An intelligent sorting method for advanced prediction of relay protection settings
By adopting intelligent sorting methods in relay protection, using deep learning and current mediators, calculating information entropy and impact measurements, and establishing a convolutional neural network model, we realize advanced prediction and protection constant value verification of key lines of the power system. It solves the problem of difficulty in protecting fault lines caused by artificial dependence in traditional methods, and improves fault response speed and accuracy.
Patent Information
- Application Number
- CN202210075658.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-01-22
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2042-01-22
AI Technical Summary
The traditional relay protection fixed value setting method relies on manual experience, and there are errors, so it is impossible to quickly and effectively protect the faulty lines. Especially after the integration of distributed new energy into the power system, the topology of the power system is complicated, and traditional random or sequential verification methods are difficult to quickly response.
Using the intelligent sorting method of forward prediction of relay protection constant value, the intelligent sorting model of protection constant value is established by obtaining the real-time flow data of the transmission line and the equipment parameters, and by obtaining the transmission line's current flow, the information entropy, current sensitivity and impact measurement are calculated, and combined with the convolutional neural network of deep learning, the intelligent sorting model of protection constant value is established. , realize advanced prediction sorting and protection value verification of key lines in the power system.
It realizes the identification of the key degree of the transmission line based on the power grid operation mode and the current dielectric number, and forms a related relay protection setting value from high to low, quickly and effectively protects the faulty line, reducing the risk of errors of manual calculation dependence.
Smart Images

Figure QLYQS_1 
Figure QLYQS_10 
Figure QLYQS_11
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of power distribution automation, and in particular to a method for intelligently sequencing relay protection setting advance predictions. Background Art
[0002] With the rapid development of new technologies, emerging DC technologies such as flexible DC and hybrid DC are constantly being applied, the scale of grid-connected new energy power generation such as distributed photovoltaics and offshore wind power is constantly expanding, and the digital revolution and energy revolution are accelerating their integration, which has brought new opportunities and challenges to the setting of power grid relay protection settings.
[0003] In terms of setting the fixed value, the setting method is based on the industry setting regulations, but because the regulations are relatively broad, different setting personnel have different understandings of the details, and their selection of fixed values will also be different. The setting based on manual experience is highly dependent on the physical and mental state of the personnel when calculating the fixed value, and there is a possibility of setting errors. In particular, various forms of distributed new energy are constantly being incorporated into the power system, which increases the complexity of the power system topology and system operation mode, making the offline setting of relay protection backup protection increasingly difficult. The traditional random or sequential verification method cannot quickly and effectively protect the safety of the fault line. Summary of the invention
[0004] The present invention proposes an intelligent sorting method for advance prediction of relay protection settings, which solves the problem in the prior art that relay settings cannot quickly and effectively protect the safety of faulty lines according to a random or sequential verification method.
[0005] The technical solution of the present invention is as follows:
[0006] A method for intelligent sorting of relay protection setting advance prediction includes the following steps:
[0007] S100: Acquire real-time power flow data and equipment parameters of the transmission line, wherein the real-time power flow data includes active power flow, reactive power flow, node injected active power, node injected reactive power and the current time of the transmission line at multiple times, and the equipment parameters include upper and lower limits on the transmission line and generator start and stop limits;
[0008] S200: Calculating an active power flow index and a reactive power flow index of a transmission line, and calculating information entropy of the transmission line according to the active power flow index and the reactive power flow index;
[0009] S300: Calculating active power flow sensitivity and reactive power flow sensitivity of the transmission line, and calculating active impact measurement index and reactive impact measurement index of the transmission line according to the active power flow sensitivity and reactive power flow sensitivity,
[0010] S400: Calculate a danger level index of the transmission line according to the information entropy, the active impact measurement index and the reactive impact measurement index, sort the lines according to the danger level index, and use the sorting result as a comprehensive identification index;
[0011] S500: Establish a protection setting intelligent sorting convolutional neural network optimization model, take the active power flow, reactive power flow, node injected active power and node injected reactive power of the transmission line as the input of the protection setting intelligent sorting convolutional neural network optimization model, take the comprehensive identification index as the output of the protection setting intelligent sorting convolutional neural network optimization model, and train the model.
[0012] Furthermore, the active power flow index is
[0013] The reactive power flow index is
[0014] In the formula, G represents the total number of generators; N represents the total number of loads; P m and Q m Respectively represent the active power and reactive power of generator m; P n and Q n Respectively represent the active power and reactive power of load n; P L (m,n) indicates that the generator m delivers active power P(m,n) to the load P n The active power flowing through the transmission line L at the time, Q L (m,n) indicates that generator m delivers reactive power Q(m,n) to load Q n The reactive power flowing through the transmission line L at time .
[0015] Furthermore, the information entropy is
[0016] In the formula, w L Represents the power system N s The active and reactive power flow betweenness of the transmission line; p(L) represents the power flow probability of line L, f PL and f QL They represent active power flow index and reactive power flow index respectively.
[0017] Furthermore, the active impact index is
[0018] in, represents the active power flow on the transmission line L at time t; ΔP L′ It represents the active power flow transfer amount of the transmission line L′; represents the distribution factor of the active power flow transfer of transmission line L′ on line L,
[0019] It represents the active power change ΔP from given node k′ to node l′ k′l′ When the transmission line L k-l Active power flow sensitivity on
[0020] represents the active power flow injected into node k′ in line L k-l The amount of change on represents the active power injected into node l′ in line L k-l The amount of change on
[0021] The reactive impact index is
[0022] in, represents the reactive power flow on the transmission line L at time t; ΔQ L′ It represents the reactive power flow transfer amount of the transmission line L′; It represents the distribution factor of reactive power flow transfer of transmission line L′ on line L;
[0023] It represents the reactive power change ΔQ from given node k′ to node l′ k′l′ When the transmission line L k-l Active power flow sensitivity on
[0024] Represents the reactive power flow injected into node k′ on line L k-l The amount of change on represents the reactive power injected into node l′ in line L k-l The amount of change on .
[0025] Furthermore, the danger level of the transmission line is 0 L =μ(1-H L )+ν(λ PL +λ QL ), where S L Represents the apparent power of the transmission line L; H s is the calculated information entropy; PL and λ QL Calculated distribution factor; μ and ν denote coefficients, respectively.
[0026] Furthermore, the protection setting value intelligent sorting convolutional neural network optimization model is:
[0027]
[0028] In the formula, X lrepresents the input data vector of the first layer, mainly referring to the active power flow and reactive power flow of each transmission line in the power system; y l is the output result; ω represents the support vector machine classification vector; b represents the bias; λ represents the coefficient of feature extraction by the convolution and pooling layers; ξ l and They represent the relaxation amount during the optimization process.
[0029] The working principle and beneficial effects of the present invention are:
[0030] 1. The present invention determines the degree of danger of chain failures in the power system by calculating the information entropy through the active power flow index and the reactive power flow index, and determines the degree of danger of active and reactive power flow exceeding the limit in the circuit system according to the distribution state of the active power flow index and the reactive power flow index; determines the impact of the active and reactive power flow transfer on the power transmission line of the power system by calculating the active impact measurement index and the reactive impact measurement index; obtains the ranking of the degree of danger of the line power flow exceeding the limit and the impact of the power flow transfer by combining the danger degree of the line power flow exceeding the limit and the impact of the power flow transfer; finally, the multi-layer regression convolutional neural network in the deep learning theory is used for regression learning to achieve advanced prediction to obtain the ranking of important lines, thereby obtaining the advanced ranking of the relay protection setting.
[0031] 2. Aiming at the problem that the traditional relay protection setting sorting is overly dependent on manual calculation, the present invention improves the traditional power flow entropy method that only considers the active power and the impact of the line active power, and proposes a relay protection setting advance prediction intelligent sorting method combining deep learning and power flow betweenness. According to the grid operation mode and based on the power flow betweenness, the criticality of the power grid transmission line is identified, and then the associated relay protection setting values with the criticality from high to low are formed, and the relay protection setting values are verified in sequence according to the criticality. The relay protection setting value is verified according to the importance to achieve fast and effective protection of the safety of the fault line.
[0032] The present invention is further described in detail below in conjunction with specific implementation modes. DETAILED DESCRIPTION
[0033] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0034] Example 1
[0035] This embodiment proposes a method for intelligent sorting of relay protection setting advance prediction, and the specific method is as follows:
[0036] 1. Obtain the existing line data of power grid automation, including design parameters such as resistance, reactance, and susceptance, or obtain the measured parameters of the line; obtain the real-time data of power grid automation flow, mainly including the active power flow, reactive power flow, node injected active power and reactive power (including generator active output and reactive output, load active power and reactive power) of the transmission line at intervals of every 2 minutes, save the obtained data into the database file, and name them in time series order. According to the equipment parameters and design drawings, obtain the system structure parameters, upper and lower limits of the transmission line, and the generator start and stop limit data as constants and save them in the database.
[0037] 2. Calculate the active and reactive power flow information entropy by using the acquired power grid data and power flow data. The calculation method of information entropy is as follows:
[0038] 1) Establish active and reactive power flow indicators and calculate the power flow indicators, as follows:
[0039]
[0040] In the formula, G represents the total number of generators; N represents the total number of loads; P m and Q m Respectively represent the active power and reactive power of generator m; P n and Q n Respectively represent the active power and reactive power of load n; P L (m,n) indicates that the generator m delivers active power P(m,n) to the load P n The active power flowing through the transmission line L at the time, Q L (m,n) indicates that generator m delivers reactive power Q(m,n) to load Q n The reactive power flowing through the transmission line L at time .
[0041] 2) According to the importance level and power flow index of the transmission line, the information entropy of the transmission line is calculated. The specific method is as follows:
[0042]
[0043] In the formula, w L Represents the power system N s The active and reactive power flow coefficient of the transmission line; p(L) represents the power flow probability of line L. P and F Q They represent active power flow index and reactive power flow index respectively.
[0044] 3. Based on the calculation results in step 2, determine the degree of balance of the power flow in the power system, and thus determine the degree of danger of cascading failures in the power system.
[0045] 4. Establish the measurement indicators of active power impact of transmission lines as follows:
[0046] In the formula, represents the reactive power flow on the transmission line L at time t; ΔP L′ It represents the active power flow transfer amount of the transmission line L′; It represents the distribution factor of the active power flow transfer of the transmission line L′ on the line L;
[0047] It represents the active power change ΔP from given node k′ to node l′ k′l′ When the transmission line L k-l Active power flow sensitivity on
[0048] The metric is solved according to the following formula:
[0049]
[0050] represents the active power flow injected into node k′ in line L k-l The amount of change on represents the active power injected into node l′ in line L k-l The amount of change on
[0051] 5. Establish the reactive power impact measurement indicators of transmission lines as follows:
[0052]
[0053] In the formula, represents the reactive power flow on the transmission line L at time t; ΔQ L′ It represents the reactive power flow transfer amount of the transmission line L′; It represents the distribution factor of reactive power flow transfer on transmission line L′.
[0054] It represents the reactive power change ΔQ from given node k′ to node l′ k′l′ When the transmission line L k-l Active power flow sensitivity on
[0055]
[0056] The metric is solved according to the following formula:
[0057]
[0058] Represents the reactive power flow injected into node k′ on line L k-l The amount of change on represents the reactive power injected into node l′ in line L k-l The amount of change on .
[0059] The impact of active and reactive power flow transfer on the power transmission lines of the power system is determined based on the active power impact index and the reactive power impact index.
[0060] 6. The transmission lines L in the power system are ranked according to their degree of danger. The ranking method is as follows:
[0061] O L =μ(1-H L )+ν(λ PL +λ QL ), where S L Represents the apparent power of the transmission line L; H s is the calculated information entropy; PL and λ QL The calculated distribution factor; μ and ν represent coefficients, which are given by the relay protection setting according to their importance.
[0062] 7. Establish a convolutional neural network optimization model for intelligent sorting of protection values:
[0063]
[0064] In the formula, X l represents the input data vector of the first layer, mainly referring to the active power flow and reactive power flow of each transmission line in the power system; y l is the output result; ω represents the support vector machine classification vector; b represents the bias; λ represents the coefficient of feature extraction by the convolution and pooling layers; ξ l and They represent the relaxation amount during the optimization process.
[0065] 8. Use convolutional neural network for regression learning, take the active power flow, reactive power flow, node injected active power and node injected reactive power of the transmission line as the input of the protection setting intelligent sorting convolutional neural network optimization model, take the comprehensive identification index as the output of the protection setting intelligent sorting convolutional neural network optimization model, and train the model. The identification accuracy result can be obtained, and the advance prediction can be achieved to obtain the important line sorting, thereby obtaining the advance sorting of the relay protection setting and realizing the relay protection setting importance verification.
[0066] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A method for intelligent sorting of relay protection setting advance prediction, characterized in that: The following steps are included: S100: Acquire real-time power flow data and equipment parameters of the transmission line, wherein the real-time power flow data includes active power flow, reactive power flow, node injected active power, node injected reactive power and the current time of the transmission line at multiple times, and the equipment parameters include upper and lower limits on the transmission line and generator start and stop limits; S200: Calculating an active power flow index and a reactive power flow index of a transmission line, and calculating information entropy of the transmission line according to the active power flow index and the reactive power flow index; S300: Calculating active power flow sensitivity and reactive power flow sensitivity of the transmission line, and calculating active impact measurement index and reactive impact measurement index of the transmission line according to the active power flow sensitivity and reactive power flow sensitivity; S4 00: Calculate the danger level index of the transmission line according to the information entropy, the active impact measurement index and the reactive impact measurement index, sort them according to the danger level index, and use the sorting result as the comprehensive identification index; S500: establishing a protection setting value intelligent sorting convolutional neural network optimization model, taking the active power flow, reactive power flow, node injected active power, and node injected reactive power of the transmission line as inputs of the protection setting value intelligent sorting convolutional neural network optimization model, taking the comprehensive identification index as output of the protection setting value intelligent sorting convolutional neural network optimization model, and training the model; The active impact index is ; in, Indicated in Transmission lines at all times The meritorious trend on Indicates transmission line Active power flow transfer amount; Indicates transmission line Active power flow transfer in the line The distribution factor on , Indicates that when a given node To Node Active power change When the transmission line Active power flow sensitivity on , Indicates the injection node The active power flow in the line The amount of change on Indicates the injection node The active power in the line The amount of change on The reactive impact index is , in, Indicated in Transmission lines at all times Reactive power flow on Indicates transmission line Reactive power flow transfer amount; Indicates transmission line Reactive power flow transfer in the line The distribution factor on ; , Indicates that when a given node To Node The reactive power change When the transmission line Active power flow sensitivity on , Indicates the injection node The reactive power flow in the line The amount of change on Indicates the injection node The reactive power in the line The amount of change on The transmission line danger index is: , where , Indicates transmission line The apparent power of is the information entropy of the transmission line; and The calculated distribution factor; and Represent the coefficients respectively.
2. According to claim 1, a relay protection setting advanced prediction intelligent sorting method is characterized in that: The active power flow index is , The reactive power flow index is , In the formula, Indicates the total number of generators; Indicates the total number of loads; and Respectively represent generator Active power and reactive power; and Respectively represent load Active power and reactive power; Indicates generator Active power transmission To load The active power flowing through the transmission line L at time , Indicates generator Transmitting reactive power To load The reactive power flowing through the transmission line L at time .
3. The method for intelligent sorting of relay protection setting advance prediction according to claim 2 is characterized in that: The information entropy is , , In the formula, Represents the power system The active and reactive power flow betweenness of the transmission lines; Indicates line Flow probability, and They represent active power flow index and reactive power flow index respectively.
4. The method for intelligent sorting of relay protection setting advance prediction according to claim 1 is characterized in that: The protection setting value intelligent sorting convolutional neural network optimization model is: , In the formula, Indicates Layer input data vector, mainly refers to the active power flow and reactive power flow of each transmission line in the power system; Output the result for it; represents the support vector machine classification vector; Indicates the offset; Represents the coefficients of the convolution and pooling layers to extract features; and They represent the relaxation amount during the optimization process.
