New energy power system comprehensive safety risk assessment method based on vector machine

Abstract

The invention discloses a comprehensive safety risk assessment method for a new energy power system based on a vector machine. The method comprises the following steps: carrying out equipment type division on the whole new energy power system according to equipment types; combining historical working data of the new energy power system to construct a correlation model of different operation scenes of multiple types of equipment of the power system; calculating to obtain a multi-service scene sample set of different types of equipment by using the correlation model; randomly allocating a basic operation environment of the new energy power system to each business scene of each type of equipment, and obtaining a comprehensive safety risk index of the new energy power system in the corresponding basic operation environment; based on the multi-service scene sample set, establishing and training a comprehensive security risk assessment model of the new energy power system by using a kernel support vector machine algorithm; and based on the comprehensive safety risk assessment model, assessing the comprehensive operation safety risk of the new energy power system. According to the method, tedious mathematical modeling is avoided, and the new energy utilization benefit is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of risk assessment of new energy power systems, and in particular discloses a comprehensive safety risk assessment method for new energy power systems based on a vector machine. Background Art

[0002] At present, more and more power grid systems are adopting clean energy such as wind power and solar energy to replace traditional fossil energy power generation. With the growth of installed capacity of various new energy sources such as wind power generation and solar power generation and the rapid expansion of the total amount of grid connection, the uncertainty of power output at the power supply end has deepened, the complexity of the power network structure has increased, and the power grids of various levels are interconnected. The number of power system components has continued to increase, and the power grid has also gradually moved towards long-distance and large-capacity transmission. The uncertainty of source-grid-load will impose more severe tests on the safe and stable operation, system regulation and system planning of the new power system. This also warns that we should strengthen the research on power grid safety issues, conduct comprehensive safety assessments of the power grid and control operation and maintenance risks in a timely manner.

[0003] Based on this, the present invention proposes a comprehensive safety risk assessment method for a new energy power system based on a vector machine. Summary of the invention

[0004] The present invention provides a new energy power system comprehensive safety risk assessment method based on vector machine, aiming to solve at least one defect existing in the above-mentioned prior art.

[0005] The present invention relates to a new energy power system comprehensive security risk assessment method based on a vector machine, comprising the following steps:

[0006] Divide the entire new energy power system into equipment types according to the equipment types;

[0007] Combined with the historical working data of the new energy power system, a correlation model of different operating scenarios of multiple types of equipment in the power system is constructed;

[0008] Using the correlation model, a sample set of multiple business scenarios for different types of equipment is calculated;

[0009] Randomly assign a basic operating environment of a new energy power system to each business scenario of each type of equipment, and obtain the comprehensive safety risk index of the new energy power system under the corresponding basic operating environment;

[0010] Based on a multi-business scenario sample set, a comprehensive security risk assessment model for the new energy power system is established and trained using the kernel support vector machine algorithm;

[0011] Based on the comprehensive safety risk assessment model, the comprehensive operation safety risk of the new energy power system is assessed;

[0012] Based on the comprehensive operational safety risk assessment results, provide new energy power system operation and maintenance decision-making recommendations.

[0013] Furthermore, the steps of classifying the equipment types of the entire new energy power system according to the equipment types include:

[0014] Construct the network topology of the new energy power system. The network topology is expressed by the formula T(N, L), where N represents different nodes in the topology and L represents different lines in the topology;

[0015] On this basis, the efficiency concentration of each node is calculated. The efficiency concentration of each node is calculated by the following formula:

[0016]

[0017] Among them, PC(i) represents the efficiency concentration of node i, PW(i) represents the power consumption of node i, PW(j) represents the power consumption of node j, and V(i) represents the topological nodes adjacent to node i and connected to each other.

[0018] Furthermore, in the step of dividing the entire new energy power system into equipment types according to the equipment types, the DBSCAN clustering algorithm is used to classify the different equipment types of the new energy power system, and the objective function of the DBSCAN clustering algorithm is defined as:

[0019]

[0020] Among them, TAR(C) represents the clustering objective function, C represents the cluster center matrix, N represents the number of nodes, CLU represents the number of clusters, and u ij represents the degree of node i’s belonging to device type j, The square representing the center point of the device type;

[0021] The distance function of the DBSCAN clustering algorithm is:

[0022] H(PC(i),PC(j))=D(PC(i)||PC(i))

[0023]

[0024] Where PC(i) represents the efficiency concentration of node i, PC(j) represents the efficiency concentration of node j, H(PC(i), PC(j)) is the cross entropy from PC(i) to PC(j), i.e. the objective function; D(PC(i)||PC(j)) is the relative entropy from PC(i) to PC(j), i.e. the KL divergence; E PC(i) represents the expected value of the concentration of effectiveness;

[0025] After that, all node data are marked as unprocessed, and a point that has not been marked and processed is randomly selected to calculate all neighboring points of the point within a radius of ε, that is, all nodes with cross entropy H(PC(i), PC(j))<ε, where ε is a custom radius; if the number of neighboring nodes of the point is not less than 10, the point is marked as a core point; a new cluster is created for the core point, and it and its neighboring points are added to the cluster; if the number of neighbors of the point is less than 10, it is marked as a noise point; for each core point in the new cluster, all its neighboring nodes within a radius of ε are searched, among which the unprocessed neighboring nodes are marked as processed and added to the current cluster; if the neighboring node has not less than 10 neighboring nodes, it is marked as a core point and its neighboring nodes are continued to be searched; the above steps are repeated until all nodes are marked as processed, and nodes that are not classified into any cluster are noise points; finally, all clustering information is output.

[0026] Furthermore, in the step of constructing a correlation model for different operation scenarios of multiple types of equipment in the power system in combination with the historical working data of the new energy power system, the joint probability distribution of multi-dimensional random variables is decomposed into a two-dimensional Archimedean Copulas function including two random variables based on the historical working data of the new energy power system. The random variables of different equipment types in multiple business scenarios are considered, and the probability density function is expressed as:

[0027] f θ (x,y)=(1+e -x +e -y +(1-θ)e -y-x ) -1

[0028] Among them, f θ (x, y) is the probability density function, θ represents the expansion parameter, when the marginal distribution of the random variable F(x) = (1 + e -x ) -1 and G(y)=(1+e -y ) -1 When they are independent of each other, θ takes the value of 0, otherwise it takes the value of 1; F(x) is the marginal distribution of variable x, and G(y) is the marginal distribution of variable y.

[0029] Furthermore, the steps of calculating and obtaining a multi-service scenario sample set of different types of devices using the correlation model include:

[0030] Using the Monte Carlo sampling method, M samples are extracted for different business scenarios of different device types, and different samples correspond to different business scenarios; for each device type i, a uniformly distributed random sample s of M number is generated i1 ,s i2 ,…,s iM, the value range is an integer between 1 and 9, representing 9 different business scenarios. Different random samples represent the marginal distribution of business scenarios of this type of equipment.

[0031] Based on the established Archimedean Copulas model, all samples of each device type are converted into business scenario samples D of the device type. ij , all business scenario samples of the device type constitute the business scenario sample set of the device type; where i represents the device type, and j represents the sample sequence number of the corresponding device type;

[0032] Specifically, for each sample s ij , using the inverse Archimedean Copulas transform, we get a two-dimensional random variable:

[0033] C(i,j)=H(F -1 (i),G -1 (j))

[0034] Substituting into the Archimedean Copulas function, we get:

[0035]

[0036] Where C(i,j) is the Copulas function of variables i and j, and H(F -1 (i),G -1 (j)) represents the cross entropy of the inverse marginal distribution function, C θ (i, j) is the Archimedean Copulas function of variables i, j, and θ represents the expansion parameter;

[0037] Afterwards, based on the marginal distribution of business scenarios for each device type, we obtain the following business scenario samples:

[0038]

[0039] Among them, D ij For business scenario samples, is the inverse transformation function of the marginal distribution of device type i.

[0040] Furthermore, the steps of randomly assigning a basic operating environment of a new energy power system to each business scenario of each type of equipment and obtaining a comprehensive safety risk index of the new energy power system under the corresponding basic operating environment include:

[0041] Use the random number generation algorithm to obtain the random equipment load state A k , Equipment flow status B k , and the combined wind and solar power output status of the equipment Q k Parameters, ensuring that the values ​​of the three state variables are within a reasonable range;

[0042] According to the equipment load status A k , Equipment flow status B k , and the combined wind and solar power output status of the equipment Q k Parameters, three indicators are selected for evaluation: equipment load loss risk, power flow over-limit risk, and wind and solar power abandonment risk;

[0043] The risk of equipment losing load is expressed as:

[0044]

[0045] Among them, R EQ Represents the risk value of equipment load loss, T M represents the total working time, P(A k ) represents the equipment load status A k The probability of C S (A k ) is the equipment load status A k The severity of the load loss consequence under load node, NL represents the load shedding amount at the load node;

[0046] The risk of power flow exceeding the limit is expressed as:

[0047]

[0048] Among them, R WL represents the risk of current exceeding the limit, T M represents the total working time, P(B k ) represents the equipment flow status B k The probability of C S (B k ) represents the equipment flow status B k The severity of the consequences of power flow exceeding the limit under the condition, NP represents the number of lines, represents the overcurrent power of overcurrent line i, P Limix represents the rated maximum overcurrent power of overcurrent circuit i;

[0049] The risk of wind and solar power curtailment is expressed as:

[0050]

[0051] Among them, P(Q k ) represents the risk probability of wind and solar power abandonment, C FG (Q k ) represents the combined output state of wind and solar energy of the equipment Q k The severity of wind and solar power abandonment under the condition of FG represents the total amount of wind and solar power abandoned, and T represents the total operating time of new energy equipment;

[0052] Finally, the equipment load state A k , Equipment flow status B k , and the combined wind and solar power output status of the equipment Q k The weighted average of the three indicators gives the operating risk value of new energy power equipment:

[0053] ζ=β 1 R FQ +β 2 R WL +β 3 R FG

[0054] Among them, ζ represents the operating risk value of energy and power equipment, β 1 , β 2 , β 3 Represent the weight coefficient, β 1 Take 0.8, β 2 Take 2.0, β 3 Take 1.0; sum the operating risk values ​​of all equipment in the new energy power system to obtain the comprehensive safety risk index ζ of the new energy power system T ; Construct a training sample set based on the historical operation data, comprehensive safety risk indicators and corresponding business scenarios.

[0055] Furthermore, based on the multi-business scenario sample set, using the kernel support vector machine algorithm, the steps of establishing and training a comprehensive security risk assessment model for the new energy power system include:

[0056] The kernel support vector machine is used to construct a comprehensive safety risk assessment model for new energy power systems. The kernel support vector machine algorithm is used to construct the risk assessment model. The kernel support vector machine algorithm uses the weighted average of linear kernel function, polynomial kernel function, radial basis function kernel function, sigmoid kernel function, cosine similarity kernel function and Laplace kernel function as the final kernel function K(x 1 ,x 2 ), its actual expression is: K(x 1 ,x 2 )

[0057] =c 1 K L (x 1 ,x 2 )+c 2 K P (x 1 ,x 2 )+c 3 K Rbf (x 1 ,x 2 )+c 4 KSig (x 1 ,x 2 )+c 5 K Cos (x 1 ,x 2 )+c 6 K Lap (x 1 ,x 2 )

[0058] K L (x 1 ,x 2 )=x 1 T x 2

[0059] K P (x 1 ,x 2 )=(γx 1 T x 2 +r) d

[0060] K Rbf (x 1 ,x 2 )=exp(-γ||x 1 -x 2 || 2 )

[0061] K Sig (x 1 ,x 2 )=tanh(γx 1 T x 2 +r)

[0062] K Cos (x 1 ,x 2 )=(x 1 T x 2 ) / (||x 1 ||*||x 2 ||)

[0063] K Lap (x 1 ,x 2 )=exp(-γ||x 1 -x 2 || 1 )

[0064] where K L (x 1 ,x2 ) represents the linear kernel function, K P (x 1 ,x 2 ) represents the polynomial kernel function, K Rbf (x 1 ,x 2 ) represents the radial basis function kernel function, K Sig (x 1 ,x 2 ) represents the Sigmoid kernel function, K Cos (x 1 ,x 2 ) represents the cosine similarity kernel function, K Lap (x 1 ,x 2 ) represents the Laplace kernel function; γ represents the scaling factor, r is the bias matrix, d represents the actual order of the polynomial kernel function, c 1 、c 2 、c 3 、c 4 、c 5 and c 6 Represents the weight of each kernel function; x 1 represents the first data set, x 2 represents the second data set; T represents the transpose operation of the vector;

[0065] The business scenarios of the training sample set and the historical operation data of the new energy power system are used as the input training data of the kernel support vector machine model, and the corresponding comprehensive security risk index of the new energy power system ζ is used. T As output training data;

[0066] Initialize the model's hidden layer weights and biases, and convert the input training data into hidden layer outputs through the final kernel function.

[0067] Furthermore, based on the comprehensive safety risk assessment model, the steps of assessing the comprehensive operational safety risk of the new energy power system include:

[0068] The business scenarios of Y types of equipment, load node load shedding, line rated maximum overcurrent power and total wind and solar power abandonment are collected and input into the input end of the kernel support vector machine model, and the comprehensive safety risk index of the new energy power system ζ is obtained from the output end. T According to the corresponding relationship between risk value and risk level, the comprehensive safety risk level of the new energy power system is determined, where 0≤ζ T <0.8 means no risk, 0.8≤ζ T <2 represents low risk, 2≤ζ T <6 represents medium risk, 6≤ζ T ≤10 indicates high risk, ζT >10 indicates severe risk.

[0069] Furthermore, based on the comprehensive operational safety risk assessment results, the steps for providing new energy power system operation and maintenance decision-making recommendations include:

[0070] If the safety assessment level of the new energy power system is no risk or low risk, it will operate according to the current system status; if the safety assessment level of the new energy power system is medium risk or high risk, the operation and maintenance process of the new energy power system will be readjusted. By adjusting the power system operation and maintenance process, it can output appropriate power to meet the actual load needs while avoiding the possibility of power flow exceeding the limit, and at the same time appropriately reduce the combined output of wind and solar energy.

[0071] The beneficial effects achieved by the present invention are:

[0072] The present invention provides a new energy power system comprehensive safety risk assessment method based on vector machine, which divides the equipment type of the entire new energy power system according to the equipment type; combines the historical working data of the new energy power system to build a correlation model of different operation scenarios of multiple types of equipment in the power system; uses the correlation model to calculate and obtain a multi-business scenario sample set of different types of equipment; randomly assigns a basic operation environment of the new energy power system to each business scenario of each type of equipment, and obtains the comprehensive safety risk index of the new energy power system under the corresponding basic operation environment; based on the multi-business scenario sample set, using the kernel support vector machine algorithm, establishes and trains a comprehensive safety risk assessment model of the new energy power system; based on the comprehensive safety risk assessment model, evaluates the comprehensive operation safety risk of the new energy power system; provides a new energy power system operation and maintenance decision suggestion according to the evaluation result of the comprehensive operation safety risk. Compared with the prior art, the present invention has the following beneficial effects:

[0073] (1) The kernel support vector machine used in the present invention is similar to other neural networks in that it does not rely on the complex relationship between parameters and outputs, but it relies on the continuous change of weights to closely associate parameters with outputs, avoiding cumbersome mathematical modeling.

[0074] (2) The present invention has high generalization and can maintain high calculation accuracy while reducing some input parameters. The operation and maintenance decision suggestions of the present invention are beneficial to the protection of the new energy power grid system, thereby improving the efficiency of new energy utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Figure 1 The present invention is a flowchart of an embodiment of a method for comprehensive safety risk assessment of a new energy power system based on a vector machine. DETAILED DESCRIPTION

[0076] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0077] like Figure 1 As shown, the first embodiment of the present invention proposes a new energy power system comprehensive security risk assessment method based on a vector machine, comprising the following steps:

[0078] Step S100: classify the entire new energy power system into equipment types according to the equipment types.

[0079] Step S200: Combine the historical working data of the new energy power system to build a correlation model for different operating scenarios of multiple types of equipment in the power system.

[0080] Step S300: Use the correlation model to calculate and obtain a multi-service scenario sample set for different types of devices.

[0081] Step S400: randomly assign a basic operating environment of a new energy power system to each business scenario of each type of equipment, and obtain a comprehensive safety risk index of the new energy power system under the corresponding basic operating environment.

[0082] Step S500: Based on a multi-business scenario sample set, a comprehensive security risk assessment model for a new energy power system is established and trained using a kernel support vector machine algorithm.

[0083] Step S600: Evaluate the comprehensive operational safety risk of the new energy power system based on the comprehensive safety risk evaluation model.

[0084] Step S700: Provide new energy power system operation and maintenance decision suggestions based on the comprehensive operation safety risk assessment results.

[0085] Further, see Figure 1 The method for comprehensive security risk assessment of new energy power system based on vector machine proposed in this embodiment, step S100 includes:

[0086] Step S110: construct a network topology structure of the new energy power system. The network topology structure is expressed by a formula T(N, L), where N represents different nodes in the topology structure, and L represents different lines in the topology structure.

[0087] Step S120: On this basis, the efficiency concentration of each node is calculated. The efficiency concentration of each node is calculated by the following formula:

[0088]

[0089] In formula (1), PC(i) represents the efficiency concentration of node i, PW(i) represents the power consumption of node i, PW(j) represents the power consumption of node j, and V(i) represents the topological nodes adjacent to node i and connected to each other.

[0090] Preferably, see Figure 1 In the comprehensive safety risk assessment method of a new energy power system based on a vector machine proposed in this embodiment, in step S100, a DBSCAN (Density-Based Spatial Clustering of Applications with Noise) clustering algorithm is used to classify different types of equipment in the new energy power system, and the objective function of the DBSCAN clustering algorithm is defined as:

[0091]

[0092] In formula (2), TAR(C) represents the clustering objective function, C represents the cluster center matrix, N represents the number of nodes, CLU represents the number of clusters, and u ij represents the degree of node i’s belonging to device type j, Represents the square of the center point of the device type.

[0093] The distance function of the DBSCAN clustering algorithm is:

[0094]

[0095] In formula (3), PC(i) represents the efficiency concentration of node i, PC(j) represents the efficiency concentration of node j, H(PC(i), PC(j)) is the cross entropy from PC(i) to PC(j), i.e., the objective function; D(PC(i)||PC(j)) is the relative entropy from PC(i) to PC(j), i.e., the KL divergence; E PC(i) Represents the expected value of effectiveness concentration.

[0096] After that, all node data are marked as unprocessed, and a point that has not been marked and processed is randomly selected to calculate all neighboring points of the point within a radius of ε, that is, all nodes with cross entropy H(PC(i), PC(j))<ε, where ε is a custom radius; if the number of neighboring nodes of the point is not less than 10, the point is marked as a core point; a new cluster is created for the core point, and it and its neighboring points are added to the cluster; if the number of neighbors of the point is less than 10, it is marked as a noise point; for each core point in the new cluster, all its neighboring nodes within a radius of ε are searched, among which the unprocessed neighboring nodes are marked as processed and added to the current cluster; if the neighboring node has not less than 10 neighboring nodes, it is marked as a core point and its neighboring nodes are continued to be searched; the above steps are repeated until all nodes are marked as processed, and nodes that are not classified into any cluster are noise points; finally, all clustering information is output.

[0097] Further, see Figure 1 In the comprehensive safety risk assessment method of a new energy power system based on a vector machine proposed in this embodiment, in step S200, based on the historical working data of the new energy power system, the joint probability distribution of the multidimensional random variables is decomposed into a two-dimensional Archimedean Copulas function including two random variables. Considering the random variables of different equipment types in multiple business scenarios, the probability density function can be expressed as:

[0098] f θ (2,y)=(1+e -x +e -y +(1-θ)e -y-x ) -1 (4)

[0099] In formula (4), f θ (x, y) is the probability density function, θ represents the expansion parameter, when the marginal distribution of the random variable F(x) = (1 + e -x ) -1 and G(y)=(1+e -y ) -1 When they are independent of each other, θ takes the value of 0, otherwise it takes the value of 1; F(x) is the marginal distribution of variable x, and G(y) is the marginal distribution of variable y.

[0100] Preferably, see Figure 1 The method for comprehensive safety risk assessment of new energy power system based on vector machine proposed in this embodiment, step S300 includes:

[0101] Step S310: Using the Monte Carlo sampling method, M samples are extracted for different business scenarios of different device types, and different samples correspond to different business scenarios; for each device type i, M uniformly distributed random samples s are generated. i1,s i2 ,…,s iM , the value range is an integer between 1 and 9, representing 9 different business scenarios. Different random samples each represent the marginal distribution of business scenarios of this type of equipment.

[0102] Step S320: Based on the established Archimedean Copulas model, all samples of each device type are converted into business scenario samples D of the device type. ij , all business scenario samples of the device type constitute the business scenario sample set of the device type; where i represents the device type and j represents the sample serial number of the corresponding device type.

[0103] Specifically, for each sample s ij , using the inverse Archimedean Copulas transform, we get a two-dimensional random variable:

[0104] C(i,j)=H(F -1 (i),G -1 (j)) (5)

[0105] Substituting into the Archimedean Copulas function, we get:

[0106]

[0107] In formulas (5) and (6), C(i,j) is the Copulas function of variables i, j; H (F -1 (i),G -1 (j)) represents the cross entropy of the inverse marginal distribution function, C θ (i,j) is the Archimedean Copulas function of variables i, j; θ represents the expansion parameter.

[0108] Afterwards, based on the marginal distribution of business scenarios for each device type, we obtain the following business scenario samples:

[0109]

[0110] In formula (7), D ij For business scenario samples, is the inverse transformation function of the marginal distribution of device type i.

[0111] Further, see Figure 1 The method for comprehensive safety risk assessment of new energy power system based on vector machine proposed in this embodiment, step S400 includes:

[0112] Step S410: Use a random number generation algorithm to obtain a random device load state A k , Equipment flow status B k, and the combined wind and solar power output status of the equipment Q k Parameters to ensure that the values ​​of the three state variables are within a reasonable range.

[0113] Equipment load status A k For example, using: A k =A min +rand(0,1)·(A max -A min ) represents the distribution process of equipment load status, where A min and A max Respectively represent the minimum and maximum values ​​of the equipment load status; rand(0,1) represents a random number uniformly distributed between 0 and 1; A k Represents the generated equipment load state. Using the same method, generate equipment flow state B k And the combined wind and solar power output status of the equipment Q k .

[0114] Step S420: According to the equipment load status A k , Equipment flow status B k , and the combined wind and solar power output status of the equipment Q k Parameters, three indicators are selected for evaluation: equipment load loss risk, power flow over-limit risk and wind and solar power abandonment risk.

[0115] The equipment load loss risk is the probability that the available power generation capacity of the equipment is less than or equal to a certain constant load demand. The equipment load loss risk can be expressed as:

[0116]

[0117] In formula (8), R EQ Represents the risk value of equipment load loss, T M represents the total working time, P(A k ) represents the equipment load status A k The probability of C S (A k ) is the equipment load status A k The severity of the load loss consequences under load conditions, NL represents the load shedding amount at the load node.

[0118] The power flow over-limit risk refers to the risk that the bus voltage, branch current or power supplied to the equipment exceeds the rated value, causing line overload and causing equipment damage and power outage. The power flow over-limit risk can be expressed as:

[0119]

[0120] In formula (9), R WL represents the risk of current exceeding the limit, TM represents the total working time, P(B k ) represents the equipment flow status B k The probability of C S (B k ) represents the equipment flow status B k The severity of the consequences of power flow exceeding the limit under the condition, NP represents the number of lines, P Li represents the overcurrent power of overcurrent line i, P Limix Represents the rated maximum overcurrent power of overcurrent circuit i.

[0121] The risk of wind and solar abandonment is the phenomenon of wind and solar abandonment caused by the combined output of wind and solar energy of the equipment exceeding the acceptable capacity of the power grid, resulting in the risk of wasting wind and solar resources. The risk of wind and solar abandonment can be expressed as:

[0122]

[0123] In formula (10), P(Q k ) represents the risk probability of wind and solar power abandonment, C FG (Q k ) represents the combined output state of wind and solar energy of the equipment Q k The severity of wind and solar power abandonment under the condition of FG It represents the total amount of wind and solar power abandoned, and T represents the total operating time of new energy equipment.

[0124] Step S430: Check the equipment load status A k , Equipment flow status B k , and the combined wind and solar power output status of the equipment Q k The weighted average of the three indicators gives the operating risk value of new energy power equipment:

[0125] ζ=β 1 R FQ +β 2 R WL +β 3 R FG (11)

[0126] In formula (11), ζ represents the operating risk value of energy and power equipment, β 1 , β 2 , β 3 Represent weight coefficients respectively, preferably, β 1 Take 0.8, β 2 Take 2.0, β 3 Take 1.0; sum the operating risk values ​​of all equipment in the new energy power system to obtain the comprehensive safety risk index ζ of the new energy power system T; Construct a training sample set based on the historical operation data, comprehensive safety risk indicators and corresponding business scenarios.

[0127] Preferably, see Figure 1 The method for comprehensive safety risk assessment of new energy power system based on vector machine proposed in this embodiment, step S500 includes:

[0128] Step S510, using a kernel support vector machine to construct a comprehensive safety risk assessment model for a new energy power system, wherein the risk assessment model is constructed using a kernel support vector machine algorithm, wherein the kernel support vector machine algorithm uses a weighted average of a linear kernel function, a polynomial kernel function, a radial basis function kernel function, a sigmoid kernel function, a cosine similarity kernel function, and a Laplace kernel function as the final kernel function K(x 1 ,x 2 ), its actual expression is:

[0129] K(x 1 ,x 2 )=c 1 K L (x 1 ,x 2 )+c 2 K P (x 1 ,x 2 )+c 3 K Rbf (x 1 ,x 2 )+c 4 K Sig

[0130] (x 1 ,x 2 )+c 5 K Cos (x 1 ,x 2 )+c 6 K Lap (x 1 ,x 2 )(12)

[0131] K L (x 1 ,x 2 )=x 1 T x 2 (13)

[0132] K P (x 1 ,x 2 )=(γx 1 T x2 +r) d (14)

[0133] K Rbf (x 1 ,x 2 ) = exp(-γ||x 1 -x 2 || 2 ) (15)

[0134] K Sig (x 1 ,x 2 ) = tanh(γx 1 T x 2 +r) (16)

[0135] K Cos (x 1 ,x 2 )=(x 1 T x 2 ) / (||x 1 ||*||x 2 ||) (17)

[0136] K Lap (x 1 ,x 2 ) = exp(-γ||x 1 -x 2 || 1 ) (18)

[0137] In formulas (12) to (18), K L (x 1 ,x 2 ) represents the linear kernel function, K P (x 1 ,x 2 ) represents the polynomial kernel function, K Rbf (x 1 ,x 2 ) represents the radial basis function kernel function, K Sig (x 1 ,x 2 ) represents the Sigmoid kernel function, K Cos (x 1 ,x 2 ) represents the cosine similarity kernel function, K Lap (x 1 ,x 2 ) represents the Laplace kernel function; γ represents the scaling factor, r is the bias matrix, d represents the actual order of the polynomial kernel function, c 1 、c2 、c 3 、c 4 、c 5 and c 6 Represents the weight of each kernel function; x 1 represents the first data set, x 2 represents the second data set; T represents the transpose operation of the vector.

[0138] Step S520: Use the business scenarios of the training sample set and the historical operation data of the new energy power system as the input training data of the kernel support vector machine model, and use the corresponding new energy power system comprehensive security risk index ζ T As output training data.

[0139] Step S530: Initialize the hidden layer weights and biases of the model, and convert the input training data into hidden layer outputs through the final kernel function.

[0140] The weights between the hidden layer and the output layer are solved using the Moore-Penrose pseudo-inverse algorithm, which is a commonly used algorithm for calculating weights and will not be described here.

[0141] Further, see Figure 1 In the comprehensive safety risk assessment method of a new energy power system based on a vector machine proposed in this embodiment, in step S600, the business scenarios of Y types of equipment, the load shedding amount of load nodes, the rated maximum overcurrent power of the line and the total amount of wind and solar power abandonment are collected and input into the input end of the core support vector machine model, and the comprehensive safety risk index ζ of the new energy power system is obtained from the output end. T According to the corresponding relationship between risk value and risk level, the comprehensive safety risk level of the new energy power system is determined, where 0≤ζ T <0.8 means no risk, 0.8≤ζ T <2 represents low risk, 2≤ζ T <6 represents medium risk, 6≤ζ T ≤10 indicates high risk, ζ T >10 indicates severe risk.

[0142] Preferably, see Figure 1, the comprehensive safety risk assessment method of the new energy power system based on the vector machine proposed in this embodiment, in step S700, if the safety assessment level of the new energy power system is no risk or low risk, it will operate according to the current system state; if the safety assessment level of the new energy power system is medium risk or high risk, the operation and maintenance process of the new energy power system will be readjusted, and the power system operation and maintenance process will be adjusted to output appropriate power, while meeting the actual load needs, avoiding the possibility of over-limit of the current, and appropriately reducing the combined output of wind and solar energy to avoid the waste of renewable energy, and finally achieving efficient and reliable power generation at the lowest cost. In addition, when a risk occurs, the power system dispatcher can dynamically adjust the weights of each risk indicator according to the actual situation, and then obtain risk control under different risk states.

[0143] The present embodiment provides a comprehensive safety risk assessment method for a new energy power system based on a vector machine. Compared with the prior art, the entire new energy power system is divided into equipment types according to equipment types; a correlation model of different operation scenarios of multiple types of equipment in the power system is constructed in combination with historical working data of the new energy power system; a multi-business scenario sample set of different types of equipment is calculated using the correlation model; a basic operating environment of the new energy power system is randomly assigned to each business scenario of each type of equipment, and a comprehensive safety risk index of the new energy power system under the corresponding basic operating environment is obtained; based on the multi-business scenario sample set, a comprehensive safety risk assessment model of the new energy power system is established and trained using a kernel support vector machine algorithm; based on the comprehensive safety risk assessment model, the comprehensive operation safety risk of the new energy power system is assessed; and according to the assessment result of the comprehensive operation safety risk, a new energy power system operation and maintenance decision recommendation is provided. Compared with the prior art, the present embodiment has the following beneficial effects:

[0144] (1) The kernel support vector machine used in this embodiment is similar to other neural networks in that it does not rely on the complex relationship between parameters and outputs, but it relies on the continuous change of weights to closely associate parameters with outputs, avoiding cumbersome mathematical modeling.

[0145] (2) This embodiment has high generalization and can maintain high calculation accuracy while reducing some input parameters. The operation and maintenance decision-making suggestions of the present invention are beneficial to the protection of the new energy power grid system, thereby improving the efficiency of new energy utilization.

[0146] Although preferred embodiments of the present invention have been described, additional changes and modifications may be made to these embodiments by those skilled in the art once the basic inventive concepts are known. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention. Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A comprehensive security risk assessment method for new energy power system based on vector machine, characterized in that: The following steps are involved: Divide the entire new energy power system into equipment types according to the equipment types; Combined with the historical working data of the new energy power system, a correlation model of different operating scenarios of multiple types of equipment in the power system is constructed; Using the correlation model, a sample set of multiple business scenarios of different types of devices is calculated; Randomly assign a basic operating environment of a new energy power system to each business scenario of each type of equipment, and obtain the comprehensive safety risk index of the new energy power system under the corresponding basic operating environment; Based on the multi-business scenario sample set, a comprehensive security risk assessment model for the new energy power system is established and trained using a kernel support vector machine algorithm; Based on the comprehensive safety risk assessment model, the comprehensive operation safety risk of the new energy power system is assessed; Based on the comprehensive operational safety risk assessment results, provide new energy power system operation and maintenance decision-making recommendations.

2. The method for comprehensive safety risk assessment of new energy power system based on vector machine according to claim 1, characterized in that: The step of classifying the entire new energy power system into equipment types according to equipment types includes: Construct the network topology of the new energy power system. The network topology is expressed by the formula T(N, L), where N represents different nodes in the topology and L represents different lines in the topology. On this basis, the efficiency concentration of each node is calculated. The efficiency concentration of each node is calculated by the following formula: Among them, PC(i) represents the efficiency concentration of node i, PW(i) represents the power consumption of node i, PW(j) represents the power consumption of node j, and V(i) represents the topological nodes adjacent to node i and connected to each other.

3. The method for comprehensive safety risk assessment of new energy power system based on vector machine according to claim 2, characterized in that: In the step of dividing the entire new energy power system into equipment types according to equipment types, the DBSCAN clustering algorithm is used to classify different equipment types of the new energy power system, and the objective function of the DBSCAN clustering algorithm is defined as: Among them, TAR(C) represents the clustering objective function, C represents the cluster center matrix, N represents the number of nodes, CLU represents the number of clusters, and u ij represents the degree of node i’s belonging to device type j, PC(i) represents the efficiency concentration of node i, The square representing the center point of the device type; The distance function of the DBSCAN clustering algorithm is: Where PC(i) represents the efficiency concentration of node i, PC(j) represents the efficiency concentration of node j, H(PC(i), PC(j)) is the cross entropy from PC(i) to PC(j), i.e. the objective function; D(PC(i)||PC(j)) is the relative entropy from PC(i) to PC(j), i.e. the KL divergence; E PC(i) represents the expected value of the concentration of effectiveness; After that, all node data are marked as unprocessed, and a point that has not been marked and processed is randomly selected to calculate all neighboring points of the point within a radius of ε, that is, all nodes with cross entropy H(PC(i), PC(j))<ε, where ε is a custom radius; if the number of neighboring nodes of the point is not less than 10, the point is marked as a core point; a new cluster is created for the core point, and it and its neighboring points are added to the cluster; if the number of neighbors of the point is less than 10, it is marked as a noise point; for each core point in the new cluster, all its neighboring nodes within a radius of ε are searched, among which the unprocessed neighboring nodes are marked as processed and added to the current cluster; if the neighboring node has not less than 10 neighboring nodes, it is marked as a core point and its neighboring nodes are continued to be searched; the above steps are repeated until all nodes are marked as processed, and nodes that are not classified into any cluster are noise points; finally, all clustering information is output.

4. The method for comprehensive safety risk assessment of new energy power system based on vector machine according to claim 1, characterized in that: In the step of building a correlation model for different operation scenarios of multiple types of equipment in the power system in combination with the historical working data of the new energy power system, the joint probability distribution of the multi-dimensional random variables is decomposed into a two-dimensional Archimedean Copulas function including two random variables based on the historical working data of the new energy power system. The random variables of different equipment types in multiple business scenarios are considered, and the probability density function is expressed as: f θ (x,y)=(1+e -x +e -y +(1-θ)e -y-x ) -1 Among them, f θ (x, y) is the probability density function, θ represents the expansion parameter, when the marginal distribution of the random variable F(x) = (1 + e -x ) -1 and G(y)=(1+e -y ) -1 When they are independent of each other, θ takes the value of 0, otherwise it takes the value of 1; F(x) is the marginal distribution of variable x, and G(y) is the marginal distribution of variable y.

5. The method for comprehensive safety risk assessment of new energy power system based on vector machine according to claim 1, characterized in that: The step of using the correlation model to calculate and obtain a multi-service scenario sample set for different types of devices includes: Using the Monte Carlo sampling method, M samples are extracted for different business scenarios of different device types, and different samples correspond to different business scenarios; for each device type i, a uniformly distributed random sample s of M number is generated i1 ,s i2 ,…,s iM , the value range is an integer between 1 and 9, representing 9 different business scenarios. Different random samples represent the marginal distribution of business scenarios of this type of equipment. Based on the established Archimedean Copulas model, all samples of each device type are converted into business scenario samples D of the device type. ij , all business scenario samples of the device type constitute the business scenario sample set of the device type; where i represents the device type, and j represents the sample sequence number of the corresponding device type; Specifically, for each sample s ij , using the inverse Archimedean Copulas transform, we get a two-dimensional random variable: C(i,j)=H(F -1 (i),G -1 (j)) Substituting into the Archimedean Copulas function, we get: Where C(i,j) is the Copulas function of variables i and j; H(F -1 (i),G -1 (j)) represents the cross entropy of the inverse marginal distribution function, C θ (i, j) is the Archimedean Copulas function of variables i, j, and θ represents the expansion parameter; Afterwards, based on the marginal distribution of business scenarios for each device type, we obtain the following business scenario samples: Among them, D ij For business scenario samples, is the inverse transformation function of the marginal distribution of device type i.

6. The method for comprehensive safety risk assessment of new energy power system based on vector machine according to claim 1, characterized in that: The steps of randomly assigning a basic operating environment of a new energy power system to each business scenario of each type of equipment and obtaining a comprehensive safety risk index of the new energy power system under the corresponding basic operating environment include: Use the random number generation algorithm to obtain the random equipment load state A k , Equipment flow status B k , and the combined wind and solar power output status of the equipment Q k Parameters, ensuring that the values ​​of the three state variables are within a reasonable range; According to the equipment load state A k , Equipment flow status B k , and the combined wind and solar power output status of the equipment Q k Parameters, three indicators are selected for evaluation: equipment load loss risk, power flow over-limit risk, and wind and solar power abandonment risk; The equipment load loss risk is expressed as: Among them, R EQ Represents the risk value of equipment load loss, T M represents the total working time, P(A k ) represents the equipment load status A k The probability of C S (A k ) is the equipment load status A k The severity of the load loss consequence under load node, NL represents the load shedding amount at the load node; The flow over-limit risk is expressed as: Among them, R WL represents the risk of current exceeding the limit, T M represents the total working time, P(B k ) represents the equipment flow status B k The probability of C S (B k ) represents the equipment flow status B k The severity of the consequences of power flow exceeding the limit under the condition, NP represents the number of lines, P Li represents the overcurrent power of overcurrent line i, P Limix represents the rated maximum overcurrent power of overcurrent circuit i; The risk of wind and solar power abandonment is expressed as: Among them, P(Q k ) represents the risk probability of wind and solar power abandonment, C FG (Q k ) represents the combined output state of wind and solar energy of the equipment Q k The severity of wind and solar power abandonment under the condition of FG represents the total amount of wind and solar power abandoned, and T represents the total operating time of new energy equipment; Finally, the equipment load state A k , Equipment flow status B k , and the combined wind and solar power output status of the equipment Q k The weighted average of the three indicators gives the operating risk value of new energy power equipment: ζ=β1R FQ +β2R WL +β3R FG Among them, ζ represents the operating risk value of energy and power equipment, β1, β2, and β3 represent weight coefficients, β1 is 0.8, β2 is 2.0, and β3 is 1.0; the operating risk values ​​of all equipment in the new energy power system are summed to obtain the comprehensive safety risk index ζ of the new energy power system T ; Construct a training sample set based on the historical operation data, comprehensive safety risk indicators and corresponding business scenarios.

7. The vector machine-based comprehensive security risk assessment method for new energy power systems according to claim 1, characterized in that: The step of establishing and training a comprehensive security risk assessment model for a new energy power system based on the multi-business scenario sample set and using a kernel support vector machine algorithm comprises: The kernel support vector machine is used to construct a comprehensive safety risk assessment model for new energy power systems. The kernel support vector machine algorithm is used to construct the risk assessment model. The kernel support vector machine algorithm uses the weighted average of the linear kernel function, polynomial kernel function, radial basis function kernel function, sigmoid kernel function, cosine similarity kernel function and Laplace kernel function as the final kernel function K(x1,x2), and its actual expression is: K(x1,x2) =c1K L (x1,x2)+c2K P (x1,x2)+c3K Rbf (x1,x2)+c4K Sig (x1,x2)+c5K Cos (x1,x2)+c6K Lap (x1,x2) K L (x1,x2)=x1 Tx 2 K P (x1,x2)=(γx1 Tx 2+r) d K Rbf (x1,x2)=exp(-γ||x1-x2|| 2 ) K Sig (x1,x2)=tanh(γx1 Tx 2+r) K Cos (x1,x2)=(x1 Tx 2) / (||x1||*||x2||) K Lap (x1,x2)=exp(-γ||x1-x2||1) Among them, K L (x1,x2) represents the linear kernel function, K P (x1,x2) represents the polynomial kernel function, K Rbf (x1,x2) represents the radial basis function kernel function, K Sig (x1,x2) represents the Sigmoid kernel function, K Cos (x1,x2) represents the cosine similarity kernel function, K Lap (x1,x2) represents the Laplace kernel function. γ represents the scaling factor, r is the bias matrix, d represents the actual order of the polynomial kernel function, c1, c2, c3, c4, c5 and c6 represent the weights of each kernel function; x1 represents the first data set, x2 represents the second data set; T represents the transpose operation of the vector; The business scenarios of the training sample set and the historical operation data of the new energy power system are used as the input training data of the kernel support vector machine model, and the corresponding comprehensive security risk index of the new energy power system ζ is used. T As output training data; The hidden layer weights and biases of the model are initialized, and the input training data is converted into hidden layer outputs through the final kernel function.

8. The method for comprehensive safety risk assessment of new energy power system based on vector machine according to claim 7, characterized in that: In the step of evaluating the comprehensive operational safety risk of the new energy power system based on the comprehensive safety risk assessment model, business scenarios of Y types of equipment, load node load shedding, line rated maximum overcurrent power and total wind and solar power abandonment are collected and input into the input end of the kernel support vector machine model, and the comprehensive safety risk index ζ of the new energy power system is obtained from the output end. T According to the corresponding relationship between risk value and risk level, the comprehensive safety risk level of the new energy power system is determined, where 0≤ζ T <0.8 means no risk, 0.8≤ζ T <2 represents low risk, 2≤ζ T <6 represents medium risk, 6≤ζ T ≤10 indicates high risk, ζ T >10 indicates severe risk.

9. The method for comprehensive safety risk assessment of new energy power system based on vector machine according to claim 8, characterized in that: In the step of providing operation and maintenance decision recommendations for the new energy power system based on the comprehensive operation safety risk assessment results, if the safety assessment level of the new energy power system is risk-free or low risk, the system is operated according to the current system state; if the safety assessment level of the new energy power system is medium risk or high risk, the operation and maintenance process of the new energy power system is readjusted. By adjusting the power system operation and maintenance process, the system outputs appropriate power, meets the actual load needs, avoids the possibility of over-limit of power flow, and appropriately reduces the combined output of wind and solar energy.

Images