Layered and partitioned information monitoring method and system based on power distribution network

Abstract

The invention discloses a layered and partitioned information monitoring method and system based on a power distribution network, and relates to the technical field of power distribution network monitoring and management, and the method comprises the steps: collecting power grid data of all nodes in the power distribution network, and carrying out the preprocessing; performing state vector modeling on the preprocessed power grid data and forming a state transition matrix, and substituting the state transition matrix and the power grid data into a BP neural network to obtain power grid prediction data; and calculating the energy potential value of each node, carrying out dynamic classification decision on all nodes, and storing all data into a relational database and managing the data. According to the method, high-quality data collection and preprocessing are realized, reliable basic data are provided for the whole monitoring system, effective dynamic modeling and prediction are realized, scientific basis and guidance are provided for real-time monitoring and management of a power grid, stable operation of the power grid under a changing condition is ensured, and the power grid quality is improved. The high-efficiency operation and stability of the power distribution network are ensured, and the intelligent development of a power system is promoted.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network monitoring and management, and in particular to a distribution network-based hierarchical and partitioned information monitoring method and system. Background Art

[0002] With the rapid development of smart grids and renewable energy, the modern management of distribution networks has gradually become the focus of power system technology development. In recent years, research on distribution network operation monitoring, status assessment and optimization control has made significant progress, especially in data acquisition and processing technology. The introduction of Internet of Things (IoT) technology enables each node in the distribution network to collect and transmit grid data in real time, providing a basis for grid status monitoring. The application of models based on machine learning and deep learning in power data analysis has improved prediction accuracy and decision-making efficiency. Algorithms such as BP neural networks have been widely used in grid load forecasting and status assessment, which has gradually improved the intelligence level of distribution network operation. The development of these technologies has provided strong support for the efficient monitoring and management of distribution networks.

[0003] The existing distribution network monitoring and management technology has improved the ability to collect and process power grid data to a certain extent, but there are still some shortcomings. Traditional power grid data processing methods often ignore the dynamic characteristics of data and cannot effectively respond to instantaneous changes in the operation of distribution networks, especially under high load or fault conditions, which will greatly reduce the timeliness and accuracy of monitoring results. Existing technologies generally rely on simple linear models or single prediction models, which limits the in-depth mining of potential patterns and complex relationships in power grid data, reduces the accuracy of monitoring and control, and is difficult to meet the growing intelligent needs of modern distribution networks. The existing technology lacks comprehensive analysis capabilities, resulting in a relatively weak decision support system. Many systems fail to effectively integrate data from different nodes and lack comprehensive status assessment capabilities. In a high-voltage environment, it is difficult for power grid managers to quickly obtain accurate decision-making information, which affects the resilience and adaptability of the power grid. At present, the system's power grid data management and storage mechanism is relatively simple, usually relying on relational databases, which is easy to cause information lag when processing large-scale data and cannot reflect the actual operating status of the power grid in a timely manner. More importantly, the existing system lacks an effective feedback mechanism in the processing and analysis stage, which makes it difficult to form a closed loop for power grid status assessment and decision support, further affecting the optimized operation of the power grid. Summary of the invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problems solved by the present invention are: the existing distribution network monitoring technology has insufficient dynamic modeling capabilities, a single data analysis model, and weak real-time decision support capabilities, as well as how to achieve efficient data processing and prediction based on state vector modeling and state transfer matrix.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: a hierarchical and partitioned information monitoring method based on a distribution network, comprising collecting and preprocessing the grid data of all nodes in the distribution network; performing state vector modeling on the preprocessed grid data and forming a state transfer matrix, substituting the state transfer matrix and the grid data into a BP neural network to obtain grid prediction data; calculating the energy potential value of each node, making dynamic classification decisions for all nodes, and storing and managing all data in a relational database.

[0007] As a preferred solution of the hierarchical and partitioned information monitoring method based on the distribution network described in the present invention, the collection of grid data of all nodes in the distribution network and preprocessing includes preliminary regional division according to the function of the distribution network, dividing the distribution network into load areas, power generation areas and energy storage areas, setting collection nodes and time windows according to the topological structure of the distribution network, collecting grid data within the time window from the grid monitoring system of the distribution network, including voltage amplitude, phase, real-time load data, real-time power generation data and real-time data of energy storage equipment of all collection nodes in each area, and preprocessing the collected grid data.

[0008] Preprocessing includes removing outliers and missing data, filling missing data using linear interpolation, and standardizing the power grid data.

[0009] As a preferred solution of the hierarchical and partitioned information monitoring method based on the distribution network described in the present invention, the state vector modeling of the preprocessed power grid data and the formation of the state transfer matrix include discretizing all power grid data in the time window into multiple intervals, each interval has a state vector, and all state vectors are combined into a state vector sequence.

[0010] The p-order Markov model is used to model the state vector of the preprocessed power grid data. The number of transitions from each state vector to another state vector in the state vector sequence is counted and the transition probability of the current transfer number is calculated, which is expressed as:

[0011]

[0012] Among them, P 1 (S i →S j ) represents the state vector S in the state sequence i To the state vector S j The transition probability of the number of transitions, C(Si →S j ) represents the state vector S in the state sequence i To the state vector S j The number of transitions, m represents the number of state vectors.

[0013] All transition probabilities are combined into a transition probability matrix.

[0014] According to the order of the Markov model, the autocorrelation coefficients of different orders are calculated and expressed as:

[0015]

[0016] Among them, r p represents the autocorrelation coefficient of order p, h represents the total number of power grid data in the time window, x a represents the ath power grid data in the time window, Represents the mean value of the power grid data.

[0017] Through the weighted transition probability matrix, the transition probability of the current state vector to the next state vector is calculated, and the next state vector P(S t+1 =S j ), expressed as:

[0018]

[0019] Among them, P(S t+1 =S j ) represents the current state vector S t The predicted state of the power grid data when the next state vector is the state vector S j The transition probability, m represents the number of state vectors, P 1 (S t →S i ) represents the current state vector S in the transition probability matrix t To the state vector S i The transition probability of the number of transitions, P 2 (S i →S j ) represents the state vector S in the weighted transition probability matrix i To the state vector S j The transition probability of the transition number.

[0020] The current state vector refers to the state vector corresponding to the interval closest to the current time in the time window, and the transition probability of transferring the current state vector to the remaining state vectors constitutes the state transfer matrix.

[0021] As a preferred scheme of the hierarchical and partitioned information monitoring method based on distribution network described in the present invention, wherein: the state transfer matrix and power grid data are substituted into the BP neural network to obtain power grid prediction data, including designing the BP neural network structure and network parameters, and using the particle swarm optimization algorithm to optimize the weights of the BP neural network.

[0022] The designed BP neural network structure includes input layer, hidden layer and output layer. The input layer includes all power grid data and state transfer matrix in the time window. Z hidden layer nodes are selected in the hidden layer, and the ReLU function is selected as the activation function. The output layer outputs the power grid prediction data of the next state vector.

[0023] Network parameters include learning rate, batch size, and number of training epochs.

[0024] Optimizing the weights of the BP neural network using the particle swarm optimization algorithm includes initializing the parameters of the particle swarm, calculating the fitness function, updating the optimal position of the particles and the global optimal position, and applying the optimized weights to the BP neural network.

[0025] The parameters of the initialized particle swarm include the initialized particle position, the initialized particle velocity and the number of initialized particle swarms. Each particle represents a weight combination of the BP neural network. Each weight combination is the weight from the input layer to the hidden layer, the hidden layer to the output layer, and the bias term.

[0026] Calculating the fitness function includes taking the weight of each particle as the connection weight of the BP neural network for feedforward calculation, calculating the error between the power grid prediction data and the actual power grid data, and using the mean square error as the fitness function.

[0027] The optimized weights are applied to the BP neural network. When the number of training rounds reaches the maximum, the weights corresponding to the global optimal position are output as the final optimized BP neural network weights. The optimized weights are used for the final training and prediction of the BP neural network to obtain the final power grid prediction data.

[0028] As a preferred solution of the hierarchical partition information monitoring method based on the distribution network described in the present invention, wherein: the calculation of the energy potential value of each node includes forming a topological matrix of the connection relationships between all nodes and nodes to define the energy field of the distribution network.

[0029] The energy field of the distribution network includes the energy potential value of each node. A preliminary energy potential value is assigned to each area of ​​the distribution network. The initial energy potential value is set to 0, indicating the neutral state of regional energy.

[0030] Obtain the resistance and reactance of each transmission line in the distribution network, calculate the impedance of each transmission line, calculate the admittance based on the impedance using the admittance formula, and obtain the real and imaginary parts of the admittance. According to the topological matrix, calculate the self-admittance and mutual admittance between each pair of nodes, and construct the admittance matrix of the self-admittance and mutual admittance of all nodes according to the topological structure of the distribution network. Each element in the admittance matrix represents the admittance from one node to another node.

[0031] Based on the admittance matrix, the power flow equation is used to calculate the active power and reactive power from each collection node to the next collection node, which can be expressed as:

[0032]

[0033] Among them, K gf (τ) represents the active power from the collection node g to the node f at the time point τ, Q gf (τ) represents the reactive power from the collection node g to the node f at the time point τ, n represents all the collection nodes in the distribution network, V g (τ) represents the voltage amplitude of node g at time point τ, V f (τ) represents the voltage amplitude of node f at time point τ, G gf (τ) represents the real part of the admittance from node g to node f at time point τ, B gf (τ) represents the imaginary part of the admittance from node g to node f at time point τ, θ gf (τ) represents the phase difference between node g and node f at time point τ.

[0034] Construct the two-dimensional state transfer matrix A(τ) between nodes at time point τ, expressed as:

[0035]

[0036] Among them, the first half of the two-dimensional state transfer matrix represents the active power between nodes, the second half of the two-dimensional state transfer matrix represents the reactive power between nodes, and n represents all collection nodes in the distribution network.

[0037] The maximum power carrying capacity of the transmission line is calculated based on the two-dimensional state transfer matrix to calculate the net active power flow ΔK at each node ε and the net reactive power flow ΔQ ε , expressed as:

[0038] ΔK ε (τ)=ΔK σ (τ)-ΔK μ (τ)

[0039] ΔQ ε (τ)=ΔQ σ (τ)-ΔQμ (τ)

[0040] Among them, ΔK ε (τ) represents the net active power flow of the node at time point τ, ΔQ ε (τ) represents the net reactive power flow of the node at time point τ, ΔK σ (τ) represents the active power inflow of the node at time point τ, ΔQ σ (τ) represents the reactive power inflow into the node at time point τ, ΔK μ (τ) represents the active power outflow of the node at time point τ, ΔQ μ (τ) represents the active power outflow and reactive power outflow of the node at time point τ.

[0041] According to the obtained net active power flow and net reactive power flow, the energy potential value J(τ) of the node is updated, which is expressed as:

[0042] J(τ)=J(τ-1)+ΔK ε (τ)+ΔQ ε (τ)

[0043] Among them, J(τ) represents the energy potential value of the node at time point τ, and J(τ-1) represents the energy potential value of the node at the previous time point of time point τ.

[0044] As a preferred scheme of the hierarchical and partitioned information monitoring method based on the distribution network described in the present invention, the dynamic classification decision of all nodes includes calculating the net flow of active power and the net flow of reactive power of each node according to the final power grid prediction data obtained, updating the energy potential value of the node, and sorting the energy potential values ​​of all nodes in descending order in the form of absolute values, setting an absolute value threshold, and judging the energy potential value of each node.

[0045] If the absolute value of the energy potential value of the current node is greater than the absolute value threshold, it means that the energy potential value deviation of the current node is large and the energy state of the current node needs to be adjusted. If the absolute value of the energy potential value of the current node is less than or equal to the absolute value threshold, the current node is skipped.

[0046] Prioritize analyzing the node with the largest absolute value of energy potential among the nodes that need to be adjusted and make dynamic classification decisions. If the energy potential value of the current node is greater than 0, it means that the load of the current node is large, and the current node is merged with the area with small remaining load. If the energy potential value of the current node is less than 0, it means that the load of the current node is small, and the current node is divided into a new area. If the energy potential value of the current node is equal to 0, it means that the load of the current node is reasonable and no operation is performed.

[0047] According to the classification decision, the regional boundary of the distribution network is adjusted, the node list and energy potential value and the two-dimensional state transfer matrix in the area are updated, the energy potential values ​​of all adjusted nodes are recalculated, and the difference between the absolute value of the energy potential value of the adjusted node and the absolute value threshold is verified.

[0048] During the adjustment process, the greedy algorithm repeatedly performs node adjustment. If the absolute values ​​of the energy potential values ​​of all nodes do not exceed the absolute value threshold, the adjustment is stopped. If the absolute values ​​of the energy potential values ​​of all nodes exceed the absolute value threshold, the adjustment continues.

[0049] After stopping the adjustment, the power flow calculation of the distribution network is carried out to verify the stability of the distribution network.

[0050] The operating status of the distribution network is fed back through the SCADA system, the energy potential and power flow of all nodes are updated in real time, and the absolute value threshold is further adjusted according to the feedback data.

[0051] As a preferred solution of the hierarchical and partitioned information monitoring method based on the distribution network described in the present invention, wherein: the storing and managing of all data in a relational database includes selecting a relational database to store and manage data and analysis results, designing a database table structure to store different types of data, setting regular backup tasks, backing up all data in the database, performing authority management on database users and encrypting and storing data.

[0052] Another object of the present invention is to provide a hierarchical and partitioned information monitoring system based on a distribution network, which can perform state vector modeling on the preprocessed power grid data and form a state transfer matrix, substitute the state transfer matrix and power grid data into the BP neural network to obtain power grid prediction data, thereby solving the problem that the current power grid monitoring technology contains insufficient static modeling and prediction capabilities.

[0053] As a preferred solution of the hierarchical and partitioned information monitoring system based on the distribution network described in the present invention, it includes: a data collection and preprocessing module, a prediction data generation module, and a data storage management module.

[0054] The data collection preprocessing module is used to collect and preprocess the grid data of all nodes in the distribution network; the prediction data generation module is used to perform state vector modeling on the preprocessed grid data and form a state transfer matrix, and substitute the state transfer matrix and grid data into the BP neural network to obtain grid prediction data; the data storage management module is used to calculate the energy potential value of each node, perform dynamic classification decisions on all nodes, and store and manage all data in a relational database.

[0055] A computer device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement a step of a hierarchical partition information monitoring method based on a distribution network.

[0056] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a hierarchical partition information monitoring method based on a distribution network.

[0057] Beneficial effects of the present invention: The hierarchical and partitioned information monitoring method based on the distribution network provided by the present invention collects the power grid data of all nodes in the distribution network and performs preprocessing, thereby realizing high-quality data collection and preprocessing, providing reliable basic data for the entire monitoring system, and helping to improve the care and decision-making efficiency of power grid operation. State vector modeling is performed on the preprocessed power grid data to form a state transfer matrix, and the state transfer matrix and power grid data are substituted into the BP neural network to obtain power grid prediction data, thereby realizing effective dynamic modeling and prediction, providing a scientific basis and guidance for real-time monitoring and management of the power grid, ensuring the stable operation of the power grid under changing conditions, calculating the energy potential value of each node, performing dynamic classification decisions on all nodes, storing all data in a relational database and managing them, promoting dynamic and flexible energy management, ensuring the efficient operation and stability of the distribution network, and promoting the intelligent development of the power system. The present invention has achieved better results in optimizing data processing, strengthening dynamic modeling, and improving intelligent decision support. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0059] Figure 1 An overall flow chart of a hierarchical and partitioned information monitoring method based on a distribution network provided in the first embodiment of the present invention.

[0060] Figure 2 An overall flow chart of a hierarchical and partitioned information monitoring system based on a distribution network is provided for the third embodiment of the present invention. DETAILED DESCRIPTION

[0061] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0062] Example 1, reference Figure 1 , as an embodiment of the present invention, provides a hierarchical partition information monitoring method based on a distribution network, comprising:

[0063] S1: Collect and pre-process the grid data of all nodes in the distribution network.

[0064] Furthermore, collecting grid data from all nodes in the distribution network and preprocessing it includes making a preliminary regional division according to the function of the distribution network, dividing the distribution network into load areas, power generation areas and energy storage areas, setting collection nodes and time windows according to the topological structure of the distribution network, and collecting grid data within the time window from the grid monitoring system of the distribution network, including the voltage amplitude, phase, real-time load data, real-time power generation data and real-time data of energy storage equipment of all collection nodes in each area, and preprocessing the collected grid data.

[0065] Preprocessing includes removing outliers and missing data, filling missing data using linear interpolation, and standardizing the power grid data.

[0066] It should be noted that the preliminary regional division includes dividing the distribution network into load areas, generation areas and energy storage areas.

[0067] Load areas include concentrated industrial, residential and commercial areas.

[0068] The power generation area includes the area centered on the power generation facilities. For example, in areas with strong winds, it can be divided into areas where wind power generation is the main focus, and the areas can be divided according to the volatility of wind power generation.

[0069] The energy storage area includes areas with a certain number of energy storage units. For example, in areas with strong renewable energy power generation capacity, a larger-scale energy storage system can be set up to cope with the load imbalance caused by power generation fluctuations.

[0070] It should also be noted that by collecting and preprocessing the grid data of all nodes in the distribution network, comprehensive and systematic data acquisition is achieved. According to the function of the distribution network, the distribution network is divided into load area, power generation area and energy storage area, and the collection nodes and time windows are set accordingly to ensure that the voltage, phase, load, power generation and energy storage data in each area are collected during the current period. This method not only provides a rich source of data for subsequent analysis, but also helps to identify the various functional areas of the distribution network, thereby improving the pertinence of data analysis; in the process of data preprocessing, the integrity and accuracy of the data are ensured by removing outliers and using linear interpolation to fill in missing data. Standardization operations unify data of different amplitudes and ranges to the same scale, making the data more comparable and easy to analyze, improving the reliability of the monitoring system, and laying a solid foundation for the effectiveness of subsequent steps. It enables managers to deal with complex power operation environments more calmly and improve the scientificity and effectiveness of decision support.

[0071] S2: The preprocessed power grid data is modeled with a state vector and a state transfer matrix is ​​formed. The state transfer matrix and power grid data are substituted into the BP neural network to obtain power grid prediction data.

[0072] Furthermore, the preprocessed power grid data is modeled with a state vector and a state transfer matrix is ​​formed, including discretizing all power grid data in a time window into a plurality of intervals, each interval having a state vector, and forming all state vectors into a state vector sequence.

[0073] The p-order Markov model is used to model the state vector of the preprocessed power grid data. The number of transitions from each state vector to another state vector in the state vector sequence is counted and the transition probability of the current transfer number is calculated, which is expressed as:

[0074]

[0075] Among them, P 1 (S i →S j ) represents the state vector S in the state sequence i To the state vector S j The transition probability of the number of transitions, C(S i →S j ) represents the state vector S in the state sequence i To the state vector S j The number of transitions, m represents the number of state vectors.

[0076] All transition probabilities are combined into a transition probability matrix.

[0077] According to the order of the Markov model, the autocorrelation coefficients of different orders are calculated and expressed as:

[0078]

[0079] Among them, r p represents the autocorrelation coefficient of order p, h represents the total number of power grid data in the time window, x a represents the ath power grid data in the time window, Represents the mean value of the power grid data.

[0080] Through the weighted transition probability matrix, the transition probability of the current state vector to the next state vector is calculated, and the next state vector P(S t+1 =S j ), expressed as:

[0081]

[0082] Among them, P(S t+1 =S j ) represents the current state vector S t The predicted state of the power grid data when the next state vector is the state vector S j The transition probability, m represents the number of state vectors, P 1 (S t →S i ) represents the current state vector S in the transition probability matrix t To the state vector S i The transition probability of the number of transitions, P 2 (S i →S j ) represents the state vector S in the weighted transition probability matrix i To the state vector S j The transition probability of the transition number.

[0083] The current state vector refers to the state vector corresponding to the interval closest to the current time in the time window, and the transition probability of transferring the current state vector to the remaining state vectors constitutes the state transfer matrix.

[0084] It should be noted that substituting the state transfer matrix and power grid data into the BP neural network to obtain power grid prediction data includes designing the BP neural network structure and network parameters, and using the particle swarm optimization algorithm to optimize the weights of the BP neural network.

[0085] The designed BP neural network structure includes input layer, hidden layer and output layer. The input layer includes all power grid data and state transfer matrix in the time window. Z hidden layer nodes are selected in the hidden layer, and the ReLU function is selected as the activation function. The output layer outputs the power grid prediction data of the next state vector.

[0086] Network parameters include learning rate, batch size, and number of training epochs.

[0087] Optimizing the weights of the BP neural network using the particle swarm optimization algorithm includes initializing the parameters of the particle swarm, calculating the fitness function, updating the optimal position of the particles and the global optimal position, and applying the optimized weights to the BP neural network.

[0088] The parameters of the initialized particle swarm include the initialized particle position, the initialized particle velocity and the number of initialized particle swarms. Each particle represents a weight combination of the BP neural network. Each weight combination is the weight from the input layer to the hidden layer, the hidden layer to the output layer, and the bias term.

[0089] Calculating the fitness function includes taking the weight of each particle as the connection weight of the BP neural network for feedforward calculation, calculating the error between the power grid prediction data and the actual power grid data, and using the mean square error as the fitness function.

[0090] The optimized weights are applied to the BP neural network. When the number of training rounds reaches the maximum, the weights corresponding to the global optimal position are output as the final optimized BP neural network weights. The optimized weights are used for the final training and prediction of the BP neural network to obtain the final power grid prediction data.

[0091] It should also be noted that by modeling the state vector of the pre-processed power grid data and forming a state transfer matrix, the problem of insufficient response to dynamic changes in traditional power grid monitoring technology is solved. By discretizing the power grid data in the time window, each time interval is assigned a state vector, thereby forming a state vector sequence; this method provides a powerful tool for the dynamic analysis of the power grid operation status and fully reflects the effective capture of the complexity of the power grid; the Markov model is used to model the state vector, count the number of state transitions, and calculate the transition probability to form a state transfer matrix, so that the power grid monitoring system can effectively describe the change process between the power grid states, which not only improves the prediction ability, but also provides a scientific basis for real-time monitoring, which can help identify potential risk states and make early warnings; by combining the state transfer matrix with the BP neural network, the data prediction process of the power grid is further optimized, and the nonlinear fitting ability is used to enhance the prediction accuracy of the future behavior of the power grid, improve the power grid's ability to respond to complex and sudden situations, ensure the safety and stability of the power system, make power dispatching management more scientific and efficient, and finally realize the intelligent operation of the distribution network.

[0092] S3: Calculate the energy potential value of each node, make dynamic classification decisions for all nodes, store all data in a relational database and manage them.

[0093] Furthermore, calculating the energy potential value of each node includes forming a topological matrix of all nodes and the connection relationships between nodes to define the energy field of the distribution network.

[0094] The energy field of the distribution network includes the energy potential value of each node. A preliminary energy potential value is assigned to each area of ​​the distribution network. The initial energy potential value is set to 0, indicating the neutral state of regional energy.

[0095] Obtain the resistance and reactance of each transmission line in the distribution network, calculate the impedance of each transmission line, calculate the admittance based on the impedance using the admittance formula, and obtain the real and imaginary parts of the admittance. According to the topological matrix, calculate the self-admittance and mutual admittance between each pair of nodes, and construct the admittance matrix of the self-admittance and mutual admittance of all nodes according to the topological structure of the distribution network. Each element in the admittance matrix represents the admittance from one node to another node.

[0096] Based on the admittance matrix, the power flow equation is used to calculate the active power and reactive power from each collection node to the next collection node, which can be expressed as:

[0097]

[0098] Among them, K gf (τ) represents the active power from the collection node g to the node f at the time point τ, Q gf (τ) represents the reactive power from the collection node g to the node f at the time point τ, n represents all the collection nodes in the distribution network, V g (τ) represents the voltage amplitude of node g at time point τ, V f (τ) represents the voltage amplitude of node f at time point τ, G gf (τ) represents the real part of the admittance from node g to node f at time point τ, B gf (τ) represents the imaginary part of the admittance from node g to node f at time point τ, θ gf (τ) represents the phase difference between node g and node f at time point τ.

[0099] Construct the two-dimensional state transfer matrix A(τ) between nodes at time point τ, expressed as:

[0100]

[0101] Among them, the first half of the two-dimensional state transfer matrix represents the active power between nodes, the second half of the two-dimensional state transfer matrix represents the reactive power between nodes, and n represents all collection nodes in the distribution network.

[0102] The maximum power carrying capacity of the transmission line is calculated based on the two-dimensional state transfer matrix to calculate the net active power flow ΔK at each node ε and the net reactive power flow ΔQε , expressed as:

[0103] ΔK ε (τ)=ΔK σ (τ)-ΔK μ (τ)

[0104] ΔQ ε (τ)=ΔQ σ (τ)-ΔQ μ (τ)

[0105] Among them, ΔK ε (τ) represents the net active power flow of the node at time point τ, ΔQ ε (τ) represents the net reactive power flow of the node at time point τ, ΔK σ (τ) represents the active power inflow of the node at time point τ, ΔQ σ (τ) represents the reactive power inflow into the node at time point τ, ΔK μ (τ) represents the active power outflow of the node at time point τ, ΔQ μ (τ) represents the active power outflow and reactive power outflow of the node at time point τ.

[0106] According to the obtained net active power flow and net reactive power flow, the energy potential value J(τ) of the node is updated, which is expressed as:

[0107] J(τ)=J(τ-1)+ΔK ε (τ)+ΔQ ε (τ)

[0108] Among them, J(τ) represents the energy potential value of the node at time point τ, and J(τ-1) represents the energy potential value of the node at the previous time point of time point τ.

[0109] It should be noted that the dynamic classification decision for all nodes includes calculating the net flow of active power and reactive power of each node based on the final grid forecast data obtained, updating the energy potential value of the node, and sorting the energy potential values ​​of all nodes in descending order in the form of absolute value, setting the absolute value threshold, and judging the energy potential value of each node.

[0110] If the absolute value of the energy potential value of the current node is greater than the absolute value threshold, it means that the energy potential value deviation of the current node is large and the energy state of the current node needs to be adjusted. If the absolute value of the energy potential value of the current node is less than or equal to the absolute value threshold, the current node is skipped.

[0111] Prioritize analyzing the node with the largest absolute value of energy potential among the nodes that need to be adjusted and make dynamic classification decisions. If the energy potential value of the current node is greater than 0, it means that the load of the current node is large, and the current node is merged with the area with small remaining load. If the energy potential value of the current node is less than 0, it means that the load of the current node is small, and the current node is divided into a new area. If the energy potential value of the current node is equal to 0, it means that the load of the current node is reasonable and no operation is performed.

[0112] According to the classification decision, the regional boundary of the distribution network is adjusted, the node list and energy potential value and the two-dimensional state transfer matrix in the area are updated, the energy potential values ​​of all adjusted nodes are recalculated, and the difference between the absolute value of the energy potential value of the adjusted node and the absolute value threshold is verified.

[0113] During the adjustment process, the greedy algorithm repeatedly performs node adjustment. If the absolute values ​​of the energy potential values ​​of all nodes do not exceed the absolute value threshold, the adjustment is stopped. If the absolute values ​​of the energy potential values ​​of all nodes exceed the absolute value threshold, the adjustment continues.

[0114] After stopping the adjustment, the power flow calculation of the distribution network is carried out to verify the stability of the distribution network.

[0115] The operating status of the distribution network is fed back through the SCADA system, the energy potential and power flow of all nodes are updated in real time, and the absolute value threshold is further adjusted according to the feedback data.

[0116] It should also be noted that storing and managing all data in a relational database includes selecting a relational database to store and manage data and analysis results, designing a database table structure to store different types of data, setting up regular backup tasks, backing up all data in the database, managing permissions for database users and encrypting and storing data.

[0117] It should also be noted that by calculating the energy potential value of each node and making dynamic classification decisions, an important basis is provided for the intelligent management of the distribution network. By constructing the topological matrix of the distribution network, defining the energy field and assigning the initial energy potential value of each area, a framework is provided for subsequent energy monitoring. The calculation of the energy potential value reflects the effectiveness of each node in the power grid in the energy distribution process, and plays a key role in identifying the problem of load imbalance. Through dynamic classification decisions, the system can quickly locate the nodes that need to be adjusted according to the energy potential values ​​of different nodes, which can not only effectively alleviate the situation of excessive or low load, but also optimize the allocation of overall power resources. Especially when the energy potential value of the current node needs to be adjusted, the system implements a fast and scientific optimization strategy, such as merging areas with large loads with areas with small loads to ensure the rational use of resources. Through continuous iteration of the greedy algorithm, it is ensured that the energy potential values ​​of all nodes are within a reasonable range, thereby ensuring the stability of the distribution network. After the adjustment, the flow calculation and the final stability verification make the whole process more reliable. By giving the SCADA system real-time feedback, the flexibility of dynamic decision-making is further enhanced, enabling managers to quickly adjust strategies according to the actual operating status, realizing the intelligent and scientific management of the distribution network.

[0118] Embodiment 2 is an embodiment of the present invention, which provides a hierarchical and partitioned information monitoring method based on a distribution network. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0119] Firstly, the experiment was carried out in a simulated distribution network system. The experimental object was a distribution network of 50 nodes, including 20 nodes in the load area, 15 nodes in the power generation area and 15 nodes in the energy storage area. The distribution structure was set based on the topological characteristics of a typical urban distribution network. The experiment collected real-time data of the distribution network nodes, and performed data processing, prediction and dynamic optimization based on the proposed method. Firstly, the experimental area was preliminarily divided. The load area was divided based on the node load power ratio exceeding 60%, the power generation area was divided based on the node actual power generation capacity exceeding 50%, and the energy storage area was based on the installed capacity and load regulation capacity of the energy storage equipment. The nodes in the grid are set to collect data every 5 minutes and the time window is 1 hour. The SCADA system collects data sets including node voltage amplitude, phase, real-time load power, real-time power generation power and energy storage equipment power in real time to form a preliminary grid data set. After data collection, the grid data is strictly preprocessed, including removing outliers and missing data. The outliers are removed by using the triple standard deviation method and the missing data are filled by the linear interpolation method. To ensure the stability of the model calculation, all grid data are standardized to the [0,1] interval to further improve the model's adaptability to data of different dimensions. In this paper, the first-order Markov model is used to model the state vector of the preprocessed power grid data. In this process, the power grid data in the time window is discretized into 10 intervals, and a state vector is generated in each interval. A sequence is formed based on the state vector. By counting the number of transitions of the state vector in the sequence, the transition probability is calculated and the state transition matrix is ​​generated. At the same time, the autocorrelation coefficients of different orders are calculated to verify the applicability of the model, and the first-order Markov model is determined as the optimal modeling order. On the basis of power grid data modeling, a BP neural network is designed and trained to realize the prediction of power grid data. The input layer of the neural network contains the power grid data and the state transition matrix. The hidden layer is set to 15 nodes, and the ReLU activation function is used. The output layer outputs the power grid prediction data for the next time step. The network parameters include learning rate 0.01, batch size 32, and maximum number of training rounds 100. In order to further optimize the network performance, the particle swarm optimization algorithm is used to adjust the weights and biases of the neural network. In the particle swarm initialization stage, the number of particles is set to 30, the maximum number of iterations is 200, and the fitness function is the mean square error between the predicted data and the actual data. After multiple rounds of training, the optimal network weight with the lowest error is obtained, and the prediction result is finally output. After the prediction is completed, the energy potential value of each node is calculated according to the predicted data.Through the distribution network topology matrix, the admittance matrix is ​​further constructed, the active power and reactive power between nodes are calculated, and a two-dimensional state transfer matrix is ​​formed. On this basis, the node energy potential value is updated, and the nodes are classified based on the threshold (0.1PU); nodes with larger loads are merged with areas with smaller loads through adjustment, and energy storage devices are used to alleviate load pressure. In the dynamic classification process, a greedy algorithm is used to iteratively adjust the node area boundaries until the absolute values ​​of the energy potential values ​​of all nodes are lower than the threshold; after the adjustment is completed, the stability of the distribution network is verified by power flow calculation, and the final results are stored in a relational database for analysis and backup; it can be obtained from the experimental results that the present invention has advantages in power grid data prediction accuracy, energy potential value adjustment and distribution network area division, which effectively improves the operation efficiency and stability of the distribution network.

[0120] Example 3, reference Figure 2 , as an embodiment of the present invention, provides a hierarchical and partitioned information monitoring system based on a distribution network, including a data collection and preprocessing module, a prediction data generation module, and a data storage and management module.

[0121] The data collection preprocessing module is used to collect and preprocess the grid data of all nodes in the distribution network; the prediction data generation module is used to model the state vector of the preprocessed grid data and form a state transfer matrix, and substitute the state transfer matrix and grid data into the BP neural network to obtain grid prediction data; the data storage management module is used to calculate the energy potential value of each node, make dynamic classification decisions for all nodes, store all data in a relational database and manage them.

[0122] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0123] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0124] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0125] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc. It should be noted that the above embodiments are only used to illustrate the technical solution of the present invention and are not limited. Although the present invention is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

[0126] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A hierarchical and partitioned information monitoring method based on a distribution network, characterized in that: include: Collect and pre-process grid data from all nodes in the distribution network; The preprocessed power grid data is modeled with state vectors and a state transfer matrix is ​​formed. The state transfer matrix and power grid data are substituted into the BP neural network to obtain power grid prediction data. Calculate the energy potential value of each node, make dynamic classification decisions for all nodes, and store and manage all data in a relational database.

2. The method for monitoring hierarchical and partitioned information based on a distribution network according to claim 1, characterized in that: The collecting and preprocessing of the grid data of all nodes in the distribution network includes performing preliminary regional division according to the function of the distribution network, dividing the distribution network into a load area, a power generation area and an energy storage area, setting the collection nodes and the time window according to the topological structure of the distribution network, collecting the grid data within the time window from the grid monitoring system of the distribution network, including the voltage amplitude, phase, real-time load data, real-time power generation data and real-time data of energy storage equipment of all collection nodes in each area, and preprocessing the collected grid data; Preprocessing includes removing outliers and missing data, filling missing data using linear interpolation, and standardizing the power grid data.

3. The method for monitoring the hierarchical and partitioned information of the distribution network according to claim 2, characterized in that: The state vector modeling of the preprocessed power grid data and forming a state transfer matrix includes discretizing all power grid data in the time window into multiple intervals, each interval has a state vector, and forming all state vectors into a state vector sequence; The p-order Markov model is used to model the state vector of the preprocessed power grid data. The number of transitions from each state vector to another state vector in the state vector sequence is counted and the transition probability of the current transfer number is calculated, which is expressed as: Among them, P1(S i →S j ) represents the state vector S in the state sequence i To the state vector S j The transition probability of the number of transitions, C(S i →S j ) represents the state vector S in the state sequence i To the state vector S j The number of transitions, m represents the number of state vectors; All transition probabilities are combined into a transition probability matrix; According to the order of the Markov model, the autocorrelation coefficients of different orders are calculated and expressed as: Among them, r p represents the autocorrelation coefficient of order p, h represents the total number of power grid data in the time window, x a represents the ath power grid data in the time window, Represents the mean value of power grid data; Through the weighted transition probability matrix, the transition probability of the current state vector to the next state vector is calculated, and the next state vector P(S t+1 =S j ), expressed as: Among them, P(S t+1 =S j ) represents the current state vector S t The predicted state of the power grid data when the next state vector is the state vector S j The transition probability, m represents the number of state vectors, P1(S t →S i ) represents the current state vector S in the transition probability matrix t To the state vector S i The transition probability of the number of transitions, P2(S i →S j ) represents the state vector S in the weighted transition probability matrix i To the state vector S j The transition probability of the number of transitions; The current state vector refers to the state vector corresponding to the interval closest to the current time in the time window, and the transition probability of transferring the current state vector to the remaining state vectors constitutes the state transfer matrix.

4. The method for monitoring the hierarchical and partitioned information of the distribution network according to claim 3, characterized in that: Substituting the state transfer matrix and the power grid data into the BP neural network to obtain the power grid prediction data includes designing the BP neural network structure and network parameters, and optimizing the weight of the BP neural network using a particle swarm optimization algorithm; The BP neural network structure is designed to include input layer, hidden layer and output layer. The input layer includes all power grid data and state transfer matrix in the time window. In the hidden layer, Z hidden layer nodes are selected and ReLU function is selected as the activation function. The output layer outputs the power grid prediction data of the next state vector. Network parameters include learning rate, batch size, and number of training rounds; Optimizing the weights of BP neural network using particle swarm optimization algorithm includes initializing the parameters of particle swarm, calculating fitness function, updating the optimal position of particles and the global optimal position, and applying the optimized weights to BP neural network; The parameters of the initialization particle swarm include the initialization particle position, initialization particle velocity and the number of initialization particle swarms. Each particle represents a weight combination of the BP neural network. Each weight combination is the weight from the input layer to the hidden layer, from the hidden layer to the output layer, and the bias term. Calculating the fitness function includes taking the weight of each particle as the connection weight of the BP neural network for feedforward calculation, calculating the error between the power grid prediction data and the actual power grid data, and using the mean square error as the fitness function; The optimized weights are applied to the BP neural network. When the number of training rounds reaches the maximum, the weights corresponding to the global optimal position are output as the final optimized BP neural network weights. The optimized weights are used for the final training and prediction of the BP neural network to obtain the final power grid prediction data.

5. The method for monitoring the hierarchical and partitioned information of the distribution network according to claim 4, characterized in that: The calculation of the energy potential value of each node includes forming a topological matrix of all the connection relationships between nodes to define the energy field of the distribution network; The energy field of the distribution network includes the energy potential value of each node, and a preliminary energy potential value is assigned to each area of ​​the distribution network. The initial energy potential value is set to 0, indicating the neutral state of regional energy; Obtain the resistance and reactance of each transmission line in the distribution network, calculate the impedance of each transmission line, calculate the admittance according to the impedance using the admittance formula, and obtain the real and imaginary parts of the admittance. According to the topological matrix, calculate the self-admittance and mutual admittance between each pair of nodes, and construct the admittance matrix with the self-admittance and mutual admittance of all nodes according to the topological structure of the distribution network. Each element in the admittance matrix represents the admittance from one node to another node. Based on the admittance matrix, the power flow equation is used to calculate the active power and reactive power from each collection node to the next collection node, which can be expressed as: Among them, K gf (τ) represents the active power from the collection node g to the node f at the time point τ, Q gf (τ) represents the reactive power from the collection node g to the node f at the time point τ, n represents all the collection nodes in the distribution network, V g (τ) represents the voltage amplitude of node g at time point τ, V f (τ) represents the voltage amplitude of node f at time point τ, G gf (τ) represents the real part of the admittance from node g to node f at time point τ, B gf (τ) represents the imaginary part of the admittance from node g to node f at time point τ, θ gf (τ) represents the phase difference between node g and node f at time point τ; Construct the two-dimensional state transfer matrix A(τ) between nodes at time point τ, expressed as: Among them, the first half of the two-dimensional state transfer matrix represents the active power between nodes, the second half of the two-dimensional state transfer matrix represents the reactive power between nodes, and n represents all the collection nodes in the distribution network; The maximum power carrying capacity of the transmission line is calculated based on the two-dimensional state transfer matrix to calculate the net active power flow ΔK at each node ε and the net reactive power flow ΔQ ε , expressed as: DK ε (t)=ΔK σ (t)-ΔK μ (t) ΔQ ε (τ)=ΔQ σ (t)-ΔQ μ (t) Among them, ΔK ε (τ) represents the net active power flow of the node at time point τ, ΔQ ε (τ) represents the net reactive power flow of the node at time point τ, ΔK σ (τ) represents the active power inflow of the node at time point τ, ΔQ σ (τ) represents the reactive power inflow into the node at time point τ, ΔK μ (τ) represents the active power outflow of the node at time point τ, ΔQ μ (τ) represents the active power outflow and reactive power outflow of the node at time point τ; According to the obtained net active power flow and net reactive power flow, the energy potential value J(τ) of the node is updated, which is expressed as: J(τ)=J(τ-1)+ΔK ε (t)+ΔQ ε (t) Among them, J(τ) represents the energy potential value of the node at time point τ, and J(τ-1) represents the energy potential value of the node at the previous time point of time point τ.

6. The method for monitoring the hierarchical and partitioned information of the distribution network according to claim 5, characterized in that: The dynamic classification decision for all nodes includes calculating the net active power flow and the net reactive power flow of each node according to the final power grid prediction data obtained, updating the energy potential value of the node, and sorting the energy potential values ​​of all nodes in descending order in the form of absolute values, setting an absolute value threshold, and judging the energy potential value of each node; If the absolute value of the energy potential value of the current node is greater than the absolute value threshold, it means that the energy potential value deviation of the current node is large and the energy state of the current node needs to be adjusted. If the absolute value of the energy potential value of the current node is less than or equal to the absolute value threshold, the current node is skipped. Prioritize the analysis of the nodes with the largest absolute value of energy potential among the nodes that need to be adjusted and make dynamic classification decisions. If the energy potential value of the current node is greater than 0, it means that the load of the current node is large, and the current node is merged with the area with small remaining load. If the energy potential value of the current node is less than 0, it means that the load of the current node is small, and the current node is divided into a new area. If the energy potential value of the current node is equal to 0, it means that the load of the current node is reasonable, and no operation is performed; According to the classification decision, the regional boundary of the distribution network is adjusted, the node list and energy potential value and the two-dimensional state transfer matrix in the region are updated, the energy potential values ​​of all nodes after adjustment are recalculated, and the difference between the absolute value of the energy potential value of the adjusted node and the absolute value threshold is verified; During the adjustment process, the greedy algorithm repeatedly performs node adjustment. If the absolute values ​​of the energy potential values ​​of all nodes do not exceed the absolute value threshold, the adjustment is stopped. If the absolute values ​​of the energy potential values ​​of all nodes exceed the absolute value threshold, the adjustment continues. After stopping the adjustment, the power flow calculation of the distribution network is carried out to verify the stability of the distribution network; The operating status of the distribution network is fed back through the SCADA system, the energy potential and power flow of all nodes are updated in real time, and the absolute value threshold is further adjusted according to the feedback data.

7. The method for monitoring hierarchical and partitioned information based on a distribution network according to claim 6, characterized in that: The storing and managing of all data in a relational database includes selecting a relational database to store and manage data and analysis results, designing a database table structure to store different types of data, setting regular backup tasks, backing up all data in the database, managing permissions for database users and encrypting and storing data.

8. A system using the hierarchical and partitioned information monitoring method based on a distribution network as claimed in any one of claims 1 to 7, characterized in that: It includes data collection and preprocessing module, prediction data generation module, and data storage management module; The data collection preprocessing is used to collect and preprocess the grid data of all nodes in the distribution network; The prediction data generation module is used to perform state vector modeling on the preprocessed power grid data and form a state transfer matrix, and substitute the state transfer matrix and power grid data into the BP neural network to obtain power grid prediction data; The data storage management module is used to calculate the energy potential value of each node, perform dynamic classification decisions on all nodes, and store and manage all data in a relational database.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the hierarchical partition information monitoring method based on the distribution network described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the hierarchical partition information monitoring method based on the distribution network described in any one of claims 1 to 7 are implemented.

Images