A power supply stability monitoring system based on fusion terminal
By using a fusion terminal and smart power meter in the power system to collect data, and using clustering model and BP neural network analysis, the problem of unpredictable power supply stability in the station area is solved, real-time monitoring of power supply stability in the station area and Automated management improves the operational safety and management efficiency of the power grid.
Patent Information
- Application Number
- CN202210866476.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-07-22
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2042-07-22
AI Technical Summary
In a single distribution station area, there are many power equipment, and there are complexity and difficulty in identifying and predicting the operating status of the power system, which makes the power supply stability difficult to predict.
Adopting a station area power supply stability monitoring system based on fusion terminals, power information is collected through intelligent power meters, concentrators and fusion terminals, and using data normalization modules, adaptive rapid search density peak algorithm clustering model and BP neural network Perform data analysis to predict the power supply stability of the station area.
Real-time monitoring and prediction of power supply stability in the Taiwan area has been realized, and an automated power grid management system has been built, which has improved the safety, reliability and economic benefits of power grid operation.
Smart Images

Figure CN115118015B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of electric power information technology, and in particular relates to a station area power supply stability monitoring system based on a fusion terminal. Background Art
[0002] The power system is an electric energy production and consumption system consisting of power plants, power transmission and transformation lines, power distribution stations and power consumption. Its function is to convert primary energy from nature into electric energy through power generation devices, and then supply electric energy to various users through transmission, transformation and distribution. In order to achieve coordination and control of various links, the power system also includes information and control systems applied to various links such as power generation, transmission, distribution and marketing, and then measures, regulates, controls, protects, communicates and dispatches the production and application process of electric energy to ensure that users obtain safe and high-quality electric energy.
[0003] The development goal of modern power systems is to realize power system automation. The fields of power system automation include automatic detection, regulation and control of production processes, automatic safety protection of systems and components, automatic transmission of network information, automatic scheduling of system production, and automated economic management of enterprises. The main goal of power system automation is to ensure the quality of power supply (frequency and voltage), ensure the safety and reliability of system operation, and improve economic benefits and management efficiency.
[0004] In the power distribution and consumption links, smart energy meters and fusion terminals can realize remote centralized power information collection, and remote power monitoring and management can be realized based on the collected power information big data. For example, in the existing power system, the fusion terminal and smart energy meters can realize automatic meter reading in the substation area, and realize automatic settlement of electricity charges in combination with the online payment system.
[0005] Power information big data can also be applied to power system operation status monitoring, etc. However, there are many power equipment in a single distribution station area, and the load status of power nodes is complex. How to identify and predict the operation status of the power system through complex power information is still a technical problem that needs to be solved urgently. Summary of the invention
[0006] In order to solve the problems of large amount of power information data in the substation, high difficulty in analysis, and unpredictable power supply stability in the substation, the present invention provides a substation power supply stability monitoring system based on a fusion terminal.
[0007] The present invention is implemented by the following technical solutions:
[0008] A power supply stability monitoring system for an area based on a fusion terminal is used to collect power information from transformers and power nodes in the area, and then monitor the operating status of the area in real time based on the collected power information to predict the power supply stability of the area.
[0009] The monitoring system provided by the present invention includes a data acquisition device and a master station server. The data acquisition device is installed on-site in the distribution station area, and the master station server is set in the cloud. The data acquisition device is connected to the master station server for communication.
[0010] Among them, the data collection equipment includes: smart energy meters, concentrators and fusion terminals. Smart energy meters are installed at the power consumption nodes of each power user to collect power information on the user side. The concentrator is used to obtain the collected data of the smart energy meters at each power consumption node. The concentrator is connected to the fusion terminal in communication, and the concentrator sends the data obtained from different power consumption nodes to the fusion terminal. The fusion terminal is also electrically connected to the substation transformer of the low-voltage power grid to obtain power information on the power grid side. The fusion terminal is connected to the main station server in communication, and the fusion terminal sends the synchronously collected user-side power information and power grid-side power information to the main station server.
[0011] The main station server runs a data normalization module, a clustering model based on an adaptive fast search density peak algorithm, and a substation operation status identification model based on a BP neural network. The data normalization module is used to normalize the collected power information, and then obtain a sample data set containing normalized data of all nodes. The clustering model is used to cluster the sample data sets of each node, determine the number of categories and cluster centers, and obtain a node attribute data set containing all node status information. The substation operation status identification model is used to predict the current power supply stability status of the substation based on the node attribute data set; the clustering model has three cluster centers, corresponding to the "underload", "stable" or "overload" status of the node. The prediction results of the substation operation status identification model are divided into two states: normal and abnormal.
[0012] As a further improvement of the present invention, the concentrator is connected to the smart meter through an RS485 serial bus interface. The fusion terminal is connected to the concentrator and the transformer in the area through Ethernet or power carrier communication. The fusion terminal is connected to the master station server through 4G, 5G mobile communication or Ethernet communication.
[0013] As a further improvement of the present invention, the power information collected by the fusion terminal on the user side includes: power supply voltage V1, power supply current I1, power factor Line loss rate ΔP. The power information on the grid side includes: equipment voltage V2, equipment current I2, real-time load P, and equipment temperature T of the transformer in the substation area.
[0014] As a further improvement of the present invention, the normalization module is used to map data of different dimensions to the (-1,1) interval using a normalization formula, and the normalization formula is as follows: Among them, x represents the measured value of the current sample data; Represents the normalized value of the current sample data; x max Indicates the theoretical upper limit of the safety threshold of the current sample data; x min Indicates the theoretical lower limit of the safety threshold of the current sample data. When a certain sample data does not have a lower limit of the safety threshold, then x min =0.
[0015] As a further improvement of the present invention, the clustering process of the clustering model is as follows:
[0016] First, obtain the sample data set of any node and calculate the Euclidean distance d between any two sample data in the sample data set. ij , the calculation formula is as follows:
[0017]
[0018] In the above formula, A represents the sample data set, N represents the number of data points in the sample data set A; x i and x j Represents two random data points in the sample data set A; dist() represents the Euclidean distance calculation function.
[0019] Secondly, according to the relationship between the Euclidean distance between the sample data and all other data points and the preset cutoff distance d0, the local density ρ of each data point in the sample data is calculated i ; The calculation formula is as follows:
[0020]
[0021] in, Represents a custom classification function used to distinguish whether the Euclidean distance between the data point and the center point is less than the cutoff distance, and satisfies:
[0022]
[0023] Then, based on the local density at each data point in the sample data set, the distance θ between the sample data and the density center is calculated. i The calculation process is as follows: Determine whether the local density of the current data point is the maximum value in the sample data set: If yes, then the current data point x i The maximum distance from other data points in the sample data set is taken as θ i , the calculation formula is: i =max j (d ij), j = N. Otherwise, the data point x in the sample data set with a local density greater than the current data point is j , and the current data point x i The minimum distance is taken as θ i ; The calculation formula is: θ i =min(d ij ), x j :ρ j >ρ i .
[0024] Finally, according to the ρ of each sample data in the sample data set i and θ i Draw a decision diagram. The horizontal coordinate of each sample point in the decision diagram is ρ i , the vertical axis is θ i ; Then determine the cluster center and the number of categories based on the decision diagram.
[0025] As a further improvement of the present invention, the substation operation status recognition model is a BP neural network with a three-layer structure; the BP neural network includes an input layer, a hidden layer and an output layer. The number of input layer nodes is equal to the number of power nodes in the current substation, the number of output layer nodes is 1; the number of hidden layer nodes = 2*the number of input layer nodes + 1.
[0026] As a further improvement of the present invention, in the BP neural network, the hyperbolic tangent function Tanh is used as the activation function between the input layer and the hidden layer. The Tanh activation function is as follows:
[0027]
[0028] The nonlinear transformation function Sigmoid is used as the activation function between the hidden layer and the output layer; the expression of the Sigmoid activation function is as follows:
[0029]
[0030] As a further improvement of the present invention, in the BP neural network, the transfer formula from the input layer to the hidden layer is:
[0031]
[0032] In the above formula, x i is the input value of the i-th input layer, i=N, N represents the number of nodes in the input layer. 1j is the output of the jth node in the hidden layer, j = 2N + 1. f1 is the Tanh activation function, ω ij is the weight between the i-th node in the input layer and the j-th node in the hidden layer, x i is the input value of the i-th node in the input layer, a j is the threshold of the jth node in the hidden layer.
[0033] The transfer formula from the hidden layer to the output layer is:
[0034]
[0035] In the above formula, y is the output value of the output layer, f2 is the Sigmoid activation function, ω j is the weight between the jth node in the hidden layer and the output layer, and b is the output layer threshold.
[0036] As a further improvement of the present invention, the training process of the BP neural network is as follows:
[0037] S1: Set the initial training parameters: including the number of iterations, target minimum error, and learning rate η.
[0038] S2: Use the pre-collected node attribute data sets under different operating conditions as training samples, and use the BP neural network to perform forward propagation on the training samples.
[0039] S3: Calculate the absolute sum of squares of the relative errors between the predicted output of the output layer and the expected output in each round of forward propagation, and determine whether the target minimum error is met:
[0040] (1) The training process of the network model is completed.
[0041] (2) Otherwise, enter the back propagation process.
[0042] S4: Use the gradient descent method to perform the back propagation process, and dynamically update the output layer weight and the weight between the input layer and the hidden layer according to the weight update formula from the hidden layer to the output layer and the weight update formula between the input layer and the hidden layer.
[0043] Among them, the weight update formula between the input layer and the hidden layer is as follows:
[0044]
[0045] In the above formula, ω′ ij represents the updated weight between the i-th node in the input layer and the j-th node in the hidden layer, represents the expected output of the current input sample in the network model, y represents the actual output of the current input sample in the network model, and E represents the relative sum of squares of the error between the expected output and the actual output.
[0046] The weight update formula between the hidden layer and the output layer is as follows:
[0047]
[0048] In the above formula, ω′ jRepresents the updated weight between the jth node in the hidden layer and the output layer.
[0049] S5: Continue to input the training samples into the BP neural network after weight update, and start the next round of forward propagation again.
[0050] S6: loop steps S3-S4 until the preset number of iterations is reached, or the absolute sum of squares of the relative errors between the predicted output of the network model and the expected output meets the requirement of the target minimum error.
[0051] As a further improvement of the present invention, a genetic algorithm is also used to optimize the threshold value during the BP neural network training process. The optimization process is as follows:
[0052] S01: The threshold of the BP neural network is converted into the corresponding chromosome individuals by real number coding, and then an initial population containing multiple chromosome individuals is randomly generated.
[0053] S02: Set the iterative termination condition of the classic genetic algorithm to be synchronized with the training phase of the BP neural network, and use the classic genetic algorithm to iteratively optimize the initial population. Each round of iteration includes the following:
[0054] S021: Calculate the fitness of each chromosome in the initial population using a preset fitness function.
[0055] S022: Use the selection operator of the classical genetic algorithm to select the initial population; the selection operator uses the elite retention operator.
[0056] S023: Use the crossover operator of the classic genetic algorithm to perform a crossover operation on the initial population. The crossover operator calculates the similarity between any two individuals in the current population through a custom similarity function, and then performs a two-point crossover on the two individuals with the lowest similarity.
[0057] S024: Use the mutation operator of the classic genetic algorithm to perform mutation operation on the initial population. The mutation operator performs single-point mutation on the chromosomes with the highest fitness and the lowest fitness according to the preset mutation probability.
[0058] S03: In each round of population iteration, the threshold of the chromosome representation with the maximum fitness is output to the BP neural network as the threshold of the BP neural network in the next training round.
[0059] The technical solution provided by the present invention has the following beneficial effects:
[0060] The present invention constructs a data collection and data analysis system based on a fusion terminal, uses the basic power information collected from each node in the substation to analyze the operating status of the substation, and predicts the stability of the power supply in the substation according to the power supply and demand relationship in the substation. The system provided by the present invention introduces clustering algorithms, BP neural networks, genetic algorithms, etc. into the system, and constructs a network model that can effectively analyze the operating status of the power grid. And the powerful data processing capabilities of the cloud are used to solve the calculation and processing of local data; thus, a system that can perform real-time analysis and online monitoring of the power supply stability of the substation is obtained, so it has outstanding practical value; it also lays the foundation for achieving the goal of fully automated power grid management in the substation. BRIEF DESCRIPTION OF THE DRAWINGS
[0061] Figure 1 A flowchart of the steps of a real-time analysis method for low-voltage power supply stability provided in Example 1 of the present invention.
[0062] Figure 2 This is a flowchart of the steps of clustering a multi-node sample data set using the adaptive fast density peak search method in Example 1 of the present invention.
[0063] Figure 3 This is a program flow chart of the BP neural network training phase constructed in Example 1 of the present invention.
[0064] Figure 4 This is a flowchart of a program for optimizing the BP neural network threshold using a classical genetic algorithm in Example 1 of the present invention.
[0065] Figure 5 This is a system architecture diagram of a substation power supply stability monitoring system based on a fusion terminal provided in Example 2 of the present invention.
[0066] Figure 6 This is the topology diagram of the substation power supply stability monitoring system based on the integrated terminal.
[0067] Figure 7 This is a network architecture diagram of the BP neural network model constructed in Example 2 of the present invention. DETAILED DESCRIPTION
[0068] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.
[0069] Example 1
[0070] This embodiment provides a real-time analysis method for low-voltage power supply stability, which is used to analyze the current power supply operation status of the power supply area based on the power monitoring data collected by the fusion terminal. Figure 1 As shown, the real-time analysis method specifically includes the following steps:
[0071] S1: Collect the power information of each power node in the area under different states as the sample data of the current node. The sample data includes: the power supply voltage V1, power supply current I1, power factor Line loss rate ΔP. Equipment voltage V2, equipment current I2, real-time load P, equipment temperature T of the transformer in the substation area. And other related data related to the operation status of the power grid.
[0072] In addition to abnormalities caused by equipment failure or other accidents in existing power grids, the most common reasons that affect the stable operation of power grids are imbalance in supply and demand, and the impact of fluctuations in power demand on the power grid on the user side. Therefore, the real-time analysis of power supply stability in this embodiment mainly analyzes the balance relationship between power supply and demand through power information on the power supply side and the user side in the power grid, and analyzes and predicts possible abnormal conditions.
[0073] Specifically, the power supply voltage V1, power supply current I1, device voltage V2, and device current I2 collected in this embodiment correspond to the voltage and current of each phase A, B, and C respectively; including: V 1A 、V 1B 、V 1C , I 1A ,I 1B ,I 1C , V 2A 、V 2B 、V 2C , I 2A ,I 2B ,I 2C .
[0074] Power factor refers to the ratio of active power to apparent power in an AC circuit. It is a coefficient that measures the efficiency of electrical equipment. A low power factor indicates that the reactive power used by the circuit for alternating magnetic field conversion is large, which reduces the utilization rate of the equipment and increases the power supply loss of the line. The power factor is related to the load properties of the power grid circuit. It is also an important technical indicator for evaluating the power system.
[0075] The line loss rate is the percentage of electric energy lost in the power grid (line loss load) to the electric energy supplied to the power network (power supply load). The line loss rate can evaluate the economic efficiency of the power system operation. The line loss rate is related to the load power factor of the power grid. The voltage supplied by the power system to the load changes with the active and reactive power transmitted by the line. When the active power transmitted by the line and the starting voltage remain unchanged, if the reactive power transmitted is more, the voltage loss of the line will be greater, and the line loss rate will be higher. Therefore, when the power factor is improved, the reactive power absorbed by the load from the system will be reduced, the voltage loss of the line will be reduced accordingly, and the line loss rate will be reduced.
[0076] The equipment temperature is an indicator for evaluating the tolerance of the transformer equipment. The transformer needs to dissipate heat during operation. When the real-time load of the transformer increases, the equipment temperature will also increase accordingly. If the ambient temperature and other conditions cause the transformer to be unable to dissipate heat effectively, the equipment temperature of the transformer will rise rapidly, which may cause the transformer to be unable to operate at the rated load, thereby affecting the normal power supply in the substation. Therefore, this embodiment also needs to take the equipment temperature of the transformer into account when analyzing the operation status of the power grid.
[0077] S2: According to the theoretical safety threshold of each data during operation, a large amount of collected sample data is normalized respectively, and then a sample data set containing all normalized sample data is obtained.
[0078] The normalization formula for sample data is as follows:
[0079]
[0080] In the above formula, x represents the measured value of the current sample data; Represents the normalized value of the current sample data; x max Indicates the theoretical upper limit of the safety threshold of the current sample data; x min represents the theoretical lower limit of the safety threshold of the current sample data; when a certain sample data does not have a lower limit of the safety threshold, then x min =0.
[0081] In this embodiment, the dimensions and dimensional units of the various data used are different, which is not conducive to the later data analysis. This embodiment eliminates the dimensional influence between different index items through normalization. After normalization, this embodiment also maps data of different dimensions to the (-1,1) interval, so that the data indicators are comparable. After the original data is processed by data standardization, each indicator is at the same order of magnitude, which is suitable for comprehensive comparative evaluation.
[0082] S3: Cluster the sample data set of each node based on the adaptive fast search density peak method, determine the cluster center and the number of categories, and obtain the node attribute data set after clustering.
[0083] like Figure 2 As shown in the figure, the process of clustering the multi-node sample data set by the adaptive fast search density peak method is as follows:
[0084] S31: Obtain a sample data set of any node and calculate the Euclidean distance d between any two sample data in the sample data set ij , the calculation formula is as follows:
[0085]
[0086] In the above formula, A represents the sample data set, N represents the number of data points in the sample data set A; i and x j Represents two random data points in the sample data set A; dist() represents the Euclidean distance calculation function.
[0087] S32: Calculate the local density ρ of each data point in the sample data according to the relationship between the Euclidean distance between the sample data and all other data points and the preset cutoff distance d0 i The calculation formula is as follows:
[0088]
[0089] in, Represents a custom classification function used to distinguish whether the Euclidean distance between the data point and the center point is less than the cutoff distance, and satisfies:
[0090]
[0091] S33: Based on the local density of each data point in the sample data set, calculate the distance θ between the sample data and the density center i The calculation process is as follows:
[0092] Determine whether the local density of the current data point is the maximum value in the sample data set:
[0093] (1) If yes, then the current data point x i The maximum distance from other data points in the sample data set is taken as θ i , the calculation formula is: i =max j (d ij ), j=N.
[0094] (2) Otherwise, the data point x in the sample data set with a local density greater than the current data point j, and the current data point x i The minimum distance is taken as θ i ; The calculation formula is: θ i =min(d ij ), x j :ρ j >ρ i .
[0095] S34: According to the ρ of each sample data in the sample data set i and θ i Draw a decision diagram. The horizontal coordinate of each sample point in the decision diagram is ρ i , the vertical axis is θ i ; Then determine the cluster center and the number of categories based on the decision diagram.
[0096] Specifically, the category of each cluster center is determined by the corresponding power consumption node state. In this embodiment, the number of categories of the node attribute data set is 3; the category of each node in the node attribute data set is divided into "underload", "stable" or "overload".
[0097] The adaptive fast density peak search method provided in this embodiment determines the cluster center by analyzing the density of different data points. The basis for clustering is that the density of the cluster center is greater than the density of its surrounding neighboring points. This method can automatically obtain the number of categories based on the collected data set without determining the initial cluster center. Instead, it uses the characteristic that the local density of the category center is always higher than the local density of its nearest neighbor point to quickly find the density peak of any shape data set, effectively distribute non-central sample points, and determine the cluster center using a decision diagram. The adaptive fast density peak search method provided in this embodiment is suitable for multi-category clustering and significantly improves the efficiency of clustering processing.
[0098] S4: Construct a BP neural network with a three-layer structure, including an input layer, a hidden layer, and an output layer. In the BP neural network, the number of input layer nodes n is equal to the number of power nodes in the current area; the number of output layer nodes is 1, and the number of hidden layer nodes is 2n+1.
[0099] In the constructed BP neural network,
[0100] The hyperbolic tangent function Tanh is used as the activation function between the input layer and the hidden layer. The Tanh activation function is as follows:
[0101]
[0102] The nonlinear transformation function Sigmoid is used as the activation function between the hidden layer and the output layer. The value range of sigmoid is (0, 1), and it is monotonically continuous and differentiable everywhere, so it can achieve binary classification output. The expression of the Sigmoid activation function is as follows:
[0103]
[0104] The transfer formula from the input layer to the hidden layer is:
[0105]
[0106] In the above formula, x i is the input value of the i-th input layer, i=N, N represents the number of nodes in the input layer; H 1j is the output of the jth node in the hidden layer, j = 2N + 1; f1 is the Tanh activation function, ω ij is the weight between the i-th node in the input layer and the j-th node in the hidden layer, x i is the input value of the i-th node in the input layer, a j is the threshold of the jth node in the hidden layer.
[0107] The transfer formula from the hidden layer to the output layer is:
[0108]
[0109] In the above formula, y is the output value of the output layer, f2 is the Sigmoid activation function, ω j is the weight between the jth node in the hidden layer and the output layer, and b is the output layer threshold.
[0110] S5: A large number of node attribute data sets under different operating conditions collected in advance are used as training samples, and the BP neural network constructed in the previous step is trained using the training samples to update the weights of the network model. At the same time, the threshold of the BP neural network is optimized in combination with the classical genetic algorithm during the training process. The trained BP neural network is used as the required substation operation status recognition model.
[0111] In this embodiment, if Figure 3 As shown in the figure, the training process of BP neural network is as follows:
[0112] S51: Set initial training parameters: including number of iterations, target minimum error, and learning rate η.
[0113] S52: Using the node attribute data sets under different operating states collected in advance as training samples, and using the BP neural network to perform forward propagation on the training samples.
[0114] S53: Calculate the absolute square sum of the relative errors E between the predicted output of the output layer and the expected output in each round of forward propagation, and determine whether the target minimum error is met:
[0115] (1) The training process of the network model is completed.
[0116] (2) Otherwise, the back propagation process is entered.
[0117] S54: The gradient descent method is used for the back propagation process, and the weight of the output layer and the weight between the input layer and the hidden layer are dynamically updated according to the weight update formula from the hidden layer to the output layer and the weight update formula between the input layer and the hidden layer.
[0118] Specifically, the weight update formula between the input layer and the hidden layer is as follows:
[0119]
[0120] In the above formula, ω′ ij Represents the updated weight between the i-th node in the input layer and the j-th node in the hidden layer; represents the expected output of the current input sample in the network model, y represents the actual output of the current input sample in the network model; E represents the relative sum of squares of the error between the expected output and the actual output.
[0121] The weight update formula between the hidden layer and the output layer is as follows:
[0122]
[0123] In the above formula, ω′ j Represents the updated weight between the jth node in the hidden layer and the output layer.
[0124] S55: continue to input the training samples into the BP neural network after weight update, and re-perform the next round of forward propagation.
[0125] S56: looping steps S53-S54 until the preset number of iterations is reached, or the absolute sum of squares of the relative errors between the predicted output of the network model and the expected output meets the requirement of the target minimum error.
[0126] S6: The real-time power information of the transformers and all power nodes in the substation is collected through the fusion terminal, and the real-time power information is normalized and clustered in turn to obtain a real-time node attribute data set. Then the real-time node attribute data set is input into the trained substation operation status recognition model, and the model outputs the real-time operation status of the current substation. Among them, the real-time operation status of the substation is divided into normal and abnormal.
[0127] In particular, in this embodiment, if Figure 4 As shown in the figure, the process of optimizing the threshold of BP neural network using the classic genetic algorithm is as follows:
[0128] S01: The threshold of the BP neural network is converted into the corresponding chromosome individuals by real number coding, and then an initial population containing multiple chromosome individuals is randomly generated.
[0129] S02: The iterative termination condition of the classical genetic algorithm is set to be synchronized with the training phase of the BP neural network, and the initial population is iteratively optimized using the classical genetic algorithm. Each round of iteration includes the following:
[0130] S021: Calculate the fitness of each chromosome in the initial population using a preset fitness function.
[0131] S022: Use the selection operator of the classical genetic algorithm to perform selection operations on the initial population.
[0132] S023: Use the crossover operator of the classic genetic algorithm to perform a crossover operation on the initial population.
[0133] S024: Use the mutation operator of the classic genetic algorithm to mutate the initial population.
[0134] S03: In each round of population iteration, the threshold of the chromosome representation with the maximum fitness is output to the BP neural network as the threshold of the BP neural network in the next training round.
[0135] In the classical genetic algorithm used in this embodiment, the selection operator uses the elite retention operator. The crossover operator calculates the similarity between any two individuals in the current population through a custom similarity function, and then performs a two-point crossover on the two individuals with the lowest similarity. The mutation operator performs single-point mutation on the chromosomes with the highest and lowest fitness according to the preset mutation probability.
[0136] In the final stage of the technical solution, this embodiment uses a BP neural network to complete the prediction task of the power supply stability status of the power grid based on the node attribute data sets of different nodes. The BP neural network adopts a step-by-step approximation method. In the continuous algorithm training process, the obtained results and the expected results are analyzed for errors, and then the weights and thresholds are modified to obtain a model that can output consistent results with the expected results step by step. In particular, in order to improve the training effect of the network model, this embodiment also introduces a genetic algorithm into the threshold update process of the network model, thereby significantly improving the convergence rate of the BP neural network model. The trained network model of this embodiment has a high response speed, so the power supply status of the substation can be dynamically evaluated based on the data collected in real time.
[0137] Example 2
[0138] On the basis of Example 1, this embodiment further provides a substation power supply stability monitoring system based on a fusion terminal, which is used to collect power information of transformers and power nodes in the substation, and then use the collected power information to perform online analysis of the substation operation status through the real-time analysis method of low-voltage substation power supply stability such as Example 1, and predict the power supply stability of the substation.
[0139] like Figure 5 As shown, the monitoring system provided in this embodiment includes a data acquisition device and a master station server. The data acquisition device is installed on-site in the distribution station area, and the master station server is set in the cloud. The data acquisition device is connected to the master station server for communication.
[0140] Among them, the data collection equipment includes: smart energy meters, concentrators and fusion terminals. Smart energy meters are installed at the power consumption nodes of each power user to collect power information on the user side. The concentrator is used to obtain the collected data of the smart energy meters at each power consumption node. The concentrator is connected to the fusion terminal in communication, and the concentrator sends the data obtained from different power consumption nodes to the fusion terminal. The fusion terminal is also electrically connected to the substation transformer of the low-voltage power grid to obtain power information on the power grid side. The fusion terminal is connected to the main station server in communication, and the fusion terminal sends the synchronously collected user-side power information and power grid-side power information to the main station server.
[0141] like Figure 6 As shown, a data normalization module, a clustering model based on an adaptive fast search density peak algorithm, and a substation operation status identification model based on a BP neural network are running in the main station server. The data normalization module is used to normalize the collected power information, and then obtain a sample data set containing normalized data of all nodes. The clustering model is used to cluster the sample data sets of each node, determine the number of categories and cluster centers, and obtain a node attribute data set containing all node status information. The substation operation status identification model is used to predict the current power supply stability status of the substation based on the node attribute data set; the clustering model has three cluster centers, corresponding to the "underload", "stable" or "overload" status of the node. The prediction results of the substation operation status identification model are divided into two states: normal and abnormal.
[0142] The concentrator is connected to the smart meter through the RS485 serial bus interface. The fusion terminal is connected to the concentrator and the transformer in the substation area through Ethernet or power carrier communication. The fusion terminal is connected to the master station server through 4G, 5G mobile communication or Ethernet communication.
[0143] In this embodiment, the power information collected by the fusion terminal on the user side includes: power supply voltage V1, power supply current I1, power factor Line loss rate ΔP. The power information on the grid side includes: equipment voltage V2, equipment current I2, real-time load P, and equipment temperature T of the transformer in the substation area.
[0144] The normalization module is used to map data of different dimensions to the (-1,1) interval using a normalization formula. The normalization formula is as follows: Among them, x represents the measured value of the current sample data; Represents the normalized value of the current sample data; x max Indicates the theoretical upper limit of the safety threshold of the current sample data; x min Indicates the theoretical lower limit of the safety threshold of the current sample data. When a certain sample data does not have a lower limit of the safety threshold, then x min =0.
[0145] In the monitoring system, the clustering model uses an adaptive fast search density peak algorithm to cluster the sample data set. The clustering process of the model is as follows:
[0146] First, obtain the sample data set of any node and calculate the Euclidean distance d between any two sample data in the sample data set. ij , the calculation formula is as follows:
[0147]
[0148] In the above formula, A represents the sample data set, N represents the number of data points in the sample data set A; x i and x j Represents two random data points in the sample data set A; dist() represents the Euclidean distance calculation function.
[0149] Secondly, according to the relationship between the Euclidean distance between the sample data and all other data points and the preset cutoff distance d0, the local density ρ of each data point in the sample data is calculated i ; The calculation formula is as follows:
[0150]
[0151] in, Represents a custom classification function used to distinguish whether the Euclidean distance between the data point and the center point is less than the cutoff distance, and satisfies:
[0152]
[0153] Then, based on the local density at each data point in the sample data set, the distance θ between the sample data and the density center is calculated. i The calculation process is as follows: Determine whether the local density of the current data point is the maximum value in the sample data set: If yes, then the current data point x i The maximum distance from other data points in the sample data set is taken as θ i , the calculation formula is: i =max j (d ij ), j = N. Otherwise, the data point x in the sample data set with a local density greater than the current data point is j, and the current data point x i The minimum distance is taken as θ i ; The calculation formula is: θ i =min(d ij ), x j :ρ j >ρ i .
[0154] Finally, according to the ρ of each sample data in the sample data set i and θ i Draw a decision diagram. The horizontal coordinate of each sample point in the decision diagram is ρ i , the vertical axis is θ i ; Then determine the cluster center and the number of categories based on the decision diagram.
[0155] The substation operation status recognition model constructed in this embodiment is a BP neural network with a three-layer structure. Figure 7 As shown, the BP neural network includes an input layer, a hidden layer, and an output layer. The number of input layer nodes is equal to the number of power nodes in the current area, and the number of output layer nodes is 1; the number of hidden layer nodes = 2*the number of input layer nodes + 1.
[0156] Among them, in the BP neural network, the hyperbolic tangent function Tanh is used as the activation function between the input layer and the hidden layer. The Tanh activation function is as follows:
[0157]
[0158] The nonlinear transformation function Sigmoid is used as the activation function between the hidden layer and the output layer; the expression of the Sigmoid activation function is as follows:
[0159]
[0160] In the BP neural network constructed in this embodiment, the transfer formula from the input layer to the hidden layer is:
[0161]
[0162] In the above formula, x i is the input value of the i-th input layer, i=N, N represents the number of nodes in the input layer. 1j is the output of the jth node in the hidden layer, j = 2N + 1. f1 is the Tanh activation function, ω ij is the weight between the i-th node in the input layer and the j-th node in the hidden layer, x i is the input value of the i-th node in the input layer, a j is the threshold of the jth node in the hidden layer.
[0163] The transfer formula from the hidden layer to the output layer is:
[0164]
[0165] In the above formula, y is the output value of the output layer, f2 is the Sigmoid activation function, ω j is the weight between the jth node in the hidden layer and the output layer, and b is the output layer threshold.
[0166] In the substation operation status recognition model provided in this embodiment, the training process of the BP neural network is as follows:
[0167] S1: Set the initial training parameters: including the number of iterations, target minimum error, and learning rate η.
[0168] S2: Use the pre-collected node attribute data sets under different operating conditions as training samples, and use the BP neural network to perform forward propagation on the training samples.
[0169] S3: Calculate the absolute sum of squares of the relative errors between the predicted output of the output layer and the expected output in each round of forward propagation, and determine whether the target minimum error is met:
[0170] (1) The training process of the network model is completed.
[0171] (2) Otherwise, enter the back propagation process.
[0172] S4: Use the gradient descent method to perform the back propagation process, and dynamically update the output layer weight and the weight between the input layer and the hidden layer according to the weight update formula from the hidden layer to the output layer and the weight update formula between the input layer and the hidden layer.
[0173] Among them, the weight update formula between the input layer and the hidden layer is as follows:
[0174]
[0175] In the above formula, ω′ ij represents the updated weight between the i-th node in the input layer and the j-th node in the hidden layer, represents the expected output of the current input sample in the network model, y represents the actual output of the current input sample in the network model, and E represents the relative sum of squares of the error between the expected output and the actual output.
[0176] The weight update formula between the hidden layer and the output layer is as follows:
[0177]
[0178] In the above formula, ω′ j Represents the updated weight between the jth node in the hidden layer and the output layer.
[0179] S5: Continue to input the training samples into the BP neural network after weight update, and start the next round of forward propagation again.
[0180] S6: loop steps S3-S4 until the preset number of iterations is reached, or the absolute sum of squares of the relative errors between the predicted output of the network model and the expected output meets the requirement of the target minimum error.
[0181] In this embodiment, the number of iterations and the error being less than a preset value are both used as conditions for terminating the iteration. When either of the two conditions is met, the recursive process of the planting algorithm training is started.
[0182] In this embodiment, in order to improve the training effect and convergence rate of the algorithm model, a genetic algorithm is also used to optimize the threshold during the BP neural network training process. The optimization process is as follows. It has been described in detail in the previous embodiment and will not be repeated in this embodiment.
[0183] Example 3
[0184] This embodiment provides a real-time analysis device for low-voltage power supply stability, which includes 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 steps of the real-time analysis method for low-voltage power supply stability in embodiment 1 are implemented, and then the prediction result of the power supply stability in the substation is analyzed based on the collected power information in the substation.
[0185] The computer device may be a smart phone, tablet computer, laptop computer, desktop computer, rack server, blade server, tower server or cabinet server (including an independent server or a server cluster composed of multiple servers) capable of executing a program. The computer device of this embodiment includes at least but is not limited to: a memory and a processor that can be interconnected through a system bus.
[0186] In this embodiment, the memory (i.e., readable storage medium) includes flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory may be an internal storage unit of a computer device, such as a hard disk or memory of the computer device. In other embodiments, the memory may also be an external storage device of a computer device, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the computer device. Of course, the memory may also include both the internal storage unit of the computer device and its external storage device. In this embodiment, the memory is generally used to store an operating system and various application software installed on the computer device. In addition, the memory may also be used to temporarily store various types of data that have been output or are to be output.
[0187] The processor may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip in some embodiments. The processor is generally used to control the overall operation of a computer device. In this embodiment, the processor is used to run program codes stored in a memory or process data.
[0188] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A power supply stability monitoring system for a substation based on a fusion terminal, which is used to collect power information of transformers and power nodes in the substation, and then monitor the operation status of the substation in real time according to the collected power information, and predict the power supply stability of the substation; it is characterized by: The monitoring system includes a data acquisition device and a main station server; the data acquisition device is in communication connection with the main station server; The data collection equipment includes: a smart energy meter, a concentrator and a fusion terminal; the smart energy meter is installed at the power consumption node of each power user and is used to collect power information on the user side; the concentrator is used to obtain the collected data of the smart energy meter at each power consumption node; the concentrator is connected to the fusion terminal in communication, and the concentrator sends the acquired data at different power consumption nodes to the fusion terminal; the fusion terminal is also electrically connected to the substation transformer of the low-voltage power grid to obtain power information on the power grid side; the fusion terminal is connected to the master station server in communication, and the fusion terminal sends the synchronously collected power information on the user side and the power information on the power grid side to the master station server; The master station server runs a data normalization module, a clustering model based on an adaptive fast search density peak algorithm, and a substation operation status identification model based on a BP neural network; the data normalization module is used to normalize the collected power information, and then obtain a sample data set containing normalized data of all nodes; the clustering model is used to cluster the sample data sets of each node, determine the number of categories and cluster centers, and obtain a node attribute data set containing all node status information; the substation operation status identification model is used to predict the current power supply stability state of the substation according to the node attribute data set; the clustering model has three cluster centers, which correspond to the "underload", "stable" or "overload" states of the node respectively; the prediction results of the substation operation status identification model are divided into two states: normal and abnormal; The clustering process of the clustering model is as follows: First, obtain the sample data set of any node and calculate the Euclidean distance d between any two sample data in the sample data set. ij , the calculation formula is as follows: In the above formula, A represents the sample data set, N represents the number of data points in the sample data set A; i and x j Represents two random data points in the sample data set A; dist() represents the Euclidean distance calculation function; Secondly, according to the relationship between the Euclidean distance between the sample data and all other data points and the preset cutoff distance d0, the local density ρ of each data point in the sample data is calculated i ; The calculation formula is as follows: in, Represents a custom classification function used to distinguish whether the Euclidean distance between the data point and the center point is less than the cutoff distance, and satisfies: Then, based on the local density at each data point in the sample data set, the distance θ between the sample data and the density center is calculated. i The calculation process is as follows: Determine whether the local density of the current data point is the maximum value in the sample data set: If yes, then the current data point x i The maximum distance from other data points in the sample data set is taken as θ i , the calculation formula is: i =max(d ij ), j = 1…N; otherwise, the data point x in the sample data set with a local density greater than the current data point j , and the current data point x i The minimum distance is taken as θ i ; The calculation formula is: θ i =min(d ij ), x j :ρ j >ρ i ; Finally, according to the ρ of each sample data in the sample data set i and θ i Draw a decision diagram; the horizontal coordinate of each sample point in the decision diagram is ρ i , the vertical axis is θ i ; Then determine the cluster center and the number of categories based on the decision diagram.
2. The power supply stability monitoring system based on the fusion terminal according to claim 1 is characterized in that: The concentrator is connected to the smart meter through an RS485 serial bus interface; the fusion terminal is connected to the concentrator and the substation voltage device through Ethernet or power carrier communication; the fusion terminal is connected to the main station server through 4G, 5G mobile communication or Ethernet communication.
3. The power supply stability monitoring system based on the integrated terminal according to claim 1, characterized in that: The power information collected by the fusion terminal on the user side includes: power supply voltage V1, power supply current I1, power factor Line loss rate ΔP; the power information on the grid side includes: equipment voltage V2, equipment current I2, real-time load P, and equipment temperature T of the substation transformer.
4. The power supply stability monitoring system based on the integrated terminal according to claim 1 is characterized in that: The normalization module is used to map data of different dimensions to the (-1,1) interval using a normalization formula. The normalization formula is as follows: In the above formula, x represents the measured value of the current sample data; Represents the normalized value of the current sample data; x max Indicates the theoretical upper limit of the safety threshold of the current sample data; x min represents the theoretical lower limit of the safety threshold of the current sample data; when a certain sample data does not have a lower limit of the safety threshold, then x min =0.
5. The power supply stability monitoring system based on the integrated terminal according to claim 1, characterized in that: The substation operation status identification model is a BP neural network with a three-layer structure; the BP neural network includes an input layer, a hidden layer and an output layer; the number of input layer nodes is equal to the number of power consumption nodes in the current substation, and the number of output layer nodes is 1; the number of hidden layer nodes = 2*the number of input layer nodes + 1.
6. The power supply stability monitoring system based on the integrated terminal according to claim 1, characterized in that: In the BP neural network, the hyperbolic tangent function Tanh is used as the activation function between the input layer and the hidden layer; The nonlinear transformation function Sigmoid is used as the activation function between the hidden layer and the output layer.
7. The power supply stability monitoring system based on the integrated terminal according to claim 1 is characterized in that: The training process of BP neural network is as follows: S1: Set the initial training parameters: including the number of iterations, target minimum error, and learning rate η; S2: Use the pre-collected node attribute data sets under different operating conditions as training samples, and use the BP neural network to perform forward propagation on the training samples; S3: Calculate the absolute sum of squares of the relative errors between the predicted output of the output layer and the expected output in each round of forward propagation, and determine whether the target minimum error is met; if so, the training process of the network model is completed, otherwise, the back propagation process is entered; S4: Use the gradient descent method to perform the back propagation process, and dynamically update the output layer weight and the weight between the input layer and the hidden layer according to the weight update formula from the hidden layer to the output layer and the weight update formula between the input layer and the hidden layer; The weight update formula between the input layer and the hidden layer is as follows: In the above formula, ω ij represents the weight between the i-th node of the input layer and the j-th node of the hidden layer before the update; ω i ' j Represents the updated weight between the i-th node in the input layer and the j-th node in the hidden layer; represents the expected output of the current input sample in the network model, y represents the actual output of the current input sample in the network model; E represents the relative error sum of the expected output and the actual output; The weight update formula between the hidden layer and the output layer is as follows: In the above formula, ω j Represents the weight between the jth node of the hidden layer and the output layer before updating; ω j ′ represents the updated weight between the jth node of the hidden layer and the output layer; S5: continue to input the training samples into the BP neural network after weight update, and re-perform the next round of forward propagation; S6: loop steps S3-S4 until the preset number of iterations is reached, or the absolute sum of squares of the relative errors between the predicted output of the network model and the expected output meets the requirement of the target minimum error.
8. The area power supply stability monitoring system based on the fusion terminal according to claim 7 is characterized in that: Genetic algorithm is also used for threshold optimization during BP neural network training. The optimization process is as follows: S01: Use real number coding to convert the threshold of the BP neural network into the corresponding chromosome individuals, and then randomly generate an initial population containing multiple chromosome individuals; S02: The iterative termination condition of the classical genetic algorithm is set to be synchronized with the training phase of the BP neural network, and the initial population is iteratively optimized using the classical genetic algorithm. Each round of iteration includes the following contents; S021: Calculate the fitness of each chromosome in the initial population using a preset fitness function; S022: using a selection operator of a classical genetic algorithm to perform a selection operation on the initial population; the selection operator uses an elite retention operator; S023: using the crossover operator of the classical genetic algorithm to perform a crossover operation on the initial population; the crossover operator calculates the similarity between any two individuals in the current population through a user-defined similarity function, and then performs a two-point crossover on the two individuals with the lowest similarity; S024: using a mutation operator of a classical genetic algorithm to perform a mutation operation on the initial population; the mutation operator performs single-point mutation on the chromosomes with the highest fitness and the chromosomes with the lowest fitness according to a preset mutation probability; S03: In each round of population iteration, the threshold of the chromosome representation with the maximum fitness is output to the BP neural network as the threshold of the BP neural network in the next training round.
