmission reliability; the established satellite communication link is used to send the data packet to the Beidou satellite in accordance with the specifications of the Beidou satellite communication mode; after receiving the data packet, the Beidou satellite performs basic processing such as demodulation and error detection; according to the address information in the header of the data packet, the data packet is forwarded to the ground data center; after receiving the data packet, the data center performs operations such as synchronization detection and address recognition, caches and reorganizes an ordered sequence of data packets; the ground data center distributes the data packet to the corresponding application server according to the destination address information in the data packet, and the application server decrypts and parses the data packet to obtain the original parameters, stores the parameters and performs subsequent monitoring, analysis, decision-making and other processing.

[0111] During the entire process, the continuity and stability of the communication link are guaranteed to prevent data packet loss or disorder. The process fully utilizes the advantages of the Beidou satellite communication network such as wide area coverage and high reliability to achieve safe and efficient transmission of power equipment operation data, providing strong support for real-time monitoring and fault diagnosis of the power system.

[0112] This embodiment effectively improves the reliability and representativeness of the data through intelligent data preprocessing and feature extraction methods, laying a solid foundation for subsequent analysis and decision-making; secondly, by using encryption algorithms, the confidentiality and integrity of the data are guaranteed from the source of the data, greatly reducing the risk of data leakage and tampering; thirdly, by comprehensively considering multiple factors such as data characteristics, equipment status, and communication resources, dynamic optimization of satellite communication modes is achieved, which significantly improves the real-time, reliability and flexibility of data transmission; more importantly, the system has built a complete closed-loop solution, which seamlessly connects data collection, secure transmission, intelligent analysis and other links, forming a coordinated connection of data flow, information flow, and business flow, providing strong data support for real-time monitoring, fault diagnosis, operation and maintenance optimization and other applications of the power system; in addition, it fully utilizes the navigation, positioning and communication capabilities of Beidou satellites, breaks through the limitations of traditional ground communications, and realizes power data collection coverage, wide connection, and easy access, promoting the construction and application of the power Internet of Things.

[0113] Example 2

[0114] See also Figure 2 As shown, the part not described in detail in this embodiment is described in Example 1, which provides an intelligent collection method using Beidou communication technology, including:

[0115] S1. Collecting operation data sets of power equipment;

[0116] S2, using the improved grid algorithm to pre-process the parameters in the running data set to obtain preliminary processed data;

[0117] S3, performing feature extraction and dimensionality reduction processing on the preliminary processed data to obtain key feature vectors;

[0118] S4, encrypting the key feature vector to obtain encrypted data;

[0119] S5. Based on the key feature vector, calculate the data importance score and real-time requirement score; and use Beidou satellite positioning to obtain the location information of the power equipment;

[0120] S6. Dynamically select the Beidou satellite communication mode based on the location information of the power equipment, the data importance score and the real-time requirement score; use the Beidou satellite communication mode to send the encrypted data to the ground data center through the Beidou communication network.

[0121] Example 3

[0122] This embodiment discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the operation mode of the intelligent collection method using Beidou communication technology provided above is implemented.

[0123] Since the electronic device introduced in this embodiment is an electronic device used to implement an intelligent collection method using Beidou communication technology in the embodiment of the present application, based on the intelligent collection method using Beidou communication technology introduced in the embodiment of the present application, the technical personnel of the field can understand the specific implementation of the electronic device of the present embodiment and its various variations, so how the electronic device implements the method in the embodiment of the present application is not described in detail here. As long as the technical personnel of the field implement the electronic device used in the intelligent collection method using Beidou communication technology in the embodiment of the present application, it belongs to the scope of protection of this application.

[0124] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0125] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technical users in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. An intelligent data collection system using Beidou communication technology, characterized in that: include: A data acquisition module, used to collect operating data sets of power equipment; A data processing module is used to pre-process the parameters in the running data set using an improved grid algorithm to obtain preliminary processed data; The feature extraction module is used to perform feature extraction and dimensionality reduction on the preliminary processed data to obtain key feature vectors; A data encryption module is used to encrypt key feature vectors to obtain encrypted data; The comprehensive transfer module calculates the data importance score and real-time requirement score based on the key feature vectors, and uses Beidou satellite positioning to obtain the location information of power equipment; A dynamic transmission module is used to dynamically select the Beidou satellite communication mode according to the location information, data importance score and real-time requirement score of the power equipment; the encrypted data is sent to the ground data center through the Beidou communication network using the Beidou satellite communication mode; each module is connected by wire and / or wireless means; The method of preprocessing the operation data includes: Construct a two-dimensional discrete grid space and define the total number of levels of the two-dimensional discrete grid space, corresponding to the grouping of parameters, and assign a unique identifier to each level; construct a grid coordinate system for each level; select parameters in any grouping as the x-axis and y-axis of the grid coordinate system; define the number of grids for each axis in the x-axis and y-axis, and adjust the number of grids to a power of 2; assign data points in the time series data of the parameters to the corresponding grid coordinate system of the corresponding level; Calculate the correlation coefficient matrix between parameters; identify directly related parameter pairs based on the correlation coefficient matrix; group the directly related parameter pairs as a group; The rules for particle motion are defined in a two-dimensional discrete lattice space, including collision rules and propagation rules; The collision rule is that when two particles meet in a grid, they follow the law of conservation of momentum and energy; by calculating the momentum and energy before and after the collision, the new velocity direction and magnitude of the particles after the collision are determined; the propagation rule is that the particles move along a straight line between adjacent grids at a constant speed, and the particles move between adjacent grids in random directions and steps, and a Coulomb force interaction force is introduced; A displacement threshold is preset, and particles are iteratively updated. For each time step, pairs of particles on all grids are detected, and the collision rule is applied to calculate the new velocity direction and magnitude after the collision. For each particle, its new position in the two-dimensional discrete grid space is updated according to the propagation rule. For each grid, the weighted average of all particles in it is calculated as the smoothed data value of the grid. If the displacement of the particle is less than the preset displacement threshold, the iteration is terminated, otherwise it enters the next iteration; finally, the smoothed data value of the grid node is mapped back to the original data space to obtain the corresponding smoothing parameter; the corresponding smoothing parameter constitutes the preliminary processed data.

2. The intelligent data collection system using Beidou communication technology according to claim 1, characterized in that: The parameters in the operating data set include basic electrical parameters, mechanical parameters and power quality parameters; the basic electrical parameters include voltage, current, power and frequency; the mechanical parameters include temperature and vibration; the power quality parameters include harmonics and negative sequence components.

3. The intelligent data collection system using Beidou communication technology according to claim 2 is characterized in that: The calculation method of the correlation coefficient matrix includes: Create an n×n matrix, where n is the number of parameters, and for each pair of parameters the time series data and , calculate the correlation coefficient between any data points ; ;in, for The data points, for The data points, for The average of the data points in for The average value of the data points in The current time and The time difference of the data points; is the attenuation rate, For the The joint weight of the data points; ;in, and is an adjustable parameter; fill the correlation coefficient into the corresponding position of the matrix, and the matrix is ​​symmetrical, and the values ​​on the diagonal are all 1; get the correlation coefficient matrix; The method of identifying directly related parameter pairs includes: Calculate statistics based on the correlation matrix ;in, is the k-order autocorrelation coefficient, is an adjustable parameter, is the Spearman rank correlation coefficient; Determine the critical value based on the preset significance level and degrees of freedom; if If it is greater than the critical value, the corresponding two parameters are directly related parameters.

4. The intelligent data collection system using Beidou communication technology according to claim 3 is characterized in that: The method of performing feature extraction and dimensionality reduction processing includes: Calculate the statistical characteristics of the preliminary processed data, including mean, variance, peak value and peak factor; perform Fourier transform on the preliminary processed data and extract frequency domain characteristics, including dominant frequency and frequency distribution; use wavelet transform to perform multi-scale decomposition on the preliminary processed data and extract wavelet coefficients of each scale; record the statistical characteristics, frequency domain characteristics and wavelet coefficients of each scale as data characteristics and form a data feature set; use the data features in the data feature set as samples to construct a random forest model; For each decision tree in the random forest model, calculate each data feature In the corresponding decision tree Importance score on ;in, Data features For Node The reduction in the Gini coefficient is Data features For Node The reduction in node impurity, Data features For Node The reduction in the residual sum of squares; , and is the weight coefficient of the corresponding item; For the random forest model, the mean of the importance scores of all decision trees is taken as the data feature The final importance score of all data features is Sort in descending order and select the first N1 data features as important features, where N1 is an integer greater than 1; use Laplace eigenmap to reduce the dimension of important features to obtain low-dimensional feature data. The low-dimensional feature data is in matrix form and is standardized, that is, each dimensional feature is subtracted from the mean and divided by the standard deviation; perform singular value decomposition on the low-dimensional feature data, extract the singular vectors corresponding to the maximum p singular values, and concatenate the p singular vectors by column to obtain the key feature vectors.

5. The intelligent data collection system using Beidou communication technology according to claim 4 is characterized in that: The method of constructing the random forest model includes: The random forest model consists of several decision trees. For each decision tree, several samples are extracted from the data feature set with replacement to form a sample set. At each node of the decision tree, m data features are randomly selected from the sample set to form a feature subset. The optimal split point of the corresponding node is selected based on the feature subset. According to the optimal split point, the current node is split into two child nodes. For the newly generated child nodes, the splitting process is repeated until the maximum tree depth set in advance is reached. Construct N decision trees, each of which uses a different sample set, and each decision tree uses a different feature subset when splitting a node; that is, complete the construction of the random forest model; The selection method of the optimal segmentation point includes: For each data feature, its value range is divided into several adjacent intervals; the value of each data feature will be mapped to the corresponding interval; for each data feature, a histogram is constructed, and for each interval in the histogram, the number of samples contained therein is counted as the statistical information of the corresponding interval; based on the statistical information of each interval, the Gini coefficient or information gain when the corresponding interval is used as the split point is calculated as a scoring indicator; all intervals of each data feature are traversed to find the interval with the largest scoring indicator as the optimal split point.

6. The intelligent data collection system using Beidou communication technology according to claim 5, characterized in that: The method of encrypting the key feature vector includes: Define an elliptic curve E and a base point G, where the elliptic curve E is an algebraic curve defined on a finite field or a binary extension field, and the base point G is a known point on the elliptic curve E; randomly select an integer d as a private key, where d belongs to the interval [1, n1-1], where n1 is the order of the finite point on the elliptic curve E, and calculate the public key Q=d×G based on the private key; represent the key feature vector as a binary string of length k, and map the binary string to a point Pm on the elliptic curve E; encrypt Pm using the public key Q to obtain an encrypted point C=[k]Pm+Q; where [k] represents a scalar multiplication operation on the elliptic curve, and the point Pm is repeatedly added k times; and the encrypted point C is used as the encrypted data.

7. The intelligent data collection system using Beidou communication technology according to claim 6 is characterized in that: The calculation method of the data importance score and the real-time requirement score includes: Based on the key feature vectors, the pre-trained machine learning model is used to detect anomalies in the data and obtain the anomaly probability; the anomaly probability is used as the data importance score; Analyze the timestamp of the original data corresponding to the key feature vector; pre-configure the corresponding timeliness window ; Based on the timestamp and timeliness window obtained from the analysis , calculate the timeliness deadline ;in The timestamp obtained by analysis; the timeliness deadline and the timestamp obtained by analysis are used to calculate the real-time requirement score ;in, is the current time, and To adjust the parameters.

8. The intelligent data collection system using Beidou communication technology according to claim 7, characterized in that: The method of dynamically selecting the Beidou satellite communication mode includes: According to the location information of the power equipment, query the currently visible Beidou satellite list, and determine the available Beidou communication mode based on the orbital parameters and signal strength of the visible satellites in the currently visible Beidou satellite list; for each available Beidou communication mode, calculate its applicability score ;in, , , , and The weight of the corresponding factors; Score the data for importance. is the corresponding power consumption in Beidou communication mode, is the communication distance, is the reliability score in Beidou communication mode; Among all available Beidou communication modes, the Beidou communication mode with the highest applicability score is selected as the final Beidou satellite communication mode.

9. An intelligent collection method using Beidou communication technology, which is implemented based on an intelligent collection system using Beidou communication technology as claimed in any one of claims 1 to 8, characterized in that: include: S1. Collecting operation data sets of power equipment; S2, using the improved grid algorithm to pre-process the parameters in the running data set to obtain preliminary processed data; S3, performing feature extraction and dimensionality reduction processing on the preliminary processed data to obtain key feature vectors; S4, encrypting the key feature vector to obtain encrypted data; S5. Based on the key feature vector, calculate the data importance score and real-time requirement score; and use Beidou satellite positioning to obtain the location information of the power equipment; S6. Dynamically select the Beidou satellite communication mode based on the location information of the power equipment, the data importance score and the real-time requirement score; use the Beidou satellite communication mode to send the encrypted data to the ground data center through the Beidou communication network.

Images