Low-voltage active transformer area line loss calculation method and system based on clustering and AdaBoost
Through clustering and AdaBoost technology, a low-voltage active table area line loss calculation model is constructed, which solves the problem of uncertainty in line loss in traditional methods in distributed power access, and achieves fast and accurate line loss calculation and management.
Patent Information
- Application Number
- CN202410861875.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-28
- Publication Date
- 2024-11-19
AI Technical Summary
The traditional low-voltage active table area line loss calculation method is difficult to adapt to the dynamic changes in the operating state of the power system, especially in the case of distributed power supply access, which leads to increased line loss uncertainty, large calculation overhead, slow convergence speed, and unavailable Realize lean management.
Using clustering and AdaBoost-based methods, by obtaining line loss historical data of low-voltage active table areas, performing clustering and standardizing processing, training weak classifiers and iterative optimization to obtain strong classifiers, and combining them into the final line loss calculation model. Use this model to perform accurate calculations.
It realizes accurate calculation and effective management of low-voltage active table area line losses, improves calculation speed and accuracy, does not need to obtain the low-voltage grid topology, and has the advantages of rapid learning and accurate reliability.
Smart Images

Figure CN118981634A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of low-voltage power distribution, and relates to a method for calculating line loss in a power distribution area, and in particular to a method and system for calculating line loss in a low-voltage active power distribution area based on clustering and AdaBoost. Background Art
[0002] As an important carrier of energy distribution, the operation efficiency and safety of the power system have received extensive attention. Especially in the distribution network, the line loss management of the low-voltage active substation, as the key node connecting users and the power grid, is of great significance for improving energy utilization efficiency, reducing operating costs and promoting sustainable energy development.
[0003] Traditional line loss calculation methods often rely on empirical formulas or static models, which are difficult to adapt to the dynamic changes in the operating status of the power system. Especially with the emergence of new loads such as distributed power sources and electric vehicle charging piles, the uncertainty and complexity of line losses in substations have increased significantly.
[0004] With the development of machine learning technology, machine learning methods have also been introduced into line loss calculation. Machine learning-based methods include calculation methods based on support vector machines, algorithms based on traditional BP neural networks, calculation methods based on improved K-means clustering and BP neural networks, etc. For example, patent application CN115423264A discloses a line loss interval adaptive evaluation method and system based on deep learning. However, in the calculation of line loss in low-voltage active areas with distributed power sources connected, there are still problems such as high algorithm calculation overhead and slow convergence speed, which results in the calculated line loss value not being a relatively optimal solution and unable to support lean line loss management. Summary of the invention
[0005] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide a method and system for calculating line loss in a low-voltage active substation based on clustering and AdaBoost, so as to solve the problem that the uncertainty of line loss in the low-voltage active substation is significantly enhanced due to the access of distributed power sources, making it difficult to accurately calculate the line loss in the low-voltage active substation.
[0006] The purpose of the present invention can be achieved by the following technical solutions:
[0007] A method for calculating line loss in a low voltage active area based on clustering and AdaBoost, the method comprising:
[0008] Collect the values of the line loss influencing factors of the low-voltage active area to be calculated, determine the sample cluster to which it belongs, and calculate the line loss rate of the low-voltage active area to be calculated based on the strong classifier of the sample cluster in the pre-trained low-voltage active area line loss calculation model;
[0009] The construction process of the low voltage active area line loss calculation model includes the following steps:
[0010] Acquire a low-voltage active area line loss historical data sample, wherein the low-voltage active area line loss historical data sample includes line loss influencing factors and corresponding line loss rates;
[0011] Based on the line loss influencing factors, all low-voltage active substation line loss historical data samples are clustered into several sample clusters;
[0012] For each sample cluster, a weak classifier is trained separately;
[0013] For each sample cluster, the AdaBoost algorithm is iteratively used to train a strong classifier for each sample cluster based on the weak classifier.
[0014] The strong classifiers of all sample clusters are combined into the final low-voltage active area line loss calculation model.
[0015] Furthermore, the line loss influencing factors include multiple ones of the station load rate, power factor, voltage, distributed photovoltaic output and sunshine intensity.
[0016] Furthermore, the specific process of clustering all low-voltage active transformer area line loss historical data samples into a plurality of sample clusters includes:
[0017] Standardize the factors affecting line loss;
[0018] The Euclidean distance between each sample is calculated, and clustering is performed based on the Euclidean distance.
[0019] Furthermore, the K-means clustering algorithm is used to cluster all low-voltage active substation line loss historical data samples into several sample clusters.
[0020] Furthermore, the expression of the standardization process is:
[0021]
[0022] In the formula, z ij represents the standardized value of the jth influencing factor of the i-th sample, x ij represents the sampling value of the jth influencing factor of the i-th sample, represents the sample average value of the jth influencing factor, s ij It represents the sample variance of the j-th influencing factor, and n is the number of samples.
[0023] Furthermore, the weak classifier is obtained by training using a neural network algorithm, an association analysis method or a Bayesian method.
[0024] Furthermore, the process of clustering the kth sample to obtain the strong classifier includes:
[0025] Initialize the weight D of each sample 0i =1 / n, where n is the number of samples;
[0026] Assume h t For the weak classifier trained in the tth iteration, calculate the error rate ε of the weak classifier after each iteration t and weight α t ;
[0027] Update the sample weights for round t+1
[0028] When the error rate ε t The iteration is terminated when <β, where β is the preset error rate threshold;
[0029] The weak classifiers and their weights of each iteration are weighted summed to obtain a strong classifier, that is, In the formula, H k is the strong classifier finally trained for the k-th sample clustering, and T is the total number of iterations.
[0030] Furthermore, the error rate ε of the weak classifier after the tth iteration is t It is expressed as:
[0031]
[0032] The weight α t It is expressed as:
[0033]
[0034] Where D ti is the weight of the i-th sample in the t-th iteration, sign() is the indicator function, when h t (x i )≠y i The value of sign() is 1 if true, and 0 otherwise.
[0035] The present invention also provides a computer-readable storage medium, comprising one or more programs for execution by one or more processors of an electronic device, wherein the one or more programs include instructions for executing the low-voltage active area line loss calculation method based on clustering and AdaBoost as described above.
[0036] The present invention also provides a low-voltage active area line loss calculation system based on clustering and AdaBoost, comprising one or more processors, a memory and one or more programs stored in the memory, wherein the one or more programs include instructions for executing the method described above.
[0037] Compared with the prior art, the present invention has the following beneficial effects:
[0038] 1. When calculating line loss, the present invention first determines the sample cluster to which it belongs based on the actual operating status of the low-voltage active area, and then calculates the line loss rate with the corresponding strong classifier, which can realize accurate calculation and effective management of the line loss of the low-voltage active area.
[0039] 2. The strong classifier of the present invention is obtained by iteratively optimizing the pre-built weak classifier by iteratively adopting the AdaBoost algorithm, and can obtain more accurate and reliable low-voltage active area line loss calculation results.
[0040] 3. The present invention uses a combination of clustering and AdaBoost technology to calculate line losses in low-voltage active substations, which has the advantages of no need to obtain low-voltage power grid topology and fast learning speed. BRIEF DESCRIPTION OF THE DRAWINGS
[0041] Figure 1 A schematic diagram of a process for constructing a line loss calculation model for a low voltage active transformer area in an embodiment of the present invention;
[0042] Figure 2 Schematic diagram of a neural network model used in an embodiment of the present invention. DETAILED DESCRIPTION
[0043] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.
[0044] Example 1
[0045] The present embodiment provides a method for calculating line loss in a low-voltage active substation based on clustering and AdaBoost, the method comprising: collecting numerical values of line loss influencing factors of the low-voltage active substation to be calculated, determining the sample cluster to which it belongs, and obtaining the line loss rate of the low-voltage active substation to be calculated by calculating the strong classifier of the sample cluster according to a pre-trained low-voltage active substation line loss calculation model.
[0046] like Figure 1 As shown, the construction process of the low-voltage active area line loss calculation model of this embodiment includes the following steps:
[0047] S1. Obtain historical data samples of line loss in low-voltage active transformer areas, wherein the historical data samples of line loss in low-voltage active transformer areas include line loss influencing factors and corresponding line loss rates.
[0048] The line loss influencing factors considered in this embodiment include the load rate of the substation, power factor, voltage, distributed photovoltaic output, sunshine intensity, etc.
[0049] S2. Based on the line loss influencing factors, all low-voltage active substation line loss historical data samples are clustered into several sample clusters.
[0050] In this embodiment, the clustering process may include:
[0051] (1) Standardize the factors affecting line loss:
[0052] Assuming the number of historical samples is n, the expression of standardization is:
[0053]
[0054] In the formula, z ij represents the standardized value of the jth influencing factor of the i-th sample, x ij represents the sampling value of the jth influencing factor of the i-th sample, x j represents the sample average value of the jth influencing factor, s ij It represents the sample variance of the j-th influencing factor, and n is the number of samples.
[0055] (2) Calculate the Euclidean distance between each sample and perform clustering based on the Euclidean distance.
[0056] After standardization, the influencing factor data in each sample can be expressed as Zi = (zi1, zi2, ..., zim), where Z i represents the i-th sample, m represents the type of influencing factor, then the Euclidean distance between samples is expressed as:
[0057]
[0058] d(Z j ,Z k )<δ are classified into the same cluster, where δ is the preset threshold.
[0059] In other embodiments, the K-means clustering algorithm may also be used to cluster all low-voltage active transformer area line loss historical data samples into a number of sample clusters.
[0060] S3. For each sample cluster, a weak classifier is trained separately.
[0061] In this embodiment, a neural network algorithm is used to construct the weak classifier. Specifically, the construction is as follows: Figure 2 In the neural network model shown, the input layer is the influencing factors of each sample in the sample cluster, and the output layer is the line loss rate. The weight coefficients in the neural network are determined by training the samples, and finally a weak classifier under the sample cluster is formed.
[0062] In addition to using the simplest neural network mentioned above, it can also be replaced by other neural network models such as MLP, CNN, LSTM, etc., and their functions are the same.
[0063] In other embodiments, the weak classifier may be trained by using other prediction methods such as association analysis or Bayesian method.
[0064] S4. For each sample cluster, the AdaBoost algorithm is iteratively used to train a strong classifier for each sample cluster based on the weak classifier.
[0065] Specifically, the process of clustering the kth samples to obtain the strong classifier includes:
[0066] First, initialize the weight D of each sample 0i =1 / n, where n is the number of samples.
[0067] Then, let h t For the weak classifier trained in the tth iteration, calculate the error rate ε of the weak classifier after each iteration t and weight α t :
[0068] The error rate ε of the weak classifier after the tth iteration t It is expressed as:
[0069]
[0070] The weight α t It is expressed as:
[0071]
[0072] Where D ti is the weight of the i-th sample in the t-th iteration, sign() is the indicator function, when h t (x i )≠y i The value of sign() is 1 if true, and 0 otherwise.
[0073] Update the sample weights for round t+1
[0074] The above update process ensures that samples that are misclassified by the weak classifier will have a higher weight in the next iteration, while the weight of samples that are correctly classified will be reduced. In this way, in each iteration, we can try to correct the errors of the previous round and gradually improve the performance of the classifier.
[0075] Finally, when the error rate ε tThe iteration is terminated when β is less than β, where β is the preset error rate threshold. The final strong classifier is obtained by weighted summation of the weak classifiers and their weights in each iteration, that is, In the formula, H k is the strong classifier finally trained for the k-th sample clustering, and T is the total number of iterations.
[0076] S5. Combine the strong classifiers of all sample clusters into the final low-voltage active area line loss calculation model.
[0077] The above method first uses a cluster analysis method to classify the different operating states of the low-voltage active area, and then uses a neural network method and an AdaBoost framework to learn and predict the line loss of the low-voltage active area, so as to realize the calculation and analysis of the line loss of the low-voltage active area. The above method uses advanced data processing technology and machine learning algorithms to achieve accurate calculation and effective management of the line loss of the low-voltage active area, and solves the problem that the uncertainty of the line loss in the low-voltage active area is significantly enhanced due to the access of distributed power sources, making it difficult for existing methods to accurately calculate the line loss of the low-voltage active area.
[0078] If the above method 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, a server, or a network device, etc.) to perform all or part of the steps of the method described in 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 and other media that can store program codes.
[0079] Example 2
[0080] This embodiment provides a low-voltage active substation line loss calculation system based on clustering and AdaBoost, characterized in that it includes one or more processors, a memory and one or more programs stored in the memory, and the one or more programs include instructions for executing the method described in Example 1.
[0081] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.
Claims
1. A method for calculating line loss in low voltage active area based on clustering and AdaBoost, characterized in that: The method includes: Collect the values of the line loss influencing factors of the low-voltage active area to be calculated, determine the sample cluster to which it belongs, and calculate the line loss rate of the low-voltage active area to be calculated based on the strong classifier of the sample cluster in the pre-trained low-voltage active area line loss calculation model; The construction process of the low voltage active area line loss calculation model includes the following steps: Acquire a low-voltage active area line loss historical data sample, wherein the low-voltage active area line loss historical data sample includes line loss influencing factors and corresponding line loss rates; Based on the line loss influencing factors, all low-voltage active substation line loss historical data samples are clustered into several sample clusters; For each sample cluster, a weak classifier is trained separately; For each sample cluster, the AdaBoost algorithm is iteratively used to train a strong classifier for each sample cluster based on the weak classifier. The strong classifiers of all sample clusters are combined into the final low-voltage active area line loss calculation model.
2. The method for calculating line loss in low voltage active area based on clustering and AdaBoost according to claim 1, characterized in that: The line loss influencing factors include multiple factors of the area load rate, power factor, voltage, distributed photovoltaic output and sunshine intensity.
3. The method for calculating line loss in low voltage active area based on clustering and AdaBoost according to claim 1, characterized in that: The specific process of clustering all low-voltage active transformer area line loss historical data samples into several sample clusters includes: Standardize the factors affecting line loss; The Euclidean distance between each sample is calculated, and clustering is performed based on the Euclidean distance.
4. The method for calculating line loss in low voltage active area based on clustering and AdaBoost according to claim 1, characterized in that: The K-means clustering algorithm is used to cluster all low-voltage active substation line loss historical data samples into several sample clusters.
5. The method for calculating line loss in low voltage active area based on clustering and AdaBoost according to claim 3, characterized in that: The expression of the standardization process is: In the formula, z ij represents the standardized value of the jth influencing factor of the i-th sample, x ij represents the sampling value of the jth influencing factor of the i-th sample, represents the sample average value of the jth influencing factor, s ij It represents the sample variance of the jth influencing factor, and n is the number of samples.
6. The method for calculating line loss in low voltage active area based on clustering and AdaBoost according to claim 1, characterized in that: The weak classifier is obtained by training using a neural network algorithm, an association analysis method or a Bayesian method.
7. The method for calculating line loss in low voltage active area based on clustering and AdaBoost according to claim 1, characterized in that: The process of clustering the kth samples to obtain the strong classifier includes: Initialize the weight D of each sample 0i =1 / n, where n is the number of samples; Assume h t For the weak classifier trained in the tth iteration, calculate the error rate ε of the weak classifier after each iteration t and weight α t ; Update the sample weights for round t+1 When the error rate ε t The iteration is terminated when <β, where β is the preset error rate threshold; The weak classifiers and their weights of each iteration are weighted summed to obtain a strong classifier, that is, In the formula, H k is the strong classifier finally trained for the k-th sample clustering, and T is the total number of iterations.
8. The method for calculating line loss in low voltage active area based on clustering and AdaBoost according to claim 7, characterized in that: The error rate ε of the weak classifier after the tth iteration t It is expressed as: The weight α t It is expressed as: Where D ti is the weight of the i-th sample in the t-th iteration, sign() is the indicator function, when h t (x i )≠y i The value of sign() is 1 if true, and 0 otherwise.
9. A computer-readable storage medium, characterized in that: It includes one or more programs for execution by one or more processors of an electronic device, and the one or more programs include instructions for executing the low-voltage active area line loss calculation method based on clustering and AdaBoost as described in any one of claims 1-8.
10. A low voltage active area line loss calculation system based on clustering and AdaBoost, characterized in that: The system comprises one or more processors, a memory and one or more programs stored in the memory, wherein the one or more programs include instructions for executing the method according to any one of claims 1 to 8.
Citation Information
Patent Citations
Power grid loss reduction optimization method and system based on clustering and deep belief network
CN111738483A
Low-voltage transformer area line loss anomaly analysis method, device and equipment
CN112598234A
